1 2018-11-05 Paul Koning <ni1d@arrl.net>
3 * doc/sourcebuild.texi (target attributes): Document new "inf"
4 effective target keyword.
6 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
8 * config/s390/s390.c (s390_register_move_cost): Increase costs for
9 moves involving the CC reg.
11 2018-11-05 Richard Biener <rguenther@suse.de>
13 PR tree-optimization/87873
14 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
16 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
17 * tree-vect-loop.c (vect_transform_loop): When splitting the
18 loop exit also create forwarder PHIs for constants.
19 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
20 Handle constant to_arg, add extra checking we match up the correct
23 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
25 * config/s390/s390.md: QImode and HImode for load on condition.
27 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
29 * config/s390/predicates.md: Fix typo.
30 * config/s390/s390.md: Allow immediates for load on condition.
32 2018-11-05 Martin Liska <mliska@suse.cz>
34 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
35 * bitmap.h (struct bitmap_usage): Likewise.
36 * ggc-common.c (SCALE): Remove.
38 (struct ggc_usage): Use SIZE_AMOUNT. And update
40 * ggc-page.c (SCALE): Remove.
42 (ggc_print_statistics): Use SIZE_AMOUNT.
43 * gimple.h (SCALE): Remove.
45 * input.c (ONE_K): Remove.
48 (STAT_LABEL): Likewise.
49 (FORMAT_AMOUNT): Likewise.
50 (dump_line_table_statistics): Use SIZE_AMOUNT.
51 * mem-stats.h (struct mem_usage): Likewise.
52 * rtl.c (dump_rtx_statistics): Likewise.
53 (rtx_alloc_counts): Change type to size_t.
54 (rtx_alloc_sizes): Likewise.
56 (dump_rtx_statistics): Sort first based on counts.
57 * tree.c (tree_nodes_cmp): New.
58 (tree_codes_cmp): New.
59 (dump_tree_statistics): Sort first based on counts.
60 * system.h (ONE_K): New.
62 (SIZE_SCALE): Likewise.
63 (SIZE_LABEL): Likewise.
64 (SIZE_AMOUNT): Likewise.
65 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
66 * tree-dfa.c (dump_dfa_stats): Likewise.
67 * tree-phinodes.c (phinodes_print_statistics): Likewise.
68 * tree-ssanames.c (ssanames_print_statistics): Likewise.
69 * tree.c (dump_tree_statistics): Likewise.
70 * vec.c (struct vec_usage): Likewise.
71 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
72 trigger a -Werror=format-overflow with
73 --enable-gather-detailed-stats.
75 2018-11-05 Martin Liska <mliska@suse.cz>
77 * mem-stats.h (mem_alloc_description::release_instance_overhead):
79 * vec.c (struct vec_usage): Register m_element_size.
80 (vec_prefix::register_overhead): New arguments: elements and
82 (vec_prefix::release_overhead): Subtract elements.
83 * vec.h (struct vec_prefix): Change signature.
84 (va_heap::reserve): Pass proper arguments.
85 (va_heap::release): Likewise.
87 2018-11-05 Martin Liska <mliska@suse.cz>
89 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
93 2018-11-05 Richard Biener <rguenther@suse.de>
95 * tree-scalar-evolution.h (final_value_replacement_loop): Update
97 * tree-scalar-evolution.c (final_value_replacement_loop): Return
98 whether anything was done.
99 (scev_const_prop): Remove constant propagation part, fold
101 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
102 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
105 2018-11-05 Jakub Jelinek <jakub@redhat.com>
107 PR tree-optimization/87859
108 * gimple-ssa-store-merging.c (struct merged_store_group): Add
109 only_constants and first_nonmergeable_order members.
110 (merged_store_group::merged_store_group): Initialize them.
111 (merged_store_group::do_merge): Clear only_constants member if
112 adding something other than INTEGER_CST store.
113 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
114 stores with order >= first_nonmergeable_order. Use
115 merged_store->only_constants instead of always recomputing it.
116 Set merged_store->first_nonmergeable_order if we've skipped any
117 stores. Attempt to merge overlapping INTEGER_CST stores that
118 we would otherwise skip.
121 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
123 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
126 * config/i386/emmintrin.h (__v16qs): New to cope with option
128 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
129 (_mm_cmplt_epi8): Likewise.
130 (_mm_cmpgt_epi8): Likewise.
132 2018-11-05 Richard Biener <rguenther@suse.de>
134 PR rtl-optimization/87852
135 * fwprop.c (use_killed_between): Only consider single-defs of the
136 use whose definition statement dominates the use.
138 2018-11-05 Martin Liska <mliska@suse.cz>
141 * doc/invoke.texi: Remove options that are
142 not disabled with -Os.
144 2018-11-05 Martin Liska <mliska@suse.cz>
147 * doc/extend.texi: Update constrain about the last argument
148 of __builtin_expect_with_probability.
150 2018-11-05 Martin Liska <mliska@suse.cz>
153 * predict.c (expr_expected_value_1): Verify
154 that last argument is a real constants and emit
157 2018-11-05 Martin Liska <mliska@suse.cz>
159 PR gcov-profile/77698
160 * ipa-profile.c (ipa_profile): Adjust hotness threshold
163 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
165 PR tree-optimization/86572
166 * builtins.c (c_strlen): Handle negative offsets in a safe way.
168 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
170 PR tree-optimization/87672
171 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
172 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
174 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
177 * cfgexpand.c (pass_expand::execute): Move the call to
178 finish_eh_generation in front of the call to expand_stack_alignment.
180 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
182 * common/config/i386/i386-common.c (processor_alias_table): Add
184 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
185 (case ${target}): Add znver2.
186 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
187 -march=native recognize znver2 processors.
188 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
189 * config/i386/i386.c (m_znver2): New definition.
190 (m_ZNVER): New definition.
191 (m_AMD_MULTIPLE): Includes m_znver2.
192 (processor_cost_table): Add znver2 entry.
193 (processor_target_table): Add znver2 entry.
194 (get_builtin_code_for_version): Set priority for
196 (processor_model): Add M_AMDFAM17H_ZNVER2.
197 (arch_names_table): Ditto.
198 (ix86_reassociation_width): Include znver2.
199 * config/i386/i386.h (TARGET_znver2): New definition.
200 (struct ix86_size_cost): Add TARGET_ZNVER2.
201 (enum processor_type): Add PROCESSOR_ZNVER2.
202 * config/i386/i386.md (define_attr "cpu"): Add znver2.
203 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
204 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
205 (ix86_adjust_cost): Add znver2.
206 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
207 * gcc/doc/extend.texi: Add details about znver2.
208 * gcc/doc/invoke.texi: Add details about znver2.
210 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
214 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
217 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
219 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
220 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
222 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
224 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
225 shrink-wrapping when -mtrack-speculation.
227 2018-11-02 Richard Biener <rguenther@suse.de>
229 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
230 (pop_cost_one_pair): Do not free pair.
231 (pop_best_coalesce): Likewise.
232 (create_coalesce_list): Initialize obstack.
233 (delete_coalesce_list): Free obstack.
234 (find_coalesce_pair): Obstack-allocate coalesce pairs.
235 (add_cost_one_coalesce): Likewise.
236 (struct live_track): Remove bitmap pointer indirections.
237 (new_live_track): Adjust.
238 (delete_live_track): Likewise.
239 (live_track_remove_partition): Likewise.
240 (live_track_add_partition): Likewise.
241 (live_track_live_p): Likewise.
242 (live_track_process_def): Likewise.
243 (live_track_clear_base_vars): Likewise.
245 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
247 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
248 * configure: Regenerate.
249 * config.in: Regenerate.
250 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
251 HAVE_AS_REGISTER_PSEUDO_OP guard.
252 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
253 !HAVE_AS_REGISTER_PSEUDO_OP support.
254 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
257 2018-11-02 Richard Biener <rguenther@suse.de>
259 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
262 2018-11-02 Richard Biener <rguenther@suse.de>
264 PR tree-optimization/87776
265 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
266 executable when iterating but running into rpo-vn-max-loop-depth
267 and not eliding the iteration.
269 2018-11-30 Jan Hubicka <jh@suse.cz>
271 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
274 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
276 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
278 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
279 Combine with rs6000_address_for_fpconvert.
280 (rs6000_address_for_fpconvert): Combine with
281 rs6000_force_indexed_or_indirect_mem.
282 (rs6000_expand_vector_init): Change function call from
283 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
284 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
285 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
286 (floatsi<mode>2_lfiwax_mem): Ditto.
287 (floatunssi<mode>2_lfiwzx): Ditto.
288 (floatunssi<mode>2_lfiwzx_mem): Ditto.
289 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
290 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
291 (fix_trunc<mode>si2_stfiwx): Ditto.
292 (fixuns_trunc<mode>si2_stfiwx): Ditto.
293 (float_<mode>si2_hw): Ditto.
294 (floatuns_<mode>si2_hw): Ditto.
295 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
296 (vsx_splat_<mode>): Ditto.
298 2018-11-01 Joseph Myers <joseph@codesourcery.com>
300 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
301 at end of assembler input text.
302 * configure: Regenerate.
304 2018-11-01 Jakub Jelinek <jakub@redhat.com>
306 PR tree-optimization/87826
307 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
308 negative or larger or equal to type's precision.
310 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
312 * opts.c (default_options_table): Do not enable
313 OPT_fdelayed_branch at -Og.
314 * doc/invoke.texi (-fdelayed-branch): Document it.
316 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
318 * optabs-libfuncs.c (build_libfunc_function_visibility):
319 New, split out from...
320 (build_libfunc_function): ... here.
321 (init_one_libfunc_visibility): New, split out from ...
322 (init_one_libfunc): ... here.
324 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
325 scratch register need not be early-clobber. Document the reason
326 why we cannot use ST<OP>.
328 2018-10-31 Joseph Myers <joseph@codesourcery.com>
331 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
332 line for second argument of AC_DEFINE_UNQUOTED.
333 * doc/install.texi (Tools/packages necessary for modifying GCC):
334 Update to autoconf 2.69 and automake 1.15.1.
335 * aclocal.m4, config.in, configure: Regenerate.
337 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
339 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
342 2018-10-31 Martin Liska <mliska@suse.cz>
345 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
346 Add new argument invalid_extension.
347 (aarch64_get_all_extension_candidates): New function.
348 (aarch64_rewrite_selected_cpu): Add NULL to function call.
349 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
351 (aarch64_get_all_extension_candidates): New function.
352 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
353 argument invalid_extension.
354 (aarch64_parse_cpu): Likewise.
355 (aarch64_print_hint_for_extensions): New function.
356 (aarch64_validate_mcpu): Provide hint about invalid extension.
357 (aarch64_validate_march): Likewise.
358 (aarch64_handle_attr_arch): Pass new argument.
359 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
360 (aarch64_handle_attr_isa_flags): Likewise.
362 2018-10-31 Richard Biener <rguenther@suse.de>
366 * tree-outof-ssa.c (insert_backedge_copies): Restrict
367 copy generation to useful cases. Place the copy before
368 the definition of the backedge value when possible.
370 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
372 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
373 * config/arc/arc.c (arc_active_insn): New function.
374 (check_store_cacheline_hazard): Likewise.
375 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
376 (arc_override_options): Disable delay slot scheduler for older
378 (arc_store_addr_hazard_p): New implementation, old one renamed to
380 (arc_store_addr_hazard_internal_p): Renamed.
381 (arc_reorg): Don't combine into brcc instructions which are part
382 of hardware hazard solution.
383 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
384 (tune_arc700): Likewise.
385 * config/arc/arc.opt (arc7xx): New tune value.
386 * config/arc/arc700.md: Improve A7 scheduler.
388 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
390 * config/arc/arc.c (arc_override_options): Remove
391 TARGET_COMPACT_CASESI.
392 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
393 (CASE_VECTOR_MODE): Likewise.
394 (CASE_VECTOR_PC_RELATIVE): Likewise.
395 (CASE_VECTOR_SHORTEN_MODE): Likewise.
396 (CASE_VECTOR_SHORTEN_MODE1): Delete.
397 (ADDR_VEC_ALIGN): Update.
398 (ASM_OUTPUT_CASE_LABEL): Undefine.
399 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
400 (TARGET_BI_BIH): Define.
401 (DEFAULT_BRANCH_INDEX): Likewise.
402 * config/arc/arc.md (casesi): Rework to accept BI/BIH
403 instructions, remove compact_casesi use case.
404 (casesi_compact_jump): Remove.
405 (casesi_dispatch): New pattern.
406 * config/arc/arc.opt: Add mbranch-index option. Deprecate
407 compact_casesi option.
408 * doc/invoke.texi: Document mbranch-index option.
410 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
412 * config/arc/arc.c (arc_get_tp): Remove function.
413 (arc_emit_call_tls_get_addr): Likewise.
414 (arc_call_tls_get_addr): New function.
415 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
416 * config/arc/arc.md (tls_load_tp_soft): Remove.
417 (tls_gd_get_addr): Likewise.
419 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
421 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
422 (mulsi3_600_lib): Remove pattern.
423 (umulsi3_highpart_600_lib_le): Likewise.
424 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
425 (umulsidi3): Remove call to umulsidi3_600_lib.
426 (umulsidi3_600_lib): Remove pattern.
427 (peephole2): Remove peephole using the above deprecated patterns.
429 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
432 * config/arm/arm.c (arm_option_check_internal): Disable the combined
433 use of -mslow-flash-data and -mword-relocations.
434 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
435 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
437 * doc/invoke.texi (-mword-relocations): Mention conflict with
439 (-mslow-flash-data): Reciprocally.
441 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
443 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
444 16-byte modes held in GP registers to use an even regno.
446 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
447 (aarch64_atomic_ldop_supported_p): Remove.
448 (aarch64_gen_atomic_ldop): Remove.
449 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
450 Fully expand LSE operations here.
451 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
452 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
453 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
454 and use ATOMIC_LDOP instead; use register_operand for the input;
455 drop the split and emit insns directly.
456 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
457 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
458 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
460 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
461 (aarch64_gen_atomic_ldop): Don't call it.
462 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
463 Use aarch64_reg_or_zero.
464 (aarch64_atomic_exchange<ALLI>): Likewise.
465 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
466 operand 0; use aarch64_reg_or_zero for input; merge ...
467 (@aarch64_atomic_swp<ALLI>): ... this and remove.
469 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
470 (aarch64_split_compare_and_swap): Use it.
471 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
472 test oldval against the proper predicate.
473 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
474 Use nonmemory_operand for expected.
475 (cas_short_expected_pred): New.
476 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
477 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
478 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
479 (aarch64_plushi_operand): New.
481 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
482 Force oldval into the rval register for TARGET_LSE; emit the compare
483 during initial expansion so that it may be deleted if unused.
484 (aarch64_gen_atomic_cas): Remove.
485 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
486 Change =&r to +r for operand 0; use match_dup for operand 2;
487 remove is_weak and mod_f operands as unused. Drop the split
489 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
490 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
491 (@aarch64_atomic_cas<GPI>): Similarly.
493 2018-10-31 Richard Biener <rguenther@suse.de>
495 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
498 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
500 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
501 * config/aarch64/aarch64.md: Include saphira.md
502 * config/aarch64/saphira.md: New file for pipeline description.
504 2018-10-30 Martin Sebor <msebor@redhat.com>
507 * gimple-ssa-sprintf.c (format_directive): Use %G to include
509 (sprintf_dom_walker::compute_format_length):
510 Avoid setting POSUNDER4K here.
511 (get_destination_size): Handle null argument values.
512 (get_user_idx_format): New function.
513 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
514 functions, including user-defined with attribute format printf.
515 Use %G to include inlining context.
518 2018-10-30 Jan Hubicka <jh@suse.cz>
520 * params.def (lto-partitions): Bump from 32 to 128.
522 2018-10-30 Jan Hubicka <jh@suse.cz>
525 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
527 (free_lang_data_in_type): Forward declare.
528 (fld_type_variant_equal_p): New function.
529 (fld_type_variant): New function
530 (fld_incomplete_types): New hash.
531 (fld_incomplete_type_of): New function
532 (fld_simplfied-type): New function.
533 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
534 (free_lang_data): Allocate and free fld_incomplete_type; update call
535 of free_lang_data_in_decl.
537 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
539 * gcov.c (output_lines): Remove duplicate line.
541 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
543 * config/rs6000/rs6000.md (bswapdi2): Force address into register
544 if not in indexed or indirect form.
545 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
546 (bswapdi2_store): Ditto.
547 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
549 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
550 Prototype for helper function.
552 2018-10-30 Martin Sebor <msebor@redhat.com>
554 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
555 (target, pragma GCC optimize, pragma GCC target): Ditto.
557 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
559 * doc/extend.texi: Fix prototype and description of
560 __builtin_expect_with_probability.
562 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
564 * cgraph.h (clone_function_name_1): Replaced by new
565 clone_function_name_numbered that takes name as string; for
566 privatize_symbol_name_1 use only.
567 (clone_function_name): Renamed to
568 clone_function_name_numbered to be explicit about numbering.
569 (clone_function_name): New two-argument function that does
570 not number its output.
571 (clone_function_name): New three-argument function that
572 takes a number to append to its output.
573 * cgraphclones.c (duplicate_thunk_for_node):
574 (clone_function_name_1): Renamed.
575 (clone_function_name_numbered): Two new functions.
576 (clone_function_name): Improved documentation.
577 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
578 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
579 * final.c (final_scan_insn_1): Use the new clone_function_name
581 * multiple_target.c (create_dispatcher_calls): Ditto.
582 (create_target_clone): Ditto.
583 * omp-expand.c (grid_expand_target_grid_body): Ditto.
584 * omp-low.c (create_omp_child_function_name): Ditto.
585 * omp-simd-clone.c (simd_clone_create): Ditto.
586 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
587 new clone_function_name without numbering.
589 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
591 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
592 Assert that the allocation size is not zero.
594 2018-10-30 Richard Biener <rguenther@suse.de>
596 PR tree-optimization/87800
597 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
598 non-induction or reduction PHIs.
600 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
602 * config/aarch64/falkor-tag-collision-avoidance.c
603 (execute_tag_collision_avoidance): Call df_note_add_problem.
605 2018-10-30 Martin Liska <mliska@suse.cz>
607 * doc/extend.texi: Fix typo in documentation
608 of __builtin_expect_with_probability.
610 2018-10-29 David Malcolm <dmalcolm@redhat.com>
613 * input.c (get_substring_ranges_for_loc): Detect if
614 linemap_resolve_location gives us a NULL map, and reject
617 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
619 * config.gcc (xstormy16-*-elf): Set tm_d_file.
621 2018-10-29 David Malcolm <dmalcolm@redhat.com>
622 Martin Sebor <msebor@redhat.com>
623 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
625 * doc/ux.texi (Quoting): New subsection, adapted from material at
626 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
627 MartinSebor and ManuelLopezIbanez.
628 (Fix-it hints): Note that fix-it hints shouldn't be marked for
631 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
634 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
637 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
639 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
641 2018-10-29 Paul Koning <ni1d@arrl.net>
643 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
645 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
647 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
648 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
649 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
650 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
651 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
652 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
653 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
654 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
655 Change 'vector' to '__vector'.
656 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
657 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
658 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
659 _mm_avg_pu16): Likewise. And, whitespace corrections.
661 2018-10-29 Richard Biener <rguenther@suse.de>
663 PR tree-optimization/87785
664 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
667 2018-10-29 Olivier Hainque <hainque@adacore.com>
669 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
671 2018-10-29 Olivier Hainque <hainque@adacore.com>
673 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
674 ports configuration macro, defaults to "gnu".
675 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
676 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
678 2018-10-29 Olivier Hainque <hainque@adacore.com>
680 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
682 2018-10-29 Richard Biener <rguenther@suse.de>
684 PR tree-optimization/87790
685 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
686 (vect_make_slp_decision): Adjust.
687 (vect_slp_analyze_bb_1): Likewise.
688 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
691 2018-10-29 Richard Biener <rguenther@suse.de>
693 PR tree-optimization/87785
694 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
696 (vect_build_slp_tree): Likewise.
697 (vect_gather_slp_loads): New function.
698 (vect_analyze_slp_instance): Gather loads separately from the
701 2018-10-29 Martin Liska <mliska@suse.cz>
703 * Makefile.in: Make dependency to json.o.
704 * doc/gcov.texi: Document new JSON format, remove
705 old intermediate format documentation.
706 * gcov.c (struct function_info): Come up with m_name and
708 (function_info::function_info): Initialize it.
709 (function_info::~function_info): Release it.
710 (main): Rename flag_intermediate_format to flag_json_format.
711 (print_usage): Describe --json-format.
712 (process_args): Set flag_json_format.
713 (output_intermediate_line): Remove.
714 (output_intermediate_json_line): Likewise.
715 (get_gcov_intermediate_filename): Return new extension
717 (output_intermediate_file): Implement JSON emission.
718 (output_json_intermediate_file): Implement JSON emission.
719 (generate_results): Use ::get_name for function name.
720 Handle JSON output file.
721 (read_graph_file): Use ::get_name instead of cplus_demangle.
722 (read_count_file): Likewise.
723 (solve_flow_graph): Likewise.
724 (add_line_counts): Likewise.
725 (accumulate_line_counts): Use new flag_json_format.
726 (output_function_details): Use ::get_name instead of cplus_demangle.
727 (output_lines): Likewise.
728 * json.cc (test_writing_literals): Add new tests.
729 * json.h (class literal): Add new boolean constructor.
731 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
733 PR rtl-optimization/87701
734 PR rtl-optimization/87780
735 * combine.c (make_more_copies): Rewrite.
737 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
739 * doc/generic.texi (ABSU_EXPR): Document.
740 * match.pd (absu(x)*absu(x) -> x*x): Handle.
741 (absu(absu(X)) -> absu(X)): Likewise.
742 (absu(-X) -> absu(X)): Likewise.
743 (absu(X) where X is nonnegative -> X): Likewise.
745 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
747 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
748 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
749 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
750 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
751 (s-tm-texi): Also check timestamp on d-target.def.
752 (generated_files): Add TM_D_H and d-target-hooks-def.h.
753 (build/genhooks.o): Also depend on D_TARGET_DEF.
754 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
756 * config/aarch64/aarch64-d.c: New file.
757 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
759 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
761 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
762 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
763 * config/arm/arm-d.c: New file.
764 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
765 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
766 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
767 * config/arm/t-arm (arm-d.o): New rule.
768 * config/default-d.c: New file.
769 * config/glibc-d.c: New file.
770 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
771 * config/i386/i386-d.c: New file.
772 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
773 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
774 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
775 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
776 * config/i386/t-i386 (i386-d.o): New rule.
777 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
778 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
779 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
780 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
781 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
782 * config/mips/mips-d.c: New file.
783 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
784 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
785 * config/mips/t-mips (mips-d.o): New rule.
786 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
787 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
788 * config/powerpcspe/powerpcspe-d.c: New file.
789 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
791 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
792 Support GNU D by using 0 as the language type.
793 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
794 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
795 * config/riscv/riscv-d.c: New file.
796 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
798 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
799 * config/riscv/t-riscv (riscv-d.o): New rule.
800 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
801 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
802 * config/rs6000/rs6000-d.c: New file.
803 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
805 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
806 Support GNU D by using 0 as the language type.
807 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
808 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
809 * config/s390/s390-d.c: New file.
810 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
811 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
812 * config/s390/t-s390 (s390-d.o): New rule.
813 * config/sparc/sparc-d.c: New file.
814 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
816 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
817 * config/sparc/t-sparc (sparc-d.o): New rule.
818 * config/t-glibc (glibc-d.o): New rule.
819 * configure: Regenerated.
820 * configure.ac (tm_d_file): New variable.
821 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
822 * doc/contrib.texi (Contributors): Add self for the D frontend.
823 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
824 * doc/install.texi (Configuration): Mention libphobos as an option for
825 --enable-shared. Mention d as an option for --enable-languages.
826 (Testing): Mention check-d as a target.
827 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
828 name suffixes. Mention d as a -x option.
829 * doc/sourcebuild.texi (Top Level): Mention libphobos.
830 * doc/standards.texi (Standards): Add section on D language.
831 * doc/tm.texi: Regenerated.
832 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
833 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
834 * dwarf2out.c (is_dlang): New function.
835 (gen_compile_unit_die): Use DW_LANG_D for D.
836 (declare_in_namespace): Return module die for D, instead of adding
837 extra declarations into the namespace.
838 (gen_namespace_die): Generate DW_TAG_module for D.
839 (gen_decl_die): Handle CONST_DECLSs for D.
840 (dwarf2out_decl): Likewise.
841 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
842 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
844 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
845 * genhooks.c: Include d/d-target.def.
847 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
850 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
851 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
852 (STACK_DYNAMIC_OFFSET): Likewise.
854 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
857 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
860 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
862 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
863 return type and other typos.
865 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
867 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
868 a shorter sequence with fewer branches.
869 (emit_final_str_compare_gpr): Ditto.
871 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
873 * config/rs6000/tmmintrin.h: New file.
874 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
876 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
878 * config/rs6000/mmintrin.h: Enable 32bit compilation.
879 * config/rs6000/xmmintrin.h: Likewise.
881 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
883 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
885 2018-10-26 Richard Biener <rguenther@suse.de>
887 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
889 (vect_mark_slp_stmts_relevant): Likewise.
890 (vect_detect_hybrid_slp_stmts): Likewise.
891 (vect_bb_slp_scalar_cost): Likewise.
892 (vect_remove_slp_scalar_calls): Likewise.
894 2018-10-26 Jan Hubicka <jh@suse.cz>
896 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
897 (warn_types_mismatch): Fix walk of DECL_NAME.
898 (odr_types_equivalent_p): Fix overactive assert.
900 2018-10-26 Richard Biener <rguenther@suse.de>
902 PR tree-optimization/87105
903 * tree-vectorizer.h (_slp_tree::refcnt): New member.
904 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
906 (vect_create_new_slp_node): Initialize refcnt to one.
908 (scalar_stmts_set_t, bst_fail): Remove.
909 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
910 (vect_build_slp_tree): Add bst_map argument and lookup
911 already created SLP nodes.
912 (vect_print_slp_tree): Handle a SLP graph, print SLP node
914 (vect_slp_rearrange_stmts): Handle a SLP graph.
915 (vect_analyze_slp_instance): Adjust and free SLP nodes from
916 the CSE map. Fix indenting.
917 (vect_schedule_slp_instance): Add short-cut.
919 2018-10-26 Martin Liska <mliska@suse.cz>
922 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
923 addr_expr and not with pointers.
925 2018-10-26 Jan Hubicka <jh@suse.cz>
927 * tree.c (free_lang_data_in_type): Only check main variants.
928 * ipa-devirt.c (warn_odr): Make static.
929 (types_same_for_odr): Drop strict variant.
930 (types_odr_comparable): Likewise.
931 (odr_or_derived_type_p): Look for main variants.
932 (odr_name_hasher::equal): Cleanup comment.
933 (odr_subtypes_equivalent): Add warn and warned arguments; check main
935 (type_variants_equivalent_p): break out from ...
936 (odr_types_equivalent): ... here; go for main variants where needed.
937 (warn_odr): ... here; turn static.
938 (warn_types_mismatch): Compare mangled names of main variants.
939 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
940 (type_with_linkage_p): Sanity check that we look at main variant.
941 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
942 * tree.h (types_same_for_odr): Drop strict argument.
944 2018-10-26 Richard Biener <rguenther@suse.de>
946 PR tree-optimization/87746
947 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
948 Simplify and fix WRT strided store groups with size not
949 equal to step in element count.
950 (vect_analyze_group_access_1): Dump the whole group.
952 2018-10-25 Carl Love <cel@us.ibm.com>
954 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
955 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
956 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
957 precicion entry for each overloaded builtin.
958 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
959 VSCEDPUO): Rename overloaded name.
960 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
961 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
962 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
963 define_expand for xscmpexqp instruction.
964 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
966 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
967 Jinsong Ji <jji@us.ibm.com>
969 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
970 function with vec_sl.
971 (_mm_slli_epi32): Likewise.
972 (_mm_slli_epi64): Likewise.
973 (_mm_srai_epi16): Replace deprecated function with vec_sra.
974 (_mm_srai_epi32): Likewise.
975 (_mm_srli_epi16): Replace deprecated function with vec_sr.
976 (_mm_srli_epi32): Likewise.
977 (_mm_srli_epi64): Likewise.
978 (_mm_sll_epi16): Replace deprecated function with vec_sl.
979 (_mm_sll_epi32): Likewise.
980 (_mm_sll_epi64): Likewise.
981 (_mm_sra_epi16): Replace deprecated function with vec_sra.
982 (_mm_sra_epi32): Likewise.
983 (_mm_srl_epi16): Replace deprecated function with vec_sr.
984 (_mm_srl_epi32): Likewise.
985 (_mm_srl_epi64): Likewise.
987 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
988 Jinsong Ji <jji@us.ibm.com>
990 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
991 comparison operators with vec_cmp* for compatibility due to
992 unfortunate history; clean up formatting and use types more
994 (_mm_sll_epi32): Likewise.
995 (_mm_sll_epi64): Likewise.
996 (_mm_srl_epi16): Likewise.
997 (_mm_srl_epi32): Likewise.
998 (_mm_srl_epi64): Likewise.
1000 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
1001 Jinsong Ji <jji@us.ibm.com>
1003 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
1004 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
1005 __vector __bool int. Use vec_cmpgt in preference to deprecated
1006 function vec_vcmpgtfp.
1007 (_mm_max_ps): Likewise.
1009 2018-10-25 Jeff Law <law@redhat.com>
1011 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
1014 2018-10-25 Martin Sebor <msebor@redhat.com>
1016 * doc/extend.texi (aligned): Expand attribute description.
1017 (Alignment): Rename section. Discuss function arguments.
1019 2018-10-25 Jan Hubicka <jh@suse.cz>
1021 * ipa-devirt.c (main_odr_variant): Remove.
1022 (hash_odr_name, types_same_for_odr, types_odr_comparable,
1023 odr_name_hasher::equal, odr_subtypes_equivalent_p):
1024 Drop use of main_odr_variant.
1025 (add_type_duplicate): Silence confused warnings on integer types.
1026 (get_odr_type): Always look for main variant.
1027 (register_odr_type): Simplify.
1029 2018-10-25 Richard Biener <rguenther@suse.de>
1031 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
1032 Initialize ng to silence error with release checking bootstrap.
1034 2018-10-25 Richard Biener <rguenther@suse.de>
1036 * tree-if-conv.c: Include tree-ssa-sccvn.h.
1037 (tree_if_conversion): Run CSE on the if-converted loop body.
1039 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
1041 * config/s390/constraints.md (ZL): New constraint.
1042 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
1044 * config/s390/s390.md (movdi_larl): Remove.
1045 (movdi_64): Add the LARL alternative.
1047 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
1050 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
1051 (RTX_CODE_HWINT_P): New macro.
1052 (rtx_code_size): Use RTX_CODE_HWINT_P ().
1054 2018-10-25 Jan Hubicka <jh@suse.cz>
1056 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
1059 2018-10-25 Richard Biener <rguenther@suse.de>
1061 PR tree-optimization/87665
1062 PR tree-optimization/87745
1063 * tree-vectorizer.h (get_earlier_stmt): Remove.
1064 (get_later_stmt): Pick up UID from the original non-pattern stmt.
1066 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
1068 * options.texi (Deprecated): Move list to Var section.
1070 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
1071 Jinsong Ji <jji@us.ibm.com>
1073 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
1074 __vector long to __vector long long.
1075 (_mm_cvtpd_ps): Likewise.
1076 (_mm_cvttpd_epi32): Likewise.
1077 (_mm_cvtpi32_pd): Likewise.
1078 (_mm_unpackhi_epi64): Likewise.
1079 (_mm_unpacklo_epi64): Likewise.
1081 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
1083 PR rtl-optimization/87720
1084 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
1086 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
1088 * gimple-ssa-isolate-paths.c
1089 (find_implicit_erroneous_behavior): Do not change code if the
1090 pass is running for warnings only.
1091 (find_explicit_erroneous_behavior): Likewise.
1093 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
1095 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
1096 Define as rs6000_mangle_decl_assembler_name.
1097 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
1098 long double to IEEE long double, switch the names of the long
1099 double built-in functions to be <func>f128 instead of <func>l.
1101 2018-10-24 Martin Sebor <msebor@redhat.com>
1103 * doc/extend.texi (nonnull): List no-argument form. Reference
1104 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
1106 2018-10-24 Richard Biener <rguenther@suse.de>
1108 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
1110 2018-10-24 Martin Liska <mliska@suse.cz>
1112 PR tree-optimization/84436
1113 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
1115 (switch_conversion::contains_linear_function_p): New.
1116 (switch_conversion::build_one_array): Support linear
1117 transformation on input.
1118 * tree-switch-conversion.h (struct switch_conversion): Add
1119 contains_linear_function_p declaration.
1121 2018-10-24 Richard Biener <rguenther@suse.de>
1123 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
1124 if its argument is CONSTANT_CLASS_P.
1126 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
1128 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
1129 it is wrong for forward declarations.
1131 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
1133 * config/s390/s390.c (s390_check_qrst_address): Add the missing
1134 SYMBOL_REF_P () check.
1136 2018-10-24 Richard Biener <rguenther@suse.de>
1138 PR tree-optimization/87105
1139 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
1140 dump classification.
1141 (vect_analyze_data_ref_accesses): Handle duplicate loads and
1142 stores by splitting the affected group after the fact.
1143 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
1144 fail the SLP build because of size constraints.
1146 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1148 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
1149 * configure: Regenerate.
1150 * config.in: Regenerate.
1151 * varasm.c (mergeable_string_section): Use readonly_data_section
1152 if linker doesn't support SHF_MERGE with alignment > 8.
1153 (mergeable_constant_section): Likewise.
1155 2018-10-24 Richard Biener <rguenther@suse.de>
1157 PR tree-optimization/84013
1158 * tree-ssa-structalias.c (struct msdi_data): New struct for
1159 marshalling data to walk_stmt_load_store_ops.
1160 (maybe_set_dependence_info): Refactor as callback for
1161 walk_stmt_load_store_ops.
1162 (compute_dependence_clique): Set restrict info on all stmt kinds.
1164 2018-10-24 Martin Liska <mliska@suse.cz>
1166 * cgraph.c (cgraph_node::dump):
1167 Remove reduntant dumps and make tp_first_run dump more compact.
1169 2018-10-24 Richard Biener <rguenther@suse.de>
1171 PR tree-optimization/87665
1172 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
1175 2018-10-12 Jeff Law <law@redhat.com>
1177 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
1180 2018-10-23 Richard Biener <rguenther@suse.de>
1182 * tree-vrp.c (add_assert_info): Guard dump_printf with
1184 * gimple-ssa-evrp-analyze.c
1185 (evrp_range_analyzer::record_ranges_from_incoming_edge):
1186 Use value_range::ignore_equivs_equal_p.
1188 2018-10-23 Richard Biener <rguenther@suse.de>
1190 PR tree-optimization/87105
1191 PR tree-optimization/87608
1192 * passes.def (pass_all_early_optimizations): Add early phi-opt
1194 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
1195 addition to debug stmts.
1196 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
1197 and abs replacement early.
1198 * tree-cfg.c (gimple_empty_block_p): Likewise.
1200 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
1203 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
1204 specified to configure.
1205 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
1207 2018-10-23 Richard Biener <rguenther@suse.de>
1209 PR tree-optimization/87700
1210 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
1212 2018-10-23 Jakub Jelinek <jakub@redhat.com>
1215 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
1216 second argument from __mmask16 to __mmask8.
1217 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
1218 _mm_mask_packs_epi32): Likewise.
1219 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
1221 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
1223 2018-10-23 Richard Biener <rguenther@suse.de>
1225 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
1227 2018-10-23 Richard Biener <rguenther@suse.de>
1229 PR tree-optimization/86144
1230 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
1231 over simd attribute.
1233 2018-10-23 Richard Biener <rguenther@suse.de>
1235 PR tree-optimization/87693
1236 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
1237 the case we do not find the taken edge.
1239 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
1240 Jinsong Ji <jji@us.ibm.com>
1242 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
1243 (_mm_store_pd): Use unaligned vector type for pointer cast.
1244 (_mm_maskmoveu_si128): Likewise.
1245 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
1246 (_mm_store_ps): Use unaligned vector type for pointer cast.
1248 2018-10-22 Paul Koning <ni1d@arrl.net>
1250 * symtab.c (symtab_node::increase_alignment): Correct max
1253 2018-10-22 Yury Gribov <tetra2005@gmail.com>
1255 PR tree-optimization/87633
1256 * match.pd: Do not generate unordered integer comparisons.
1258 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1260 PR rtl-optimization/87600
1261 * combine.c: Add include of expr.h.
1262 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
1263 register to a pseudo.
1264 (make_more_copies): New function, add a copy to a new pseudo after
1265 the moves from hard registers into pseudos.
1266 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
1267 later. Call make_more_copies.
1269 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
1271 * lra-constraints.c (process_alt_operands): New local array,
1272 matching_early_clobber. Check matching_early_clobber before
1273 decrementing reject, and set matching_early_clobber after.
1275 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1278 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
1279 call output_addr_const and hope for the best.
1281 2018-10-22 Richard Biener <rguenther@suse.de>
1283 * gimple-ssa-evrp-analyze.c
1284 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
1285 smarter about what ranges to use.
1286 * tree-vrp.c (add_assert_info): Dump here.
1287 (register_edge_assert_for_2): Instead of here at multiple but
1290 * gcc.dg/tree-ssa/evrp12.c: New testcase.
1291 * gcc.dg/predict-6.c: Adjust.
1292 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
1293 * gcc.dg/tree-ssa/vrp02.c: Likewise.
1294 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
1296 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
1297 Richard Biener <rguenther@suse.de>
1299 * bitmap.h: Update data structure documentation, including a
1300 description of bitmap views as either linked-lists or splay trees.
1301 (struct bitmap_element_def): Update comments for splay tree bitmaps.
1302 (struct bitmap_head_def): Likewise.
1303 (bitmap_list_view, bitmap_tree_view): New prototypes.
1304 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
1306 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
1307 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
1308 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
1309 released bitmap element here.
1310 (bitmap_element_free): Remove.
1311 (bitmap_elt_clear_from): Work on splay tree bitmaps.
1312 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
1313 this function similar ones such that linked-list bitmap implementation
1314 functions are grouped.
1315 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
1316 and moved for grouping.
1317 (bitmap_list_insert_element_after): Renamed from
1318 bitmap_elt_insert_after, and moved for grouping.
1319 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
1320 (bitmap_tree_link_left, bitmap_tree_link_right,
1321 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
1322 bitmap_tree_link_element, bitmap_tree_unlink_element,
1323 bitmap_tree_find_element): New functions for splay-tree bitmap
1325 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
1326 Renamed and moved, see above entries.
1327 (bitmap_tree_listify_from): New function to convert part of a splay
1328 tree bitmap to a linked-list bitmap.
1329 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
1330 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
1331 (bitmap_find_bit): Remove.
1332 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
1333 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
1334 Handle splay tree bitmaps.
1335 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
1336 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
1337 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
1338 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
1339 bitmap_intersect_compl_p, bitmap_ior_and_compl,
1340 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
1341 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
1342 corresponding changes to use linked-list specific bitmap_element
1343 manipulation functions as applicable for efficiency.
1344 (bitmap_tree_to_vec): New function.
1345 (debug_bitmap_elt_file): New function split out from ...
1346 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
1347 (bitmap_print): Likewise.
1349 PR tree-optimization/63155
1350 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
1352 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
1355 2018-10-22 Martin Liska <mliska@suse.cz>
1357 PR tree-optimization/87686
1359 2018-08-29 Martin Liska <mliska@suse.cz>
1361 * tree-switch-conversion.c (switch_conversion::expand):
1362 Strenghten assumption about gswitch statements.
1364 2018-10-22 Martin Liska <mliska@suse.cz>
1366 * ipa-icf.c (sem_item::compare_attributes): Remove.
1367 (sem_item::compare_referenced_symbol_properties): Use
1368 attribute_list_equal instead.
1369 (sem_function::equals_wpa): Likewise.
1370 * ipa-icf.h: Remove compare_attributes.
1372 2018-10-22 Richard Biener <rguenther@suse.de>
1375 * mem-stats.h (mem_usage::operator==): Fix pasto.
1377 2018-10-22 Richard Biener <rguenther@suse.de>
1379 PR tree-optimization/87640
1380 * tree-vrp.c (set_value_range_with_overflow): Decompose
1382 (extract_range_from_binary_expr_1): Adjust.
1384 2018-10-22 Martin Jambor <mjambor@suse.cz>
1386 * tree-eh.h (stmt_could_throw_p): Add function parameter.
1387 (stmt_can_throw_external): Likewise.
1388 (stmt_can_throw_internal): Likewise.
1389 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
1390 (lower_eh_constructs_2): Likewise.
1391 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
1392 (stmt_can_throw_external): Likewise.
1393 (stmt_can_throw_internal): Likewise.
1394 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
1395 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
1396 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
1397 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
1398 (pass_lower_eh_dispatch::execute): Pass cfun to
1399 stmt_can_throw_external.
1400 (cleanup_empty_eh): Likewise.
1401 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
1402 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
1403 stmt_can_throw_external instead of pushing it to cfun.
1404 (symbol_table::create_edge): Likewise.
1405 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
1406 stmt_can_throw_internal.
1407 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
1408 to stmt_could_throw_p.
1409 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
1410 stmt_can_throw_internal.
1411 (pass_store_merging::execute): Likewise.
1412 * gimple-ssa-strength-reduction.c
1413 (find_candidates_dom_walker::before_dom_children): Pass cfun to
1415 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
1416 stmt_can_throw_internal.
1417 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
1418 to stmt_can_throw_external.
1419 (check_stmt): Pass cfun to stmt_could_throw_p.
1420 (check_stmt): Pass cfun to stmt_can_throw_external.
1421 (pass_nothrow::execute): Likewise.
1422 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
1423 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
1424 stmt_can_throw_internal.
1425 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
1426 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
1427 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
1428 * tree-complex.c (expand_complex_libcall): Pass cfun to
1429 stmt_could_throw_p and to stmt_can_throw_internal.
1430 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
1431 * tree-inline.c (copy_edges_for_bb): Likewise.
1432 (maybe_move_debug_stmts_to_successors): Likewise.
1433 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
1435 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
1436 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
1437 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
1438 stmt_can_throw_internal.
1439 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
1440 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
1442 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
1443 stmt_can_throw_internal.
1444 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
1445 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
1447 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
1448 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
1449 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1450 (convert_mult_to_fma_1): Likewise.
1451 (convert_to_divmod): Likewise.
1452 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
1453 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
1454 * tree-ssa-propagate.c
1455 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
1456 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
1457 (maybe_optimize_range_tests): Likewise.
1458 (linearize_expr_tree): Likewise.
1459 (reassociate_bb): Likewise.
1460 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
1461 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
1462 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
1463 (handle_char_store): Likewise.
1464 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
1465 stmt_can_throw_internal.
1466 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
1468 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
1469 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
1470 (vectorizable_simd_clone_call): Likewise.
1471 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
1472 (gimple_stringop_fixed_value): Likewise.
1474 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
1476 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
1477 literal pool references.
1478 (s390_check_qrst_address): Adapt to the new behavior of
1479 s390_loadrelative_operand_p ().
1481 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1484 * config/i386/sse.md (*andnot<mode>3_bcst): New.
1486 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1489 * config/i386/sse.md (*<code><mode>3_bcst): New.
1491 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1494 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
1495 V4DI, V16SI and V8DI.
1496 (*sub<mode>3<mask_name>_bcst): New.
1497 (*add<mode>3<mask_name>_bcst): Likewise.
1499 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
1500 Jinsong Ji <jji@us.ibm.com>
1502 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
1503 __m64 with __vector unsigned long long for compatibility.
1504 (_mm_movemask_epi8): Likewise.
1505 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
1506 (_mm_cvttps_pi32): Likewise.
1507 (_mm_cvtpi32_ps): Likewise.
1508 (_mm_cvtps_pi16): Likewise.
1509 (_mm_loadh_pi): Likewise.
1510 (_mm_storeh_pi): Likewise.
1511 (_mm_movehl_ps): Likewise.
1512 (_mm_movelh_ps): Likewise.
1513 (_mm_loadl_pi): Likewise.
1514 (_mm_storel_pi): Likewise.
1515 (_mm_movemask_ps): Likewise.
1516 (_mm_shuffle_pi16): Likewise.
1518 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1521 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
1522 __builtin_ia32_vfnmsubpd512_mask.
1523 (_mm512_mask_fnmsub_round_pd): Likewise.
1524 (_mm512_fnmsub_pd): Likewise.
1525 (_mm512_mask_fnmsub_pd): Likewise.
1526 (_mm512_maskz_fnmsub_round_pd): Use
1527 __builtin_ia32_vfnmsubpd512_maskz.
1528 (_mm512_maskz_fnmsub_pd): Likewise.
1529 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
1530 (_mm512_mask_fnmsub_round_ps): Likewise.
1531 (_mm512_fnmsub_ps): Likewise.
1532 (_mm512_mask_fnmsub_ps): Likewise.
1533 (_mm512_maskz_fnmsub_round_ps): Use
1534 __builtin_ia32_vfnmsubps512_maskz.
1535 (_mm512_maskz_fnmsub_ps): Likewise.
1536 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
1537 __builtin_ia32_vfnmsubpd256_mask.
1538 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
1539 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1540 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
1541 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1542 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1543 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
1544 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
1545 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
1546 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
1547 __builtin_ia32_vfnmsubpd.
1548 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
1549 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
1550 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1551 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1552 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1553 * config/i386/i386-builtin.def: Add
1554 __builtin_ia32_vfnmsubpd256_mask,
1555 __builtin_ia32_vfnmsubpd256_maskz,
1556 __builtin_ia32_vfnmsubpd128_mask,
1557 __builtin_ia32_vfnmsubpd128_maskz,
1558 __builtin_ia32_vfnmsubps256_mask,
1559 __builtin_ia32_vfnmsubps256_maskz,
1560 __builtin_ia32_vfnmsubps128_mask,
1561 __builtin_ia32_vfnmsubps128_maskz,
1562 __builtin_ia32_vfnmsubpd512_mask,
1563 __builtin_ia32_vfnmsubpd512_maskz,
1564 __builtin_ia32_vfnmsubps512_mask,
1565 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1566 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1567 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1568 __builtin_ia32_vfnmsubpd256.
1569 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1570 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1571 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1573 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1575 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1577 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1579 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1582 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1583 __builtin_ia32_vfnmaddpd512_mask.
1584 (_mm512_mask_fnmadd_round_pd): Likewise.
1585 (_mm512_fnmadd_pd): Likewise.
1586 (_mm512_mask_fnmadd_pd): Likewise.
1587 (_mm512_maskz_fnmadd_round_pd): Use
1588 __builtin_ia32_vfnmaddpd512_maskz.
1589 (_mm512_maskz_fnmadd_pd): Likewise.
1590 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1591 (_mm512_mask_fnmadd_round_ps): Likewise.
1592 (_mm512_fnmadd_ps): Likewise.
1593 (_mm512_mask_fnmadd_ps): Likewise.
1594 (_mm512_maskz_fnmadd_round_ps): Use
1595 __builtin_ia32_vfnmaddps512_maskz.
1596 (_mm512_maskz_fnmadd_ps): Likewise.
1597 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1598 __builtin_ia32_vfnmaddpd256_mask.
1599 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1600 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1601 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1602 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1603 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1604 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1605 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1606 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1607 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1608 __builtin_ia32_vfnmaddpd.
1609 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1610 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1611 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1612 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1613 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1614 * config/i386/i386-builtin.def: Add
1615 __builtin_ia32_vfnmaddpd256_mask,
1616 __builtin_ia32_vfnmaddpd256_maskz,
1617 __builtin_ia32_vfnmaddpd128_mask,
1618 __builtin_ia32_vfnmaddpd128_maskz,
1619 __builtin_ia32_vfnmaddps256_mask,
1620 __builtin_ia32_vfnmaddps256_maskz,
1621 __builtin_ia32_vfnmaddps128_mask,
1622 __builtin_ia32_vfnmaddps128_maskz,
1623 __builtin_ia32_vfnmaddpd512_mask,
1624 __builtin_ia32_vfnmaddpd512_maskz,
1625 __builtin_ia32_vfnmaddps512_mask,
1626 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1627 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1628 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1629 __builtin_ia32_vfnmaddpd256.
1630 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1631 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1632 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1634 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1636 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1638 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1640 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1643 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1644 __builtin_ia32_vfmsubpd512_mask.
1645 (_mm512_mask_fmsub_round_pd): Likewise.
1646 (_mm512_fmsub_pd): Likewise.
1647 (_mm512_mask_fmsub_pd): Likewise.
1648 (_mm512_maskz_fmsub_round_pd): Use
1649 __builtin_ia32_vfmsubpd512_maskz.
1650 (_mm512_maskz_fmsub_pd): Likewise.
1651 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1652 (_mm512_mask_fmsub_round_ps): Likewise.
1653 (_mm512_fmsub_ps): Likewise.
1654 (_mm512_mask_fmsub_ps): Likewise.
1655 (_mm512_maskz_fmsub_round_ps): Use
1656 __builtin_ia32_vfmsubps512_maskz.
1657 (_mm512_maskz_fmsub_ps): Likewise.
1658 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1659 __builtin_ia32_vfmsubpd256_mask.
1660 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1661 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1662 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1663 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1664 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1665 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1666 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1667 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1668 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1669 __builtin_ia32_vfmsubpd.
1670 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1671 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1672 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1673 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1674 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1675 * config/i386/i386-builtin.def: Add
1676 __builtin_ia32_vfmsubpd256_mask,
1677 __builtin_ia32_vfmsubpd256_maskz,
1678 __builtin_ia32_vfmsubpd128_mask,
1679 __builtin_ia32_vfmsubpd128_maskz,
1680 __builtin_ia32_vfmsubps256_mask,
1681 __builtin_ia32_vfmsubps256_maskz,
1682 __builtin_ia32_vfmsubps128_mask,
1683 __builtin_ia32_vfmsubps128_maskz,
1684 __builtin_ia32_vfmsubpd512_mask,
1685 __builtin_ia32_vfmsubpd512_maskz,
1686 __builtin_ia32_vfmsubps512_mask,
1687 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1688 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1689 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1690 __builtin_ia32_vfmsubpd256.
1691 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1692 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1693 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1695 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1697 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1699 (fmai_vmfmsub_<mode><round_name>): Likewise.
1701 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1703 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1704 Remove plus. Renamed to ...
1705 (*sub<mode>3<mask_name>_bcst): This.
1706 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1707 (*add<mode>3<mask_name>_bcst): This.
1709 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1712 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1714 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1717 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1718 (_mm_or_epi32): Likewise.
1719 (_mm256_xor_epi32): Likewise.
1720 (_mm_xor_epi32): Likewise.
1721 (_mm256_or_epi64): Likewise.
1722 (_mm_or_epi64): Likewise.
1723 (_mm256_xor_epi64): Likewise.
1724 (_mm_xor_epi64): Likewise.
1726 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1729 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1731 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1734 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1736 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1738 * doc/ux.texi: Move @section directly after @node.
1740 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1744 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1745 depend closely nested inside of loop with ordered clause with
1748 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1750 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1751 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1752 * doc/ux.texi: New file.
1754 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1756 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1757 be the first CR field allocated.
1759 2018-10-19 Richard Biener <rguenther@suse.de>
1762 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1763 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1765 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1768 * config/i386/sse.md
1769 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1770 (*add<mode>3<mask_name>_bcst_2): Likewise.
1772 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1774 * config/i386/sse.md
1775 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1776 Replace nonimmediate_operand with register_operand.
1777 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1779 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1782 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1784 PR rtl-optimization/87596
1785 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1786 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1787 for instructions in FROM..TO range.
1789 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1791 * cfgexpand.c (expand_one_var): Use specific wording in error message
1792 for non-local frame variables.
1793 * stor-layout.c (layout_decl): Do not issue a warning for them.
1795 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1797 * haifa-sched.c (priority): Add force_recompute parameter.
1798 (apply_replacement): Call priority () with force_recompute = true.
1799 (restore_pattern): Likewise.
1801 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1803 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1804 HOST_BITS_PER_WIDE_INT.
1805 (test_vector_ops_duplicate): Likewise.
1807 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1810 * config/i386/sse.md (VF_AVX512): New.
1811 (avx512bcst): Likewise.
1812 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1814 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1816 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1819 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1821 * doc/invoke.texi (-dumpversion): Improve grammar.
1822 (-dumpfullversion): Make more consistent with -dumpversion.
1824 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1826 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1827 Set JUMP_LABEL to the jump insn.
1828 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1829 Predict emitted jump and add label to jump insn.
1831 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1833 PR tree-optimization/87562
1834 * input.c (get_substring_ranges_for_loc): Use
1835 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1836 getting the linemap for the endpoint. Verify that it's either
1837 in the same linemap as the start point's spelling location, or
1838 at least in the same file.
1840 2018-10-18 Richard Biener <rguenther@suse.de>
1842 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1843 feed width-specific load/store costs through ix86_vec_cost.
1844 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1845 (k8_cost): Likewise.
1846 (bdver_cost): Likewise.
1847 (znver1_cost): Likewise.
1848 (btver1_cost): Likewise.
1849 (btver2_cost): Likewise.
1851 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1853 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1854 to simplify subreg of vec_merge.
1856 2018-10-18 Richard Biener <rguenther@suse.de>
1858 * config/i386/i386.c: Fix costing of vector FMA.
1860 2018-10-18 Richard Biener <rguenther@suse.de>
1862 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1864 (ix86_builtin_vectorization_cost): For vec_construct properly
1865 cost insertion into SSE regs.
1866 (...): Adjust calls to ix86_vec_cost.
1868 2018-10-18 Richard Biener <rguenther@suse.de>
1872 2018-02-07 Richard Biener <rguenther@suse.de>
1874 PR tree-optimization/84204
1875 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1878 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1881 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1883 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1884 VEC_MERGE of a VEC_DUPLICATE.
1886 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1888 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1889 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1890 * doc/standards.texi (C Language): Document C2X.
1891 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1892 (rl78_option_override): Handle "GNU C2X" language name.
1894 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1896 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1897 Document C17 as published in 2018.
1899 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1902 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1903 bail out if both sides do not have the same storage order.
1905 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1907 * bitmap.c (bitmap_head::dump): New.
1908 * bitmap.h (bitmap_head): Add dump().
1909 * gimple-ssa-evrp-analyze.c
1910 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1911 (evrp_range_analyzer::set_ssa_range_info): Same.
1912 (evrp_range_analyzer::record_ranges_from_phis): Same.
1913 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1914 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1915 * gimple-ssa-sprintf.c (get_int_range): Same.
1916 (format_integer): Same.
1917 (sprintf_dom_walker::handle_gimple_call): Same.
1918 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1919 (ipcp_vr_lattice::top_p): Same.
1920 (ipcp_vr_lattice::bottom_p): Same.
1921 (ipcp_vr_lattice::set_to_bottom): Same.
1922 (ipa_vr_operation_and_type_effects): Same.
1923 (propagate_vr_across_jump_function): Same.
1924 (ipcp_store_vr_results): Same.
1925 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1926 (ipa_print_node_jump_functions_for_edge): Same.
1927 (ipa_get_value_range): Same.
1928 (ipa_compute_jump_functions_for_edge): Same.
1929 (ipa_write_jump_function): Same.
1930 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1931 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1933 * vr-values.c (set_value_range_to_nonnegative): Same.
1934 (set_value_range_to_truthvalue): Same.
1935 (vr_values::get_value_range): Same.
1936 (vr_values::set_defs_to_varying): Same.
1937 (vr_values::update_value_range): Same.
1938 (symbolic_range_based_on_p): Same.
1939 (vr_values::op_with_boolean_value_range_p): Same.
1940 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1941 (vr_values::extract_range_from_ssa_name): Same.
1942 (vr_values::extract_range_from_binary_expr): Same.
1943 (vr_values::extract_range_from_unary_expr): Same.
1944 (vr_values::extract_range_from_cond_expr): Same.
1945 (vr_values::extract_range_from_comparison): Same.
1946 (vr_values::check_for_binary_op_overflow): Same.
1947 (vr_values::extract_range_basic): Same.
1948 (vr_values::extract_range_from_assignment): Same.
1949 (compare_ranges): Same.
1950 (compare_range_with_value): Same.
1951 (vr_values::adjust_range_with_scev): Same.
1952 (vrp_valueize): Same.
1953 (vrp_valueize_1): Same.
1954 (vr_values::get_vr_for_comparison): Same.
1955 (vr_values::compare_name_with_value): Same.
1956 (vr_values::compare_names): Same.
1957 (vr_values::vrp_evaluate_conditional): Same.
1958 (find_case_label_ranges): Same.
1959 (vr_values::vrp_visit_switch_stmt): Same.
1960 (vr_values::extract_range_from_phi_node): Same.
1961 (vr_values::simplify_div_or_mod_using_ranges): Same.
1962 (vr_values::simplify_bit_ops_using_ranges): Same.
1963 (test_for_singularity): Same.
1964 (range_fits_type_p): Same.
1965 (vr_values::simplify_cond_using_ranges_1): Same.
1966 (vr_values::simplify_switch_using_ranges): Same.
1967 (vr_values::simplify_float_conversion_using_ranges): Same.
1968 (vr_values::two_valued_val_range_p): Same.
1969 (vr_values::add_equivalence): Move to value_range::equiv_add.
1970 * vr-values.h (vr_values::add_equivalence): Remove.
1971 (VR_INITIALIZER): Remove.
1972 * tree-vrp.c (value_range::set): New.
1973 (value_range::equiv_add): New.
1974 (value_range::value_range): New.
1975 (value_range::deep_copy): New.
1976 (value_range::check): New.
1977 (value_range::equal_p): New.
1978 (value_range::ignore_equivs_equal_p): New.
1979 (value_range::operator==): New.
1980 (value_range::operator!=): New.
1981 (value_range::symbolic_p): New.
1982 (value_range::numeric_p): New.
1983 (value_range::set_undefined): New.
1984 (value_range::set_varying): New.
1985 (value_range::may_contain_p): New.
1986 (value_range::equiv_clear): New.
1987 (value_range::singleton_p): New.
1988 (value_range::intersect): New.
1989 (value_range::dump): New.
1990 (value_range::set_and_canonicalize): New.
1991 (set_value_range): Adjust for value_range API.
1992 (set_value_range_to_undefined): Same.
1993 (set_value_range_to_varying): Same.
1994 (set_and_canonicalize_value_range): Same.
1995 (set_value_range_to_nonnull): Same.
1996 (set_value_range_to_null): Same.
1997 (range_is_null): Same.
1998 (range_is_nonnull): Same.
1999 (range_int_cst_p): Same.
2000 (range_int_cst_singleton_p): Same.
2001 (symbolic_range_p): Same.
2002 (range_includes_zero_p): Same.
2003 (value_range_constant_singleton): Same.
2004 (vrp_set_zero_nonzero_bits): Same.
2005 (ranges_from_anti_range): Same.
2006 (extract_range_into_wide_ints): Same.
2007 (extract_range_from_multiplicative_op): Same.
2008 (set_value_range_with_overflow): Same.
2009 (extract_range_from_binary_expr_1): Same.
2010 (extract_range_from_unary_expr): Same.
2011 (dump_value_range): Same.
2012 (debug_value_range): Same.
2013 (vrp_prop::check_array_ref): Same.
2014 (vrp_prop::check_mem_ref): Same.
2015 (vrp_prop::vrp_initialize): Same.
2016 (vrp_prop::visit_stmt): Same.
2017 (intersect_ranges): Same.
2018 (vrp_prop::visit_phi): Same.
2019 (vrp_prop::vrp_finalize): Same.
2020 (determine_value_range_1): Same.
2021 (determine_value_range): Same.
2022 (vrp_intersect_ranges_1): Rename to...
2023 (vrp_intersect_1): this.
2024 (vrp_intersect_ranges): Rename to...
2025 (value_range::intersect_helper): ...this.
2026 (vrp_meet_1): Rename to...
2027 (value_range::union_helper): ...this.
2028 (vrp_meet): Rename to...
2029 (value_range::union_): ...this.
2030 (copy_value_range): Remove.
2031 * tree-vrp.h (struct value_range): Rewrite into a proper class.
2032 (value_range::vrtype): New.
2033 (value_range::type): New.
2034 (value_range::equiv): New.
2035 (value_range::min): New.
2036 (value_range::max): New.
2037 (value_range::varying_p): New.
2038 (value_range::undefined_p): New.
2039 (value_range::null_p): New.
2040 (value_range::equiv_add): New.
2041 (copy_value_range): Remove.
2043 2018-10-17 David Malcolm <dmalcolm@redhat.com>
2045 * Makefile.in (SELFTEST_TARGETS): New.
2046 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
2047 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
2048 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
2050 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
2051 (selftest-c++-gdb, selftest-c++-valgrind): Move to
2053 * configure: Regenerate.
2054 * configure.ac (selftest_languages): New.
2056 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2058 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
2059 overflow wraps argument.
2060 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
2061 wide_int_range_multiplicative_op.
2062 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
2063 overflow wraps argument.
2064 (wide_int_range_multiplicative_op): Same.
2065 (wide_int_range_lshift): Same.
2066 (wide_int_range_div): Same.
2067 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
2068 (wide_int_range_lshift): Same.
2069 (wide_int_range_div): Same.
2071 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
2073 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
2074 use sign as argument.
2075 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
2076 wide_int_range_shift_undefined_p.
2078 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2080 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
2082 (@despeculate_copy<ALLI_TI:mode>): ... This.
2083 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
2086 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2088 * config.gcc: Obsolete *-*-solaris2.10*.
2089 * doc/install.texi (Specific, *-*-solaris2*): Document it.
2091 2018-10-12 Jeff Law <law@redhat.com>
2093 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
2094 reg + sym +- const_int addressing modes.
2096 2018-10-15 David Malcolm <dmalcolm@redhat.com>
2098 * common.opt (fdiagnostics-minimum-margin-width=): New option.
2099 * diagnostic-show-locus.c (layout::layout): Apply the minimum
2101 (layout::start_annotation_line): Only print up to 3 of the
2102 margin character, to avoid touching the left-hand side.
2103 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
2104 minimum margin width, as set by test_diagnostic_context's ctor.
2105 (selftest::test_fixit_insert_containing_newline): Likewise.
2106 (selftest::test_fixit_insert_containing_newline_2): Likewise.
2107 (selftest::test_line_numbers_multiline_range): Clear
2108 dc.min_margin_width.
2109 * diagnostic.c (diagnostic_initialize): Initialize
2111 * diagnostic.h (struct diagnostic_context): Add field
2113 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
2114 * opts.c (common_handle_option): Handle
2115 OPT_fdiagnostics_minimum_margin_width_.
2116 * selftest-diagnostic.c
2117 (selftest::test_diagnostic_context::test_diagnostic_context):
2118 Initialize min_margin_width to 6.
2119 * toplev.c (general_init): Initialize global_dc->min_margin_width.
2121 2018-10-15 David Malcolm <dmalcolm@redhat.com>
2123 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
2124 Fix usage of "error_at_rich_loc" in the comment.
2126 2018-10-15 Renlin Li <renlin.li@arm.com>
2129 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
2130 if-conversioned loop when it contains ifn with types not
2131 supported by backend.
2132 * internal-fn.c (expand_direct_optab_fn): Add an assert.
2133 (direct_internal_fn_supported_p): New helper function.
2134 * internal-fn.h (direct_internal_fn_supported_p): Declare.
2136 2018-10-15 Jakub Jelinek <jakub@redhat.com>
2139 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2140 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2141 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2143 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2145 PR tree-optimization/87022
2146 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
2147 bits in dist vector rather than the first one.
2149 2018-10-15 Richard Biener <rguenther@suse.de>
2152 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
2153 (visit_loadstore): When a used restrict tag escaped verify that
2154 the points-to solution of "other" pointers do not include
2156 (compute_dependence_clique): If a used restrict tag escaped
2157 communicated that down to visit_loadstore.
2159 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
2161 * config/s390/s390.c (s390_expand_vec_init): Force vector element
2162 into reg if it isn't a general operand.
2164 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2167 * config/i386/sse.md (*vec_dupv2di): Add register source to
2170 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2173 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2174 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
2175 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2176 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2178 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
2180 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
2181 (notice_args_size): Set it in the current trace if no insn that can
2182 throw internally has been seen yet.
2183 (connect_traces): When connecting args_size between traces, allow the
2184 incoming values not to match if there is an insn setting it before the
2185 first insn that can throw internally; in that case, force the creation
2186 of a CFI note on this latter insn.
2188 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
2190 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
2191 the base class, because using the injected-class-name was not clearly
2192 specified until DR 176.
2194 2018-10-12 Paul Koning <ni1d@arrl.net>
2196 * config/pdp11/pdp11.md (doloop_end): New expander.
2197 (doloop_end_insn): renamed from "doloop_end".
2198 (addqi3): New pattern.
2199 (subqi3): New pattern.
2200 * config/pdp11/predicates.md (incdec_operand): New predicate.
2202 2018-10-12 Yury Gribov <tetra2005@gmail.com>
2205 * real.c (format_helper::can_represent_integral_type_p): New function
2206 * real.h (format_helper::can_represent_integral_type_p): Ditto.
2207 * match.pd: New pattern.
2209 2018-10-12 Alexandre Oliva <oliva@adacore.com>
2211 * configure.ac: Introduce --enable-large-address-aware
2212 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
2213 * doc/install.texi: Document it.
2214 * configure, config.in: Rebuilt.
2215 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
2216 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
2217 (LINK_SPEC): Insert it.
2218 * config/i386/mingw-w64.h: Likewise.
2220 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
2222 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
2224 PR rtl-optimization/87600
2225 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
2227 2018-10-12 Paul Koning <ni1d@arrl.net>
2229 * doc/md.texi (doloop_end): Document that the pattern code may
2230 need to check operand mode.
2232 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
2234 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
2235 to zero-extend between int and floating-point registers.
2236 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
2237 ldp into floating-point registers. Add type and arch attributes.
2238 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
2239 Use f_loads for type attribute.
2241 2018-10-11 Martin Sebor <msebor@redhat.com>
2243 * doc/extend.texi (attribute packed): Correct typos.
2245 2018-10-11 Martin Sebor <msebor@redhat.com>
2247 * doc/extend.texi (attribute flatten): Mention interaction with
2250 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
2253 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
2254 Set new_decl virtual flag to zero.
2256 2018-10-11 Martin Sebor <msebor@redhat.com>
2259 * doc/extend.texi (attribute format_arg): Discuss using multiple
2260 attributes on a single function.
2262 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
2264 PR tree-optimization/86829
2265 * match.pd (sin (atan (x))): New simplification rules.
2266 (cos (atan (x))): Likewise.
2267 * real.c (build_sinatan_real): New function.
2268 * real.h (build_sinatan_real): Prototype.
2270 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2272 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
2274 (fold_mergeeo_helper): New helper function.
2275 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
2276 intrinsics. Correct some whitespace indentation issues.
2278 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
2281 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
2282 Use HOST_WIDE_INT_1U for shift.
2284 2018-10-11 Doug Rupp <rupp@adacore.com>
2285 Olivier Hainque <hainque@adacore.com>
2287 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
2288 Pass --relax to the linker for RTPs.
2289 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
2291 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
2292 Jan Hubicka <jh@suse.cz>
2293 Martin Jambor <mjambor@suse.cz>
2295 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
2296 the same elements are repeated rather than printing all of them.
2297 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
2298 "repeated" elements.
2299 * read-rtl-function.c (test_loading_repeat): New function.
2300 (read_rtl_function_c_tests): Call test_loading_repeat.
2301 * rtl-tests.c (test_dumping_repeat): New function.
2302 (rtl_tests_c_tests): Call test_dumping_repeat.
2304 2018-10-11 Richard Biener <rguenther@suse.de>
2306 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
2307 bdver?_cost): Unify to ...
2308 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
2309 * config/i386/i386.c (processor_cost_table): Adjust.
2311 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
2314 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
2315 the thunk when expanding to GIMPLE.
2317 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2319 * varasm.c (mergeable_string_section): Don't try to move zero-length
2320 strings to the merge section.
2322 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
2325 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
2327 2018-10-10 Jakub Jelinek <jakub@redhat.com>
2330 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
2331 to special_args set.
2333 2018-10-10 Richard Biener <rguenther@suse.de>
2335 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
2336 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
2337 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
2338 and recursing and pattern terminating the recursion on SSE
2339 vector width using ix86_expand_reduc.
2340 (reduc_sminmax_scal_<mode>): Split into part reducing to half
2341 width and recursing and SSE2 vector variant doing the final
2342 reduction with ix86_expand_reduc.
2343 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
2344 with terminating the recursion at AVX level, splitting that
2347 2018-10-09 David Malcolm <dmalcolm@redhat.com>
2349 * genmatch.c (error_cb): Rename to...
2350 (diagnostic_cb): ...this, converting int params to enums.
2351 (fatal_at): Update for renaming.
2352 (warning_at): Likewise.
2354 * input.c (selftest::ebcdic_execution_charset::apply):
2355 Update for renaming of...
2356 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
2358 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
2359 converting level and reason to enums.
2360 (class selftest::lexer_error_sink): Rename to...
2361 (class selftest::lexer_test_options): ...this, renaming field
2362 "m_errors" to "m_diagnostics".
2363 (selftest::lexer_test_options::apply): Update for renaming of...
2364 (selftest::lexer_test_options::on_error): ...this, renaming to...
2365 (selftest::lexer_test_options::on_diagnostic): ...this
2366 converting level and reason to enums.
2367 (selftest::test_lexer_string_locations_raw_string_unterminated):
2368 Update for renamings.
2369 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
2372 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
2374 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
2375 * config/rs6000/pmmintrin.h: New file.
2377 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
2379 PR tree-optimization/86659
2380 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
2382 2018-10-09 Richard Biener <rguenther@suse.de>
2384 PR tree-optimization/63155
2385 * tree-ssa-structalias.c: Include tree-ssa.h.
2386 (get_constraint_for_ssa_var): For undefs return nothing_id.
2387 (find_func_aliases): Cleanup PHI handling.
2389 2018-10-09 Richard Biener <rguenther@suse.de>
2391 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
2394 2018-10-09 Martin Liska <mliska@suse.cz>
2396 * asan.c (asan_emit_stack_protection): If a stack variable
2397 is located in a same file as current function, then emit
2398 line info into variable definition string.
2400 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2402 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
2405 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2407 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
2410 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2412 PR tree-optimization/86659
2413 * gimple-match.h (struct gimple_match_op): Add reverse field.
2414 (gimple_match_op::set_op): New overloaded method.
2415 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
2416 the REF_REVERSE_STORAGE_ORDER flag on the value.
2417 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
2418 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
2420 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
2423 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
2424 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
2426 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
2429 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
2430 Defined with __builtin_ia32_vfmaddsubpd512_mask.
2432 2018-10-08 Richard Biener <rguenther@suse.de>
2434 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
2435 cost the same as AVX128 ones.
2437 2018-10-08 Paul Koning <ni1d@arrl.net>
2439 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
2440 (expand_block_move): New function.
2441 * config/pdp11/pdp11.c (output_block_move): Remove.
2442 (expand_block_move): New function.
2443 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
2444 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
2445 (*movmemhi1): Remove.
2447 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
2449 * config/s390/2827.md: Increase latencies for some FP instructions.
2451 2018-10-08 Richard Biener <rguenther@suse.de>
2453 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
2455 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
2456 * tree-vectorizer.h (dump_stmt_cost): Adjust.
2457 (add_stmt_cost): Dump return value of the hook.
2459 2018-10-08 Richard Biener <rguenther@suse.de>
2461 PR tree-optimization/63155
2462 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
2463 (ssa_propagation_engine::ssa_propagate): Remove redundant
2464 bitmap bit clearing.
2466 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
2468 PR rtl-optimization/86939
2469 PR rtl-optimization/87479
2470 * ira.h (non_conflicting_reg_copy_p): New prototype.
2471 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
2472 (make_hard_regno_dead): Don't add conflicts for register
2473 ignore_reg_for_conflicts.
2474 (make_object_dead): Likewise.
2475 (non_conflicting_reg_copy_p): New function.
2476 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
2477 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
2478 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
2479 (make_hard_regno_dead): Don't add conflicts for register
2480 ignore_reg_for_conflicts. Remove special conflict handling of
2481 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
2482 check_pic_pseudo_p and update callers.
2483 (mark_pseudo_dead): Don't add conflicts for register
2484 ignore_reg_for_conflicts.
2485 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
2487 2018-10-05 Andrew Waterman <andrew@sifive.com>
2488 Jim Wilson <jimw@sifive.com>
2490 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
2491 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
2492 new pattern using HONOR_SNANS that emits one extra instruction.
2494 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2496 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
2497 patterns): Merge SI and DI patterns to a GPR pattern.
2498 (unnamed define_insn and define_split for record form of that): Merge
2499 to a single define_insn_and_split pattern.
2501 2018-10-05 David Malcolm <dmalcolm@redhat.com>
2504 * input.c (expand_location_to_spelling_point): Add param "aspect"
2505 and use rather than hardcoding LOCATION_ASPECT_CARET.
2506 (get_substring_ranges_for_loc): Handle the case of a single token
2507 within a macro expansion.
2508 * input.h (expand_location_to_spelling_point): Add "aspect" param,
2509 defaulting to LOCATION_ASPECT_CARET.
2511 2018-10-05 Paul Koning <ni1d@arrl.net>
2513 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
2514 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
2515 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2516 (pdp11_guard_type): New function.
2518 2018-10-05 Paul Koning <ni1d@arrl.net>
2520 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
2521 * config/pdp11/pdp11.opt (mfloat32): Remove.
2523 * doc/invoke.texi (pdp11 -mfloat32): Remove:
2524 (pdp11 -mfloat64): Remove.
2526 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2528 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
2529 (*cmp<mode>_cc_i387): Ditto.
2530 (*cmpu<mode>_cc_i387): Ditto.
2531 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2532 * config/i386/i386.c (ix86_expand_fp_compare): Remove
2534 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
2535 Emit x86_sahf_1 pattern.
2536 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
2537 (ix86_expand_carry_flag_compare): Ditto.
2539 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2541 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
2542 to reg_or_0_operand. Add "C" constraint.
2543 (*cmpxf_cc_i387): Ditto.
2544 (*cmp<mode>_i387): Change operand 2 predicate
2545 to nonimm_or_0_operand. Add "C" constraint.
2546 (*cmp<mode>_cc_i387): Ditto.
2547 (*cmp<mode>_0_i387): Remove insn pattern.
2548 (*cmp<mode>_0_cc_i387): Ditto.
2550 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2552 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2553 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2554 from vector_move_operand. Update all uses.
2556 2018-10-05 Martin Sebor <msebor@redhat.com>
2558 PR tree-optimization/87490
2559 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2562 2018-10-05 Richard Biener <rguenther@suse.de>
2564 PR tree-optimization/63155
2565 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2566 vertical space in dumpfiles.
2567 * tree-ssa-propagate.h
2568 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2569 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2570 (ssa_edge_worklist_back): Likewise.
2571 (curr_order): Likewise.
2572 (cfg_blocks_get): Remove abstraction.
2573 (cfg_blocks_add): Likewise.
2574 (cfg_blocks_empty_p): Likewise.
2575 (add_ssa_edge): Add to current or next worklist based on
2577 (add_control_edge): Likewise.
2578 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2580 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2581 iteration from CFG and SSA edge worklist so we process
2582 everything in RPO order, prioritizing forward progress
2584 (ssa_prop_init): Allocate new worklists, do not dump
2586 (ssa_prop_fini): Free new worklists.
2588 2018-10-05 Richard Biener <rguenther@suse.de>
2590 * tree-core.h (tree_block::abstract_flag): Remove.
2591 (tree_block::block_num): Make full 32bits.
2592 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2593 * tree.h (BLOCK_ABSTRACT): Remove.
2594 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2595 resulting from BLOCK_ABSTRACT being always false.
2596 (gen_inlined_subroutine_die): Likewise.
2597 (gen_block_die): Likewise.
2598 * tree.c (block_ultimate_origin): Likewise.
2599 * tree-pretty-print.c (dump_block_node): Remove code dealing
2600 with BLOCK_ABSTRACT.
2601 * tree-ssa-live.c (dump_scope_block): Likewise.
2602 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2603 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2605 2018-10-05 Richard Biener <rguenther@suse.de>
2607 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2608 is asked for initialize mode to the component mode of the
2611 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2614 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2615 assembler for -mavx.
2616 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2618 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2621 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2623 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2624 to be DImode. When using mffscrn, force the operand to a register.
2626 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2628 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2629 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2630 X87MODEF mode iterator.
2631 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2632 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2633 X87MODEF mode iterator.
2635 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2637 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2638 -Wno-prio-ctor-dtor.
2640 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2642 * Makefile.in (OBJS): Add opt-problem.o.
2643 * dump-context.h: Include "selftest.h.
2644 (selftest::temp_dump_context): New forward decl.
2645 (class dump_context): Make friend of class
2646 selftest::temp_dump_context.
2647 (dump_context::dump_loc_immediate): New decl.
2648 (class dump_pretty_printer): Move here from dumpfile.c.
2649 (class temp_dump_context): Move to namespace selftest.
2650 (temp_dump_context::temp_dump_context): Add param
2651 "forcibly_enable_dumping".
2652 (selftest::verify_dumped_text):
2653 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2654 (selftest::verify_item):
2655 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2656 (ASSERT_IS_TREE): Likewise.
2657 (ASSERT_IS_GIMPLE): Likewise.
2658 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2660 (dump_context::dump_loc_immediate): ...this new function.
2661 (class dump_pretty_printer): Move to dump-context.h.
2662 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2663 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2664 (temp_dump_context::temp_dump_context): Move to "selftest"
2665 namespace. Add param "forcibly_enable_dumping", and use it to
2666 conditionalize the use of m_pp;
2667 (selftest::verify_dumped_text): Make non-static.
2668 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2669 (selftest::verify_item): Make non-static.
2670 (ASSERT_IS_TEXT): Move to dump-context.h.
2671 (ASSERT_IS_TREE): Likewise.
2672 (ASSERT_IS_GIMPLE): Likewise.
2673 (selftest::test_capture_of_dump_calls): Pass "true" for new
2674 param of temp_dump_context.
2675 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2676 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2678 * opt-problem.cc: New file.
2679 * opt-problem.h: New file.
2680 * optinfo-emit-json.cc
2681 (selftest::test_building_json_from_dump_calls): Pass "true" for
2682 new param of temp_dump_context.
2683 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2684 (optinfo::emit_for_opt_problem): New function.
2685 (optinfo::emit): Clarity which emit_item is used.
2686 * optinfo.h (optinfo::get_dump_location): New accessor.
2687 (optinfo::emit_for_opt_problem): New decl.
2688 (optinfo::emit): Make const.
2689 * selftest-run-tests.c (selftest::run_tests): Call
2690 selftest::opt_problem_cc_tests.
2691 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2692 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2693 bool to opt_result, converting fprintf messages to
2694 opt_result::failure_at calls. Add "stmt" param for use by the
2696 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2697 (runtime_alias_check_p): Convert return type from bool to
2698 opt_result, converting dump_printf calls to
2699 opt_result::failure_at, using the statement DDR_A for their
2701 (find_data_references_in_stmt): Convert return type from bool to
2702 opt_result, converting "return false" to opt_result::failure_at
2704 * tree-data-ref.h: Include "opt-problem.h".
2705 (dr_analyze_innermost): Convert return type from bool to opt_result,
2706 and add a const gimple * param.
2707 (find_data_references_in_stmt): Convert return type from bool to
2709 (runtime_alias_check_p): Likewise.
2710 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2711 dr_analyze_innermost.
2712 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2713 Convert return type from bool to opt_result, adding a message for
2714 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2715 (vect_analyze_data_ref_dependence): Convert return type from bool
2716 to opt_result. Change sense of return type from "false"
2717 effectively meaning "no problems" to "false" meaning a problem,
2718 so that "return false" becomes "return opt_result::success".
2719 Convert "return true" calls to opt_result::failure_at, using
2720 the location of statement A rather than vect_location.
2721 (vect_analyze_data_ref_dependences): Convert return type from bool
2723 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2724 calls to opt_result::failure_at, using the stmt location rather
2726 (vect_verify_datarefs_alignment): Convert return type from bool
2728 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2729 into multiple more-tightly-scoped copies.
2730 (vect_analyze_data_refs_alignment): Convert return type from bool
2732 (vect_analyze_data_ref_accesses): Likewise, converting a
2733 "return false" to a "return opt_result::failure_at", adding a
2735 (vect_prune_runtime_alias_test_list): Convert return type from
2736 bool to opt_result, converting dump_printf_loc to
2737 opt_result::failure_at. Add a %G to show the pertinent statement,
2738 and use the stmt's location rather than vect_location.
2739 (vect_find_stmt_data_reference): Convert return type from
2740 bool to opt_result, converting dump_printf_loc to
2741 opt_result::failure_at, using stmt's location.
2742 (vect_analyze_data_refs): Convert return type from bool to
2743 opt_result. Convert "return false" to "return
2744 opt_result::failure_at", adding messages as needed.
2745 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2746 type from bool to opt_result.
2747 (vect_determine_vf_for_stmt): Likewise.
2748 (vect_determine_vectorization_factor): Likewise, converting
2749 dump_printf_loc to opt_result::failure_at, using location of phi
2750 rather than vect_location.
2751 (vect_analyze_loop_form_1): Convert return type from bool to
2752 opt_result, converting dump_printf_loc calls, retaining the use of
2754 (vect_analyze_loop_form): Convert return type from loop_vec_info
2755 to opt_loop_vec_info.
2756 (vect_analyze_loop_operations): Convert return type from bool to
2757 opt_result, converting dump_printf_loc calls, using the location
2758 of phi/stmt rather than vect_location where available. Convert
2759 various "return false" to "return opt_result::failure_at" with
2760 "unsupported phi" messages.
2761 (vect_get_datarefs_in_loop): Convert return type from bool to
2762 opt_result. Add a message for the
2763 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2764 (vect_analyze_loop_2): Convert return type from bool to
2765 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2766 each "goto again;", adding new messages where needed.
2767 Add "unsupported grouped {store|load}" messages.
2768 (vect_analyze_loop): Convert return type from loop_vec_info to
2770 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2772 * tree-vect-stmts.c (process_use): Likewise, converting
2773 dump_printf_loc call and using stmt location, rather than
2775 (vect_mark_stmts_to_be_vectorized): Likeise.
2776 (vect_analyze_stmt): Likewise, adding a %G.
2777 (vect_get_vector_types_for_stmt): Convert return type from bool to
2778 opt_result, converting dump_printf_loc calls and using stmt
2779 location, rather than vect_location.
2780 (vect_get_mask_type_for_stmt): Convert return type from tree to
2781 opt_tree, converting dump_printf_loc calls and using stmt location.
2782 * tree-vectorizer.c: Include "opt-problem.h.
2783 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2784 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2785 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2786 enabled, use it to report at the top level "couldn't vectorize
2787 loop" followed by the problem.
2788 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2789 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2791 (vect_analyze_stmt): Likewise.
2792 (vect_get_vector_types_for_stmt): Likewise.
2793 (tree vect_get_mask_type_for_stmt): Likewise.
2794 (vect_analyze_data_ref_dependences): Likewise.
2795 (vect_enhance_data_refs_alignment): Likewise.
2796 (vect_analyze_data_refs_alignment): Likewise.
2797 (vect_verify_datarefs_alignment): Likewise.
2798 (vect_analyze_data_ref_accesses): Likewise.
2799 (vect_prune_runtime_alias_test_list): Likewise.
2800 (vect_find_stmt_data_reference): Likewise.
2801 (vect_analyze_data_refs): Likewise.
2802 (vect_analyze_loop): Convert return type from loop_vec_info to
2804 (vect_analyze_loop_form): Likewise.
2805 (vect_analyze_slp): Convert return type from bool to opt_result.
2807 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2809 * doc/invoke.texi (-fopt-info): Document new "internals"
2811 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2812 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2814 (optinfo_verbosity_options): Add "internals".
2815 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2816 (dump_context::apply_dump_filter_p): New member function.
2817 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2818 explicitly masking the dump_kind.
2819 (dump_context::begin_scope): Increment the scope depth first. Use
2820 apply_dump_filter_p rather than explicitly masking the dump_kind.
2821 (dump_context::emit_item): Use apply_dump_filter_p rather than
2822 explicitly masking the dump_kind.
2823 (dump_dec): Likewise.
2824 (dump_hex): Likewise.
2825 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2826 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2827 (opt_info_switch_p): Update handling of default
2828 MSG_OPTIMIZED_LOCATIONS to cope with default of
2829 MSG_PRIORITY_USER_FACING.
2830 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2831 masking the dump_kind.
2832 (selftest::test_capture_of_dump_calls): Update test_dump_context
2833 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2834 than MSG_ALL. Generalize scope test to be run at all four
2835 combinations of with/without MSG_PRIORITY_USER_FACING and
2836 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2837 for each of the two values.
2838 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2839 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2840 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2841 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2842 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2843 with MSG_PRIORITY_*.
2844 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2845 dump messages as MSG_PRIORITY_USER_FACING.
2846 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2847 about the interaction with MSG_PRIORITY_*.
2849 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2851 * varasm.c (output_constant): Add new parameter merge_strings.
2852 Make strings properly zero terminated in merge string sections.
2853 (mergeable_string_section): Don't fail if the last char is non-zero.
2854 (assemble_variable_contents): Handle merge string sections.
2855 (assemble_variable): Likewise.
2856 (assemble_constant_contents): Likewise.
2857 (output_constant_def_contents): Likewise.
2858 (output_constructor_array_range,
2859 output_constructor_regular_field): Adjust call to output_constant.
2860 (output_object_block): Adjust call to assemble_constant_contents
2861 and assemble_variable_contents.
2863 2018-10-04 Martin Liska <mliska@suse.cz>
2866 * cgraphunit.c (process_function_and_variable_attributes):
2867 Warn about a function with alias attribute and a body.
2869 2018-10-04 Martin Liska <mliska@suse.cz>
2872 * multiple_target.c (redirect_to_specific_clone): New function.
2873 (ipa_target_clone): Use it.
2874 * tree-inline.c: Fix comment.
2876 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2878 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2880 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2881 (gcc::dump_manager::register_pass): New member function, adapted
2882 from loop body in gcc::pass_manager::register_pass, adding a
2883 call to update_dfi_for_opt_info.
2884 (gcc::dump_manager::opt_info_enable_passes): Store the
2885 -fopt-info options into the new fields. Move the loop
2887 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2889 * dumpfile.h (struct opt_pass): New forward decl.
2890 (gcc::dump_manager::register_pass): New decl.
2891 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2892 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2893 "m_optinfo_flags", and "m_optinfo_filename".
2894 * passes.c (gcc::pass_manager::register_pass): Move all of the
2895 dump-handling code to gcc::dump_manager::register_pass.
2897 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2899 PR rtl-optimization/87466
2900 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2901 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2902 * doc/tm.texi: Regenerate.
2903 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2904 * lra-lives.c (process_bb_lives): Likewise.
2905 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2908 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2910 * params.c (add_params): Fix initialization.
2912 2018-10-04 Martin Liska <mliska@suse.cz>
2914 PR gcov-profile/84107
2915 * tree-profile.c (init_ic_make_global_vars):
2916 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2917 Come up with new ic_tuple* variables. Emit
2918 __gcov_indirect_call{,_topn} variables.
2919 (gimple_gen_ic_profiler): Access the variable
2921 (gimple_gen_ic_func_profiler): Access
2922 __gcov_indirect_call.callee field.
2923 (gimple_init_gcov_profiler): Use ptr_type_node.
2924 * value-prof.c (gimple_ic): Use ptr_type_node.
2926 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2928 PR tree-optimization/85787
2929 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2930 into this function and add support for detecting multiple phis.
2931 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2933 2018-10-04 Martin Liska <mliska@suse.cz>
2936 * ipa-inline.c (inline_to_all_callers_1):
2937 Call ultimate_alias_target for node being inlined.
2939 2018-10-03 Jeff Law <law@redhat.com>
2941 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2943 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2944 * tree.h (get_typenode_from_name): Prototype.
2946 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2948 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2949 Change operand 2 predicate to nonimmediate_operand.
2950 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2952 2018-10-03 Martin Sebor <msebor@redhat.com>
2953 Jeff Law <law@redhat.com>
2955 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2957 (get_string_length): Detect unterminated arrays.
2958 (format_string): Same.
2959 (format_directive): Warn about unterminated arrays.
2960 (handle_gimple_call): Mark statements with no_warning as needed.
2962 2018-10-03 Jim Wilson <jimw@sifive.com>
2964 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2965 also define __riscv_abi_rve. Delete trailing white space.
2967 2018-10-03 Paul Koning <ni1d@arrl.net>
2969 Enable LRA register allocator for PDP11.
2970 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2973 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2974 * config/pdp11/pdp11.opt (-mlra): New option.
2975 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2977 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2979 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2980 (*<absneg:code>extend<mode>xf2): Ditto.
2982 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2984 PR tree-optimization/87415
2985 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2988 2018-10-02 Jeff Law <law@redhat.com>
2990 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2991 an unterminated string is discovered. Bubble up range
2992 even for unterminated strings.
2993 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2994 indicates the string was not terminated via NONSTR.
2996 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2998 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2999 pointer conversions.
3000 Do not do anything special for anti-ranges.
3002 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
3004 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
3005 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
3007 2018-10-03 Martin Liska <mliska@suse.cz>
3009 PR gcov-profile/86109
3010 * coverage.c (coverage_begin_function): Do not
3011 mark lambdas as artificial.
3012 * tree-core.h (struct GTY): Remove tm_clone_flag
3013 and introduce new lambda_function.
3014 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
3016 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3019 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
3020 P8_VECTOR and VSX are enabled.
3022 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
3024 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3025 0x3907 as CPU model number.
3027 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
3029 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
3030 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
3031 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
3032 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
3033 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
3034 * config/s390/s390.md: Likewise. Rename also the cpu attribute
3035 value from arch12 to z14.
3037 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
3039 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
3041 (isinf<mode>2): Ditto.
3043 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
3045 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
3046 before emitting fxam. Perform calculations in XFmode.
3048 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
3050 * match.pd (((X /[ex] A) +- B) * A): New transformation.
3052 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
3055 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
3056 * tree.c (signed_or_unsigned_type_for): Handle complex.
3058 2018-10-02 Jeff Law <law@redhat.com>
3060 * gimple-fold.c (get_range_strlen): Remove dead code.
3062 2018-10-02 Martin Sebor <msebor@redhat.com>
3063 Jeff Law <law@redhat.com>
3065 * builtins.c (unterminated_array): Add new arguments.
3066 If argument is not terminated, bubble up size and exact
3068 (expand_builtin_strnlen): Detect, avoid expanding
3069 and diagnose unterminated arrays.
3070 (c_strlen): Fill in offset of start of unterminated strings.
3071 * builtins.h (unterminated_array): Update prototype.
3073 2018-10-02 Richard Biener <rguenther@suse.de>
3075 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
3076 of haddv4df, first reduce to SSE width and exploit the fact
3077 that we only need element zero with the reduction result.
3078 (reduc_plus_scal_v2df): Likewise.
3080 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
3082 * dojump.h (do_jump): Delete.
3083 (do_jump_1): Likewise.
3084 (split_comparison): Move around.
3085 * dojump.c (do_jump): Make static.
3086 (do_jump_1): Likewise.
3087 (jumpifnot): Move around.
3088 (jumpifnot_1): Likewise.
3090 (jumpif_1): Likewise.
3091 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
3093 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
3095 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
3096 insns in the delay slot and add_insn_after for the jump insn.
3098 2018-10-02 Richard Biener <rguenther@suse.de>
3100 * tree-inline.c (expand_call_inline): Use the location of
3101 the callee declaration for the inline-entry marker.
3102 * final.c (notice_source_line): Remove special-casing of
3103 NOTE_INSN_INLINE_ENTRY.
3105 2018-10-01 Carl Love <cel@us.ibm.com>
3108 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
3109 (__builtin_mtfsb0): New.
3110 (__builtin_mtfsb1): New.
3111 ( __builtin_set_fpscr_rn): New.
3112 (__builtin_set_fpscr_drn): New.
3113 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
3114 (rs6000_expand_set_fpscr_rn_builtin): Add.
3115 (rs6000_expand_set_fpscr_drn_builtin): Add.
3116 (rs6000_expand_builtin): Add case statement entries for
3117 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
3118 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
3119 RS6000_BUILTIN_MFFSL.
3120 (rs6000_init_builtins): Add ftype initialization and def_builtin
3121 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
3122 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
3123 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
3124 rs6000_mffscdrn): Add define_insn.
3125 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
3126 * doc/extend.texi: Add documentation for the builtins.
3128 2018-10-01 Richard Biener <rguenther@suse.de>
3130 PR tree-optimization/87465
3131 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
3132 causing branch miscounts.
3134 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3136 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
3137 aarch64_option_default_param): New.
3138 (params.h): Include.
3139 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
3140 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
3141 stack-clash protection validation code.
3143 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3145 * params.c (validate_param): New.
3146 (add_params): Use it.
3147 (set_param_value): Refactor param validation into validate_param.
3148 (diagnostic.h): Include.
3149 * diagnostic.h (diagnostic_ready_p): New.
3151 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3153 * params.c (set_param_value):
3154 Add index of parameter being validated.
3155 * common/common-target.def (option_validate_param): New.
3156 * common/common-targhooks.h (default_option_validate_param): New.
3157 * common/common-targhooks.c (default_option_validate_param): New.
3158 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
3159 * doc/tm.texi: Regenerate.
3161 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3164 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3165 Add validation for stack-clash parameters and set defaults.
3167 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3170 * configure.ac: Add stack-clash-protection-guard-size.
3171 * doc/install.texi: Document it.
3172 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
3173 * params.def: Update comment for guard-size.
3174 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
3175 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
3176 * configure: Regenerate.
3178 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3181 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
3182 STACK_DYNAMIC_OFFSET): New.
3183 * config/aarch64/aarch64.c (aarch64_layout_frame):
3184 Update outgoing args size.
3185 (aarch64_stack_clash_protection_alloca_probe_range,
3186 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3188 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3191 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
3193 * target.def (stack_clash_protection_alloca_probe_range): New.
3194 (stack_clash_protection_final_dynamic_probe): Remove.
3195 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
3196 (default_stack_clash_protection_final_dynamic_probe): Remove.
3197 * targhooks.c: Likewise.
3198 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3199 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
3200 * doc/tm.texi: Regenerate.
3202 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3205 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
3206 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
3207 aarch64_clamp_to_uimm12_shift): New.
3208 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
3209 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
3211 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3214 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
3216 2018-10-01 Jeff Law <law@redhat.com>
3217 Richard Sandiford <richard.sandiford@linaro.org>
3218 Tamar Christina <tamar.christina@arm.com>
3221 * config/aarch64/aarch64.md
3222 (probe_stack_range): Add k (SP) constraint.
3223 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
3224 STACK_CLASH_MAX_UNROLL_PAGES): New.
3225 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
3226 stack probes for stack clash.
3227 (aarch64_allocate_and_probe_stack_space): New.
3228 (aarch64_expand_prologue): Use it.
3229 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
3230 (aarch64_sub_sp): Add emit_move_imm optional param.
3232 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
3234 PR tree-optimization/87261
3235 * match.pd: Remove trailing whitespace.
3236 Add (x & y) | ~(x | y) -> ~(x ^ y),
3237 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
3239 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3241 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
3243 (ashlsi3_insn): Update instruction constraints.
3244 (ashrsi3_insn): Likewise.
3245 (rotrsi3): Likewise.
3246 (add_shift): Likewise.
3247 * config/arc/constraints.md (Csz): New 32 bit constraint. It
3248 avoids placing in the limm field small constants which, otherwise,
3249 could end into a small instruction.
3251 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3253 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
3254 destination register is not odd-even.
3255 (umaddsidi4_split): Likewise.
3257 2018-10-01 Richard Biener <rguenther@suse.de>
3259 * tree-inline.c (expand_call_inline): Store origin of fn
3260 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
3261 * tree.c (block_ultimate_origin): Simplify and do some
3264 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
3266 * config/i386/mmx.md (EMMS): New int iterator.
3267 (emms): New int attribute.
3268 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
3269 EMMS int iterator. Explicitly declare clobbers.
3270 (mmx_emms): Remove expander.
3272 * config/i386/predicates.md (emms_operation): Remove predicate.
3273 (vzeroall_pattern): New predicate.
3274 (vzeroupper_pattern): Rename from vzeroupper_operation.
3275 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
3276 vzeroupper_pattern and vzeroall_pattern predicates.
3278 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
3280 PR rtl-optimization/86939
3281 * ira-lives.c (make_hard_regno_born): Rename from this...
3282 (make_hard_regno_live): ... to this. Remove update to conflict
3283 information. Update function comment.
3284 (make_hard_regno_dead): Add conflict information update. Update
3286 (make_object_born): Rename from this...
3287 (make_object_live): ... to this. Remove update to conflict information.
3288 Update function comment.
3289 (make_object_dead): Add conflict information update. Update function
3291 (mark_pseudo_regno_live): Call make_object_live.
3292 (mark_pseudo_regno_subword_live): Likewise.
3293 (mark_hard_reg_dead): Update function comment.
3294 (mark_hard_reg_live): Call make_hard_regno_live.
3295 (process_bb_node_lives): Likewise.
3296 * lra-lives.c (make_hard_regno_born): Rename from this...
3297 (make_hard_regno_live): ... to this. Remove update to conflict
3298 information. Remove now uneeded check_pic_pseudo_p argument.
3299 Update function comment.
3300 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
3301 to conflict information. Update function comment.
3302 (mark_pseudo_live): Remove update to conflict information. Update
3304 (mark_pseudo_dead): Add conflict information update.
3305 (mark_regno_live): Call make_hard_regno_live.
3306 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
3307 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
3309 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
3312 * config/i386/i386.c (construct_container): Use TImode for
3313 BLKmode values in 2 integer registers.
3315 2018-09-29 Jeff Law <law@redhat.com>
3317 * builtins.c (unterminated_array): Pass in c_strlen_data * to
3318 c_strlen rather than just a tree *.
3319 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
3320 Update recursive calls appropriately. If caller did not provide a
3321 suitable data pointer, create a local one. When a non-terminated
3322 string is discovered, bubble up information about the string via the
3323 c_strlen_data object.
3324 * builtins.h (c_strlen): Update prototype.
3325 (c_strlen_data): New structure.
3326 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
3327 For a type 2 call, if c_strlen indicates a non-terminated string
3328 use the length of the non-terminated string.
3329 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
3331 2018-09-29 Jakub Jelinek <jakub@redhat.com>
3334 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
3335 __m512d type for __A argument rather than __m512.
3337 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
3339 * match.pd (simple_comparison): Don't optimize if either operand is
3340 a function pointer when target needs function pointer canonicalization.
3342 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3344 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
3345 power5 .. power9 to remove indirection.
3346 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
3347 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
3348 ASM_CPU_476_SPEC): Delete.
3349 (ASM_CPU_SPEC): Adjust.
3350 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
3351 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
3353 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3355 * config.in: Delete HAVE_AS_DCI.
3356 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
3357 * config/rs6000/rs6000.h: Ditto.
3358 * configure.ac: Delete HAVE_AS_DCI.
3359 * configure: Regenerate.
3361 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3363 * config.in (HAVE_AS_LWSYNC): Delete.
3364 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3365 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
3367 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3368 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
3370 * configure.ac: Delete HAVE_AS_LWSYNC.
3371 * configure: Regenerate.
3373 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3374 Pierre-Marie de Rodat <derodat@adacore.com>
3376 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
3377 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
3378 (cgraph_node::create_thunk): Add indirect_offset parameter.
3379 (thunk_adjust): Likewise.
3380 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
3381 and initialize the corresponding field with it.
3382 (cgraph_node::dump): Dump indirect_offset field.
3383 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
3384 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
3385 (thunk_adjust): Add indirect_offset parameter and deal with it.
3386 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
3387 pass it to thunk_adjust. Do not call the target hook if it's non-zero
3388 or if the thunk is external or local. Fix formatting. Do not chain
3389 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
3390 if any, in the GIMPLE representation.
3391 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
3392 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
3393 (input_node): Read indirect_offset field.
3394 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
3395 call to thunk_adjust.
3396 * tree-nested.c (struct nesting_info): Add thunk_p field.
3397 (create_nesting_tree): Set it.
3398 (convert_all_function_calls): Copy static chain from targets to thunks.
3399 (finalize_nesting_tree_1): Return early for thunks.
3400 (unnest_nesting_tree_1): Do not finalize thunks.
3401 (gimplify_all_functions): Do not gimplify thunks.
3403 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3405 * opt-suggestions.c (option_proposer::build_option_suggestions):
3406 Release "option_values".
3408 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3410 * coverage.c (get_coverage_counts): Convert problem-reporting dump
3411 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
3412 * dumpfile.c (kind_as_string): New function.
3413 (dump_loc): Rather than a hardcoded prefix of "note: ", use
3414 kind_as_string to vary the prefix based on dump_kind.
3415 (selftest::test_capture_of_dump_calls): Update for above.
3417 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3419 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
3420 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
3422 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3424 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
3425 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
3426 INVALID_REGNUM instead of FPSR_REG.
3427 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
3428 * config/i386/i386.md: Update comment of FP compares.
3429 (fldenv): Do not clobber FPSR_REG.
3431 2018-09-28 Richard Biener <rguenther@suse.de>
3433 * tree.h (BLOCK_ORIGIN): New.
3434 * omp-expand.c (grid_expand_target_grid_body): Assign
3435 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
3436 * tree-inline.c (remap_block): Likewise.
3437 * auto-profile.c (get_function_decl_from_block): Simplify
3438 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
3439 * langhooks.c (lhd_print_error_function): Likewise.
3440 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
3442 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
3443 * tree.c (block_nonartificial_location): Likewise.
3444 (block_ultimate_origin): Likewise.
3445 * tree-pretty-print.c (percent_K_format): Likewise. Remove
3446 no longer needed LTO case.
3448 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
3449 Jan Hubicka <jh@suse.cz>
3450 Martin Jambor <mjambor@suse.cz>
3452 * simplify-rtx.c (simplify_merge_mask): New function.
3453 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
3454 same masks are used in op1 or op2.
3455 (test_vec_merge): New function.
3456 (test_vector_ops): Call test_vec_merge.
3458 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3460 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
3461 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
3462 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
3463 (cypress_costs): Set it.
3464 (supersparc_costs): Likewise.
3465 (hypersparc_costs): Likewise.
3466 (leon_cost): Likewise.
3467 (leon3_costs): Likewise.
3468 (sparclet_costs): Likewise.
3469 (ultrasparc_costs): Likewise.
3470 (ultrasparc_costs): Likewise.
3471 (niagara_costs): Likewise.
3472 (niagara2_costs): Likewise.
3473 (niagara3_costs): Likewise.
3474 (niagara4_costs): Likewise.
3475 (niagara7_costs): Likewise.
3476 (m8_costs): Likewise.
3477 (TARGET_CAN_FOLLOW_JUMP): Define.
3478 (pass_work_around_errata::gate): Minor tweak.
3479 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
3480 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
3481 Automaitcally clear MASK_FSMULD mask for V7 processors.
3482 (sparc_can_follow_jump): New static function.
3483 (output_ubranch): Deal with CROSSING_JUMP_P.
3484 (sparc_use_sched_lookahead): Rewrite using switch statement.
3485 (sparc_issue_rate): Reorder.
3486 (sparc_branch_cost): New function.
3488 2018-09-27 Martin Sebor <msebor@redhat.com>
3490 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
3491 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
3492 (int_fits_type_p): Same.
3494 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
3496 * config/i386/i386.md (FPCR_REG): Remove.
3497 (UNSPEC_FLDCW): Remove.
3498 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
3499 (x86_fldcw_1): Remove insn pattern.
3500 (fnstenv): Do not clobber FPCR_REG.
3502 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
3503 (CALL_USED_REGISTERS): Ditto.
3504 (REG_ALLOC_ORDER): Ditto.
3505 (REG_CLASS_CONTENTS): Ditto.
3506 (HI_REGISTER_NAMES): Ditto.
3507 (ADDITIONAL_REGISTER_NAMES): Use defines instead
3508 of numerical constants.
3509 * config/i386/i386.c (regclass_map): Remove fpsr register.
3510 (dbx_register_map): Ditto.
3511 (dbx64_register_map): Ditto.
3512 (svr4_dbx_register_map): Ditto.
3513 (print_reg): Do not handle FPCR_REG.
3515 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
3518 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
3519 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
3520 Delete, always treat as true.
3521 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
3522 Ditto. Simplify remaining code.
3523 * config/powerpcspe/powerpcspe.h: Ditto.
3524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
3525 Simplify remaining code.
3526 (rs6000_expand_builtin): Ditto.
3527 * config/rs6000/rs6000.h: Ditto.
3528 * configure.ac: Ditto.
3529 * configure: Regenerate.
3531 2018-09-27 Martin Liska <mliska@suse.cz>
3533 * coverage.c (get_coverage_counts): Revert the formatting
3534 of missing profile opt info.
3536 2018-09-27 Richard Biener <rguenther@suse.de>
3540 * dwarf2out.c (set_block_origin_self): Do not mark outermost
3541 block as we do not output that.
3542 (gen_inlined_subroutine_die): Elide the originally outermost
3543 block, matching what we do for concrete instances.
3544 (decls_for_scope): Add parameter specifying whether to recurse
3547 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
3548 Tom de Vries <tom@codesourcery.com>
3552 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3553 STORE_FLAG_VALUE == 1.
3555 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3557 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3558 constant definitions.
3559 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3560 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3561 ("speculation_barrier"): New expander definition.
3563 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3565 PR gcov-profile/86957
3566 * common.opt: New warning option -Wmissing-profile.
3567 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3568 * doc/invoke.texi: Document -Wmissing-profile.
3570 2018-09-26 Jim Wilson <jimw@sifive.com>
3572 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3573 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3575 2018-09-26 Martin Sebor <msebor@redhat.com>
3577 * tree.c (zerop): Change return type to bool.
3578 (integer_zerop, integer_onep, integer_each_onep): Same.
3579 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3580 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3581 (real_onep, real_minus_onep, chain_index): Same.
3582 (print_type_hash_statistics, type_list_equal): Same.
3583 * tree.h (zerop): Same.
3584 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3585 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3586 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3587 (real_onep, real_minus_onep, chain_index): Same.
3588 (print_type_hash_statistics, type_list_equal): Same.
3590 2018-09-26 Jim Wilson <jimw@sifive.com>
3592 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3594 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3597 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3598 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3601 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3603 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3605 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3607 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3608 and FP_SECOND_SSE_REGS.
3609 (REG_CLASS_NAMES): Ditto.
3610 (REG_CLASS_CONTENTS): Ditto.
3611 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3612 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3613 (ix86_preferred_output_reload_class): Ditto.
3614 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3615 clobber constraint to "=&f".
3616 (fix_truncdi_i387): Ditto.
3617 (lrintxfdi2): Ditto.
3618 (fistdi2_<rounding>): Ditto.
3619 (fpremxf4_i387): Change "=u" constraint to "=f".
3620 (fprem1xf4_i387): Ditto.
3622 (fptanxf4_i387): Ditto.
3623 (fxtractxf3_i387): Ditto.
3624 (fscalexf4_i387): Ditto.
3625 (atan2xf3): Change "u" constraint to "f".
3626 (fyl2xxf3_i387): Ditto.
3627 (fyl2xp1xf3_i387): Ditto.
3629 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3632 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3633 for removed I387_MASK_PM entity.
3636 2018-09-26 Jeff Law <law@redhat.com>
3638 2018-09-26 Alexey Neyman <stilor@att.net>
3640 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3641 headers are no longer pulled in by <isl/val.h>.
3643 2018-09-26 Richard Biener <rguenther@suse.de>
3646 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3647 or concrete instance DIE to the tree. Create abstract origin
3648 attributes also for concrete instances.
3650 2018-09-26 Alexey Neyman <stilor@att.net>
3652 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3653 headers are no longer pulled in by <isl/val.h>.
3655 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3657 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3658 Use new helper functions.
3659 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3660 Use new helper functions.
3661 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3662 aarch_mm_needs_release): New declarations.
3663 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3664 aarch_mm_needs_release): New.
3666 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3668 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3669 (arm32_output_mi_thunk): Deal with long calls.
3671 2018-09-26 Richard Biener <rguenther@suse.de>
3675 * tree-inline.c (expand_call_inline): When the location
3676 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3677 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3678 the inserted BLOCK to make inlined_function_outer_scope_p
3680 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3681 coords for reserved locations.
3683 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3685 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3686 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3687 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3688 (*sibcall_nonlocal_sysv<mode>): Ditto.
3689 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3690 (<bd>_<mode>): Ditto.
3691 (<bd>tf_<mode>): Ditto.
3693 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3695 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3696 control string as a list of templates instead of as C code.
3697 (*altivec_movti): Ditto.
3698 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3700 2018-09-25 Jim Wilson <jimw@sifive.com>
3702 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3703 when target symbol is weak.
3705 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3708 * builtins.c (unterminated_array): Simplify.
3709 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3710 where pointer arithmetic is safe.
3712 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3715 * config/rs6000/altivec.md (altivec_vspltb): Use
3716 const_0_to_15_operand instead of u5bit_cint_operand.
3717 (*altivec_vspltb_internal): Ditto.
3718 (altivec_vspltb_direct): Ditto.
3719 (altivec_vsplth): Use const_0_to_7_operand instead of
3721 (*altivec_vsplth_internal): Ditto.
3722 (altivec_vsplth_direct): Ditto.
3723 (altivec_vspltw): Use const_0_to_3_operand instead of
3725 (*altivec_vspltw_internal): Ditto.
3726 (altivec_vspltw_direct): Ditto.
3727 (altivec_vspltsf): Ditto.
3728 (*altivec_vspltsf_internal): Ditto.
3729 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3730 various splats with the proper size immediate. Reorder the various
3731 cases by ascending size of immediate, and put all such together.
3733 2018-09-25 Richard Biener <rguenther@suse.de>
3736 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3738 (maybe_create_die_with_external_ref): Do not create
3739 DW_TAG_imported_unit here.
3740 (add_abstract_origin_attribute): Handle external BLOCK refs.
3741 (dwarf2out_abstract_function): Simplify LTO case.
3742 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3743 rather than using maybe_create_die_with_external_ref.
3745 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3748 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3749 (frndintxf2_mask_pm_i387): Ditto.
3750 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3751 Enable for !flag_trapping_math.
3752 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3753 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3754 Change operand 1 predicate to nonimmediate_operand.
3755 (attr "i387_cw"): Remove mask_pm.
3756 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3757 (enum ix86_entity): Remove I387_MASK_PM.
3758 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3759 handle I387_MASK_PM.
3760 (ix86_mode_needed): Ditto.
3761 (ix86_mode_after): Ditto.
3762 (ix86_mode_entry): Ditto.
3763 (ix86_mode_exit): Ditto.
3764 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3766 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3768 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3769 to_update_switch_stmts to vNULL instead of calling create on them
3772 2018-09-25 Richard Biener <rguenther@suse.de>
3774 PR tree-optimization/87402
3775 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3776 (visit_phi): Re-instantiate handling of supposed to be VARYING
3777 but non-VARYING backedge value.
3779 2018-09-25 Richard Biener <rguenther@suse.de>
3782 * dwarf2out.c (struct sym_off_pair): New.
3783 (external_die_map): New global.
3784 (lookup_decl_die): When in LTO create DIEs lazily from the
3786 (lookup_block_die): New function, create DIEs lazily in LTO.
3787 (equate_block_to_die): New function.
3788 (dwarf2out_die_ref_for_decl): During WPA get the association
3789 from the external DIE map.
3790 (dwarf2out_register_external_die): Record mapping into the
3792 (maybe_create_die_with_external_ref): New function split out from
3793 DIE generation part of old dwarf2out_register_external_die.
3794 (add_abstract_origin_attribute): Do not return the DIE. When
3795 in LTO reference externals directly.
3796 (dwarf2out_abstract_function): When in LTO ignore calls for
3797 decls with external DIEs (already present abstract instances).
3798 (gen_call_site_die): Adjust.
3799 (add_high_low_attributes): Likewise.
3800 (gen_lexical_block_die): Likewise.
3801 (gen_inlined_subroutine_die): Likewie.
3802 (gen_block_die): Likewise.
3803 (dwarf2out_inline_entry): Likewise.
3804 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3807 2018-09-25 Martin Liska <mliska@suse.cz>
3809 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3810 integers and not by a float value.
3812 2018-09-25 Martin Liska <mliska@suse.cz>
3815 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3817 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3818 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3819 * godump.c (go_format_type): Remove extra parenthesis.
3821 2018-09-25 Martin Liska <mliska@suse.cz>
3823 * alias.c (set_dest_equal_p): Remove unused function.
3824 * config/i386/i386.c (def_builtin_pure2): Likewise.
3825 * diagnostic-show-locus.c (class layout): Remove
3827 (layout::layout): Likewise here.
3828 * dump-context.h (class temp_dump_context): Likewise.
3829 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3830 (add_AT_loclistsptr): Likewise.
3831 (add_AT_offset): Likewise.
3832 (get_AT_hi_pc): Likewise.
3833 (is_comdat_die): Likewise.
3834 (type_is_enum): Likewise.
3835 (ceiling): Likewise.
3836 (add_AT_vms_delta): Likewise.
3837 (is_class_die): Likewise.
3838 * edit-context.c (class line_event): Remove unused field.
3839 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3841 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3842 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3843 (ok_for_base_p_nonstrict): Likewise.
3844 * tree-chrec.c (is_not_constant_evolution): Likewise.
3845 (chrec_fold_poly_cst): Likewise.
3846 * tree-if-conv.c (has_pred_critical_p): Likewise.
3847 * tree-ssa-coalesce.c (print_exprs): Likewise.
3848 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3849 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3850 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3851 (value_range_nonnegative_p): Likewise.
3853 2018-09-25 Martin Liska <mliska@suse.cz>
3855 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3856 Do not handle "GNU Pascal".
3857 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3859 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3860 from documentation. Likewise.
3861 * dbxout.c (dbxout_range_type): Likewise.
3862 * doc/cpp.texi: Likewise.
3863 * doc/extend.texi: Likewise.
3864 * doc/frontends.texi: Likewise.
3865 * doc/invoke.texi: Remove Pascal entry.
3866 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3867 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3869 2018-09-25 Martin Liska <mliska@suse.cz>
3872 * doc/invoke.texi: Document all parameters and remove default
3875 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3878 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3879 contains HOST_WIDE_INTs when computing its size.
3881 2018-09-24 Jim Wilson <jimw@sifive.com>
3884 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3886 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3888 2018-09-24 Andrew Pinski <apinski@marvell.com>
3890 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3891 access prev before checking it for NULLness in the
3892 AARCH64_FUSE_CMP_BRANCH case.
3894 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3897 * config/i386/i386.c (rest_of_insert_endbranch): Set
3898 endbr_queued_at_entrance to true and don't insert ENDBR if
3899 x86_function_profiler will be called.
3900 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3902 * config/i386/i386.h (machine_function): Add
3903 endbr_queued_at_entrance.
3905 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3907 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3908 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3910 (check_attr_test): Use alternative_mask.
3911 (get_attr_value): Likewise.
3912 (compute_alternative_mask): Use alternative_mask and XWINT.
3913 (make_alternative_compare): Use alternative_mask.
3914 (attr_alt_subset_p): Use XWINT.
3915 (attr_alt_subset_of_compl_p): Likewise.
3916 (attr_alt_intersection): Use alternative_mask and XWINT.
3917 (attr_alt_union): Likewise.
3918 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3919 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3920 (simplify_test_exp): Use alternative_mask and XWINT.
3921 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3922 number calculation to support 64 bits. Generate code that
3923 checks 64-bit masks.
3924 (main): Use alternative_mask.
3925 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3927 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3930 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3931 RETURN+USE when returning via %r14.
3933 2018-09-24 Martin Liska <mliska@suse.cz>
3935 * gcov.c (output_lines): Print colorization legend
3936 for both flag_use_colors and flag_use_hotness_colors.
3939 2018-09-24 Martin Liska <mliska@suse.cz>
3941 * coverage.c (get_coverage_counts): Use warning_at
3942 with current_function_decl location. Use %qD in warning
3945 2018-09-24 Martin Liska <mliska@suse.cz>
3947 * memory-block.h (memory_block_pool::release): Annotate with
3948 valgrind that the memory is not accessible.
3950 2018-09-24 Martin Liska <mliska@suse.cz>
3953 * asan.c: Make asan_handled_variables extern.
3955 * cfgexpand.c (expand_stack_vars): Make sure
3956 a representative is unpoison if another
3957 variable in the partition is handled by
3958 use-after-scope sanitization.
3960 2018-09-24 Richard Biener <rguenther@suse.de>
3962 PR tree-optimization/63155
3963 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3964 the worklist when the edge of the respective argument isn't
3967 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3969 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3970 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3971 (MASK_CLASS_P): Update for rename.
3972 (MAYBE_MASK_CLASS_P): Ditto.
3973 (REG_CLASS_NAMES): Update.
3974 (REG_CLASS_CONTENT): Update.
3975 * config/i386/i386.c (regclass_map): Update for MASK_REG
3976 and ALL_MASK_REGS rename.
3977 * config/i386/constraints.md (Yk): Update for rename.
3980 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3982 * config/i386/i386.h (enum reg_class): Remove
3983 EVEX_SSE_REGS and MOD4_SSE_REGS.
3984 (REG_CLASS_NAMES): Update.
3985 (REG_CLASS_CONTENT): Update.
3986 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3987 registers as ALL_SSE_REGS.
3988 (ix86_additional_allocno_class_p): Remove.
3989 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3990 (ix86_register_priority): Lower priority of EVEX SSE registers.
3991 Use IN_RANGE macro where appropriate.
3992 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3993 AVX-5124VNNIW checks.
3994 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3995 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3996 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3997 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3998 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3999 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
4000 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
4001 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
4002 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
4003 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
4004 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
4005 * config/i386/constraints.md (Yh): Remove.
4007 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
4009 * config/i386/i386.c (regclass_map): Declare integer REX registers
4012 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
4014 * doc/service.texi (Service): Switch the fsf.org link to https.
4016 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
4019 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
4020 Define to speculation_safe_value_not_needed.
4022 2018-09-21 Florian Weimer <fweimer@redhat.com>
4025 * doc/extend.texi (Common Function Attributes): Mention that
4026 noreturn suppresses tail call optimization.
4028 2018-09-21 Jeff Law <law@redhat.com>
4030 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
4031 vr_values::cleanup_edges_and_switches.
4032 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
4034 (identify_jump_threads): Remove EDGE_IGNORE handling.
4035 (execute_vrp): Move handling of to_remove_edges and
4036 to_update_switch_stmts into vr_values class member functions.
4037 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
4038 (to_update_switch_stmts): Likewise.
4039 * vr-values.c: Include cfghooks.h.
4040 (vr_values::vr_values): Initialize to_remove_edges and
4041 to_update_switch_stmts.
4042 (vr_values::~vr_values): Verify to_remove_edges and
4043 to_update_switch_stmts are empty.
4044 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
4045 (vr_values::cleanup_edges_and_switches): New member function.
4046 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
4047 function. Add new data members.
4049 2018-09-21 David Malcolm <dmalcolm@redhat.com>
4051 PR tree-optimization/87309
4052 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
4053 calls with pflags and alt_flags.
4054 (selftest::test_capture_of_dump_calls): Add test of interaction of
4055 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
4057 2018-09-21 Olivier Hainque <hainque@adacore.com>
4059 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
4061 2018-09-21 Olivier Hainque <hainque@adacore.com>
4063 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
4065 2018-09-21 Olivier Hainque <hainque@adacore.com>
4067 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
4068 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
4069 (VXWORKS_LIBS_RTP): Minor reordering.
4071 2018-09-21 Olivier Hainque <hainque@adacore.com>
4073 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
4074 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
4076 2018-09-21 Olivier Hainque <hainque@adacore.com>
4078 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
4079 (PTRDIFF_TYPE): Likewise.
4081 2018-09-21 Olivier Hainque <hainque@adacore.com>
4083 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
4084 triplet, similar to support for VxWorks7.
4085 * config/vxworks-dummy.h: Provide a default definition
4086 of TARGET_VXWORKS64 to 0.
4088 2018-09-21 Olivier Hainque <hainque@adacore.com>
4090 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
4091 * config/vxworks-dummy.h: here.
4093 2018-09-21 Olivier Hainque <hainque@adacore.com>
4095 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
4097 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
4098 Bo Zhou <zbo.zhou@hisilicon.com>
4100 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
4101 * config/aarch64/aarch64-tune.md: Regenerated.
4102 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
4103 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
4104 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
4106 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
4107 Julian Brown <julian@codesourcery.com>
4109 * builtins.c (get_builtin_sync_mem): Handle address spaces.
4111 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4113 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
4114 if the call takes a static chain.
4116 2018-09-21 Martin Liska <mliska@suse.cz>
4118 * auto-profile.c (autofdo_source_profile::read): Do not
4120 (read_profile): Do not add working sets.
4121 (read_autofdo_file): Remove sum_all.
4122 (afdo_callsite_hot_enough_for_early_inline): Remove const
4124 * coverage.c (struct counts_entry): Remove gcov_summary.
4125 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
4126 do not support GCOV_TAG_PROGRAM_SUMMARY.
4127 (get_coverage_counts): Remove summary and expected
4129 * coverage.h (get_coverage_counts): Likewise.
4130 * doc/gcov-dump.texi: Remove -w option.
4131 * gcov-dump.c (dump_working_sets): Remove.
4132 (main): Do not support '-w' option.
4133 (print_usage): Likewise.
4134 (tag_summary): Likewise.
4135 * gcov-io.c (gcov_write_summary): Do not dump
4137 (gcov_read_summary): Likewise.
4138 (gcov_histo_index): Remove.
4139 (gcov_histogram_merge): Likewise.
4140 (compute_working_sets): Likewise.
4141 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
4143 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
4144 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
4145 (GCOV_HISTOGRAM_SIZE): Remove.
4146 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
4147 (struct gcov_summary): Simplify rapidly just
4148 to runs and sum_max fields.
4149 (gcov_histo_index): Remove.
4150 (NUM_GCOV_WORKING_SETS): Likewise.
4151 (compute_working_sets): Likewise.
4152 * gcov-tool.c (print_overlap_usage_message): Remove
4153 trailing empty line.
4154 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
4155 (output_lines): Remove program related line.
4156 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
4157 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
4159 (input_profile_summary): Do not read it.
4160 (merge_profile_summaries): And do not merge it.
4161 (input_symtab): Do not call removed function.
4162 * modulo-sched.c (sms_schedule): Do not print sum_max.
4163 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
4164 removed when histogram method was invented.
4165 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
4167 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
4169 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
4170 and dump selected value.
4171 * profile.c (add_working_set): Remove.
4172 (get_working_sets): Likewise.
4173 (find_working_set): Likewise.
4174 (get_exec_counts): Do not work with working sets.
4175 (read_profile_edge_counts): Do not inform as sum_max is removed.
4176 (compute_branch_probabilities): Likewise.
4177 (compute_value_histograms): Remove argument for call of
4178 get_coverage_counts.
4179 * profile.h: Do not make gcov_summary const.
4181 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
4183 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
4185 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4187 PR tree-optimization/86990
4188 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
4189 Check that the entire merged store group is made of constants only for
4192 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
4194 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
4195 (VTABLE_VERIFICATION_SPEC): Likewise.
4196 (SANITIZER_EARLY_SPEC): Likewise.
4197 (SANITIZER_SPEC): Likewise.
4198 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
4199 * doc/invoke.texi (Link Options): Document -r.
4201 2018-09-20 Richard Biener <rguenther@suse.de>
4204 * gimplify.c (gimplify_expr): Retain alignment of
4205 addressable lvalue in dereference.
4207 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
4210 * configure.ac: Check for .loc is_stmt support.
4211 * configure, config.in: Rebuilt.
4212 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
4215 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4217 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
4218 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
4221 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4223 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
4225 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
4228 2018-09-20 Marek Polacek <polacek@redhat.com>
4230 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
4232 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4234 PR tree-optimization/87288
4235 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
4236 into account when determining PEELING_FOR_NITERS.
4238 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4240 PR tree-optimization/86877
4241 * tree-vect-loop.c (vect_analyze_loop_2): Call
4242 vect_verify_datarefs_alignment.
4244 2018-09-19 Marek Polacek <polacek@redhat.com>
4246 * doc/invoke.texi: Document -Wclass-conversion.
4248 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
4250 * config/pa/pa.c (pa_adjust_priority): Delete.
4251 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
4253 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
4254 (atomic_storehi): Likewise.
4255 (atomic_storesi): Likewise.
4256 (atomic_loaddi): Restore compare and swap exchange loop code.
4258 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4260 PR rtl-optimization/86902
4261 * combine.c (try_combine): When changing the CC mode used, don't change
4262 an unrelated mode in other_insn to that new CC mode.
4264 2018-09-19 David Malcolm <dmalcolm@redhat.com>
4266 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
4267 with %T in place of calls to dump_generic_expr.
4268 (prune_runtime_alias_test_list): Likewise.
4269 (create_runtime_alias_checks): Likewise.
4270 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
4271 (vect_analyze_data_ref_dependence): Likewise.
4272 (vect_slp_analyze_data_ref_dependence): Likewise.
4273 (vect_record_base_alignment): Likewise. Use %G in place of call
4274 to dump_gimple_stmt.
4275 (vect_compute_data_ref_alignment): Likewise.
4276 (verify_data_ref_alignment): Likewise.
4277 (vect_find_same_alignment_drs): Likewise.
4278 (vect_analyze_group_access_1): Likewise.
4279 (vect_analyze_data_ref_accesses): Likewise.
4280 (dependence_distance_ge_vf): Likewise.
4281 (dump_lower_bound): Likewise.
4282 (vect_prune_runtime_alias_test_list): Likewise.
4283 (vect_find_stmt_data_reference): Likewise.
4284 (vect_analyze_data_refs): Likewise.
4285 (vect_create_addr_base_for_vector_ref): Likewise.
4286 (vect_create_data_ref_ptr): Likewise.
4287 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
4288 (vect_can_advance_ivs_p): Likewise.
4289 (vect_update_ivs_after_vectorizer): Likewise.
4290 (vect_gen_prolog_loop_niters): Likewise.
4291 (vect_prepare_for_masked_peels): Likewise.
4292 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4293 (vect_determine_vectorization_factor): Likewise.
4294 (vect_is_simple_iv_evolution): Likewise.
4295 (vect_analyze_scalar_cycles_1): Likewise.
4296 (vect_analyze_loop_operations): Likewise.
4297 (report_vect_op): Likewise.
4298 (vect_is_slp_reduction): Likewise.
4299 (check_reduction_path): Likewise.
4300 (vect_is_simple_reduction): Likewise.
4301 (vect_create_epilog_for_reduction): Likewise.
4302 (vect_finalize_reduction:): Likewise.
4303 (vectorizable_induction): Likewise.
4304 (vect_transform_loop_stmt): Likewise.
4305 (vect_transform_loop): Likewise.
4306 (optimize_mask_stores): Likewise.
4307 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
4308 (vect_split_statement): Likewise.
4309 (vect_recog_over_widening_pattern): Likewise.
4310 (vect_recog_average_pattern): Likewise.
4311 (vect_determine_min_output_precision_1): Likewise.
4312 (vect_determine_precisions_from_range): Likewise.
4313 (vect_determine_precisions_from_users): Likewise.
4314 (vect_mark_pattern_stmts): Likewise.
4315 (vect_pattern_recog_1): Likewise.
4316 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4317 (vect_record_max_nunits): Likewise.
4318 (vect_build_slp_tree_1): Likewise.
4319 (vect_build_slp_tree_2): Likewise.
4320 (vect_print_slp_tree): Likewise.
4321 (vect_analyze_slp_instance): Likewise.
4322 (vect_detect_hybrid_slp_stmts): Likewise.
4323 (vect_detect_hybrid_slp_1): Likewise.
4324 (vect_slp_analyze_operations): Likewise.
4325 (vect_slp_analyze_bb_1): Likewise.
4326 (vect_transform_slp_perm_load): Likewise.
4327 (vect_schedule_slp_instance): Likewise.
4328 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4329 (vect_mark_stmts_to_be_vectorized): Likewise.
4330 (vect_init_vector_1): Likewise.
4331 (vect_get_vec_def_for_operand): Likewise.
4332 (vect_finish_stmt_generation_1): Likewise.
4333 (vect_check_load_store_mask): Likewise.
4334 (vectorizable_call): Likewise.
4335 (vectorizable_conversion): Likewise.
4336 (vectorizable_operation): Likewise.
4337 (vectorizable_load): Likewise.
4338 (vect_analyze_stmt): Likewise.
4339 (vect_is_simple_use): Likewise.
4340 (vect_get_vector_types_for_stmt): Likewise.
4341 (vect_get_mask_type_for_stmt): Likewise.
4342 * tree-vectorizer.c (increase_alignment): Likewise.
4344 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
4346 * doc/rtl.texi: Adjust vec_select description.
4347 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
4348 non-constant selectors.
4350 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
4352 * config/aarch64/aarch64-protos.h
4353 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
4354 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
4355 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
4356 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
4357 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
4358 (AARCH64_FL_PROFILE): Move index so flags are ordered.
4359 (AARCH64_ISA_RCPC8_4): New flag.
4360 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
4361 to aarch64_offset_9bit_signed_unscaled_p.
4362 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
4364 * config/aarch64/constraints.md (Ust): New constraint.
4365 * config/aarch64/predicates.md.
4366 (aarch64_9bit_offset_memory_operand): New predicate.
4367 (aarch64_rcpc_memory_operand): New predicate.
4369 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
4371 PR rtl-optimization/87361
4372 * rtlanal.c (nonzero_bits1): Revert accidental change.
4374 2018-09-19 Richard Biener <rguenther@suse.de>
4376 PR tree-optimization/87349
4377 PR tree-optimization/87342
4378 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
4380 2018-09-18 Marek Polacek <polacek@redhat.com>
4382 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
4383 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
4385 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4387 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
4389 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4391 PR rtl-optimization/86882
4392 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
4394 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
4396 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
4397 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
4399 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
4402 * doc/invoke.texi (Link Options): Fix formatting and grammar.
4404 2018-09-18 Richard Biener <rguenther@suse.de>
4407 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
4408 (compute_samebase_partition_bases): Likewise.
4409 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
4410 (gimple_can_coalesce_p): Simplify.
4412 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
4414 Handle a library implementation of ffs calling __builtin_ffs.
4415 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
4416 (mmix_init_libfuncs): New function: make __builtin_ffs expand
4419 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4421 * diagnostic-show-locus.c (class layout_range): Add field
4423 (layout_range::layout_range): Add "original_idx" param and use it
4424 to initialize new field.
4425 (make_range): Use 0 for original_idx.
4426 (layout::layout): Pass in index to calls to
4427 maybe_add_location_range.
4428 (layout::maybe_add_location_range): Add param "original_idx" and
4429 pass it on to layout_range.
4430 (layout::print_any_labels): Pass on range->m_original_idx to
4432 (gcc_rich_location::add_location_if_nearby): Use 0 for
4434 * gcc-rich-location.h (text_range_label::get_text): Update for new
4436 (range_label_for_type_mismatch::get_text): Likewise.
4438 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4440 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
4442 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4444 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
4445 format_string_diagnostic_t.
4446 (fmtwarn_n): Likewise.
4447 * substring-locations.c
4448 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
4449 (format_warning_n_va): Convert to...
4450 (format_string_diagnostic_t::emit_warning_n_va): ...this.
4451 (format_warning_va): Convert to...
4452 (format_string_diagnostic_t::emit_warning_va): ...this.
4453 (format_warning_at_substring): Convert to...
4454 (format_string_diagnostic_t::emit_warning): ...this.
4455 (format_warning_at_substring_n): Convert to...
4456 (format_string_diagnostic_t::emit_warning_n): ...this.
4457 * substring-locations.h (class format_string_diagnostic_t): New
4459 (format_warning_va): Convert to
4460 format_string_diagnostic_t::emit_warning_va.
4461 (format_warning_n_va): Convert to
4462 format_string_diagnostic_t::emit_warning_n_va.
4463 (format_warning_at_substring): Convert to
4464 format_string_diagnostic_t::emit_warning.
4465 (format_warning_at_substring_n): Convert to
4466 format_string_diagnostic_t::emit_warning_n.
4468 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
4469 Bernd Schmidt <bernds_cb1@t-online.de>
4471 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
4474 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4476 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
4477 operand 0 predicate to nonimmediate operand.
4478 (rint<mode>2_frndint): Remove insn pattern.
4479 (rint<mode>2): Change operand 1 predicate to general_operand.
4480 Extend operand 1 to XFmode and generate rintxf2 insn.
4481 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
4482 Do not use X87MODEF mode macro.
4483 (frndintxf2_<rounding>_i387): Rename from
4484 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
4485 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
4486 to XFmode and generate significandxf3 insn.
4488 2018-09-17 Richard Biener <rguenther@suse.de>
4490 PR tree-optimization/87328
4491 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
4492 visiting unexecutable backedges when not iterating.
4493 (do_rpo_vn): Mark all edges not executable even when not
4496 2018-09-17 Martin Jambor <mjambor@suse.cz>
4499 * doc/invoke.texi (Warning Options): Likewise.
4501 2018-09-17 Richard Biener <rguenther@suse.de>
4503 PR tree-optimization/87301
4504 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
4505 clean EH info from leftover copy assignments.
4507 2018-09-17 Martin Liska <mliska@suse.cz>
4509 PR gcov-profile/85871
4510 * gcov.c (output_intermediate_file): Fix out of bounds
4513 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
4515 * config/arc/arc.c: Object attributes for core4 not reflected
4517 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
4520 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
4522 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
4524 2018-09-17 Martin Liska <mliska@suse.cz>
4526 * doc/gcov.texi: Document new option --use-hotness-colors.
4527 * gcov.c (struct source_info): Declare new field.
4528 (source_info::source_info): Set default for maximum_count.
4529 (print_usage): Add new -q option.
4530 (process_args): Process it.
4531 (accumulate_line_info): Save src->maximum_count.
4532 (output_line_beginning): Make color line number if
4533 flag_use_hotness_colors is set.
4534 (output_line_details): Pass default argument value.
4535 (output_lines): Pass src->maximum_count.
4537 2018-09-17 Martin Liska <mliska@suse.cz>
4539 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
4540 Use processor_names table.
4541 * config/i386/i386.c (ix86_default_align): Use
4542 processor_cost_table for alignment values.
4543 (ix86_option_override_internal): Use processor_names.
4544 (ix86_function_specific_print): Likewise.
4545 * config/i386/i386.h (struct processor_costs):
4546 Add alignment values.
4547 (struct ptt): Remove and replace with const char *.
4548 * config/i386/x86-tune-costs.h (struct processor_costs):
4549 Declare default alignments for all costs.
4551 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4553 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4554 symbolics or VR_VARYING ranges for ABS_EXPR.
4555 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4556 when range will wrap.
4558 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4561 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4562 before and after a JUMP_TABLE_DATA.
4564 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4567 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4568 when one operand is a function pointer. Use POINTER_TYPE_P and
4569 FUNC_OR_METHOD_TYPE_P.
4570 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4571 * fold-const.c (build_range_check): Likewise.
4572 * match.pd (simple_comparison): Likewise.
4574 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4577 * spellcheck.c (get_edit_distance_cutoff): New function.
4578 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4579 (selftest::test_get_edit_distance_one_way): ...this.
4580 (selftest::test_get_edit_distance_unit): Rename to...
4581 (selftest::test_get_edit_distance_both_ways): ...this.
4582 (selftest::test_edit_distances): Move tests to this new function,
4583 and test some more pairs of strings. Update for above renaming.
4584 (selftest::get_old_cutoff): New function.
4585 (selftest::test_get_edit_distance_cutoff): New function.
4586 (selftest::assert_suggested_for): New function.
4587 (ASSERT_SUGGESTED_FOR): New macro.
4588 (selftest::assert_not_suggested_for): New function.
4589 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4590 (selftest::test_suggestions): New function.
4591 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4592 tests to selftest::test_edit_distances and call it. Add calls to
4593 selftest::test_get_edit_distance_cutoff and
4594 selftest::test_suggestions.
4595 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4596 (best_match::consider): Replace hard-coded cutoff calculation with
4598 (best_match::get_cutoff): New declaration.
4599 (best_match::get_best_meaningful_candidate): Likewise.
4601 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4603 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4605 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4608 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4610 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4611 terminated string literal.
4613 2018-09-14 Martin Sebor <msebor@redhat.com>
4615 * builtins.c (unterminated_array): Handle ARRAY_REF.
4616 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4617 * builtins.h (unterminated_array): Declare extern.
4618 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4620 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4623 2018-09-14 Martin Sebor <msebor@redhat.com>
4624 Jeff Law <law@redhat.com>
4626 * builtins.c (unterminated_array): New.
4627 (expand_builtin_strcpy): Adjust.
4628 (expand_builtin_strcpy_args): Detect unterminated arrays.
4629 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4630 unterminated arrays.
4631 * gimple-fold.h (get_maxval_strlen): Add argument.
4632 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4634 * gimple-fold.c (get_range_strlen): Add argument.
4635 (get_maxval_strlen): Adjust.
4636 * gimple-fold.h (get_range_strlen): Add argument.
4638 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4640 * config/i386/movdirintrin.h: Fix copyright year.
4642 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4644 * reg-stack.c: Include regs.h.
4645 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4646 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4647 FIRST_STACK_REG, not DFmode.
4648 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4649 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4650 (change stack): Default register mode to the reg_raw_mode of
4651 FIRST_STACK_REG, not DFmode.
4652 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4653 (*swapxf): Rename from swapxf.
4655 2018-09-14 Carl Love <cel@us.ibm.com>
4657 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4658 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4660 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4663 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4666 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4669 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4671 2018-09-14 Jason Merrill <jason@redhat.com>
4673 Fix --enable-gather-detailed-mem-stats.
4674 * hash-table.c (hash_table_usage): Change from variable to function.
4675 * hash-table.h: Adjust.
4676 * Makefile.in: Add missing dependencies on hash-table.h.
4678 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4680 PR tree-optimization/87259
4682 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4683 execute_cse_reciprocals_1 has tried transforming.
4685 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4687 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4688 VR_VARYING for PLUS/MINUS_EXPR.
4690 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4692 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4695 2018-09-14 Richard Biener <rguenther@suse.de>
4698 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4699 bits for the merged partition.
4701 2018-09-13 Martin Sebor <msebor@redhat.com>
4702 Bernd Edlinger <bernd.edlinger@hotmail.de>
4704 * builtins.h (c_srlen): Add argument.
4705 * builtins.c (warn_string_no_nul): New function.
4706 (c_strlen): Add argument and use it. Update recursive calls.
4707 Pass DECL argument to string_constant to get info on non
4708 terminated strings. Update *NONSTR as needed.
4709 (fold_builtin_strlen): Add argument to calls to c_strlen.
4710 Warn for unterminated arrays.
4711 (warn_string_no_null): Add prototype.
4712 * expr.c (string_constant): Update arguments. Update recursive
4713 calls appropriately. Detect missing NUL terminator and outermost
4714 declaration its missing in.
4715 Improve checks for arrays with nonzero lower bound or elements
4716 that are not a single byte. Simplify offset computation.
4717 Simplify checks for non-NUL terminated strings.
4718 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4719 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4721 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4723 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4725 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4726 argument. Fix range checks.
4727 * fold-const.h (c_getstr): Adjust protoype.
4728 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4729 string is constant but contains no NUL byte.
4731 * expr.c (string_constant): Adjust function comment.
4732 Remove bogus check for zero termination.
4734 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4736 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4737 (get_constant_size): Don't make STRING_CSTs larger than they are.
4738 (check_string_literal): New check function for STRING_CSTs.
4739 (output_constant): Use it.
4741 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4744 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4746 2018-09-13 Richard Biener <rguenther@suse.de>
4748 PR tree-optimization/87263
4749 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4750 (struct unwind_state): Add max_rpo field.
4751 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4752 Compute max_rpo, the max RPO number a block can be backwards reached
4753 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4754 separating it from the iterating mode.
4756 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4758 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4759 (rfs_decision): New scheduling decision.
4761 2018-09-13 Richard Biener <rguenther@suse.de>
4764 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4765 (vn_nary_op_insert_pieces_predicated): Do not write useless
4766 valid_dominated_by_p entry outside of the allocated storage.
4768 2018-09-13 Omar Sandoval <osandov@osandov.com>
4769 Tom de Vries <tdevries@suse.de>
4772 * dwarf2out.c (is_c): New function.
4773 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4775 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4778 * config/aarch64/aarch64.md (*aarch64_bfxil):
4780 * config/aarch64/constraints.md (Ulc): Define.
4781 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4783 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4786 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4788 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4789 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4790 aarch64_layout_frame call.
4791 (aarch64_expand_epilogue): Likewise.
4792 (aarch64_initial_elimination_offset): Likewise.
4793 (aarch64_get_separate_components): Likewise.
4794 (aarch64_use_return_insn_p): Likewise.
4795 (aarch64_layout_frame): Remove unneeded check.
4797 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4799 * configure.ac: Only append
4800 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4801 gcc_config_arguments if it was never reconfigured or last reconfigure
4802 was with different arguments.
4803 * configure: Regenerated.
4805 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4806 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4809 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4810 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4812 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4814 PR tree-optimization/87287
4815 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4816 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4817 * match.pd (X % C == 0): ... here. New optimization.
4819 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4822 * expr.h (maybe_optimize_mod_cmp): Declare.
4823 * expr.c (mod_inv): New function.
4824 (maybe_optimize_mod_cmp): New function.
4825 (do_store_flag): Use it.
4826 * cfgexpand.c (expand_gimple_cond): Likewise.
4828 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4829 Julian Brown <julian@codesourcery.com>
4832 * gimplify.c (gimplify_scan_omp_clauses): Set
4833 target_firstprivatize_array_bases in OpenACC parallel and kernels
4834 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4835 OpenACC data regions.
4837 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4839 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4840 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4842 2018-09-12 Richard Biener <rguenther@suse.de>
4844 PR tree-optimization/87280
4845 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4846 edge but unreachable target.
4847 (do_rpo_vn): For conservatively handling a PHI only mark
4848 the backedge executable but not the block reachable.
4850 2018-09-12 Richard Biener <rguenther@suse.de>
4852 PR tree-optimization/87266
4853 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4856 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4858 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4860 ("trunc<BFP:mode><DFP_ALL:mode>2")
4861 ("trunc<DFP_ALL:mode><BFP:mode>2")
4862 ("extend<BFP:mode><DFP_ALL:mode>2")
4863 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4864 according to the target operand type.
4866 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4867 Andreas Krebbel <krebbel@linux.ibm.com>
4869 PR tree-optimization/86844
4870 * gimple-ssa-store-merging.c
4871 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4872 there are any overlapping stores in between them, make sure they are
4873 also coalesced or we give up completely.
4875 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4878 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4879 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4881 2018-09-12 Tom de Vries <tdevries@suse.de>
4883 * common.opt (gdescribe-dies): Add option.
4884 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4885 attribute for artifical and nameless decls.
4886 (dwarf2out_register_external_die): Add description attribute to
4887 external reference die.
4888 (add_desc_attribute): New functions.
4889 (gen_subprogram_die): Add description attribute to
4890 DW_TAG_call_site_parameter.
4891 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4892 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4893 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4895 (@item -gdescribe-dies): Add.
4897 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4899 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4900 (extract_range_from_binary_expr_1: Call
4901 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4902 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4905 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4907 * gimple-ssa-warn-alloca.c
4908 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4909 field for ALLOCA_BOUND_*_LARGE.
4911 2018-09-11 Nathan Sidwell <nathan@acm.org>
4913 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4915 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4917 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4918 for clobbers. Remove obsolete comment.
4920 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4922 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4923 mpxchk, mpxld and mpxst types.
4924 (define_attr length_immediate): Remove all processing of mpx types.
4925 (define_attr prefix_0f): Ditto.
4926 (define_attr memory): Ditto.
4928 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4930 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4931 (log<mode>2): Change operand 1 predicate to general_operand.
4932 Extend operand 1 to XFmode and generate logxf3 insn.
4933 (log10<mode>2): Change operand 1 predicate to general_operand.
4934 Extend operand 1 to XFmode and generate log10xf3 insn.
4935 (log2<mode>2): Change operand 1 predicate to general_operand.
4936 Extend operand 1 to XFmode and generate log2xf3 insn.
4937 (fyl2xp1_extend<mode>xf3_i387): Remove.
4938 (log1p<mode>2): Change operand 1 predicate to general_operand.
4939 Extend operand 1 to XFmode and generate log1pxf3 insn.
4940 (fxtract_extend<mode>xf3_i387): Remove.
4941 (logb<mode>2): Change operand 1 predicate to general_operand.
4942 Extend operand 1 to XFmode and generate logbxf3 insn.
4943 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4944 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4945 (significand<mode>2): Change operand 1 predicate to general_operand.
4946 Extend operand 1 to XFmode and generate significandxf3 insn.
4948 2018-09-11 Nathan Sidwell <nathan@acm.org>
4950 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4951 (load_specs): Use fatal_error.
4952 (DELETE_IF_ORDINARY, process_command): Use error.
4953 (execute, run_attempt): Use fatal_error.
4955 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4957 * diagnostic-core.h (sorry_at): New prototype.
4958 * diagnostic.c (sorry_at): New function.
4960 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4962 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4963 by zero as VR_UNDEFINED.
4965 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4967 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4968 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4969 (<sincos>mode2): New expander.
4970 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4971 (sincos -> sin, cos splitters): Remove splitter patterns.
4972 (sincos<mode>3): Change operand 2 predicate to general_operand.
4973 Extend operand 2 to XFmode and generate sincosxf3 insn.
4974 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4975 Change operand 3 predicate to const1_operand.
4976 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4977 (tanxf2): Update operands in the call to fptanxf4_i387.
4978 (tan<mode>2): Change operand 1 predicate to general_operand.
4979 Extend operand 1 to XFmode and generate tanxf3 insn.
4980 (atan2xf3): Rename from *fpatanxf3_i387.
4981 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4982 (atan2xf3): Remove expander.
4983 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4984 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4985 (atan<mode>2): Change operand 1 predicate to general_operand.
4986 Extend operand 1 to XFmode and generate atanxf3 insn.
4988 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4990 * config/i386/i386.md (x87/SSE constant load splitter): Use
4991 memory_operand instead of nonimmediate_operand for input operand
4994 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4996 * config/i386/i386.md (float partial SSE register stall splitter): Move
4997 splitter near its instruction pattern.
4998 (float_extend partial SSE register stall splitter): Ditto.
4999 (float_truncate partial SSE register stall splitter): Ditto.
5001 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
5004 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
5005 to speculation_safe_value_not_needed.
5008 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
5009 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
5010 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
5011 leaf_function_p, instead use has_hard_reg_initial_val.
5013 2018-09-09 Nathan Sidwell <nathan@acm.org>
5015 * gcc.h (pfatal_with_name): Don't declare here.
5016 * gcc.c (pfatal_with_name): Make static.
5018 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
5020 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
5023 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
5025 PR rtl-optimization/85458
5026 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
5027 priority hook to reduce the priority of EXPR.
5029 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
5031 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
5032 DImode for x87 on 32bit targets. Conditionally disable x87 modes
5033 with X87_ENABLE_FLOAT. Remove preparation code.
5034 (*float<SWI48:mode><MODEF:mode>2): Rename from
5035 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
5036 math using "enabled" attribute.
5037 (*floatdi<MODEF:mode>2_i387): Rename from
5038 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
5039 enable for 32bit targets only.
5040 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
5042 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
5043 as operand 1 predicate. Rewrite as define_insn_and_split.
5044 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
5046 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
5048 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
5049 to fallthru to FLOAT case.
5051 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5054 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
5055 around folding of vec_sl to handle out of range shift values.
5057 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
5059 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
5060 Update callers to gen_fix_trunc<mode>_i387_fisttp
5061 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
5062 nonimmediate_operand.
5063 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
5064 and corresponding splitters.
5065 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
5066 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
5067 (fix_truncdi_i387_with_temp): Remove insn pattern
5068 and corresponding splitters.
5069 (fix_trunc<mode>_i387): Change operand 0 predicate to
5070 nonimmediate_operand.
5071 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
5072 and corresponding splitters.
5073 (*fistdi2_1): Remove.
5075 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
5076 (lrintxfdi2): Remove expander. Reimplement as define_insn.
5077 (*fist<mode>2_1): Remove.
5078 (fist<mode>2): Ditto.
5079 (fist<mode>2_with_temp): Remove insn pattern and corresponding
5081 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
5082 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
5083 (fistdi2_<rounding>): Change operand 0 predicate to
5084 nonimmediate_operand.
5085 (fistdi2_<rounding>_with_temp): Remove insn pattern
5086 and corresponding splitters.
5087 (fist<mode>2_<rounding>): Change operand 0 predicate to
5088 nonimmediate_operand.
5089 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
5090 and corresponding splitters.
5092 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
5094 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
5096 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
5099 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5101 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5102 early gimple folding of vec_splat().
5103 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
5104 * gimple-fold.h: Add an extern define for tree_vec_extract().
5106 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
5108 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
5109 wrappers around TREE_TYPE comparisons.
5111 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5114 * config/s390/predicates.md: Add nonsym_memory_operand.
5115 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
5116 contains a SYMBOL_REF, load it into an intermediate pseudo.
5117 (s390_emit_compare_and_swap): Legitimize operand.
5118 * config/s390/s390.md: Use the new nonsym_memory_operand
5119 with UNSPECV_CAS patterns.
5121 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5124 * config/s390/s390-passes.def: New file.
5125 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
5127 (make_pass_s390_early_mach): Add declaration.
5128 * config/s390/s390.c (make_pass_s390_early_mach):
5129 (s390_option_override): Remove dynamic registration.
5130 * config/s390/t-s390: Add s390-passes.def.
5132 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5134 * config/s390/s390.c (s390_decompose_constant_pool_ref):
5135 Remove UNSPEC_LTREL_BASE check.
5136 (annotate_constant_pool_refs): Likewise.
5137 (find_constant_pool_ref): Likewise.
5138 (find_ltrel_base): Removed.
5139 (replace_ltrel_base): Removed.
5140 (s390_mainpool_finish): Remove replace_ltrel_base call.
5141 (s390_chunkify_start): Remove pending LTREL_BASE logic.
5142 (s390_chunkify_finish): Remove replace_ltrel_base call.
5143 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
5145 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
5148 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
5149 to speculation_safe_value_not_needed.
5151 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
5152 Bernd Schmidt <bernds_cb1@t-online.de>
5154 * config/nvptx/nvptx-opts.h: New file.
5155 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
5156 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
5158 (TARGET_SM35): New macro.
5159 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
5161 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
5163 (misa=): New option.
5164 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
5166 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
5168 * config/i386/i386.md (truncdfsf2): Remove expander.
5169 (truncdfsf2_with_temp): Ditto.
5170 (truncxf<mode>2): Ditto.
5171 (*truncdfsf_fast_mixed): Remove insn pattern.
5172 (*truncdfsf_fast_i387): Ditto.
5173 (*truncdfsf_mixed): Ditto.
5174 (*truncdfsf_i387): Ditto.
5175 (*truncdfsf2_i387_1): Ditto.
5176 (*truncxfsf2_mixed): Ditto.
5177 (*truncxfdf2_mixed): Ditto.
5178 (*truncxf<mode>2_i387_noop): Ditto. Update callers
5179 to call gen_truncxf<mode>2 instead.
5180 (*truncxf<mode>2_i387): Remove.
5181 (reg->reg splitters): Remove splitter pattern.
5182 (reg->mem splitters): Ditto.
5184 (truncdfsf2): New insn pattern.
5185 (truncxf<mode>2): Ditto.
5187 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5189 * tree-ssa-math-opts.c (is_mult_by): New function.
5190 (is_square_of): Use the above.
5191 (optimize_recip_sqrt): New function.
5192 (pass_cse_reciprocals::execute): Use the above.
5194 2018-09-05 Richard Biener <rguenther@suse.de>
5197 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
5198 to zero-init the emplaced vec.
5200 2018-09-05 Martin Liska <mliska@suse.cz>
5202 PR tree-optimization/87205
5203 * tree-switch-conversion.c (pass_lower_switch::execute):
5204 Group cases for switch statements.
5206 2018-09-05 Richard Biener <rguenther@suse.de>
5208 PR tree-optimization/87217
5209 * tree-ssa-sccvn.c (vuse_valueize): New.
5210 (vn_reference_lookup_pieces): Use it.
5211 (vn_reference_lookup): Likewise.
5213 2018-09-05 Nathan Sidwell <nathan@acm.org>
5216 * stor-layout.c (place_field): Scan forwards to check last
5217 bitfield when ms_bitfield_placement is in effect.
5219 2018-09-05 Richard Biener <rguenther@suse.de>
5222 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
5225 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
5226 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
5228 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
5229 * config.gcc (extra_objs): Build it.
5230 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
5232 * config/aarch64/aarch64-passes.def
5233 (pass_tag_collision_avoidance): New pass.
5234 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
5235 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
5236 (aarch64_classify_address): Remove static qualifier.
5237 (aarch64_address_info, aarch64_address_type): Move to...
5238 * config/aarch64/aarch64-protos.h: ... here.
5239 (make_pass_tag_collision_avoidance): New function.
5240 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
5243 2018-09-05 Martin Liska <mliska@suse.cz>
5245 * doc/gcov.texi: Update documentation of humar
5247 * gcov.c (format_count): Print one decimal place, it provides
5248 more fine number of situations like '1G' vs. '1.4G'.
5250 2018-09-05 Martin Liska <mliska@suse.cz>
5253 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
5254 * optc-gen.awk: Allow 'Var' for Deprecated options in order
5255 to generate a MASK value.
5257 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
5260 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
5261 if frame pointer isn't used.
5262 (compute_frame_pointer_to_fb_displacement): Likewise.
5264 2018-09-04 Jakub Jelinek <jakub@redhat.com>
5267 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
5268 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
5269 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
5270 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
5271 and OPTION_MASK_ISA_XSAVEC_UNSET.
5273 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
5275 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
5276 XOR operations in NAND case.
5278 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5280 * wide-int-range.cc (wide_int_range_convert): New.
5281 * wide-int-range.h (wide_int_range_convert): New.
5282 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
5283 code into wide_int_range_convert.
5284 (extract_range_into_wide_ints): Do not munge anti range constants
5285 into the entire domain. Just return the range back.
5287 2018-09-04 Martin Liska <mliska@suse.cz>
5289 * genmatch.c (output_line_directive): Add new argument
5291 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
5293 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5295 * doc/invoke.texi (Option Summary): Add whitespace.
5297 * doc/invoke.texi (Option Summary): Add -Waligned-new.
5299 2018-09-04 Richard Biener <rguenther@suse.de>
5301 PR tree-optimization/87211
5302 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
5303 backedge value we're supposed to treat as VARYING also number
5304 the PHI to VARYING in case it got a different value-number already.
5306 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5308 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
5309 (extract_range_from_binary_expr_1): Do not call
5310 vrp_can_optimize_bit_op.
5311 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
5313 (wide_int_range_get_mask_and_bounds): New.
5314 (wide_int_range_optimize_bit_op): New.
5315 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
5316 (wide_int_range_bit_and): Same.
5317 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
5318 (wide_int_range_optimize_bit_op): New.
5319 (wide_int_range_get_mask_and_bounds): New.
5321 2018-09-04 Richard Biener <rguenther@suse.de>
5323 PR tree-optimization/87176
5324 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
5325 variable. When value-numbering a virtual PHI node make sure
5326 to not value-number to the backedge value.
5328 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5330 * doc/extend.texi (Long Long, Hex Floats): Document support for
5331 long long and hex floats in more recent versions of ISO C++.
5333 2018-09-03 Richard Biener <rguenther@suse.de>
5335 PR tree-optimization/87177
5336 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
5339 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5341 * bb-reorder.c (edge_order): Convert to C-qsort-style
5342 tri-state comparator.
5343 (reorder_basic_blocks_simple): Change std::stable_sort to
5346 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5348 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
5349 tri-state comparator.
5350 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
5352 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5354 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
5355 (mergesort): ... here as maximum count for using netsort.
5356 (gcc_qsort): Set nlim to 3 if stable sort is requested.
5357 (gcc_stablesort): New.
5358 * system.h (gcc_stablesort): Declare.
5360 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5362 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
5363 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
5364 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
5366 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
5368 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
5369 lxsdx and stxsdx alternatives.
5370 (*mov<mode>_hardfloat64): Ditto.
5371 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
5373 2018-09-03 Richard Biener <rguenther@suse.de>
5375 PR tree-optimization/87200
5376 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
5379 2018-09-03 Martin Liska <mliska@suse.cz>
5381 PR tree-optimization/87201
5382 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5383 Fix parenthesis in an expression.
5385 2018-09-03 Richard Biener <rguenther@suse.de>
5387 PR tree-optimization/87197
5388 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
5389 visited. CSE the VN_INFO hashtable lookup.
5391 PR tree-optimization/87169
5392 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
5393 iterating make sure there's no extra backedges from irreducible
5394 regions feeding the header. Mark the destination block
5397 2018-09-03 Martin Liska <mliska@suse.cz>
5400 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
5401 * common/common-targhooks.c (default_get_valid_option_values):
5403 * common/common-targhooks.h (default_get_valid_option_values):
5405 * common/config/i386/i386-common.c: Move processor_target_table
5407 (ix86_get_valid_option_values): New function.
5408 (TARGET_GET_VALID_OPTION_VALUES): New macro.
5409 * config/i386/i386.c (struct ptt): Move to i386-common.c.
5410 (PTA_*): Move all defined masks into i386-common.c.
5411 (ix86_function_specific_restore): Use new processor_cost_table.
5412 * config/i386/i386.h (struct ptt): Moved from i386.c.
5413 (struct pta): Likewise.
5414 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
5415 * doc/tm.texi.in: Likewise.
5416 * opt-suggestions.c (option_proposer::suggest_option):
5417 Pass prefix to build_option_suggestions.
5418 (option_proposer::get_completions): Likewise.
5419 (option_proposer::build_option_suggestions): Use the new target
5421 * opts.c (struct option_help_tuple): New struct.
5422 (print_filtered_help): Use the new target hook.
5424 2018-09-03 Martin Liska <mliska@suse.cz>
5427 * predict.c (set_even_probabilities): Add likely_edges
5428 argument and handle cases where we have precisely one
5430 (combine_predictions_for_bb): Catch also likely_edges.
5431 (tree_predict_by_opcode): Handle gswitch statements.
5432 * tree-cfg.h (find_case_label_for_value): New declaration.
5433 (find_taken_edge_switch_expr): Likewise.
5434 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5435 Find pivot in decision tree based on probabily, not by number of
5438 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
5440 * doc/standards.texi (Standards): Update Objective-C reference.
5442 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5444 * doc/install.texi (Prerequisites): Update link for MPC.
5446 2018-09-01 Michael Matz <matz@suse.de>
5448 PR tree-optimization/87074
5449 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
5450 PHIs for outer-loop uses.
5452 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5454 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
5455 * doc/invoke.texi (C Dialect Options): Ditto.
5457 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5459 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
5461 2018-08-31 Richard Biener <rguenther@suse.de>
5463 PR tree-optimization/87168
5464 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
5465 (rpo_elim::eliminate_avail): When OP was not visited it must
5468 2018-08-31 David Malcolm <dmalcolm@redhat.com>
5470 * tree-vrp.c (copy_value_range): Convert param "from" from
5471 "value_range *" to "const value_range *".
5472 (range_is_null): Likewise for param "vr".
5473 (range_int_cst_p): Likewise.
5474 (range_int_cst_singleton_p): Likewise.
5475 (symbolic_range_p): Likewise.
5476 (value_ranges_intersect_p): Likewise for both params.
5477 (value_range_nonnegative_p): Likewise for param "vr".
5478 (value_range_constant_singleton): Likewise.
5479 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
5480 (extract_range_into_wide_ints): Likewise for param "vr".
5481 (extract_range_from_multiplicative_op): Likewise for params "vr0"
5483 (vrp_can_optimize_bit_op): Likewise.
5484 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
5486 (extract_range_from_unary_expr): Likewise.
5487 (debug_value_range): Likewise for param "vr".
5488 (value_range::dump): Add "const" qualifier.
5489 (vrp_prop::check_array_ref): Convert local "vr" from
5490 "value_range *" to "const value_range *".
5491 (vrp_prop::check_mem_ref): Likewise.
5492 (vrp_prop::visit_stmt): Likewise for local "old_vr".
5493 (vrp_intersect_ranges_1): Likewise for param "vr_1".
5494 (vrp_intersect_ranges): Likewise.
5495 (simplify_stmt_for_jump_threading): Likewise for local "vr".
5496 (vrp_prop::vrp_finalize): Likewise.
5497 * tree-vrp.h (value_range::dump): Add "const" qualifier.
5498 (vrp_intersect_ranges): Add "const" qualifier to params as above.
5499 (extract_range_from_unary_expr): Likewise.
5500 (value_range_constant_singleton): Likewise.
5501 (symbolic_range_p): Likewise.
5502 (copy_value_range): Likewise.
5503 (extract_range_from_binary_expr_1): Likewise.
5504 (range_int_cst_p): Likewise.
5505 (vrp_set_zero_nonzero_bits): Likewise.
5506 (range_int_cst_singleton_p): Likewise.
5508 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
5510 * config/aarch64/arm_neon.h (vabsd_s64): New.
5511 (vnegd_s64): Likewise.
5513 2018-08-31 Martin Jambor <mjambor@suse.cz>
5515 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
5517 2018-08-31 Martin Liska <mliska@suse.cz>
5519 * ipa-icf.c (sem_item::add_type): Use
5520 sem_item::m_type_hash_cache.
5521 * ipa-icf.h: Move the cache from sem_item_optimizer
5524 2018-08-31 Nathan Sidwell <nathan@acm.org>
5526 * doc/extend.texi (Backwards Compatibility): Remove implicit
5527 extern C leeway of () being (...).
5529 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5531 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
5533 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
5537 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
5539 2018-08-31 Jakub Jelinek <jakub@redhat.com>
5542 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
5543 gen_int_mode. Formatting fixes.
5545 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
5547 * target.def (custom_function_descriptors): Improve documentation.
5548 * doc/tm.texi.in (Trampolines): Expand discussion of function
5549 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
5550 beginning of the section.
5551 * doc/tm.texi: Regenerated.
5553 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5555 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5556 base class in the initializer list. This is a workaround for
5557 building with older GCC.
5558 (class auto_bb_flag): Likewise.
5560 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5562 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5563 (altivec_vcmpequ<VI_char>_p): Remove star.
5564 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5566 (expand_strncmp_vec_sequence): New function.
5567 (emit_final_str_compare_vec): New function.
5568 (expand_strn_compare): Add support for vector strncmp.
5569 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5570 length specification to bytes.
5571 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5572 (vcmpnezb_p): New pattern.
5573 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5574 for option -mstring-compare-inline-limit.
5576 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5578 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5579 (PTA_SKYLAKE): Add PTA_AES.
5580 (PTA_GOLDMONT): Likewise.
5582 2018-08-29 Jan Hubicka <jh@suse.cz>
5585 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5586 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5588 2018-08-29 Jan Hubicka <jh@suse.cz>
5590 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5592 (hash_tree): Do not visit TYPE_STUB_DECL.
5593 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5594 stream TYPE_STUB_DECL.
5595 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5596 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5597 after free_lang_data.
5598 (type_in_anonymous_namespace_p): Likewise.
5600 2018-08-29 Jan Hubicka <jh@suse.cz>
5602 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5603 comment that it has to be even number.
5604 (class sreal): Change m_sig type to int32_t.
5605 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5606 int64_t for temporary calculations.
5607 (sreal_verify_basics): Drop one bit from minimum and maximum.
5609 2018-08-30 Richard Biener <rguenther@suse.de>
5611 PR tree-optimization/87147
5612 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5613 (visit_phi): When the degenerate result is from the backedge and
5614 we didn't visit its definition yet drop to VARYING.
5615 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5617 2018-08-29 Jan Hubicka <jh@suse.cz>
5619 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5621 (hash_tree): Likewise.
5623 2018-08-29 Jan Hubicka <jh@suse.cz>
5625 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5626 and TYPE_NEXT_REF_TO.
5628 2018-08-29 Jan Hubicka <jh@suse.cz>
5630 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5631 comment that it has to be even number.
5632 (class sreal): Change m_sig type to int32_t.
5633 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5634 int64_t for temporary calculations.
5635 (sreal_verify_basics): Drop one bit from minimum and maximum.
5637 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5639 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5641 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5644 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5645 if to_add is negative.
5647 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5650 * builtins.c (c_strlen): Improve range checks.
5652 2018-08-29 Martin Sebor <msebor@redhat.com>
5653 Jeff Law <law@redhat.com>
5655 PR tree-optimization/86714
5656 PR tree-optimization/86711
5657 * builtins.c (c_strlen): Add arguments to call to string_constant.
5658 * expr.c (string_constant): Add argument. Detect missing nul
5659 terminator and outermost declaration it's missing in.
5660 * expr.h (string_constant): Add argument.
5661 * fold-const.c (read_from_constant_string): Add arguments to call to
5663 (c_getstr): Likewise.
5664 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5666 * tree-ssa-strlen.c (get_stridx): Likewise.
5668 2018-08-29 Jan Hubicka <jh@suse.cz>
5670 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5671 Do not stream DECL_VINDEX.
5672 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5673 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5674 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5676 2018-08-29 Richard Biener <rguenther@suse.de>
5678 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5679 virtual operands that are not default defs to honor region
5681 (rpo_vn_valueize): Remove ineffective code here.
5683 2018-08-29 Richard Biener <rguenther@suse.de>
5685 PR tree-optimization/87132
5686 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5687 when skipping defs reachable over backedges.
5689 2018-08-29 Richard Biener <rguenther@suse.de>
5691 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5692 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5693 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5694 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5695 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5696 (vn_reference_lookup_3): Remove use of const_parms.
5697 (free_rpo_vn): Do not free const_parms.
5698 (do_rpo_vn): Do not call init_const_parms.
5699 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5700 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5701 (call_may_clobber_ref_p_1): Likewise.
5703 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5706 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5708 (-ftree-scev-cprop): Document.
5710 2018-08-29 Jan Hubicka <jh@suse.cz>
5712 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5714 (sreal constructor): Update.
5715 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5716 sreal:operator/): Update.
5718 2018-08-29 Martin Liska <mliska@suse.cz>
5720 * tree-switch-conversion.c (switch_conversion::expand):
5721 Strenghten assumption about gswitch statements.
5723 2018-08-29 Richard Biener <rguenther@suse.de>
5725 PR tree-optimization/87117
5726 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5727 re-value-number released SSA VDEFs.
5729 2018-08-29 Richard Biener <rguenther@suse.de>
5731 PR tree-optimization/87126
5732 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5734 2018-08-28 Jim Wilson <jimw@sifive.com>
5736 * config/riscv/pic.md: Rewrite.
5737 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5739 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5740 (SOFTF, default_load, softload, softstore): New.
5742 2018-08-28 Jeff Law <law@redhat.com>
5744 * fold-const.c (fold_binary_loc): Remove recently added assert.
5746 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5748 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5749 to OP parmeter of generated function.
5751 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5753 PR tree-optimization/87009
5754 * match.pd: Add boolean optimizations.
5756 2018-08-28 Martin Sebor <msebor@redhat.com>
5759 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5760 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5761 (pass_walloca::gate): Use it.
5762 (alloca_call_type): Same.
5763 (pass_walloca::execute): Same.
5764 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5766 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5768 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5769 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5772 2018-08-28 Richard Biener <rguenther@suse.de>
5774 PR tree-optimization/87124
5775 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5776 constants before looking up avail.
5778 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5781 * calls.c (maybe_warn_nonstring_arg): Punt early if
5782 warn_stringop_overflow is zero. Don't call get_range_strlen
5783 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5784 Swap comparison operands to have constants on rhs. Only use
5785 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5786 increment lenrng[0].
5788 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5790 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5791 use of tree_to_shwi. Remove duplicated test for the size being
5792 a whole number of bytes.
5794 2018-08-28 Richard Biener <rguenther@suse.de>
5796 PR tree-optimization/87117
5797 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5798 Handle removed stmt without LHS (GIMPLE_NOP).
5800 2018-08-28 Richard Biener <rguenther@suse.de>
5802 PR tree-optimization/87117
5803 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5804 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5806 2018-08-28 Richard Biener <rguenther@suse.de>
5808 PR tree-optimization/87117
5809 * tree-ssa-pre.c (compute_avail): Do not make expressions
5810 with predicated values available.
5811 (get_expr_value_id): Assert we do not run into predicated value
5814 2018-08-28 Richard Biener <rguenther@suse.de>
5816 PR tree-optimization/87117
5817 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5818 get virtual operands.
5819 (get_expr_operands): Handle STRING_CST like other decls.
5821 2018-08-28 Martin Liska <mliska@suse.cz>
5823 * tree.h: Update documentation of fndecl_built_in_p
5827 2018-08-27 Jeff Law <law@redhat.com>
5828 PR tree-optimization/87110
5829 * tree-ssa-dse.c (compute_trims): Handle non-constant
5832 2018-08-27 Martin Sebor <msebor@redhat.com>
5834 PR tree-optimization/86914
5835 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5837 2018-08-27 Martin Sebor <msebor@redhat.com>
5839 PR tree-optimization/87112
5840 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5841 the type of the bound argument.
5843 2018-08-27 Jeff Law <law@redhat.com>
5845 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5846 type does not have a TYPE_SIZE_UNIT.
5848 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5850 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5851 with include of backend.h.
5853 2018-08-27 Richard Biener <rguenther@suse.de>
5855 PR tree-optimization/86927
5856 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5857 use const cond reduction code.
5859 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5861 PR tree-optimization/85758
5862 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5864 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5867 * diagnostic-show-locus.c (class layout_range): Update for
5868 conversion of show_caret_p to a tri-state.
5869 (layout_range::layout_range): Likewise.
5870 (make_range): Likewise.
5871 (layout::maybe_add_location_range): Likewise.
5872 (layout::should_print_annotation_line_p): Don't show annotation
5873 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5874 (layout::get_state_at_point): Update for conversion of
5875 show_caret_p to a tri-state. Bail out early for
5876 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5877 underlining or source colorization.
5878 (gcc_rich_location::add_location_if_nearby): Update for conversion
5879 of show_caret_p to a tri-state.
5880 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5881 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5883 (selftest::test_one_liner_labels): Likewise.
5884 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5885 conversion of show_caret_p to a tri-state.
5886 * pretty-print.c (text_info::set_location): Likewise.
5887 * pretty-print.h (text_info::set_location): Likewise.
5888 * substring-locations.c (format_warning_n_va): Likewise.
5889 * tree-diagnostic.c (default_tree_printer): Likewise.
5890 * tree-pretty-print.c (newline_and_indent): Likewise.
5892 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5895 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5896 line above for line-insertion fix-it hints.
5897 (selftest::test_fixit_insert_containing_newline): Update the
5898 expected results, and add a test with line-numbering enabled.
5900 2018-08-27 Martin Liska <mliska@suse.cz>
5903 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5904 params with DECL_HAS_VALUE_EXPR_P.
5906 2018-08-27 Martin Liska <mliska@suse.cz>
5908 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5909 selected expansion strategy.
5911 2018-08-27 Martin Liska <mliska@suse.cz>
5913 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5914 * builtins.c (is_builtin_fn): Likewise.
5915 * attribs.c (diag_attr_exclusions): Use new function
5916 fndecl_built_in_p and remove check for FUNCTION_DECL if
5918 (builtin_mathfn_code): Likewise.
5919 (fold_builtin_expect): Likewise.
5920 (fold_call_expr): Likewise.
5921 (fold_builtin_call_array): Likewise.
5922 (fold_call_stmt): Likewise.
5923 (set_builtin_user_assembler_name): Likewise.
5924 (is_simple_builtin): Likewise.
5925 * calls.c (gimple_alloca_call_p): Likewise.
5926 (maybe_warn_nonstring_arg): Likewise.
5927 * cfgexpand.c (expand_call_stmt): Likewise.
5928 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5929 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5930 (cgraph_node::verify_node): Likewise.
5931 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5932 (cgraph_node::create_clone): Likewise.
5933 * config/arm/arm.c (arm_insert_attributes): Likewise.
5934 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5935 * dse.c (scan_insn): Likewise.
5936 * expr.c (expand_expr_real_1): Likewise.
5937 * fold-const.c (operand_equal_p): Likewise.
5938 (fold_binary_loc): Likewise.
5939 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5940 * gimple-low.c (lower_stmt): Likewise.
5941 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5942 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5944 * gimple.c (gimple_build_call_from_tree): Likewise.
5945 (gimple_call_builtin_p): Likewise.
5946 (gimple_call_combined_fn): Likewise.
5947 * gimplify.c (gimplify_call_expr): Likewise.
5948 (gimple_boolify): Likewise.
5949 (gimplify_modify_expr): Likewise.
5950 (gimplify_addr_expr): Likewise.
5951 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5952 * ipa-cp.c (determine_versionability): Likewise.
5953 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5954 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5955 * ipa-split.c (visit_bb): Likewise.
5956 (split_function): Likewise.
5957 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5958 * lto-cgraph.c (input_node): Likewise.
5959 * lto-streamer-out.c (write_symbol): Likewise.
5960 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5961 (lower_omp_1): Likewise.
5962 * predict.c (strip_predict_hints): Likewise.
5963 * print-tree.c (print_node): Likewise.
5964 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5965 * trans-mem.c (is_tm_irrevocable): Likewise.
5966 (is_tm_load): Likewise.
5967 (is_tm_simple_load): Likewise.
5968 (is_tm_store): Likewise.
5969 (is_tm_simple_store): Likewise.
5970 (is_tm_abort): Likewise.
5971 (tm_region_init_1): Likewise.
5972 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5973 * tree-cfg.c (verify_gimple_call): Likewise.
5974 (move_stmt_r): Likewise.
5975 (stmt_can_terminate_bb_p): Likewise.
5976 * tree-eh.c (lower_eh_constructs_2): Likewise.
5977 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5978 * tree-inline.c (remap_gimple_stmt): Likewise.
5979 (copy_bb): Likewise.
5980 (estimate_num_insns): Likewise.
5981 (fold_marked_statements): Likewise.
5982 * tree-sra.c (scan_function): Likewise.
5983 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5984 (optimize_stack_restore): Likewise.
5985 (pass_fold_builtins::execute): Likewise.
5986 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5987 (mark_all_reaching_defs_necessary_1): Likewise.
5988 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5989 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5990 (pass_forwprop::execute): Likewise.
5991 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5992 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5993 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5994 * tree-ssa-strlen.c (get_string_length): Likewise.
5995 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5996 (find_func_aliases_for_call): Likewise.
5997 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5998 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5999 * tree-tailcall.c (find_tail_calls): Likewise.
6000 * tree.c (need_assembler_name_p): Likewise.
6001 (free_lang_data_in_decl): Likewise.
6002 (get_call_combined_fn): Likewise.
6003 * ubsan.c (is_ubsan_builtin_p): Likewise.
6004 * varasm.c (incorporeal_function_p): Likewise.
6005 * tree.h (DECL_BUILT_IN): Remove and replace with
6007 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
6008 (fndecl_built_in_p): New.
6010 2018-08-27 Martin Liska <mliska@suse.cz>
6012 PR tree-optimization/86847
6013 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
6014 Dump also subtree probability.
6015 (switch_decision_tree::do_jump_if_equal): New function.
6016 (switch_decision_tree::emit_case_nodes): Handle special
6017 situations in balanced tree that can be emitted much simpler.
6018 Fix calculation of probabilities that happen in tree expansion.
6019 * tree-switch-conversion.h (struct cluster): Add
6021 (struct simple_cluster): Likewise.
6022 (struct case_tree_node): Add new function has_child.
6023 (do_jump_if_equal): New.
6025 2018-08-27 Martin Liska <mliska@suse.cz>
6027 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
6028 Add new argument to bit_test_cluster constructor.
6029 (bit_test_cluster::emit): Set bits really number of values
6031 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
6032 probability argument.
6033 * tree-switch-conversion.h (struct bit_test_cluster):
6034 Add m_handles_entire_switch.
6036 2018-08-27 Martin Liska <mliska@suse.cz>
6038 PR tree-optimization/86702
6039 * tree-switch-conversion.c (jump_table_cluster::emit):
6040 Make probabilities even for values in jump table
6041 according to number of cases handled.
6042 (switch_decision_tree::compute_cases_per_edge): Pass
6043 argument to reset_out_edges_aux function.
6044 (switch_decision_tree::analyze_switch_statement): Likewise.
6045 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
6048 2018-08-27 Martin Liska <mliska@suse.cz>
6050 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
6051 cfun argument explicitly.
6052 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
6053 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
6054 function gimple_switch_default_bb.
6055 (convert_switch_statements):
6057 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
6058 * stmt.c (label_to_block_fn): Use label_to_block and pass
6059 cfun argument explicitly and use gimple_switch_label_bb.
6060 (expand_case): Likewise.
6061 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
6062 cfun argument explicitly. Likewise.
6063 (make_edges_bb): Likewise.
6064 (make_cond_expr_edges): Likewise.
6065 (get_cases_for_edge): Likewise.
6066 (make_gimple_switch_edges): Likewise.
6067 (label_to_block_fn): Likewise.
6068 (label_to_block): Likewise.
6069 (make_goto_expr_edges): Likewise.
6070 (make_gimple_asm_edges): Likewise.
6071 (main_block_label): Likewise.
6072 (group_case_labels_stmt): Likewise.
6073 (find_taken_edge_computed_goto): Likewise.
6074 (find_taken_edge_switch_expr): Likewise.
6075 (gimple_verify_flow_info): Likewise.
6076 (gimple_redirect_edge_and_branch): Likewise.
6077 (gimple_switch_label_bb): New function.
6078 (gimple_switch_default_bb): Likewise.
6079 (gimple_switch_edge): Likewise.
6080 (gimple_switch_default_edge): Likewise.
6081 * tree-cfg.h (label_to_block_fn): Remove and replace ...
6082 (label_to_block): ... with this.
6083 (gimple_switch_label_bb): New.
6084 (gimple_switch_default_bb): Likewise.
6085 (gimple_switch_edge): Likewise.
6086 (gimple_switch_default_edge): Likewise.
6087 * tree-cfgcleanup.c (convert_single_case_switch): Use
6088 new gimple functions and pass new argument to label_to_block.
6089 (cleanup_control_flow_bb):
6090 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
6091 cfun argument explicitly.
6092 (make_eh_edges): Likewise.
6093 (redirect_eh_dispatch_edge): Likewise.
6094 (lower_resx): Likewise.
6095 (lower_eh_dispatch): Likewise.
6096 (maybe_remove_unreachable_handlers): Likewise.
6097 (unsplit_eh): Likewise.
6098 (cleanup_empty_eh): Likewise.
6099 (verify_eh_edges): Likewise.
6100 (verify_eh_dispatch_edge): Likewise.
6101 * tree-ssa-dom.c (record_edge_info): Likewise.
6102 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
6103 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
6104 (thread_through_normal_block): Likewise.
6105 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
6106 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
6107 * tree-switch-conversion.c (switch_conversion::collect): Use new
6109 (switch_conversion::check_final_bb): Likewise.
6110 (switch_conversion::gather_default_values): Pass new argument
6112 (switch_conversion::build_constructors): Likewise.
6113 (switch_decision_tree::compute_cases_per_edge): Use new
6114 gimple_switch_edge function.
6115 (switch_decision_tree::analyze_switch_statement): Pass new argument
6117 (switch_decision_tree::try_switch_expansion): Use
6118 gimple_switch_default_edge.
6119 * tree-vrp.c (find_switch_asserts): Pass new argument
6121 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
6122 (vr_values::simplify_switch_using_ranges): Likewise.
6124 2018-08-27 Richard Biener <rguenther@suse.de>
6126 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
6127 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
6129 * tree-ssa-sccvn.h (struct vn_pval): New structure.
6130 (struct vn_nary_op_s): Add unwind_to member. Add
6131 predicated_values flag and put result into a union together
6132 with a linked list of vn_pval.
6133 (struct vn_ssa_aux): Add name member to make maintaining
6134 a map of SSA name to vn_ssa_aux possible. Remove no longer
6135 needed info, dfsnum, low, visited, on_sccstack, use_processed
6136 and range_info_anti_range_p members.
6137 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
6138 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
6140 (vn_valueize): New global.
6141 (vn_context_bb): Likewise.
6142 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
6143 VN_INFO_PTR_INFO): Remove.
6144 * tree-ssa-sccvn.c: ... (rewrite)
6145 (pass_fre::execute): For -O2+ initialize loops and run
6146 RPO VN in optimistic mode (iterating). For -O1 and -Og
6147 run RPO VN in non-optimistic mode.
6148 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
6149 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
6150 * doc/invoke.texi (sccvn-max-scc-size): Remove.
6151 (rpo-vn-max-loop-depth): Document.
6152 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
6153 when valuezing the VUSE signals we walked out of the region.
6154 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
6155 (phi_translate): Set VN context block to use for availability
6157 (compute_avail): Likewise.
6158 (pre_valueize): New function.
6159 (pass_pre::execute): Adjust to the RPO VN API.
6161 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
6162 (propagate_constants_for_unrolling): Remove.
6163 (tree_unroll_loops_completely): Perform value-numbering
6164 on the unrolled bodies loop parent.
6166 2018-08-27 Richard Biener <rguenther@suse.de>
6168 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
6169 for partial antic compute.
6171 2018-08-27 Jakub Jelinek <jakub@redhat.com>
6173 PR rtl-optimization/87065
6174 * combine.c (simplify_if_then_else): Formatting fix.
6175 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
6177 (known_cond): Don't return const_true_rtx for vector modes. Use
6178 CONST0_RTX instead of const0_rtx. Formatting fixes.
6180 2018-08-27 Martin Liska <mliska@suse.cz>
6182 PR gcov-profile/87069
6183 * gcov.c (process_file): Record files already processed
6184 and warn about a file being processed multiple times.
6186 2018-08-27 Martin Liska <mliska@suse.cz>
6189 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6190 Set default values for x_aarch64_*_string strings.
6191 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
6192 prefix. For -mabi do not print '=ABI' in help and use
6193 <option_value> format for -msve-vector-bits and -moverride
6196 2018-08-26 Jeff Law <law@redhat.com>
6198 * config/mips/frame-header-opt.c: Include "backend.h" rather than
6201 2018-08-26 Marek Polacek <polacek@redhat.com>
6203 PR c++/87029, Implement -Wredundant-move.
6204 * doc/invoke.texi: Document -Wredundant-move.
6206 2018-08-25 Martin Sebor <msebor@redhat.com>
6208 PR tree-optimization/87059
6209 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
6210 to the same type as the other.
6211 * fold-const.c (fold_binary_loc): Assert expectation.
6213 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
6215 * config/darwin.c (machopic_legitimize_pic_address): Clean up
6216 extraneous parentheses, dead code section and formatting.
6218 2018-08-24 David Malcolm <dmalcolm@redhat.com>
6221 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
6222 wide enough for jumps in the line-numbering to be visible.
6223 (layout::print_gap_in_line_numbering): New member function.
6224 (layout::calculate_line_spans): When using line numbering, merge
6225 line spans that are only 1 line apart.
6226 (diagnostic_show_locus): When printing line numbers, show gaps in
6227 line numbering directly, rather than printing headers.
6228 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
6229 line-numbering with multiple line spans.
6230 (selftest::test_fixit_insert_containing_newline_2): Add test of
6231 line-numbering, in which the spans are close enough to be merged.
6233 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6235 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
6236 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
6237 * tree-vrp.c (range_is_nonnull): Remove.
6238 (range_includes_zero_p): Accept value_range instead of min/max.
6239 (extract_range_from_binary_expr_1): Do not early bail on
6241 Use range_includes_zero_p instead of range_is_nonnull.
6242 (extract_range_from_unary_expr): Use range_includes_zero_p instead
6243 of range_is_nonnull.
6244 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
6245 special case VR_ANTI_RANGE.
6246 (vrp_finalize): Same.
6247 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
6249 (range_is_nonnull): Remove.
6250 * vr-values.c (vrp_stmt_computes_nonzero): Use
6251 range_includes_zero_p instead of range_is_nonnull.
6252 (extract_range_basic): Pass value_range to range_includes_zero_p
6253 instead of range_is_nonnull.
6255 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
6257 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
6258 * emit-rtl.h (rtl_data): Remove return_bnd.
6259 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
6260 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
6261 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
6262 (POINTER_BOUNDS_MODE): Remove definition.
6263 (make_pointer_bounds_mode): Remove.
6264 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
6265 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
6266 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
6267 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
6268 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
6269 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
6270 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
6272 * config/i386/i386-modes.def (BND32, BND64): Remove.
6273 * config/i386/i386.c (dbx_register_map): Remove bound registers.
6274 (dbx64_register_map): Ditto.
6275 (svr4_dbx_register_map): Ditto.
6276 (indirect_thunk_bnd_needed): Remove.
6277 (indirect_thunks_bnd_used): Ditto.
6278 (indirect_return_bnd_needed): Ditto.
6279 (indirect_return_via_cx_bnd): Ditto.
6280 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
6281 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
6282 (output_indirect_thunk): Ditto. Remove need_prefix argument.
6283 (output_indirect_thunk_function): Remove handling of
6284 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
6285 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
6286 (ix86_save_reg): Remove handling of crtl->return_bnd.
6287 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
6288 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
6289 and UNSPEC_BNDLX_ADDR.
6290 (ix86_output_indirect_branch_via_reg): Remove handling of
6291 indirect_thunk_prefix_bnd.
6292 (ix86_output_indirect_branch_via_push): Ditto.
6293 (ix86_output_function_return): Ditto.
6294 (ix86_output_indirect_function_return): Ditto.
6295 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
6296 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
6297 (CALL_USED_REGISTERS): Ditto.
6298 (REG_ALLOC_ORDER): Update for removal of bound registers.
6299 (HI_REGISTER_NAMES): Ditto.
6300 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
6301 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
6302 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
6303 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
6304 (FIRST_PSEUDO_REG): Update.
6305 (BND): Remove mode iterator.
6306 * config/i386/predicates.md (bnd_mem_operator): Remove.
6308 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6310 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
6313 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6315 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
6316 the case in which the permute needs only a single element and
6317 repeats for every vector of the result. Extend that case to
6318 handle variable-length vectors.
6319 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
6321 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
6324 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
6325 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
6327 2018-08-24 Richard Biener <rguenther@suse.de>
6329 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
6330 bb_flags_allocated members.
6331 (auto_flag): New RAII class for allocating flags.
6332 (auto_edge_flag): New RAII class for allocating edge flags.
6333 (auto_bb_flag): New RAII class for allocating bb flags.
6334 * cfgloop.c (verify_loop_structure): Allocate temporary edge
6336 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
6337 in favor of temporarily allocated BB flag.
6338 * hsa-brig.c: Re-order includes.
6339 * hsa-dump.c: Likewise.
6340 * hsa-regalloc.c: Likewise.
6341 * print-rtl.c: Likewise.
6342 * profile-count.c: Likewise.
6344 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
6347 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
6350 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6353 * wide-int-range.cc (wide_int_range_div): Do not ignore result
6354 from wide_int_range_multiplicative_op.
6356 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6358 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
6361 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
6363 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
6366 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
6368 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
6369 full GENERIC expression used for replacement.
6371 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
6373 * tree-vrp.c (abs_extent_range): Remove.
6374 (extract_range_into_wide_ints): Pass wide ints by reference.
6375 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
6376 Pass wide ints by reference in all calls to
6377 extract_range_into_wide_ints.
6378 * wide-int-range.cc (wide_int_range_div): New.
6379 * wide-int-range.h (wide_int_range_div): New.
6380 (wide_int_range_includes_zero_p): New.
6381 (wide_int_range_zero_p): New.
6383 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
6385 * config/aarch64/aarch64.md (arches): New enum.
6386 (arch): New enum attr.
6387 (arch_enabled): New attr.
6388 (enabled): Now uses arch_enabled only.
6389 (simd, sve, fp16): Removed attribute.
6390 (fp): Attr now defined in terms of 'arch'.
6391 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
6392 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
6393 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
6394 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
6395 attributes into 'arch'.
6396 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
6397 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
6398 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
6399 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
6400 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
6401 'simd' attribute into 'arch'.
6402 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
6403 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
6404 Convert use of 'fp' attribute to 'arch'.
6405 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
6406 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
6408 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
6409 (different modes) Merge 'fp' and 'simd' into 'arch'.
6410 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
6413 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
6415 PR rtl-optimization/87026
6416 * expmed.c (canonicalize_comparison): If we can no longer create
6417 pseudoregisters, don't.
6419 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
6422 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
6424 * config/arm/arm.c (speculation_barrier_libfunc): New static
6426 (arm_init_libfuncs): Initialize it.
6427 (arm_emit_speculation_barrier): New function.
6428 * config/arm/arm.md (speculation_barrier): Call
6429 arm_emit_speculation_barrier for architectures that do not have
6431 (speculation_barrier_insn): Only match on Armv7 or later.
6433 2018-08-23 Richard Biener <rguenther@suse.de>
6436 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
6439 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6441 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
6442 of single-vector TBLs.
6443 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
6446 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6449 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
6450 aarch64_evpc_tbl guard.
6452 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
6454 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
6457 2018-08-22 Martin Sebor <msebor@redhat.com>
6460 * tree-pretty-print.c (pretty_print_string): Add argument.
6461 (dump_generic_node): Call to pretty_print_string with string size.
6463 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
6465 PR rtl-optimization/86771
6466 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
6467 of two SETs into those two SETs, one to be placed at i2, if that SETs
6468 destination is modified between i2 and i3.
6470 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6472 PR tree-optimization/86725
6473 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
6475 (vect_analyze_scalar_cycles_1): Check it.
6477 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6479 PR tree-optimization/86725
6480 * tree-vect-loop.c (vect_is_simple_reduction): When treating
6481 an outer loop phi as a double reduction, make sure that the
6482 single user of the phi result is an inner loop phi.
6484 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6486 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
6487 grouped stores with gaps to a strided group.
6489 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6491 * tree-vect-stmts.c (get_group_load_store_type)
6492 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
6493 first statement in a group.
6495 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6497 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
6498 the sequence used in gcc/gcc.c.
6500 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6503 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
6506 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6508 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
6509 Darwin10-specific unwinder-shim.
6510 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
6511 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
6512 New to cater for Darwin10 Rosetta.
6514 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6516 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
6517 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
6519 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6524 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
6525 (dwarf2out_switch_text_section): Generate a local label for the
6526 second function sub-section and apply it as the second FDE start
6528 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
6529 second sub-section start.
6531 2018-08-22 Richard Biener <rguenther@suse.de>
6533 PR tree-optimization/86988
6534 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
6536 2018-08-22 Richard Biener <rguenther@suse.de>
6538 PR tree-optimization/86945
6539 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
6541 2018-08-22 Alexandre Oliva <oliva@adacore.com>
6543 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
6544 a comment about how uses of r2 for .sdata2 come about.
6546 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
6548 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
6550 2018-08-21 Marek Polacek <polacek@redhat.com>
6552 PR c++/86981, Implement -Wpessimizing-move.
6553 * doc/invoke.texi: Document -Wpessimizing-move.
6555 2018-08-21 Jan Hubicka <jh@suse.cz>
6557 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6558 * tree.h (is_redundant_typedef): Remove.
6559 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6561 2018-08-21 Jan Hubicka <jh@suse.cz>
6563 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6566 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6568 * expmed.c (extract_low_bits): Reject invalid subregs early.
6570 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6573 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6576 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6578 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6579 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6580 * config/vxworks.c: Likewise.
6582 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6584 * config/vxworks.c: Set targetm.have_ctors_dtors
6585 if HAVE_INITFINI_ARRAY_SUPPORT.
6586 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6587 if HAVE_INITFINI_ARRAY_SUPPORT.
6589 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6591 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6592 default search path for VxWorks < 7.
6594 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6596 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6597 (get_format_string): Refer to c_getstr.
6599 2018-08-21 Tom de Vries <tdevries@suse.de>
6601 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6602 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6603 (debuginfo_early_stop): Declare.
6604 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6605 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6606 (debuginfo_early_stop): New function.
6607 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6608 and debuginfo_early_stop.
6609 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6610 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6611 (general_init): Call debuginfo_early_init.
6612 (finalize): Call debuginfo_fini.
6613 (do_compile): Call debuginfo_init.
6614 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6616 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6618 2018-08-21 Tom de Vries <tdevries@suse.de>
6620 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6621 flag_dump_noaddr and flag_dump_unnumbered.
6623 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6625 * wide-int-range.cc (wide_int_range_abs): New.
6626 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6627 * wide-int-range.h (wide_int_range_abs): New.
6628 (wide_int_range_min_max): New.
6629 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6630 case to call wide_int_range_abs.
6631 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6632 (extract_range_from_abs_expr): Delete.
6634 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6637 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6638 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6641 2018-08-20 Nathan Sidwell <nathan@acm.org>
6642 Jeff Law <law@redhat.com>
6644 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6645 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6647 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6650 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6651 (decl_attributes): Likewise.
6652 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6654 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6655 * diagnostic-core.h (class auto_diagnostic_group): New class.
6656 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6657 (diagnostic_report_diagnostic): Handle the first diagnostics within
6659 (emit_diagnostic): Add auto_diagnostic_group instance.
6661 (inform_n): Likewise.
6662 (warning): Likewise.
6663 (warning_at): Likewise.
6664 (warning_n): Likewise.
6665 (pedwarn): Likewise.
6666 (permerror): Likewise.
6668 (error_n): Likewise.
6669 (error_at): Likewise.
6671 (fatal_error): Likewise.
6672 (internal_error): Likewise.
6673 (internal_error_no_backtrace): Likewise.
6674 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6675 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6676 * diagnostic.h (struct diagnostic_context): Add fields
6677 "diagnostic_group_nesting_depth",
6678 "diagnostic_group_emission_count", "begin_group_cb",
6680 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6681 Add auto_diagnostic_group instance(s).
6682 (find_explicit_erroneous_behavior): Likewise.
6683 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6684 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6685 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6686 (gimplify_va_arg_expr): Likewise.
6687 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6688 (HSA_SORRY_AT): Likewise.
6689 * ipa-devirt.c (compare_virtual_tables): Likewise.
6690 (warn_odr): Likewise.
6691 * multiple_target.c (expand_target_clones): Likewise.
6692 * opts-common.c (cmdline_handle_error): Likewise.
6693 * reginfo.c (globalize_reg): Likewise.
6694 * substring-locations.c (format_warning_n_va): Likewise.
6695 * tree-inline.c (expand_call_inline): Likewise.
6696 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6697 * tree-ssa-loop-niter.c
6698 (do_warn_aggressive_loop_optimizations): Likewise.
6699 * tree-ssa-uninit.c (warn_uninit): Likewise.
6700 * tree.c (warn_deprecated_use): Likewise.
6702 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6705 * config/i386/i386.md (eh_return): Always update EH return
6706 address in word_mode.
6708 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6710 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6711 TARGET_SPLIT_COMPLEX_ARG is defined.
6713 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6715 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6717 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6720 * expr.c (expand_assignment): Assert that bitpos is positive.
6721 (store_field): Likewise
6722 (expand_expr_real_1): Make sure that bitpos is positive.
6723 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6726 2018-08-20 Nathan Sidwell <nathan@acm.org>
6728 * Makefile.in (CPP_ID_DATA_H): Delete.
6729 (CPP_INTERNAL_H): Don't add it.
6730 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6731 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6733 2018-08-20 Richard Biener <rguenther@suse.de>
6735 PR tree-optimization/78655
6736 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6737 pointer + offset nonnull if either operand is nonnull work.
6739 2018-08-20 Tom de Vries <tdevries@suse.de>
6741 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6742 unless the referenced die describes the added property using
6743 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6744 Otherwise, add a DW_AT_location to the referenced die.
6746 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6749 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6750 register_operand when calling ix86_set_reg_reg_cost.
6751 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6752 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6753 predicate and to 1 otherwise.
6755 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6757 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6758 emit a diagnostic that it is not supported and reset the option.
6759 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6760 supported and consume the option. (ASM_FINAL_SPEC): New.
6762 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6764 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6767 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6769 C-SKY port: Documentation
6771 * doc/extend.texi (C-SKY Function Attributes): New section.
6772 * doc/invoke.texi (Option Summary): Add C-SKY options.
6773 (C-SKY Options): New section.
6774 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6776 2018-08-17 Jojo <jijie_rong@c-sky.com>
6777 Huibin Wang <huibin_wang@c-sky.com>
6778 Sandra Loosemore <sandra@codesourcery.com>
6779 Chung-Lin Tang <cltang@codesourcery.com>
6781 C-SKY port: Backend implementation
6783 * config/csky/*: New.
6784 * common/config/csky/*: New.
6786 2018-08-17 Jojo <jijie_rong@c-sky.com>
6787 Huibin Wang <huibin_wang@c-sky.com>
6788 Sandra Loosemore <sandra@codesourcery.com>
6789 Chung-Lin Tang <cltang@codesourcery.com>
6790 Andrew Jenner <andrew@codesourcery.com>
6792 C-SKY port: Configury
6794 * config.gcc (csky-*-*): New.
6795 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6796 * configure: Regenerated.
6798 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6800 * dump-context.h: Include "dumpfile.h".
6801 (dump_context::dump_printf_va): Convert final param from va_list
6802 to va_list *. Convert from ATTRIBUTE_PRINTF to
6803 ATTRIBUTE_GCC_DUMP_PRINTF.
6804 (dump_context::dump_printf_loc_va): Likewise.
6805 * dumpfile.c: Include "stringpool.h".
6806 (make_item_for_dump_printf_va): Delete.
6807 (make_item_for_dump_printf): Delete.
6808 (class dump_pretty_printer): New class.
6809 (dump_pretty_printer::dump_pretty_printer): New ctor.
6810 (dump_pretty_printer::emit_items): New member function.
6811 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6813 (dump_pretty_printer::emit_item): New member function.
6814 (dump_pretty_printer::stash_item): New member function.
6815 (dump_pretty_printer::format_decoder_cb): New member function.
6816 (dump_pretty_printer::decode_format): New member function.
6817 (dump_context::dump_printf_va): Reimplement in terms of
6818 dump_pretty_printer.
6819 (dump_context::dump_printf_loc_va): Convert final param from va_list
6821 (dump_context::begin_scope): Reimplement call to
6822 make_item_for_dump_printf.
6823 (dump_printf): Update for change to dump_printf_va.
6824 (dump_printf_loc): Likewise.
6825 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6826 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6827 with %T, %E, and %G.
6828 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6829 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6830 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6831 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6832 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6833 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6834 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6835 within a dump_printf_loc call to "%wu".
6836 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6837 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6838 missing space after "=".
6839 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6840 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6841 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6842 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6843 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6844 duplicate "vectorized" from message.
6846 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6848 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6849 polyNxK_t element's TYPE_STRING_FLAG.
6851 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6853 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6854 (they were unnamed before). Fix comments.
6856 2018-08-17 Nathan Sidwell <nathan@acm.org>
6858 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6860 2018-08-17 Richard Biener <rguenther@suse.de>
6862 PR tree-optimization/86841
6863 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6865 2018-08-17 Martin Liska <mliska@suse.cz>
6867 * common.opt: Remove Warn, Init and Report for options with
6868 Ignore/Deprecated flag. Warning is done automatically for
6870 * config/i386/i386.opt: Likewise.
6871 * config/ia64/ia64.opt: Likewise.
6872 * config/rs6000/rs6000.opt: Likewise.
6873 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6874 Remove usage of flag_check_pointer_bounds.
6875 * lto-wrapper.c (merge_and_complain): Do not handle
6876 OPT_fcheck_pointer_bounds.
6877 (append_compiler_options): Likewise.
6878 * opt-functions.awk: Do not handle Deprecated.
6879 * optc-gen.awk: Check that Var, Report and Init are not
6880 used for an option with Ignore/Deprecated flag.
6881 * opts-common.c (decode_cmdline_option): Do not report
6883 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6885 * opts.h (struct cl_option): Remove cl_deprecated flag.
6886 (CL_ERR_DEPRECATED): Remove error enum value.
6888 2018-08-17 Richard Biener <rguenther@suse.de>
6891 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6892 across a va-arg-pack using call adjust its return value accordingly.
6894 2018-08-16 Martin Sebor <msebor@redhat.com>
6896 PR tree-optimization/86853
6897 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6898 (struct fmtresult): Add member and initialize it in ctors.
6899 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6900 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6901 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6902 (parse_directive): Handle %C same as %c.
6903 (sprintf_dom_walker::compute_format_length): Adjust.
6904 (is_call_safe): Adjust.
6906 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6908 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6909 for determining how to count the elements.
6910 * builtins.h (c_strlen): Adjust prototype.
6911 * expr.c (string_constant): Add new parameter mem_size.
6912 Set *mem_size appropriately.
6913 * expr.h (string_constant): Adjust protoype.
6914 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6915 * gimple-fold.h (get_range_strlen): Adjust prototype.
6916 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6917 (format_string): Call get_string_length with eltsize.
6919 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6921 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6922 to emit the span, rather than setting it as the prefix.
6924 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6926 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6927 "margin_char" parameter, defaulting to space. Use it in place
6928 of pp_space for the initial part of the margin.
6929 (layout::print_leading_fixits): Use '+' when filling the margin
6930 of line-insertion fix-it hints.
6932 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6934 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6937 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6939 * config/rs6000/altivec.md: Don't set length attribute to the default
6941 * config/rs6000/darwin.md: Ditto.
6942 * config/rs6000/dfp.md: Ditto.
6943 * config/rs6000/htm.md: Ditto.
6944 * config/rs6000/rs6000.md: Ditto.
6945 * config/rs6000/sync.md: Ditto.
6946 * config/rs6000/vsx.md: Ditto.
6948 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6950 * config/rs6000/altivec.md: Don't set length attribute to the default
6951 value, for branch instructions.
6952 * config/rs6000/darwin.md: Ditto.
6953 * config/rs6000/rs6000.md: Ditto.
6955 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6957 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6958 (unnamed conditional branch define_insn): Set length to 4 or 8
6959 depending on offset.
6960 (<bd>_<mode>): Similar, for alternative 0.
6961 (<bd>tf_<mode>): Ditto.
6963 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6965 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6967 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6969 * doc/rtl.texi: Replace old RTX class names with new names.
6972 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6974 * expmed.h (canonicalize_comparison): New declaration.
6975 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6976 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6977 * optabs.c (prepare_cmp_insn): Likewise.
6978 * rtl.h (unsigned_condition_p): New function which checks if a
6979 comparison operator is unsigned.
6981 2018-08-16 Nathan Sidwell <nathan@acm.org>
6983 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6984 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6986 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6989 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6990 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6991 (mov<mov>): ..this and enable unconditionally.
6993 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6995 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6997 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6999 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
7000 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
7001 "Common" with "Target".
7003 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
7005 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
7006 * doc/invoke.texi (mmitigate-rop): Remove.
7007 * config/i386/i386.c: Do not include "regrename.h".
7008 (ix86_rop_should_change_byte_p, reg_encoded_number)
7009 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
7011 (ix86_reorg): Remove call to ix86_mitigate_rop.
7012 * config/i386/i386.md (attr "modrm_class"): Remove.
7013 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
7014 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
7015 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
7017 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7019 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
7020 allow folding of mergeh() and mergel() for the float and double types.
7021 (fold_mergehl_helper): Rework to handle building a permute tree
7024 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
7026 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
7029 2018-08-15 David Malcolm <dmalcolm@redhat.com>
7031 * common.opt (fdiagnostics-show-labels): New option.
7032 * diagnostic-show-locus.c (class layout_range): Add field
7034 (class layout): Add field "m_show_labels_p".
7035 (layout_range::layout_range): Add param "label" and use it to
7037 (make_range): Pass in NULL for new "label" param of layout_range's
7039 (layout::layout): Initialize m_show_labels_p.
7040 (layout::maybe_add_location_range): Pass in loc_range->m_label
7041 when constructing layout_range instances.
7042 (struct line_label): New struct.
7043 (layout::print_any_labels): New member function.
7044 (layout::print_line): Call it if label-printing is enabled.
7045 (selftest::test_one_liner_labels): New test.
7046 (selftest::test_diagnostic_show_locus_one_liner): Call it.
7047 * diagnostic.c (diagnostic_initialize): Initialize
7048 context->show_labels_p.
7049 * diagnostic.h (struct diagnostic_context): Add field
7051 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7052 -fno-diagnostics-show-labels.
7053 * dwarf2out.c (gen_producer_string): Add
7054 OPT_fdiagnostics_show_labels to the ignored options.
7055 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
7057 (gcc_rich_location::maybe_add_expr): Likewise.
7058 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
7059 label" param, defaulting to NULL.
7060 (gcc_rich_location::add_expr): Add "label" param.
7061 (gcc_rich_location::maybe_add_expr): Likewise.
7062 (class text_range_label): New class.
7063 (class range_label_for_type_mismatch): New class.
7064 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
7065 of format_warning_va.
7066 (fmtwarn_n): Likewise for new params of format_warning_n_va.
7067 * lto-wrapper.c (merge_and_complain): Add
7068 OPT_fdiagnostics_show_labels to the "pick one setting" options.
7069 (append_compiler_options): Likewise to the dropped options.
7070 (append_diag_options): Likewise to the passed-on options.
7071 * opts.c (common_handle_option): Handle the new option.
7072 * selftest-diagnostic.c
7073 (test_diagnostic_context::test_diagnostic_context): Enable
7075 * substring-locations.c: Include "gcc-rich-location.h".
7076 (format_warning_n_va): Add "fmt_label" and "param_label" params
7077 and use them as appropriate.
7078 (format_warning_va): Add "fmt_label" and "param_label" params,
7079 passing them on to format_warning_n_va.
7080 (format_warning_at_substring): Likewise.
7081 (format_warning_at_substring_n): Likewise.
7082 * substring-locations.h (format_warning_va): Add "fmt_label" and
7083 "param_label" params.
7084 (format_warning_n_va): Likewise.
7085 (format_warning_at_substring): Likewise.
7086 (format_warning_at_substring_n): Likewise.
7087 * toplev.c (general_init): Initialize global_dc->show_labels_p.
7089 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
7092 * builtins.c (expand_builtin_memcmp): Do not expand the call
7093 when overflow is detected.
7095 2018-08-15 Martin Sebor <msebor@redhat.com>
7097 PR tree-optimization/71625
7098 * config/aarch64/aarch64-builtins.c
7099 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
7101 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
7103 * config/s390/s390.c (s390_reorg): Remove loop.
7105 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
7108 (darwin_function_switched_text_sections): Delete.
7109 * gcc/config/darwin.h
7110 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
7112 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
7115 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
7116 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
7117 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
7119 2018-08-15 Martin Liska <mliska@suse.cz>
7121 PR tree-optimization/86925
7122 * predict.c (expr_expected_value_1): When taking
7123 later predictor, assign also probability.
7124 Use fold_build2_initializer_loc in order to fold
7125 the expression in -frounding-math.
7127 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
7129 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
7131 (expand_vec_perm_1): Try the new method.
7133 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
7136 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
7137 Check whether lra_live_max_point is 0 before dividing.
7139 2018-08-14 Martin Sebor <msebor@redhat.com>
7141 PR tree-optimization/86650
7142 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
7143 (vrp_prop::check_mem_ref): Same.
7145 2018-08-13 Liu Hao <lh_mouse@126.com>
7147 * pretty-print.c (eat_esc_sequence): Swap the foreground and
7148 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
7149 and clear it thereafter, as it only works for DBCS.
7151 2018-08-13 Liu Hao <lh_mouse@126.com>
7153 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
7154 handle returned by _get_osf_handle().
7156 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7158 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
7159 for folding vec_perm.
7161 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7163 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
7164 Add support for gimple-folding of vec_pack() and vec_unpack()
7167 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7169 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
7170 vec_xst variants to the list.
7171 (rs6000_gimple_fold_builtin): Add support for folding unaligned
7172 vector loads and stores.
7174 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
7176 * config.gcc (rs6000-ibm-aix4.x): Delete.
7177 (rs6000-ibm-aix5.1): Delete.
7178 (rs6000-ibm-aix5.2): Delete.
7179 (rs6000-ibm-aix5.3): Delete.
7180 * config/rs6000/aix43.h: Delete.
7181 * config/rs6000/aix51.h: Delete.
7182 * config/rs6000/aix52.h: Delete.
7183 * config/rs6000/t-aix43: Delete.
7185 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
7187 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7189 (s390_decompose_address): Factor out constant pool ref
7192 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7194 * config/nds32/nds32-predicates.c
7195 (nds32_can_use_bclr_p): Change return type as bool.
7196 (nds32_can_use_bset_p): Ditto.
7197 (nds32_can_use_btgl_p): Ditto.
7198 (nds32_can_use_bitci_p): Ditto.
7199 * config/nds32/nds32-protos.h
7200 (nds32_can_use_bclr_p): Change declaration.
7201 (nds32_can_use_bset_p): Ditto.
7202 (nds32_can_use_btgl_p): Ditto.
7203 (nds32_can_use_bitci_p): Ditto.
7205 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7207 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
7208 Support -msched-prolog-epilog option.
7209 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
7211 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7213 * common/config/nds32/nds32-common.c
7214 (nds32_option_optimization_table): Enalbe -malways-align.
7216 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7218 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
7220 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
7221 OPT_misr_secure_ case.
7222 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
7223 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
7224 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
7226 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
7227 (nds32_isr_info): New field security_level.
7228 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
7229 * config/nds32/nds32.md (return_internal): Consider critical attribute.
7230 * config/nds32/nds32.opt (misr-secure): New option.
7231 * config/nds32/nds32_init.inc: New file.
7232 * config/nds32/nds32_isr.h: New file.
7234 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
7236 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
7237 Update comment for atomic instructions.
7238 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
7239 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
7241 (atomic_loaddi): Revise fence expansion to only emit fence prior to
7242 load for __ATOMIC_SEQ_CST model.
7243 (atomic_loaddi_1): Remove float register target.
7244 (atomic_storedi): Handle CONST_INT values.
7245 (atomic_storedi_1): Remove float register source. Add special case
7247 (memory_barrier): New expander and insn.
7249 2018-08-11 Jakub Jelinek <jakub@redhat.com>
7251 PR tree-optimization/86835
7252 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
7253 new_stmt after def_gsi, make sure to insert new_square_stmt after
7254 that stmt, not 2 stmts before it.
7256 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
7259 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
7262 2018-08-10 Martin Liska <mliska@suse.cz>
7265 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
7267 * builtins.c (expand_builtin_expect_with_probability):
7269 (expand_builtin_expect_with_probability): New function.
7270 (build_builtin_expect_predicate): Add new argumnet probability
7271 for BUILT_IN_EXPECT_WITH_PROBABILITY.
7272 (fold_builtin_expect):
7275 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
7276 * builtins.h (fold_builtin_expect): Set new argument.
7277 * doc/extend.texi: Document __builtin_expect_with_probability.
7278 * doc/invoke.texi: Likewise.
7279 * gimple-fold.c (gimple_fold_call): Pass new argument.
7280 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
7281 also BUILT_IN_EXPECT_WITH_PROBABILITY.
7282 * predict.c (get_predictor_value): New function.
7283 (expr_expected_value): Add new argument probability. Assume
7284 that predictor and probability are always non-null.
7285 (expr_expected_value_1): Likewise. For __builtin_expect and
7286 __builtin_expect_with_probability set probability. Handle
7287 combination in binary expressions.
7288 (tree_predict_by_opcode): Simplify code by simply calling
7289 get_predictor_value.
7290 (pass_strip_predict_hints::execute): Add handling of
7291 BUILT_IN_EXPECT_WITH_PROBABILITY.
7292 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
7294 * tree.h (DECL_BUILT_IN_P): New function.
7296 2018-08-10 Martin Liska <mliska@suse.cz>
7298 PR tree-optimization/85799
7299 * passes.def: Add argument for pass_strip_predict_hints.
7300 * predict.c (class pass_strip_predict_hints): Add new argument
7302 (strip_predictor_early): New function.
7303 (pass_strip_predict_hints::execute): Call the function to
7305 (strip_predict_hints): New function.
7306 * predict.def: Fix comment.
7308 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
7310 * Makefile.in: Clarify which tm.texi to copy over to assert the
7311 right to grant a GFDL license for all.
7313 2018-08-09 Jeff Law <law@redhat.com>
7315 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
7318 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
7320 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
7323 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7325 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
7326 reductions for variable-length vectors.
7328 2018-08-09 David Malcolm <dmalcolm@redhat.com>
7331 * common.opt (fdiagnostics-show-line-numbers): New option.
7332 * diagnostic-show-locus.c (class layout): Add fields
7333 "m_show_line_numbers_p" and "m_linenum_width";
7334 (num_digits): New function.
7335 (test_num_digits): New function.
7336 (layout::layout): Initialize new fields. Update m_x_offset
7337 logic to handle any left margin.
7338 (layout::print_source_line): Print line number when requested.
7339 (layout::start_annotation_line): New member function.
7340 (layout::print_annotation_line): Call it.
7341 (layout::print_leading_fixits): Likewise.
7342 (layout::print_trailing_fixits): Likewise. Update calls to
7343 move_to_column for new parameter.
7344 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
7345 it to potentially call start_annotation_line.
7346 (layout::show_ruler): Call start_annotation_line.
7347 (selftest::test_line_numbers_multiline_range): New selftest.
7348 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
7349 and selftest::test_line_numbers_multiline_range.
7350 * diagnostic.c (diagnostic_initialize): Initialize
7351 show_line_numbers_p.
7352 * diagnostic.h (struct diagnostic_context): Add field
7353 "show_line_numbers_p".
7354 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7355 -fno-diagnostics-show-line-numbers.
7356 * dwarf2out.c (gen_producer_string): Add
7357 OPT_fdiagnostics_show_line_numbers to the ignored options.
7358 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
7359 one setting" options.
7360 (append_compiler_options): Likewise to the dropped options.
7361 (append_diag_options): Likewise to the passed-on options.
7362 * opts.c (common_handle_option): Handle the new option.
7363 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
7365 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
7367 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
7368 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
7369 third argument of type "const signed char" to descriptions of
7370 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
7371 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
7372 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
7373 __builtin_bcdsub_ov functions.
7375 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7377 PR tree-optimization/86858
7378 * tree-vect-loop.c (vect_is_simple_reduction): Restore
7379 flow_bb_inside_loop_p calls.
7381 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7383 PR tree-optimization/86871
7384 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
7385 instead of gimple_assign_lhs.
7387 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
7390 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
7391 register constraint to operand 0.
7392 (add<mode>3_carryinC): Likewise.
7393 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
7395 2018-08-09 Martin Liska <mliska@suse.cz>
7398 * common.opt: Remove extra line.
7400 2018-08-09 Martin Liska <mliska@suse.cz>
7402 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
7403 at the end of a line, make first letter capital and end up
7404 a sentence with a dot.
7405 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
7406 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
7407 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
7408 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
7409 (PARAM_MAX_ISL_OPERATIONS): Likewise.
7410 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
7411 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
7412 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
7413 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
7414 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
7415 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
7416 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
7417 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
7418 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
7419 (PARAM_TREE_REASSOC_WIDTH): Likewise.
7420 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
7421 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
7422 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
7424 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
7427 * config/s390/s390.c (s390_option_override_internal): Reduce the
7428 stack-clash-protection-probe-interval param if it would be too big
7431 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
7434 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
7435 * config/m68k/m68k.c (handle_move_double): Don't call
7436 m68k_final_prescan_insn.
7437 (m68k_adjust_decorated_operand): Renamed from
7438 m68k_final_prescan_insn, remove first and third operand and
7440 (print_operand): Call it.
7441 (print_operand_address): Call it.
7443 2018-08-08 Nathan Sidwell <nathan@acm.org>
7445 * diagnostic.c (diagnostic_report_current_module): Use
7446 linemap_included_from & linemap_included_from_linemap.
7448 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
7450 * config/aarch64/aarch64-cores.def: Add phecda core.
7451 * config/aarch64/aarch64-tune.md: Regenerate.
7452 * doc/invoke.texi: Add phecda core.
7454 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7457 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
7459 * config/s390/s390.md ("movti"): Add more alternatives for
7460 constant to GPR copies.
7462 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7464 * config/s390/s390.c: Fix whitespace damage throughout the file.
7465 * config/s390/s390.h: Likewise.
7466 * config/s390/tpf.h: Likewise.
7468 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7470 * config/s390/s390.c (s390_loadrelative_operand_p):
7471 Remove TARGET_CPU_ZARCH usages.
7472 (s390_rtx_costs): Likewise.
7473 (s390_legitimate_constant_p): Likewise.
7474 (s390_cannot_force_const_mem): Likewise.
7475 (legitimate_reload_constant_p): Likewise.
7476 (s390_preferred_reload_class): Likewise.
7477 (legitimize_pic_address): Likewise.
7478 (legitimize_tls_address): Likewise.
7479 (s390_split_branches): Removed.
7480 (s390_add_execute): Removed.
7481 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
7482 (s390_mainpool_start): Likewise.
7483 (s390_mainpool_finish): Likewise.
7484 (s390_mainpool_cancel): Removed.
7485 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
7486 (s390_chunkify_cancel): Likewise.
7487 (s390_return_addr_rtx): Likewise.
7488 (s390_register_info): Remove split_branches_pending_p uages.
7489 (s390_optimize_register_info): Likewise.
7490 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
7491 split_branches_pending_p usages.
7492 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
7493 (s390_load_got): Likewise.
7494 (s390_expand_split_stack_prologue): Likewise.
7495 (output_asm_nops): Likewise.
7496 (s390_function_profiler): Likewise.
7497 (s390_emit_call): Likewise.
7498 (s390_conditional_register_usage): Likewise.
7499 (s390_optimize_prologue): Likewise.
7500 (s390_reorg): Remove TARGET_CPU_ZARCH and
7501 split_branches_pending_p usages.
7502 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
7504 (s390_output_indirect_thunk_function): Likewise.
7505 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
7506 (TARGET_CPU_ZARCH_P): Removed.
7507 (struct machine_function): Remove split_branches_pending_p.
7508 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
7510 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7512 * common/config/s390/s390-common.c (processor_flags_table):
7514 * config.gcc: Remove with_arch/with_tune support.
7515 * config/s390/2064.md: Remove cpu attribute comparisons.
7516 * config/s390/driver-native.c (s390_host_detect_local_cpu):
7518 * config/s390/linux.h (ASM_SPEC):
7519 Remove -march support.
7520 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
7521 Use a table to get an arch level.
7522 * config/s390/s390-opts.h (enum processor_type):
7524 * config/s390/s390.c
7525 (processor_table): Remove entries, add arch_level values.
7526 (s390_issue_rate): Remove cases.
7527 (s390_option_override): Adjust
7528 s390_option_override_internal() call.
7529 (s390_option_override_internal): Remove deprecation warning.
7530 (s390_valid_target_attribute_tree): Adjust
7531 s390_option_override_internal() call.
7532 * config/s390/s390.h (struct s390_processor):
7533 Share with s390-c.c, add arch_level field.
7534 * config/s390/s390.md:
7535 Remove occurrences in cpu attribute.
7536 * config/s390/s390.opt: Remove -march/-mtune support.
7537 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
7538 * doc/invoke.texi: Remove deprecation warning.
7540 2018-08-08 Luis Machado <luis.machado@linaro.org>
7542 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
7544 (qdf24xx_tunings): Set vector cost structure to
7545 qdf24xx_vector_cost.
7547 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
7548 <register_sextend>: Set to 3.
7550 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7553 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7554 * config/aarch64/aarch64-simd.md
7555 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7556 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7557 (aarch64_frecpx<mode>): ...this new pattern.
7558 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7559 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7561 2018-08-07 Martin Liska <mliska@suse.cz>
7564 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7565 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7566 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7567 * doc/extend.texi: Document that malloc attribute adds
7570 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7573 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7574 Define to speculation_safe_value_not_needed.
7576 2018-08-06 Jeff Law <law@redhat.com>
7578 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7579 the vr_values instance to cprop_into_stmt.
7580 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7581 (cprop_operand): Also query EVRP to determine if OP is a constant.
7583 2018-08-06 Nathan Sidwell <nathan@acm.org>
7585 * diagnostic.c (diagnostic_report_current_module): Reroll
7586 included-at loop. Translate text.
7588 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7590 * function-tests.c (selftest::test_expansion_to_rtl): Call
7591 free_after_compilation.
7593 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7595 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7597 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7599 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7600 loops with memory block operations from getting unrolled.
7602 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7605 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7606 Define to speculation_safe_value_not_needed.
7608 2018-08-06 Jeff Law <law@redhat.com>
7610 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7613 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7616 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7617 with all enabled __intN types.
7619 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7621 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7623 * alias.c (record_set): Check for clobber high.
7624 * cfgexpand.c (expand_gimple_stmt): Likewise.
7625 * combine-stack-adj.c (single_set_for_csa): Likewise.
7626 * combine.c (find_single_use_1): Likewise.
7627 (set_nonzero_bits_and_sign_copies): Likewise.
7628 (get_combine_src_dest): Likewise.
7629 (is_parallel_of_n_reg_sets): Likewise.
7630 (try_combine): Likewise.
7631 (record_dead_and_set_regs_1): Likewise.
7632 (reg_dead_at_p_1): Likewise.
7633 (reg_dead_at_p): Likewise.
7634 * dce.c (deletable_insn_p): Likewise.
7635 (mark_nonreg_stores_1): Likewise.
7636 (mark_nonreg_stores_2): Likewise.
7637 * df-scan.c (df_find_hard_reg_defs): Likewise.
7638 (df_uses_record): Likewise.
7639 (df_get_call_refs): Likewise.
7640 * dwarf2out.c (mem_loc_descriptor): Likewise.
7641 * haifa-sched.c (haifa_classify_rtx): Likewise.
7642 * ira-build.c (create_insn_allocnos): Likewise.
7643 * ira-costs.c (scan_one_insn): Likewise.
7644 * ira.c (equiv_init_movable_p): Likewise.
7645 (rtx_moveable_p): Likewise.
7646 (interesting_dest_for_shprep): Likewise.
7647 * jump.c (mark_jump_label_1): Likewise.
7648 * postreload-gcse.c (record_opr_changes): Likewise.
7649 * postreload.c (reload_cse_simplify): Likewise.
7650 (struct reg_use): Add source expr.
7651 (reload_combine): Check for clobber high.
7652 (reload_combine_note_use): Likewise.
7653 (reload_cse_move2add): Likewise.
7654 (move2add_note_store): Likewise.
7655 * print-rtl.c (print_pattern): Likewise.
7656 * recog.c (decode_asm_operands): Likewise.
7657 (store_data_bypass_p): Likewise.
7658 (if_test_bypass_p): Likewise.
7659 * regcprop.c (kill_clobbered_value): Likewise.
7660 (kill_set_value): Likewise.
7661 * reginfo.c (reg_scan_mark_refs): Likewise.
7662 * reload1.c (maybe_fix_stack_asms): Likewise.
7663 (eliminate_regs_1): Likewise.
7664 (elimination_effects): Likewise.
7665 (mark_not_eliminable): Likewise.
7666 (scan_paradoxical_subregs): Likewise.
7667 (forget_old_reloads_1): Likewise.
7668 * reorg.c (find_end_label): Likewise.
7669 (try_merge_delay_insns): Likewise.
7670 (redundant_insn): Likewise.
7671 (own_thread_p): Likewise.
7672 (fill_simple_delay_slots): Likewise.
7673 (fill_slots_from_thread): Likewise.
7674 (dbr_schedule): Likewise.
7675 * resource.c (update_live_status): Likewise.
7676 (mark_referenced_resources): Likewise.
7677 (mark_set_resources): Likewise.
7678 * rtl.c (copy_rtx): Likewise.
7679 * rtlanal.c (reg_referenced_p): Likewise.
7680 (single_set_2): Likewise.
7681 (noop_move_p): Likewise.
7682 (note_stores): Likewise.
7683 * sched-deps.c (sched_analyze_reg): Likewise.
7684 (sched_analyze_insn): Likewise.
7686 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7688 * cse.c (invalidate_reg): New function extracted from...
7689 (invalidate): ...here.
7690 (canonicalize_insn): Check for clobber high.
7691 (invalidate_from_clobbers): invalidate clobber highs.
7692 (invalidate_from_sets_and_clobbers): Likewise.
7693 (count_reg_usage): Check for clobber high.
7694 (insn_live_p): Likewise.
7695 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7696 (cselib_invalidate_regno): Check for clobber in setter.
7697 (cselib_invalidate_rtx): Pass through setter.
7698 (cselib_invalidate_rtx_note_stores):
7699 (cselib_process_insn): Check for clobber high.
7700 * cselib.h (cselib_invalidate_rtx): Add operand.
7702 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7704 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7705 (mark_not_eliminable): Likewise.
7706 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7707 * lra-lives.c (process_bb_lives): Check for clobber high.
7708 * lra.c (new_insn_reg): Remember clobber highs.
7709 (collect_non_operand_hard_regs): Check for clobber high.
7710 (lra_set_insn_recog_data): Likewise.
7711 (add_regs_to_insn_regno_info): Likewise.
7712 (lra_update_insn_regno_info): Likewise.
7714 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7716 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7717 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7719 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7721 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7722 (copy_insn_1): Likewise.
7723 (gen_hard_reg_clobber_high): New gen function.
7724 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7725 * genemit.c (gen_exp): Likewise.
7726 (gen_emit_seq): Pass through info.
7727 (gen_insn): Check for CLOBBER_HIGH.
7728 (gen_expand): Pass through info.
7729 (gen_split): Likewise.
7730 (output_add_clobbers): Likewise.
7731 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7732 (remove_clobbers): Likewise.
7733 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7735 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7737 * doc/rtl.texi (clobber_high): Add.
7738 (parallel): Add in clobber high
7739 * rtl.c (rtl_check_failed_code3): Add function.
7740 * rtl.def (CLOBBER_HIGH): Add expression.
7741 * rtl.h (RTL_CHECKC3): Add macro.
7742 (rtl_check_failed_code3): Add declaration.
7743 (XC3EXP): Add macro.
7745 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7748 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7749 cfun->machine->max_used_stack_alignment if needed.
7751 2018-08-04 Martin Sebor <msebor@redhat.com>
7753 PR tree-optimization/86571
7754 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7757 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7760 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7763 2018-08-03 Jeff Law <law@redhat.com>
7766 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7767 Define to speculation_safe_value_not_needed.
7769 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7771 * doc/gcov.texi (-x): Remove duplicate "to".
7772 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7773 (-Wif-not-aligned): Remove duplicate "is".
7774 (-flto): Remove duplicate "the".
7775 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7776 duplicate "v5.00.b".
7777 (MSP430 Options): Remove duplicate "and" from the description
7778 of "-mgprel-sec=regexp".
7779 (x86 Options): Remove duplicate copies of "vmldLog102" and
7780 vmlsLog104 from description of "-mveclibabi=type".
7782 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7784 * internal-fn.h (first_commutative_argument): Declare.
7785 * internal-fn.c (first_commutative_argument): New function.
7786 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7787 restrictions for pattern statements. Use first_commutative_argument
7788 to look for commutative operands in calls to internal functions.
7790 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7792 * Makefile.in (wide-int-range.o): New.
7793 * tree-vrp.c: Move all the wide_int_* functions to...
7794 * wide-int-range.cc: ...here.
7795 * tree-vrp.h: Move all the wide_int_* prototypes to...
7796 * wide-int-range.h: ...here.
7798 2018-08-03 Tom de Vries <tdevries@suse.de>
7800 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7802 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7803 * except.c (output_function_exception_table): Do early exit if
7804 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7806 2018-08-03 Martin Liska <mliska@suse.cz>
7808 * predict.c (dump_prediction): Change to 2 digits
7811 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7813 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7815 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7817 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7819 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7820 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7821 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7822 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7823 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7825 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7827 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7828 pp_take_prefix when saving the existing prefix.
7829 * diagnostic.c (diagnostic_append_note): Likewise.
7830 * langhooks.c (lhd_print_error_function): Likewise.
7831 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7832 param's type. Free the existing prefix.
7833 (pp_take_prefix): New function.
7834 (pretty_printer::pretty_printer): Drop the prefix parameter.
7835 Rename the length parameter to match the comment.
7836 (pretty_printer::~pretty_printer): Free the prefix.
7837 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7839 (struct pretty_printer): Drop the "const" from "prefix" field's
7840 type and clarify memory management.
7841 (pp_set_prefix): Drop the "const" from the 2nd param.
7842 (pp_take_prefix): New decl.
7844 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7846 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7847 for word_mode_ok here instead of passing as argument.
7848 (expand_block_compare): Change select_block_compare_mode() call.
7849 (expand_strncmp_gpr_sequence): New function.
7850 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7852 2018-08-02 Jeff Law <law@redhat.com>
7855 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7856 Define to speculation_safe_value_not_needed.
7859 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7860 Define to speculation_safe_value_not_needed.
7862 2018-08-02 Tom de Vries <tdevries@suse.de>
7865 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7866 function. Return UI_TARGET unconditionally.
7867 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7868 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7870 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7872 * genemit.c (print_overload_test): Fix typo.
7874 2018-08-02 Richard Biener <rguenther@suse.de>
7876 PR tree-optimization/86816
7877 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7878 which checks for value availability before querying it.
7879 (gvn_uses_equal): Use it.
7880 (same_succ_hash): Likewise.
7881 (gimple_equal_p): Likewise.
7883 2018-08-02 Nick Clifton <nickc@redhat.com>
7886 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7887 Define to speculation_safe_value_not_needed.
7890 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7891 Define to speculation_safe_value_not_needed.
7894 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7895 Define to speculation_safe_value_not_needed.
7898 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7899 Define to speculation_safe_value_not_needed.
7902 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7903 Define to speculation_safe_value_not_needed.
7906 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7907 Define to speculation_safe_value_not_needed.
7910 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7911 Define to speculation_safe_value_not_needed.
7914 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7915 Define to speculation_safe_value_not_needed.
7918 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7919 speculation_safe_value_not_needed.
7922 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7923 to speculation_safe_value_not_needed.
7925 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7927 * doc/md.texi: Expand the documentation of instruction names
7928 to mention port-local uses. Document '@' in pattern names.
7929 * read-md.h (overloaded_instance, overloaded_name): New structs.
7931 (md_reader::handle_overloaded_name): New member function.
7932 (md_reader::get_overloads): Likewise.
7933 (md_reader::m_first_overload): New member variable.
7934 (md_reader::m_next_overload_ptr): Likewise.
7935 (md_reader::m_overloads_htab): Likewise.
7936 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7937 m_next_overload_ptr and m_overloads_htab.
7938 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7939 (get_mode_token, get_code_token, get_int_token): New functions.
7940 (map_attr_string): Add an optional argument that passes back
7941 the associated iterator.
7942 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7943 (md_reader::handle_overloaded_name, add_overload_instance): New
7945 (apply_iterators): Handle '@' names. Report an error if '@'
7946 is used without iterators.
7947 (initialize_iterators): Initialize the new iterator_group fields.
7948 * genopinit.c (handle_overloaded_code_for)
7949 (handle_overloaded_gen): New functions.
7950 (main): Use them to print declarations of maybe_code_for_* and
7951 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7952 * genemit.c (print_overload_arguments, print_overload_test)
7953 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7954 (main): Use it to print definitions of maybe_code_for_* and
7955 maybe_gen_* functions.
7956 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7957 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7958 instead of explicit mode checks.
7959 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7960 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7961 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7962 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7963 (aarch64_expand_compare_and_swap): Likewise
7964 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7965 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7966 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7967 (aarch64_constant_pool_reload_icode): Delete.
7968 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7969 instead of aarch64_constant_pool_reload_icode. Use
7970 code_for_aarch64_reload_mov instead of explicit mode checks.
7971 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7972 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7973 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7974 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7975 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7976 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7977 (aarch64_atomic_load_op_code): Delete.
7978 (aarch64_emit_atomic_load_op): Likewise.
7979 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7980 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7981 instead of aarch64_emit_atomic_load_op.
7982 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7983 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7984 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7985 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7986 character before the pattern name.
7987 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7988 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7989 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7990 (aarch64_frecps<mode>): Likewise.
7991 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7992 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7993 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7994 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7995 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7997 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7999 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
8000 Allow HFmode constants if TARGET_FP_F16INST.
8002 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
8005 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
8006 No longer check last store for clobber of address register.
8008 2018-08-02 Martin Liska <mliska@suse.cz>
8010 PR gcov-profile/86817
8011 * gcov.c (process_all_functions): New function.
8013 (process_file): Move functions processing to
8014 process_all_functions.
8016 2018-08-02 David Malcolm <dmalcolm@redhat.com>
8018 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
8019 "const" to the "gimple *" and "rtx_insn *" parameters.
8020 * dumpfile.h (dump_user_location_t::dump_user_location_t):
8022 (dump_location_t::dump_location_t): Likewise.
8024 2018-08-01 Martin Sebor <msebor@redhat.com>
8026 PR tree-optimization/86650
8027 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
8028 rather than a "gcall *". Directly pass the data of interest
8029 to percent_K_format, rather than building a temporary CALL_EXPR
8031 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
8032 (gimple_fold_builtin_strncat): Adjust.
8033 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
8034 gcall* argument with gimple*.
8035 * gimple-ssa-warn-restrict.c (check_call): Same.
8036 (wrestrict_dom_walker::before_dom_children): Same.
8037 (builtin_access::builtin_access): Same.
8038 (check_bounds_or_overlap): Same
8039 (maybe_diag_overlap): Same.
8040 (maybe_diag_offset_bounds): Same.
8041 * tree-diagnostic.c (default_tree_printer): Move usage of
8042 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
8044 * tree-pretty-print.c (percent_K_format): Add argument.
8045 * tree-pretty-print.h: Add argument.
8046 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
8047 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
8048 (maybe_diag_stxncpy_trunc): Same.
8049 (handle_builtin_stxncpy): Same.
8050 (handle_builtin_strcat): Same.
8052 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8054 * match.pd: Optimise pointer range checks.
8056 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8058 PR tree-optimization/86758
8059 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
8060 to remove pattern statements.
8062 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8064 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
8065 result of dfs_enumerate_from when constructing stmt_vec_infos,
8066 instead of additionally calling get_loop_body.
8068 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8070 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
8072 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
8073 When creating an iv, assert that the step is not known to be zero.
8074 (vect_setup_realignment): Update call accordingly.
8075 * tree-vect-stmts.c (vectorizable_store): Likewise.
8076 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
8078 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8080 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
8081 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
8082 (vectorizable_reduction): Likewise.
8083 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8084 (vect_detect_hybrid_slp_stmts): Likewise.
8085 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
8087 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
8089 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
8090 (wide_int_set_zero_nonzero_bits): ...this.
8091 (zero_nonzero_bits_from_vr): Rename to...
8092 (vrp_set_zero_nonzero_bits): ...this.
8093 (extract_range_from_multiplicative_op_1): Abstract wide int
8095 (wide_int_range_multiplicative_op): ...here.
8096 (extract_range_from_binary_expr_1): Extract wide int binary
8097 operations into their own functions.
8098 (wide_int_range_lshift): New.
8099 (wide_int_range_can_optimize_bit_op): New.
8100 (wide_int_range_shift_undefined_p): New.
8101 (wide_int_range_bit_xor): New.
8102 (wide_int_range_bit_ior): New.
8103 (wide_int_range_bit_and): New.
8104 (wide_int_range_trunc_mod): New.
8105 (extract_range_into_wide_ints): New.
8106 (vrp_shift_undefined_p): New.
8107 (extract_range_from_multiplicative_op): New.
8108 (vrp_can_optimize_bit_op): New.
8109 * tree-vrp.h (value_range::dump): New.
8110 (wide_int_range_multiplicative_op): New.
8111 (wide_int_range_lshift):New.
8112 (wide_int_range_shift_undefined_p): New.
8113 (wide_int_range_bit_xor): New.
8114 (wide_int_range_bit_ior): New.
8115 (wide_int_range_bit_and): New.
8116 (wide_int_range_trunc_mod): New.
8117 (zero_nonzero_bits_from_bounds): Rename to...
8118 (wide_int_set_zero_nonzero_bits): ...this.
8119 (zero_nonzero_bits_from_vr): Rename to...
8120 (vrp_set_zero_nonzero_bits): ...this.
8121 (range_easy_mask_min_max): Rename to...
8122 (wide_int_range_can_optimize_bit_op): this.
8123 * vr-values.c (simplify_bit_ops_using_ranges): Rename
8124 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
8126 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8128 * tree-vectorizer.h (vect_orig_stmt): New function.
8129 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
8130 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
8131 (vect_create_epilog_for_reduction): Likewise.
8132 (vectorizable_live_operation): Likewise.
8133 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8134 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
8135 * tree-vect-stmts.c (vectorizable_call): Likewise.
8136 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
8138 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8140 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
8142 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
8143 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
8144 (vect_transform_loop): Likewise.
8145 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8147 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8149 * tree-vectorizer.h (vect_schedule_slp): Return void.
8150 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8151 (vect_schedule_slp): Likewise.
8153 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8155 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
8157 (vect_transform_loop): Update calls accordingly. Schedule SLP
8158 instances before the main loop, if any exist.
8160 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8162 PR tree-optimization/86749
8163 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
8164 If the lhs is used in a COND_EXPR, check that it is being used
8165 as the "then" or "else" value.
8167 2018-08-01 Tom de Vries <tdevries@suse.de>
8170 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
8171 speculation_safe_value_not_needed.
8173 2018-08-01 Richard Biener <rguenther@suse.de>
8175 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
8178 2018-08-01 Martin Liska <mliska@suse.cz>
8180 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
8181 format how successful transformation is dumped.
8182 (gimple_mod_pow2_value_transform): Likewise.
8183 (gimple_mod_subtract_transform): Likewise.
8184 (gimple_stringops_transform): Likewise.
8186 2018-08-01 Martin Liska <mliska@suse.cz>
8189 * value-prof.c (interesting_stringop_to_profile_p):
8190 Simplify the code and add BUILT_IN_MEMMOVE.
8191 (gimple_stringops_transform): Likewise.
8193 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
8195 * config/aarch64/aarch64-simd.md
8196 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
8197 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
8198 use GPI iterator instead of SI mode.
8200 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8202 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
8203 rs6000_speculation_barrier.
8204 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
8205 new barrier pattern name.
8207 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8209 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
8210 (speculation_barrier): New insn.
8212 2018-08-01 Richard Biener <rguenther@suse.de>
8215 * graphite.h: Include isl/id.h and isl/space.h to allow build
8218 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
8221 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
8222 mode for COFF targets.
8223 * defaults.h (TARGET_COFF): Define.
8224 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
8225 TARGET_COFF): Define.
8226 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
8227 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
8228 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
8230 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8232 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
8233 * gimple-streamer-out.c (output_bb): Save it.
8234 * lto-streamer-in.c (input_struct_function_base): Restore
8235 instance discriminator if available. Create map on demand.
8236 * lto-streamer-out.c (output_struct_function_base): Save it if
8238 * final.c (decl_to_instance_map): Document LTO strategy.
8240 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8241 Olivier Hainque <hainque@adacore.com>
8243 * debug.h (decl_to_instance_map_t): New type.
8244 (decl_to_instance_map): Declare.
8245 (maybe_create_decl_to_instance_map): New inline function.
8246 * final.c (bb_discriminator, last_bb_discriminator): New statics,
8247 to track basic block discriminators.
8248 (final_start_function_1): Initialize them.
8249 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
8251 (decl_to_instance_map): New variable.
8252 (map_decl_to_instance, maybe_set_discriminator): New functions.
8253 (notice_source_line): Set discriminator.
8255 2018-07-31 Ian Lance Taylor <iant@golang.org>
8257 * targhooks.c (default_have_speculation_safe_value): Add
8260 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8262 * dump-context.h: Include "pretty-print.h".
8263 (dump_context::refresh_dumps_are_enabled): New decl.
8264 (dump_context::emit_item): New decl.
8265 (class dump_context): Add fields "m_test_pp" and
8267 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8268 (temp_dump_context::get_dumped_text): New decl.
8269 (class temp_dump_context): Add field "m_pp".
8270 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
8271 (dump_context::refresh_dumps_are_enabled): ...and add a test for
8273 (set_dump_file): Update for above change.
8274 (set_alt_dump_file): Likewise.
8275 (dump_loc): New overload, taking a pretty_printer *.
8276 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
8277 to any test pretty-printer.
8278 (make_item_for_dump_gimple_stmt): New function, adapted from
8279 optinfo::add_gimple_stmt.
8280 (dump_context::dump_gimple_stmt): Call it, and use the result,
8281 eliminating the direct usage of dump_file and alt_dump_file in
8282 favor of indirectly using them via emit_item.
8283 (make_item_for_dump_gimple_expr): New function, adapted from
8284 optinfo::add_gimple_expr.
8285 (dump_context::dump_gimple_expr): Call it, and use the result,
8286 eliminating the direct usage of dump_file and alt_dump_file in
8287 favor of indirectly using them via emit_item.
8288 (make_item_for_dump_generic_expr): New function, adapted from
8290 (dump_context::dump_generic_expr): Call it, and use the result,
8291 eliminating the direct usage of dump_file and alt_dump_file in
8292 favor of indirectly using them via emit_item.
8293 (make_item_for_dump_printf_va): New function, adapted from
8294 optinfo::add_printf_va.
8295 (make_item_for_dump_printf): New function.
8296 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
8297 and use the result, eliminating the direct usage of dump_file and
8298 alt_dump_file in favor of indirectly using them via emit_item.
8299 (make_item_for_dump_dec): New function.
8300 (dump_context::dump_dec): Call it, and use the result,
8301 eliminating the direct usage of dump_file and alt_dump_file in
8302 favor of indirectly using them via emit_item.
8303 (make_item_for_dump_symtab_node): New function, adapted from
8304 optinfo::add_symtab_node.
8305 (dump_context::dump_symtab_node): Call it, and use the result,
8306 eliminating the direct usage of dump_file and alt_dump_file in
8307 favor of indirectly using them via emit_item.
8308 (dump_context::begin_scope): Reimplement, avoiding direct usage
8309 of dump_file and alt_dump_file in favor of indirectly using them
8311 (dump_context::emit_item): New member function.
8312 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8313 Set up test pretty-printer on the underlying context. Call
8314 refresh_dumps_are_enabled.
8315 (temp_dump_context::~temp_dump_context): Call
8316 refresh_dumps_are_enabled.
8317 (temp_dump_context::get_dumped_text): New member function.
8318 (selftest::verify_dumped_text): New function.
8319 (ASSERT_DUMPED_TEXT_EQ): New macro.
8320 (selftest::test_capture_of_dump_calls): Run all tests twice, with
8321 and then without optinfo enabled. Add uses of
8322 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
8323 * dumpfile.h: Update comment for the dump_* API.
8324 * optinfo-emit-json.cc
8325 (selftest::test_building_json_from_dump_calls): Update for new
8326 param for temp_dump_context ctor.
8327 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
8328 and "m_owned" field.
8329 (optinfo_item::~optinfo_item): Likewise.
8330 (optinfo::add_item): New member function.
8331 (optinfo::emit): Update comment.
8332 (optinfo::add_string): Delete.
8333 (optinfo::add_printf): Delete.
8334 (optinfo::add_printf_va): Delete.
8335 (optinfo::add_gimple_stmt): Delete.
8336 (optinfo::add_gimple_expr): Delete.
8337 (optinfo::add_tree): Delete.
8338 (optinfo::add_symtab_node): Delete.
8339 (optinfo::add_dec): Delete.
8340 * optinfo.h (class dump_context): New forward decl.
8341 (optinfo::add_item): New decl.
8342 (optinfo::add_string): Delete.
8343 (optinfo::add_printf): Delete.
8344 (optinfo::add_printf_va): Delete.
8345 (optinfo::add_gimple_stmt): Delete.
8346 (optinfo::add_gimple_expr): Delete.
8347 (optinfo::add_tree): Delete.
8348 (optinfo::add_symtab_node): Delete.
8349 (optinfo::add_dec): Delete.
8350 (optinfo::add_poly_int): Delete.
8351 (optinfo_item::optinfo_item): Remove "owned" param.
8352 (class optinfo_item): Remove field "m_owned".
8354 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8357 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
8358 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
8359 requested variable alignment.
8360 (expand_one_ssa_partition): Likewise.
8361 (expand_one_var): Likewise.
8363 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8365 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8366 to speculation_safe_value_not_needed.
8368 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8370 * targhooks.h (speculation_safe_value_not_needed): New prototype.
8371 * targhooks.c (speculation_safe_value_not_needed): New function.
8372 * target.def (have_speculation_safe_value): Update documentation.
8373 * doc/tm.texi: Regenerated.
8375 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8377 * config/aarch64/iterators.md (ALLI_TI): New iterator.
8378 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
8380 (despeculate_copy<ALLI:mode>_insn): New insn.
8381 (despeculate_copyti_insn): New insn.
8382 (despeculate_simple<ALLI:mode>): New insn
8383 (despeculate_simpleti): New insn.
8384 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
8386 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
8387 aarch64_speculation_safe_value.
8388 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
8390 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8392 * config/aarch64/aarch64-speculation.cc: New file.
8393 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
8394 before pass_reorder_blocks.
8395 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
8397 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
8398 X14 and X15 when tracking speculation.
8399 * config/aarch64/aarch64.md (register name constants): Add
8400 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
8401 (unspec): Add UNSPEC_SPECULATION_TRACKER.
8402 (speculation_barrier): New insn attribute.
8403 (cmp<mode>): Allow SP in comparisons.
8404 (speculation_tracker): New insn.
8405 (speculation_barrier): Add speculation_barrier attribute.
8406 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
8407 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
8408 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
8410 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8412 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
8413 aarch64_track_speculation is true.
8414 (tb<optab><mode>1): Likewise.
8415 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
8416 generate CB[N]Z when tracking speculation.
8417 (aarch64_split_compare_and_swap): Likewise.
8418 (aarch64_split_atomic_op): Likewise.
8420 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8422 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
8424 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8426 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
8427 (speculation_barrier): New insn.
8429 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8431 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
8432 * config/arm/arm.md (speculation_barrier): New expand.
8433 (speculation_barrier_insn): New pattern.
8435 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8437 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
8438 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
8439 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
8440 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
8442 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
8443 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
8444 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
8445 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
8446 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
8447 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
8448 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
8449 * builtins.c (expand_speculation_safe_value): New function.
8450 (expand_builtin): Call it.
8451 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
8452 * doc/extend.texi: Document __builtin_speculation_safe_value.
8453 * doc/md.texi: Document "speculation_barrier" pattern.
8454 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
8455 TARGET_HAVE_SPECULATION_SAFE_VALUE.
8456 * doc/tm.texi: Regenerated.
8457 * target.def (have_speculation_safe_value, speculation_safe_value): New
8459 * targhooks.c (default_have_speculation_safe_value): New function.
8460 (default_speculation_safe_value): New function.
8461 * targhooks.h (default_have_speculation_safe_value): Add prototype.
8462 (default_speculation_safe_value): Add prototype.
8464 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8466 * dump-context.h (dump_context::dump_loc): New decl.
8467 * dumpfile.c (dump_context::dump_loc): New member function.
8468 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
8469 and dump_gimple_stmt.
8470 (dump_context::dump_gimple_expr_loc): Likewise, using
8472 (dump_context::dump_generic_expr_loc): Likewise, using
8474 (dump_context::dump_printf_loc_va): Likewise, using
8476 (dump_context::begin_scope): Explicitly using the global function
8477 "dump_loc", rather than the member function.
8479 2018-07-31 Martin Sebor <msebor@redhat.com>
8481 PR tree-optimization/86741
8482 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
8484 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
8486 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
8487 depend on whether prefetch instructions will be emitted or not.
8488 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
8489 will be emitted or not.
8490 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
8491 (TARGET_SETMEM_PFD): New macros.
8493 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8495 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
8496 (NULL_STMT_VEC_INFO): Delete.
8497 (stmt_vec_info::operator*): Likewise.
8498 (stmt_vec_info::operator gimple *): Likewise.
8499 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
8500 of NULL_STMT_VEC_INFO.
8501 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8502 (vect_reassociating_reduction_p): Likewise.
8503 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8504 (vectorizable_store): Likewise.
8505 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
8506 (vec_info::free_stmt_vec_infos): Likewise.
8508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8510 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
8511 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
8513 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8515 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
8516 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8517 (vec_info::free_stmt_vec_info): New private member functions.
8518 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
8519 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
8520 * tree-parloops.c (gather_scalar_reductions): Remove calls to
8521 set_stmt_vec_info_vec and free_stmt_vec_infos.
8522 * tree-vect-loop.c (_loop_vec_info): Remove call to
8523 set_stmt_vec_info_vec.
8524 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
8525 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
8526 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
8527 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8528 (vec_info::free_stmt_vec_info): ...these new functions. Remove
8529 assignments in {vec_info::,}new_stmt_vec_info that are redundant
8530 with the clearing in the xcalloc.
8531 (stmt_vec_info_vec): Delete.
8532 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
8533 (vectorize_loops): Likewise.
8534 (vec_info::~vec_info): Remove argument from call to
8535 free_stmt_vec_infos.
8536 (vec_info::add_stmt): Remove vinfo argument from call to
8539 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8541 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
8542 rather than a gimple stmt.
8543 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
8544 information for pattern statements when passed the original
8545 statement; instead wait to be passed the pattern statement itself.
8546 Don't call set_vinfo_for_stmt here.
8547 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
8548 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
8549 stmt_vec_infos here.
8550 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8551 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8552 stmt_vec_infos entry.
8554 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8556 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8557 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8558 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8559 * tree-vect-stmts.c (vectorizable_call): Likewise.
8560 (vectorizable_simd_clone_call): Likewise.
8562 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8564 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8565 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8566 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8567 * tree-vect-loop.c (vect_transform_loop): Likewise.
8568 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8569 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8571 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8573 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8574 (vect_dr_stmt): Delete.
8575 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8576 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8578 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8579 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8580 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8581 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8582 (vect_analyze_data_refs): Likewise.
8583 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8585 (vect_find_same_alignment_drs): Likewise.
8586 (vect_slp_analyze_node_dependences): Update calls accordingly.
8587 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8588 instead of DR_VECT_AUX.
8589 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8590 of a vector data references. Use vec_info::lookup_dr instead of
8592 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8593 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8594 instead of DR_VECT_AUX.
8596 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8598 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8600 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8602 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8603 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8604 (vect_gen_prolog_loop_niters): Likewise.
8606 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8608 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8609 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8610 (vect_known_alignment_in_bytes, vect_dr_behavior)
8611 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8612 instead of data_references. Update calls to other routines for
8613 which the same change has been made.
8614 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8615 dr_vec_infos instead of stmt_vec_infos.
8616 (vect_analyze_data_ref_dependence): Update call accordingly.
8617 (vect_slp_analyze_data_ref_dependence)
8618 (vect_record_base_alignments): Use DR_VECT_AUX.
8619 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8620 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8621 (vector_alignment_reachable_p, vect_get_data_access_cost)
8622 (vect_peeling_supportable, vect_analyze_group_access_1)
8623 (vect_analyze_group_access, vect_analyze_data_ref_access)
8624 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8625 (vect_compile_time_alias, vect_small_gap_p)
8626 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8627 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8628 instead of data_references. Update calls to other routines for
8629 which the same change has been made.
8630 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8631 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8632 (vect_slp_analyze_and_verify_node_alignment)
8633 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8634 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8635 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8637 (_vect_peel_info::dr): Replace with...
8638 (_vect_peel_info::dr_info): ...this new field.
8639 (vect_peeling_hash_get_most_frequent)
8640 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8641 (vect_peeling_hash_get_lowest_cost):
8642 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8643 routines for which the same change has been made.
8644 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8646 * tree-vect-loop-manip.c (get_misalign_in_elems)
8647 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8649 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8650 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8651 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8652 (get_group_load_store_type, get_negative_load_store_type)
8653 (vect_get_data_ptr_increment, vectorizable_store)
8654 (vectorizable_load): Likewise.
8655 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8656 Update calls to other routines for which the same change has been made.
8658 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8660 * tree-vectorizer.h (vec_info::move_dr): New member function.
8661 (dataref_aux): Rename to...
8662 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8663 (_stmt_vec_info::dr_aux): Update accordingly.
8664 (_stmt_vec_info::data_ref_info): Delete.
8665 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8666 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8667 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8668 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8670 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8671 (STMT_VINFO_DR_INFO): New macro.
8672 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8673 (set_dr_misalignment): Update after rename of dataref_aux.
8674 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8675 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8676 initialization of STMT_VINFO_DATA_REF.
8677 * tree-vectorizer.c (vec_info::move_dr): New function.
8678 * tree-vect-patterns.c (vect_recog_bool_pattern)
8679 (vect_recog_mask_conversion_pattern)
8680 (vect_recog_gather_scatter_pattern): Use it.
8681 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8682 the "dr" and "stmt" fields of dr_vec_info instead of
8683 STMT_VINFO_DATA_REF.
8685 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8687 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8688 (is_pattern_stmt_p): Use it.
8689 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8690 on pattern statements.
8692 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8694 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8695 original stmt as a stmt_vec_info rather than a gimple stmt.
8696 (vect_pattern_recog_1): Take the statement directly as a
8697 stmt_vec_info, rather than via a gimple_stmt_iterator.
8698 Update call to vect_mark_pattern_stmts.
8699 (vect_pattern_recog): Update calls accordingly.
8701 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8703 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8704 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8705 a vect_def_type for the first argument.
8706 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8707 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8708 operand if it isn't defined by a vectorized statement.
8709 (vect_build_gather_load_calls): Remove the mask_dt argument and
8710 update calls to vect_get_vec_def_for_stmt_copy.
8711 (vectorizable_bswap): Likewise the dt argument.
8712 (vectorizable_call): Update calls to vectorizable_bswap and
8713 vect_get_vec_def_for_stmt_copy.
8714 (vectorizable_simd_clone_call, vectorizable_assignment)
8715 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8716 (vectorizable_comparison): Update calls to
8717 vect_get_vec_def_for_stmt_copy.
8718 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8720 (vect_get_loop_based_defs): Remove dt argument and update call
8721 to vect_get_vec_def_for_stmt_copy.
8722 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8723 and vect_get_vec_def_for_stmt_copy.
8724 (vectorizable_load): Update calls to vect_build_gather_load_calls
8725 and vect_get_vec_def_for_stmt_copy.
8726 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8727 (vectorizable_reduction, vectorizable_live_operation): Update calls
8728 to vect_get_vec_def_for_stmt_copy.
8730 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8732 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8733 and gimple stmt arguments with a stmt_vec_info.
8734 (vect_record_base_alignments): Update calls accordingly.
8735 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8736 and gimple stmt arguments with a stmt_vec_info.
8737 (vect_build_slp_tree_1): Remove vinfo argument and update call
8738 to vect_record_max_nunits.
8739 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8740 and vect_record_max_nunits.
8742 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8744 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8745 file and take a stmt_vec_info instead of a gimple stmt.
8746 (supportable_widening_operation, vect_finish_replace_stmt)
8747 (vect_finish_stmt_generation, vect_get_store_rhs)
8748 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8749 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8750 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8751 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8752 (vect_create_data_ref_ptr, bump_vector_ptr)
8753 (vect_permute_store_chain, vect_setup_realignment)
8754 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8755 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8756 (vectorizable_reduction, vectorizable_induction)
8757 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8758 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8759 than gimple stmts as arguments.
8760 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8761 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8762 (can_group_stmts_p, vect_check_gather_scatter)
8763 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8764 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8765 (vect_permute_load_chain, vect_shift_permute_load_chain)
8766 (vect_transform_grouped_load)
8767 (vect_record_grouped_load_vectors): Likewise.
8768 * tree-vect-loop.c (vect_fixup_reduc_chain)
8769 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8770 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8771 (vectorizable_reduction, vectorizable_induction)
8772 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8773 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8774 (vect_get_load_store_mask): Likewise.
8775 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8776 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8777 * tree-vect-stmts.c (vect_mark_relevant)
8778 (is_simple_and_all_uses_invariant)
8779 (exist_non_indexing_operands_for_use_p, process_use)
8780 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8781 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8782 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8783 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8784 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8785 (get_negative_load_store_type, get_load_store_type)
8786 (vect_check_load_store_mask, vect_check_store_rhs)
8787 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8788 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8789 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8790 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8791 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8792 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8793 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8794 (supportable_widening_operation): Likewise.
8796 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8798 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8799 a stmt_vec_info instead of a gcall.
8800 (vect_check_gather_scatter): Update call accordingly.
8801 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8803 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8804 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8805 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8806 instead of a gimple stmt.
8807 (vect_transform_loop): Update calls accordingly.
8808 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8809 stmt_vec_infos instead of gimple stmts.
8810 (vect_analyze_slp_instance): Update use accordingly.
8811 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8812 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8813 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8814 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8815 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8816 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8817 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8818 down to subroutines.
8820 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8822 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8823 of the worklist from a vector of gimple stmts to a vector of
8825 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8826 (vect_mark_stmts_to_be_vectorized): Likewise
8828 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8830 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8831 statement before passing it to vect_analyze_stmt.
8832 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8833 the chain of phi vector definitions. Track the exit phi via its
8835 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8836 STMT_VINFO_REDUC_DEF.
8837 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8838 stmt_vec_infos to handle the statement chains.
8839 (vect_get_slp_defs): Record the first statement in the node
8840 using a stmt_vec_info.
8841 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8842 statements here and pass their stmt_vec_info down to subroutines.
8843 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8844 down to vect_finish_stmt_generation.
8845 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8846 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8847 stmt_vec_infos to subroutines.
8848 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8851 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8853 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8854 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8855 (vect_permute_store_chain, vect_setup_realignment)
8856 (vect_permute_load_chain, vect_shift_permute_load_chain)
8857 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8858 stmts internally, and when passing values to other vectorizer routines.
8859 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8860 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8861 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8862 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8863 (vectorizable_reduction, vectorizable_induction)
8864 (vectorizable_live_operation, vect_transform_loop_stmt)
8865 (vect_transform_loop): Likewise.
8866 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8867 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8868 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8869 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8870 (vect_slp_analyze_node_operations_1): Likewise.
8871 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8872 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8873 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8874 (vect_finish_stmt_generation_1, get_group_load_store_type)
8875 (get_load_store_type, vect_build_gather_load_calls)
8876 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8877 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8878 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8879 (vectorizable_store, vectorizable_load, vectorizable_condition)
8880 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8881 (supportable_widening_operation): Likewise.
8882 (vect_get_vector_types_for_stmt): Likewise.
8883 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8885 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8887 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8888 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8889 (vect_permute_store_chain, vect_permute_load_chain)
8890 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8891 repeated stmt_vec_info lookups.
8892 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8893 (vect_update_ivs_after_vectorizer): Likewise.
8894 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8895 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8896 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8897 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8898 (vect_bb_slp_scalar_cost): Likewise.
8899 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8901 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8903 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8904 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8905 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8906 to get gassigns and gcalls, rather than operating on generc gimple
8908 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8909 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8910 (vectorizable_load, vect_analyze_stmt): Likewise.
8911 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8913 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8915 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8916 return stmt_vec_infos rather than gimple stmts. Do not accept
8918 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8920 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8921 Update use of get_later_stmt.
8922 (vect_get_constant_vectors): Update call accordingly.
8923 (vect_schedule_slp_instance): Likewise
8924 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8925 (vect_slp_analyze_instance_dependence): Likewise.
8926 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8928 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8930 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8931 (stmt_info_for_cost::stmt_info): ...this new field.
8932 (add_stmt_costs): Update accordingly.
8933 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8934 (vect_get_known_peeling_cost): Likewise.
8935 (vect_estimate_min_profitable_iters): Likewise.
8936 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8938 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8940 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8941 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8942 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8944 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8946 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8948 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8949 a gimple stmt to a stmt_vec_info.
8950 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8952 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8954 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8955 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8956 (_loop_vec_info::reduction_chains): Likewise.
8957 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8959 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8961 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8963 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8964 a gimple stmt to a stmt_vec_info.
8965 (_stmt_vec_info::next_element): Likewise.
8966 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8967 (vect_slp_analyze_and_verify_node_alignment)
8968 (vect_analyze_group_access_1, vect_analyze_group_access)
8969 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8970 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8971 (vect_supportable_dr_alignment): Update accordingly.
8972 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8973 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8974 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8975 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8976 * tree-vect-slp.c (vect_build_slp_tree_1)
8977 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8978 (vect_split_slp_store_group, vect_analyze_slp_instance)
8979 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8980 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8981 (get_group_load_store_type, get_load_store_type)
8982 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8983 (vect_transform_stmt, vect_remove_stores): Likewise.
8985 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8987 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8989 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8990 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8991 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8992 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8993 (vector_alignment_reachable_p, vect_get_data_access_cost)
8994 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8995 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8996 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8997 (vect_analyze_group_access_1, vect_analyze_group_access)
8998 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8999 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
9000 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
9001 result of vect_dr_stmt and use the stmt_vec_info instead of
9002 the associated gimple stmt.
9003 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
9004 (vect_gen_prolog_loop_niters): Likewise.
9005 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
9007 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9009 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
9010 to a vec<stmt_vec_info>.
9011 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
9012 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
9014 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
9015 to a vec<stmt_vec_info>.
9016 (bst_traits::value_type, bst_traits::value_type): Likewise.
9017 (bst_traits::hash): Update accordingly.
9018 (vect_get_and_check_slp_defs): Change the stmts parameter from
9019 a vec<gimple *> to a vec<stmt_vec_info>.
9020 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
9021 (vect_build_slp_tree): Likewise.
9022 (vect_build_slp_tree_2): Likewise. Update uses of
9023 SLP_TREE_SCALAR_STMTS.
9024 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
9025 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
9026 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
9027 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
9028 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
9029 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
9030 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
9031 (vect_get_constant_vectors, vect_get_slp_defs)
9032 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
9033 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
9034 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
9035 instead of gimple stmts.
9036 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
9037 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
9038 (vect_slp_analyze_instance_dependence): Update uses of
9039 SLP_TREE_SCALAR_STMTS.
9040 (vect_slp_analyze_and_verify_node_alignment): Likewise.
9041 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
9042 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
9043 (get_initial_defs_for_reduction): Likewise.
9044 (vect_create_epilog_for_reduction): Likewise.
9045 (vectorize_fold_left_reduction): Likewise.
9046 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
9047 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
9048 (can_vectorize_live_stmts): Likewise.
9050 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9052 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
9053 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
9054 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
9056 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
9058 (gather_scalar_reductions): Update after above interface changes.
9059 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
9060 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
9062 (vect_force_simple_reduction): Likewise.
9063 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
9064 LOOP_VINFO_REDUCTIONS.
9065 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9067 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9069 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
9070 a gimple stmt to a stmt_vec_info.
9071 * tree-vect-loop.c (vect_active_double_reduction_p)
9072 (vect_force_simple_reduction, vectorizable_reduction): Update
9075 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9077 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
9078 vec<gimple *> to a vec<stmt_vec_info>.
9079 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
9080 the reduction_phis argument from a vec<gimple *> to a
9082 (vectorizable_reduction): Likewise the phis local variable that
9083 is passed to vect_create_epilog_for_reduction. Update for new type
9084 of SLP_TREE_VEC_STMTS.
9085 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
9086 (vectorizable_live_operation): Likewise.
9087 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
9088 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
9090 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9092 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
9093 a gimple stmt to a stmt_vec_info.
9094 (vectorizable_condition, vectorizable_live_operation)
9095 (vectorizable_reduction, vectorizable_induction): Pass back the
9096 vectorized statement as a stmt_vec_info.
9097 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
9098 use of STMT_VINFO_VEC_STMT.
9099 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
9100 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
9101 as stmt_vec_infos rather than gimple stmts.
9102 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
9104 (vectorizable_live_operation): Likewise.
9105 (vectorizable_reduction, vectorizable_induction): Likewise,
9106 updating use of STMT_VINFO_VEC_STMT.
9107 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
9108 of STMT_VINFO_VEC_STMT.
9109 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
9110 (vectorizable_simd_clone_call, vectorizable_conversion)
9111 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9112 (vectorizable_store, vectorizable_load, vectorizable_condition)
9113 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
9114 from a gimple stmt to a stmt_vec_info.
9115 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
9116 pointer to a stmt_vec_info to the vectorizable_* routines.
9118 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9120 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
9121 a gimple stmt to a stmt_vec_info.
9122 (is_pattern_stmt_p): Update accordingly.
9123 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
9124 (vect_record_grouped_load_vectors): Likewise.
9125 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
9126 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
9127 (vect_model_reduction_cost): Likewise.
9128 (vect_create_epilog_for_reduction): Likewise.
9129 (vectorizable_reduction, vectorizable_induction): Likewise.
9130 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9131 Return the stmt_vec_info for the pattern statement.
9132 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
9133 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
9134 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9135 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
9136 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
9137 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
9138 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
9139 (free_stmt_vec_info, vect_is_simple_use): Likewise.
9141 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9143 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
9144 (vect_finish_stmt_generation): Likewise.
9145 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9146 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
9147 (vect_build_gather_load_calls): Use the return value of the above
9148 functions instead of a separate call to vinfo_for_stmt. Use narrow
9149 scopes for the input gimple stmt and wider scopes for the associated
9150 stmt_vec_info. Use vec_info::lookup_def when setting these
9151 stmt_vec_infos from an SSA_NAME definition.
9152 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
9153 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
9154 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9155 (vectorizable_store, vectorizable_load, vectorizable_condition)
9156 (vectorizable_comparison): Likewise.
9157 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
9158 (vectorizable_reduction): Likewise.
9160 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9162 * tree-vectorizer.h (vect_is_simple_use): Add an optional
9163 stmt_vec_info * parameter before the optional gimple **.
9164 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9165 (process_use, vect_get_vec_def_for_operand_1): Update callers.
9166 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
9167 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9168 (vectorizable_live_operation): Likewise.
9169 * tree-vect-patterns.c (type_conversion_p): Likewise.
9170 (vect_look_through_possible_promotion): Likewise.
9171 (vect_recog_rotate_pattern): Likewise.
9172 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
9174 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9176 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
9177 a typedef to a wrapper class.
9178 (NULL_STMT_VEC_INFO): New macro.
9179 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
9180 (stmt_vec_info::operator*): New function.
9181 (stmt_vec_info::operator gimple *): Likewise.
9182 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
9183 (add_stmt_costs): Likewise.
9184 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
9185 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
9186 (vect_get_known_peeling_cost): Likewise.
9187 (vect_estimate_min_profitable_iters): Likewise.
9188 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9189 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
9190 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
9191 (vectorizable_store, free_stmt_vec_infos): Likewise.
9192 (new_stmt_vec_info): Change return type of xcalloc to
9195 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9197 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
9198 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
9199 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
9200 a single_imm_use-based sequence.
9201 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
9203 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9205 * tree-vectorizer.h (vec_info::lookup_def): Declare.
9206 * tree-vectorizer.c (vec_info::lookup_def): New function.
9207 * tree-vect-patterns.c (vect_get_internal_def): Use it.
9208 (vect_widened_op_tree): Likewise.
9209 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9210 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
9211 (vectorizable_reduction): Likewise.
9212 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
9214 (vect_is_slp_reduction): Update calls accordingly. Use
9215 vec_info::lookup_def.
9216 (vect_is_simple_reduction): Likewise
9217 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
9219 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9221 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
9222 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
9223 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
9225 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
9226 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
9227 (vect_update_vf_for_slp, vect_analyze_loop_operations)
9228 (vect_is_slp_reduction, vectorizable_induction)
9229 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
9230 * tree-vect-patterns.c (vect_init_pattern_stmt):
9231 (vect_determine_min_output_precision_1, vect_determine_precisions)
9232 (vect_pattern_recog): Likewise.
9233 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
9234 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
9235 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
9236 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9237 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
9238 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
9239 info field from a loop to a loop_vec_info.
9241 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9243 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
9244 (vec_info::add_stmt): Declare.
9245 * tree-vectorizer.c (vec_info::add_stmt): New function.
9246 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
9247 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
9248 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
9249 (vectorizable_induction): Likewise.
9250 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
9251 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9252 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
9253 (vectorizable_load): Likewise.
9254 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9255 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
9256 (vect_recog_gather_scatter_pattern): Likewise.
9257 (append_pattern_def_seq): Likewise. Remove a check that is
9258 performed by add_stmt itself.
9260 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9262 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
9263 which make_ssa_name was called with new_stmt before new_stmt
9266 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9268 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
9270 (vect_is_slp_reduction): ...here...
9271 (vect_is_simple_reduction): ...and here. Remove repetition of tests
9272 that are already known to be false.
9274 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9276 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
9277 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
9278 STMT_VINFO_NUM_SLP_USES when it's true.
9279 (vect_free_slp_instance): Add a final_p parameter and pass it to
9281 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
9282 (vect_analyze_slp_instance): Likewise.
9283 (vect_slp_analyze_operations): Likewise.
9284 (vect_slp_analyze_bb_1): Likewise.
9285 * tree-vectorizer.c (vec_info): Likewise.
9286 * tree-vect-loop.c (vect_transform_loop): Likewise.
9288 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9290 * tree-vect-loop.c (vectorizable_reduction): Assert that the
9291 function is not called for second and subsequent members of
9294 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9296 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
9297 cases for nested loops from here to ...
9298 (vect_create_epilog_for_reduction): ...here. Only call
9299 vect_is_simple_use for inner-loop reductions.
9301 2018-07-31 Martin Liska <mliska@suse.cz>
9303 PR gcov-profile/85338
9304 PR gcov-profile/85350
9305 PR gcov-profile/85372
9306 * profile.c (struct location_triplet): New.
9307 (struct location_triplet_hash): Likewise.
9308 (output_location): Do not output a BB that
9309 is already recorded for a line.
9310 (branch_prob): Use streamed_locations.
9312 2018-07-31 Martin Liska <mliska@suse.cz>
9314 PR gcov-profile/85370
9315 * coverage.c (coverage_begin_function): Do not mark target
9316 clones as artificial functions.
9318 2018-07-31 Martin Liska <mliska@suse.cz>
9320 PR gcov-profile/83813
9321 PR gcov-profile/84758
9322 PR gcov-profile/85217
9323 PR gcov-profile/85332
9324 * profile.c (branch_prob): Do not record GOTO expressions
9325 for GIMPLE statements which locations are already streamed.
9327 2018-07-31 Olivier Hainque <hainque@adacore.com>
9329 * gcc.c (handle_spec_function): Accept a soft_matched_part
9330 argument, as do_spec_1. Pass it down to ...
9331 (eval_spec_function): Accept a soft_matched_part argument,
9332 and pass it down to ...
9333 (do_spec_2): Accept a soft_matched_part argument, and pass
9334 it down to do_spec_1.
9335 (do_spec_1): Pass soft_matched_part to handle_spec_function.
9336 (handle_braces): Update call to handle_spec_function.
9337 (driver::set_up_specs): Update calls to do_spec_2.
9338 (compare_debug_dump_opt_spec_function): Likewise.
9339 (compare_debug_self_opt_spec_function): Likewise.
9341 2018-07-31 Olivier Hainque <hainque@adacore.com>
9343 * common.opt (nolibc): New option.
9344 * doc/invoke.texi (Link Options): Document it.
9345 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
9346 * config/alpha/linux.h: Likewise.
9347 * config/arc/elf.h: Likewise.
9348 * config/arm/uclinux-elf.h: Likewise.
9349 * config/arm/unknown-elf.h: Likewise.
9350 * config/avr/avrlibc.h: Likewise.
9351 * config/bfin/bfin.h: Likewise.
9352 * config/bfin/linux.h: Likewise.
9353 * config/bfin/uclinux.h: Likewise.
9354 * config/darwin.h: Likewise.
9355 * config/darwin10.h: Likewise.
9356 * config/darwin12.h: Likewise.
9357 * config/gnu-user.h: Likewise.
9358 * config/lm32/uclinux-elf.h: Likewise.
9359 * config/pa/pa-hpux11.h: Likewise.
9360 * config/pa/pa64-hpux.h: Likewise.
9361 * config/sparc/sparc.h: Likewise.
9363 2018-07-31 Olivier Hainque <hainque@adacore.com>
9365 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
9366 undefined variables.
9368 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9371 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
9374 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9376 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9377 terminated string literal.
9379 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9381 PR rtl-optimization/85160
9382 * combine.c (is_just_move): New function.
9383 (try_combine): Allow combining two instructions into two if neither of
9384 the original instructions was a move.
9386 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
9389 * doc/extend.texi (Global Register Variables): Discourage use of type
9391 (Local Register Variables): Likewise.
9393 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
9395 PR tree-optimization/86506
9396 * hwint.h (ceil_log2): Resync with hwint.c implementation.
9398 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
9401 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
9402 hard_regno, make sure no insn between `from` and `to` clobbers it.
9404 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
9405 Tom de Vries <tdevries@suse.de>
9407 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
9408 (PTX_DEFAULT_RUNTIME_DIM): ... this.
9409 (nvptx_goacc_validate_dims): Set default worker and gang dims to
9410 PTX_DEFAULT_RUNTIME_DIM.
9411 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
9413 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
9415 * config/pa/pa.c (pa_output_addr_vec): Align address table.
9416 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
9417 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
9419 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
9421 * config/rs6000/constraints.md (wG constraint): Delete, no longer
9423 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
9424 predicate to reflect toc fusion has been deleted.
9425 (toc_fusion_mem_raw): Delete, no longer used.
9426 (toc_fusion_mem_wrapped): Likewise.
9427 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
9429 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
9430 Delete, no longer used.
9431 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
9432 meant to be used for toc fusion.
9433 (rs6000_debug_print_mode): Delete toc fusion debugging.
9434 (rs6000_debug_reg_global): Likewise.
9435 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
9436 fusion and secondary reload support that were never used.
9437 (rs6000_option_override_internal): Delete TOC fusion, that was only
9438 partially defined, and it did not work unless you also used the
9440 (rs6000_legitimate_address_p): Delete TOC fusion support.
9441 (rs6000_opt_masks): Likewise.
9442 (fusion_wrap_memory_address): Delete function, no longer used.
9443 (fusion_split_address); Delete TOC fusion support.
9444 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
9445 longer used with toc fusion being deleted.
9446 (TARGET_TOC_FUSION_FP): Likewise.
9447 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
9449 (toc fusion spliter): Delete TOC fusion support.
9450 (toc_fusionload_<mode>): Likewise.
9451 (toc_fusionload_di): Likewise.
9452 (fusion_gpr_load_<mode>): Delete generator function, this insn no
9453 longer needs to be named. Rename predicate to delete TOC fusion.
9454 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9455 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9456 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9457 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9458 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
9460 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9462 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
9463 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
9464 __int128 in built-in function prototypes.
9465 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
9466 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
9468 2018-07-27 Martin Sebor <msebor@redhat.com>
9470 PR tree-optimization/86696
9471 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
9472 types, including enums.
9473 (handle_char_store): Be prepared for the above function to fail.
9475 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
9477 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
9478 when optimization level is lower than 2 or optimize for size.
9480 2018-07-26 Martin Sebor <msebor@redhat.com>
9482 PR tree-optimization/86043
9483 PR tree-optimization/86042
9484 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
9485 (get_string_cst_length): Rename...
9486 (get_min_string_length): ...to this. Add argument.
9487 (handle_char_store): Extend to handle multi-character stores by
9489 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
9490 * tree.h (initializer_zerop): Add argument.
9492 2018-07-26 Jakub Jelinek <jakub@redhat.com>
9495 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
9496 declare target to variables if they have always,{to,from,tofrom} map
9499 2018-07-26 Martin Liska <mliska@suse.cz>
9502 * lto-wrapper.c: Add linker_output as prefix
9503 for ltrans_output_file.
9505 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
9507 PR rtl-optimization/85805
9508 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
9509 value for hard registers if that was written in the same mode.
9511 2018-07-26 Martin Liska <mliska@suse.cz>
9513 PR gcov-profile/86536
9514 * gcov.c (format_gcov): Use printf format %.*f directly
9515 and do not handle special values.
9517 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9519 * common/config/arc/arc-common.c (arc_option_optimization_table):
9520 Update default optimizations for size.
9522 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9524 * config/arc/arc.md (movsf_insn): Add short instruction selection.
9525 * config/arc/constraints.md (CfZ): New constraint.
9526 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
9527 (subsf3_fpu): Likewise.
9528 (cmpsf_fpu): Likewise.
9529 (cmpsf_fpu_uneq): Likewise.
9531 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9533 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
9535 (arc_is_uncached_mem_p): uncached applies to both the variable and
9538 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9540 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
9543 2018-07-25 David Malcolm <dmalcolm@redhat.com>
9545 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
9546 field "m_scopes" from vec to auto_vec.
9548 2018-07-25 Martin Liska <mliska@suse.cz>
9550 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9553 2018-07-25 Richard Biener <rguenther@suse.de>
9556 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9557 special wrt context_die late.
9558 (gen_subprogram_die): Re-use DIEs in local scope.
9560 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9562 PR tree-optimization/86644
9563 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9565 2018-07-25 Martin Liska <mliska@suse.cz>
9568 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9569 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9571 2018-07-25 Martin Liska <mliska@suse.cz>
9574 * params.def: Explain ASan abbreviation and provide
9575 a documentation link.
9577 2018-07-24 Martin Sebor <msebor@redhat.com>
9579 PR tree-optimization/86622
9580 PR tree-optimization/86532
9581 * builtins.h (string_length): Declare.
9582 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9583 (check_access): Be prepared for non-constant length ranges.
9584 (string_length): Make extern.
9585 * expr.c (string_constant): Only handle the minor non-constant
9586 array index. Use string_constant to compute the length of
9587 a generic string constant.
9589 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9591 PR tree-optimization/86618
9592 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9593 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9595 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9597 PR tree-optimization/86636
9598 * json.cc (json::object::set): Fix comment. Add assertions.
9599 (json::array::append): Move here from json.h. Add comment and an
9601 (json::string::string): Likewise.
9602 * json.h (json::array::append): Move to json.cc.
9603 (json::string::string): Likewise.
9604 * optinfo-emit-json.cc
9605 (optrecord_json_writer::impl_location_to_json): Assert that we
9606 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9607 wrapper around it. Expand the location once, rather than three
9609 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9610 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9612 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9613 places. Fix some overlong lines.
9615 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9617 * config/aarch64/aarch64-simd.md
9618 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9619 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9620 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9621 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9622 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9623 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9624 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9625 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9626 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9628 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9631 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9632 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9633 and abs_d == d, do the power of two handling if profitable.
9635 2018-07-24 Richard Biener <rguenther@suse.de>
9637 * match.pd: Add BIT_FIELD_REF canonicalizations.
9639 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9642 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9644 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9646 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9647 terminated STRING_CST object.
9649 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9651 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9653 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9655 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9656 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9657 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9658 the elements into a register.
9659 (rs6000_split_v4si_init_di_reg): Delete.
9660 (rs6000_split_v4si_init): Delete.
9661 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9662 (vsx_init_v4si): Rewrite as a define_expand.
9664 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9666 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9667 zero_extend argument from memory): New.
9669 2018-07-22 Martin Sebor <msebor@redhat.com>
9672 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9673 diagnosing calls with unknown arguments unless -Walloca-larger-than
9674 is restricted to less than PTRDIFF_MAX bytes.
9676 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9678 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9680 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9682 * pretty-print.c (text_info::set_location): Remove redundant
9683 "line_table" parameter from call to rich_location::set_range.
9685 2018-07-20 Martin Sebor <msebor@redhat.com>
9688 * builtins.c (expand_builtin_alloca): Adjust.
9689 * calls.c (alloc_max_size): Simplify.
9690 * cgraphunit.c (cgraph_node::expand): Adjust.
9691 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9693 (frame_larger_than_size): Same.
9694 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9695 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9697 * doc/invoke.texi (GCC Command Options): Document option arguments.
9698 Explain byte-size arguments and suffixes.
9699 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9700 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9701 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9702 * doc/options.texi (UInteger): Expand.
9703 (Host_Wide_Int, ByteSize): Document new properties.
9704 * final.c (final_start_function_1): Include sizes in an error message.
9705 * function.c (frame_offset_overflow): Same.
9706 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9707 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9708 Diagnose unbounded alloca calls only for limits of less than
9710 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9711 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9713 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9714 only for limits of less than PTRDIFF_MAX.
9715 * langhooks-def.h (lhd_handle_option): Change function argument
9717 * langhooks.c (lhd_handle_option): Same.
9718 * langhooks.h (handle_option): Same.
9719 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9721 (var_type, var_type_struct): Same.
9722 (var_set): Handle ByteSize flag.
9723 * optc-gen.awk: Add comments to output to ease debugging. Make
9724 use of HOST_WIDE_INT where appropriate.
9725 * opts-gen-save.awk: Use %lx to format unsigned long.
9726 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9727 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9728 arguments. Parse bytes-size suffixes.
9729 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9730 (enum_value_to_arg): Same.
9731 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9732 (handle_option): Adjust.
9733 (generate_option): Change function argument to HOST_WIDE_INT.
9734 (cmdline_handle_error): Adjust.
9735 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9736 (set_option): Change function argument to HOST_WIDE_INT.
9737 (option_enabled): Handle cl_host_wide_int.
9738 (get_option_state): Handle CLVC_SIZE.
9739 (control_warning_option): Same.
9740 * opts.c (common_handle_option): Change function argument to
9741 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9742 OPT_Wvla_larger_than_.
9743 * opts.h (enum cl_var_type): Add an enumerator.
9744 * stor-layout.c (layout_decl): Print a more meaningful warning.
9745 * toplev.c (output_stack_usage): Adjust.
9747 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9749 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9750 call to inline_expand_builtin_string_cmp.
9751 (expand_builtin_strcmp): Likewise.
9752 (expand_builtin_strncmp): Likewise.
9753 (inline_string_cmp): Delete the last parameter, change char_type_node
9754 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9756 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9757 the inlining expansion on target where the type of the call has same or
9758 narrower precision than unsigned char.
9760 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9762 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9763 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9764 * common.opt (fsave-optimization-record): New option.
9765 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9766 * doc/invoke.texi (-fsave-optimization-record): New option.
9767 * dumpfile.c: Include "optinfo-emit-json.h".
9768 (struct kv_pair): Move to coretypes.h.
9769 (optgroup_options): Make non-static.
9770 (dump_context::end_scope): Call
9771 optimization_records_maybe_pop_dump_scope.
9772 * dumpfile.h (optgroup_options): New decl.
9773 * json.cc: New file.
9775 * optinfo-emit-json.cc: New file.
9776 * optinfo-emit-json.h: New file.
9777 * optinfo.cc: Include "optinfo-emit-json.h".
9778 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9779 (optinfo_enabled_p): Check optimization_records_enabled_p.
9780 (optinfo_wants_inlining_info_p): Likewise.
9781 * optinfo.h: Update comment.
9782 * profile-count.c (profile_quality_as_string): New function.
9783 * profile-count.h (profile_quality_as_string): New decl.
9784 (profile_count::quality): New accessor.
9785 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9786 and optinfo_emit_json_cc_tests.
9787 * selftest.h (selftest::json_cc_tests): New decl.
9788 (selftest::optinfo_emit_json_cc_tests): New decl.
9789 * toplev.c: Include "optinfo-emit-json.h".
9790 (compile_file): Call optimization_records_finish.
9791 (do_compile): Call optimization_records_start.
9792 * tree-ssa-live.c: Include optinfo.h.
9793 (remove_unused_scope_block_p): Retain inlining information if
9794 optinfo_wants_inlining_info_p returns true.
9796 2018-07-20 Richard Biener <rguenther@suse.de>
9799 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9800 to cover -flto-partition=none.
9802 2018-07-20 Martin Liska <mliska@suse.cz>
9804 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9805 (get_decl_source_range): Remove unused function.
9807 2018-07-20 Richard Biener <rguenther@suse.de>
9809 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9810 (struct vn_phi_s): Likewise.
9811 (struct vn_reference_s): Likewise.
9812 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9813 for searching the slot of an entry known to be in the hash itself.
9814 (vn_phi_hasher::equal): Likewise.
9815 (vn_reference_hasher::equal): Likewise.
9816 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9818 (optimistic_info, current_info): Remove, keeping only valid_info.
9819 (vn_reference_lookup_1): Remove fallback lookup.
9820 (vn_reference_lookup_2): Likewise.
9821 (vn_nary_op_lookup_1): Likewise.
9822 (vn_phi_lookup): Likewise.
9823 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9825 (vn_reference_insert): Adjust, chain the inserted hash element
9826 at last_inserted_ref.
9827 (vn_reference_insert_pieces): Likewise.
9828 (visit_reference_op_call): Likewise.
9829 (vn_nary_op_insert_into): Chain the inserted hash element at
9831 (vn_nary_op_insert_pieces): Adjust.
9832 (vn_nary_op_insert): Likewise.
9833 (vn_nary_op_insert_stmt): Likewise.
9834 (vn_phi_insert): Adjust, chain the inserted hash element at
9836 (process_scc): Remove clearing and copying the optimistic
9837 table. Instead remove elements inserted during an optimistic
9838 iteration from the single table we maintain.
9839 (init_scc_vn): Adjust.
9840 (free_scc_vn): Likewise.
9841 (sccvn_dom_walker::record_cond): Likewise.
9842 (sccvn_dom_walker::after_dom_children): Likewise.
9844 2018-07-19 Martin Sebor <msebor@redhat.com>
9846 PR tree-optimization/84047
9847 PR tree-optimization/83776
9848 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9849 (check_array_bounds): Call it.
9851 2018-07-19 Martin Sebor <msebor@redhat.com>
9853 * align.h (align_flags): Use member initialization.
9855 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9857 * Makefile.in (OBJS): Add optinfo.o.
9858 * coretypes.h (class symtab_node): New forward decl.
9859 (struct cgraph_node): New forward decl.
9860 (class varpool_node): New forward decl.
9861 * dump-context.h: New file.
9862 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9864 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9865 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9866 (set_alt_dump_file): Likewise.
9867 (dump_context::~dump_context): New dtor.
9868 (dump_gimple_stmt): Move implementation to...
9869 (dump_context::dump_gimple_stmt): ...this new member function.
9870 Add the stmt to any pending optinfo, creating one if need be.
9871 (dump_gimple_stmt_loc): Move implementation to...
9872 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9873 Start a new optinfo and add the stmt to it.
9874 (dump_gimple_expr): Move implementation to...
9875 (dump_context::dump_gimple_expr): ...this new member function.
9876 Add the stmt to any pending optinfo, creating one if need be.
9877 (dump_gimple_expr_loc): Move implementation to...
9878 (dump_context::dump_gimple_expr_loc): ...this new member function.
9879 Start a new optinfo and add the stmt to it.
9880 (dump_generic_expr): Move implementation to...
9881 (dump_context::dump_generic_expr): ...this new member function.
9882 Add the tree to any pending optinfo, creating one if need be.
9883 (dump_generic_expr_loc): Move implementation to...
9884 (dump_context::dump_generic_expr_loc): ...this new member
9885 function. Add the tree to any pending optinfo, creating one if
9887 (dump_printf): Move implementation to...
9888 (dump_context::dump_printf_va): ...this new member function. Add
9889 the text to any pending optinfo, creating one if need be.
9890 (dump_printf_loc): Move implementation to...
9891 (dump_context::dump_printf_loc_va): ...this new member function.
9892 Start a new optinfo and add the stmt to it.
9893 (dump_dec): Move implementation to...
9894 (dump_context::dump_dec): ...this new member function. Add the
9895 value to any pending optinfo, creating one if need be.
9896 (dump_context::dump_symtab_node): New member function.
9897 (dump_context::get_scope_depth): New member function.
9898 (dump_context::begin_scope): New member function.
9899 (dump_context::end_scope): New member function.
9900 (dump_context::ensure_pending_optinfo): New member function.
9901 (dump_context::begin_next_optinfo): New member function.
9902 (dump_context::end_any_optinfo): New member function.
9903 (dump_context::s_current): New global.
9904 (dump_context::s_default): New global.
9905 (dump_scope_depth): Delete global.
9906 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9907 (dump_symtab_node): New function.
9908 (get_dump_scope_depth): Reimplement in terms of dump_context.
9909 (dump_begin_scope): Likewise.
9910 (dump_end_scope): Likewise.
9911 (selftest::temp_dump_context::temp_dump_context): New ctor.
9912 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9913 (selftest::verify_item): New function.
9914 (ASSERT_IS_TEXT): New macro.
9915 (ASSERT_IS_TREE): New macro.
9916 (ASSERT_IS_GIMPLE): New macro.
9917 (selftest::test_capture_of_dump_calls): New test.
9918 (selftest::dumpfile_c_tests): Call it.
9919 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9920 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9921 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9922 descriptive comment.
9923 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9924 (dump_node, dump_bb): Move these unrelated decls.
9925 (class dump_manager): Add leading comment.
9926 * optinfo.cc: New file.
9927 * optinfo.h: New file.
9929 2018-07-19 Michael Collison <michael.collison@arm.com>
9930 Richard Henderson <rth@redhat.com>
9932 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9933 (subti): Handle op1 zero.
9934 (subvti4, usub4ti4): New.
9935 (*sub<GPI>3_compare1_imm): New.
9936 (sub<GPI>3_carryinCV): New.
9937 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9938 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9940 2018-07-19 Michael Collison <michael.collison@arm.com>
9941 Richard Henderson <rth@redhat.com>
9943 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9944 (addti3): Create simpler code if low part is already known to be 0.
9945 (addvti4, uaddvti4): New.
9946 (*add<GPI>3_compareC_cconly_imm): New.
9947 (*add<GPI>3_compareC_cconly): New.
9948 (*add<GPI>3_compareC_imm): New.
9949 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9950 handle constants within this pattern..
9951 (*add<GPI>3_compareV_cconly_imm): New.
9952 (*add<GPI>3_compareV_cconly): New.
9953 (*add<GPI>3_compareV_imm): New.
9954 (add<GPI>3_compareV): New.
9955 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9956 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9957 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9958 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9959 with 'comparison' operator.
9960 (*add<GPI>3_compareV_cconly_imm): Ditto.
9961 (*add<GPI>3_compareV_cconly): Ditto.
9962 (*add<GPI>3_compareV_imm): Ditto.
9963 (add<GPI>3_compareV): Ditto.
9964 (add<mode>3_carryinC): Ditto.
9965 (*add<mode>3_carryinC_zero): Ditto.
9966 (*add<mode>3_carryinC): Ditto.
9967 (add<mode>3_carryinV): Ditto.
9968 (*add<mode>3_carryinV_zero): Ditto.
9969 (*add<mode>3_carryinV): Ditto.
9971 2018-07-19 Michael Collison <michael.collison@arm.com>
9972 Richard Henderson <rth@redhat.com>
9974 * config/aarch64/aarch64-modes.def (CC_V): New.
9975 * config/aarch64/aarch64-protos.h
9976 (aarch64_addti_scratch_regs): Declare
9977 (aarch64_subvti_scratch_regs): Declare.
9978 (aarch64_expand_subvti): Declare.
9979 (aarch64_gen_unlikely_cbranch): Declare
9980 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9981 for signed overflow using CC_Vmode.
9982 (aarch64_get_condition_code_1): Handle CC_Vmode.
9983 (aarch64_gen_unlikely_cbranch): New function.
9984 (aarch64_addti_scratch_regs): New function.
9985 (aarch64_subvti_scratch_regs): New function.
9986 (aarch64_expand_subvti): New function.
9988 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9990 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9992 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9993 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9996 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9999 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10000 address check not strict.
10002 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
10004 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
10006 (store_pair_lanes<mode>): Likewise.
10007 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
10008 enum value 'ADDR_QUERY_LDP_STP_N'.
10009 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
10010 (aarch64_print_address_internal): Add declaration.
10011 (aarch64_print_ldpstp_address): Remove.
10012 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
10013 (aarch64_print_operand): Change printing of 'y'.
10014 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
10015 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
10016 'true' rather than '1'.
10017 * config/aarch64/constraints.md (Uml): Likewise.
10018 (Uml): Rename to Umn.
10021 2018-07-19 Richard Biener <rguenther@suse.de>
10023 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
10025 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
10026 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
10027 (vn_reference_hasher): Likewise.
10028 (struct vn_tables_s): Remove obstack and alloc-pool members.
10029 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
10030 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
10031 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
10032 (vn_reference_insert_pieces): Likewise.
10033 (alloc_vn_nary_op_noinit): Adjust.
10034 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
10035 (vn_phi_eq): Adjust.
10036 (shared_lookup_phiargs): Remove.
10037 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
10038 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
10039 (visit_reference_op_call): Likewise.
10040 (copy_nary, copy_phi, copy_reference): Remove.
10041 (process_scc): Rewind the obstack when iterating. Do not
10042 copy the elements to valid_info but just move them from one
10043 hashtable to the other.
10044 (allocate_vn_table): Adjust.
10045 (free_vn_table): Likewise.
10046 (init_scc_vn): Likewise.
10047 (free_scc_vn): Likewise.
10049 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
10052 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
10053 indirect_return as function type attribute.
10054 (ix86_attribute_table): Change indirect_return to function
10056 * doc/extend.texi: Update indirect_return attribute.
10058 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
10060 * wide-int.h (widest2_int): New.
10061 * gimple-fold.c (arith_overflowed_p): Use it.
10062 * tree.h (widest2_int_cst): New.
10063 * tree-vrp.c (wide_int_binop_overflow): Rename from
10064 vrp_int_const_binop.
10065 Rewrite to work on trees.
10066 (extract_range_from_multiplicative_op_1): Abstract code to...
10067 (wide_int_range_min_max): ...here.
10068 (wide_int_range_cross_product): ...and here.
10069 (extract_range_from_binary_expr_1): Abstract overflow code to...
10070 (wide_int_range_mult_wrapping): ...here.
10071 * tree-vrp.h (wide_int_range_cross_product): New.
10072 (wide_int_range_mult_wrapping): New.
10074 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
10075 Julia Koval <julia.koval@intel.com>
10077 * config/i386/x86-tune-costs.h (skylake_memcpy,
10078 skylake_memset): Replace rep_prefix with unrolling for size 512.
10080 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
10082 PR middle-end/86544
10083 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
10084 comparision with EQ_EXPR in last stmt.
10086 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
10088 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
10089 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
10090 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
10091 previously known as "PowerPC AltiVec Built-in Functions". Move
10092 some material to new subsubsections "PowerPC AltiVec Built-in
10093 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
10095 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
10096 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
10097 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
10098 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
10100 2018-07-18 Richard Biener <rguenther@suse.de>
10102 PR tree-optimization/86557
10103 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
10106 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
10108 * config/s390/s390.c (s390_function_profiler): Generate CFI.
10110 2018-07-17 Jeff Law <law@redhat.com>
10112 * config/arm/arm.c (get_label_padding): Update for recent
10113 changes to label_to_alignment.
10115 PR tree-optimization/86010
10116 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
10118 * config/mips/mips.c (vr4130_align_insns): Update for recent
10119 changes to label_to_alignment.
10121 * config/frv/frv.c (frv_label_align): Update for recent changes
10122 to label_to_alignment.
10124 * config/nios2/nios2.c (nios2_label_align): Update for recent
10125 changes which dropped ALIGN_LABELS_LOG.
10127 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
10129 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
10130 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
10132 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10134 * config/arc/arc.c (arc_label_align): Use align_labels instead of
10135 deprecated align_labels_log.
10137 2018-07-17 Richard Biener <rguenther@suse.de>
10140 * dwarf2out.c (init_sections_and_labels): Always generate
10141 a debug_line_str_section for early LTO debug.
10142 (dwarf2out_finish): Reset debug_line_str_hash output early.
10143 Bump counter for extra dwarf5 .debug_loc labels to not conflict
10145 (dwarf2out_early_finish): Output debug_line_str.
10147 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10149 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
10150 index register on z196 or later.
10152 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10154 * config/s390/s390.c (s390_default_align): Set default function
10156 (s390_override_options_after_change): Call s390_default align.
10157 (s390_option_override_internal): Call s390_default_align.
10158 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
10160 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10162 PR middle-end/86542
10163 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
10164 to _looptemp_ clauses, other than the first two.
10166 2018-07-17 Martin Liska <mliska@suse.cz>
10168 * opts.c: Do not enable OPT_falign_* for -Os.
10170 2018-07-17 Martin Liska <mliska@suse.cz>
10172 * align.h (MAX_CODE_ALIGN): New.
10173 (MAX_CODE_ALIGN_VALUE): New.
10174 * common/config/i386/i386-common.c (ix86_handle_option):
10175 (MAX_CODE_ALIGN): Moved to align.h.
10176 * final.c (MAX_CODE_ALIGN): Likewise.
10177 * opts.c (parse_and_check_align_values):
10178 (MAX_CODE_ALIGN): Likewise.
10179 (MAX_CODE_ALIGN_VALUE): Likewise.
10181 2018-07-17 Martin Liska <mliska@suse.cz>
10183 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
10184 in order to fulfil coding style.
10185 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
10186 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10187 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10188 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
10189 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
10190 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
10191 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
10192 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
10193 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10195 2018-07-17 Martin Liska <mliska@suse.cz>
10197 * align.h: New file.
10198 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
10200 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
10201 align_flags of label_to_alignment.
10202 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
10204 * config/m68k/m68k.c: Do not use removed align_labels_value and
10206 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
10207 (LOOP_ALIGN): Likewise.
10208 (LABEL_ALIGN): Likewise.
10209 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10210 Remove not used macro.
10211 (rs6000_loop_align): Change return type to align_flags.
10212 (rs6000_loop_align_max_skip): Remove.
10213 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
10214 Change return type to align_flags.
10215 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10216 Remove not used macro.
10217 (rs6000_loop_align): Change return type to align_flags.
10218 (rs6000_loop_align_max_skip): Remove.
10219 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
10220 * config/rx/rx-protos.h (rx_align_for_label): Make it
10222 * config/rx/rx.c (rx_align_for_label): Change return type
10224 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
10226 into align_flags class.
10227 (LABEL_ALIGN): Likewise.
10228 (LOOP_ALIGN): Likewise.
10229 * config/s390/s390.c (s390_label_align): Use align_flags
10231 (s390_asm_output_function_label): Likewise.
10232 * config/sh/sh.c (sh_override_options_after_change):
10233 Use align_flags class directly without macros.
10234 (find_barrier): Likewise.
10235 (barrier_align): Likewise.
10236 (sh_loop_align): Likewise.
10237 * config/spu/spu.c (spu_option_override):
10238 Use align_flags_tuple::get_value instead of removed macros.
10239 (spu_sched_init): Likewise.
10240 * config/spu/spu.h (GTY): Likewise.
10241 * config/visium/visium.c (visium_option_override):
10242 Set "8" as default secondary alignment.
10243 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
10244 in order to guarantee secondary alignment of 8.
10245 * coretypes.h: Include align.h header file.
10246 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
10247 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
10248 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
10249 * doc/tm.texi.in: Likewise.
10250 * final.c (struct label_alignment): Remove not used structure.
10251 (LABEL_ALIGN): Change type to align_flags.
10252 (LOOP_ALIGN): Likewise.
10253 (JUMP_ALIGN): Likewise.
10254 (default_loop_align_max_skip): Remove.
10255 (default_label_align_max_skip): Likewise.
10256 (default_jump_align_max_skip): Likewise.
10257 (default_label_align_after_barrier_max_skip):
10258 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
10259 (LABEL_TO_MAX_SKIP): Remove.
10260 (label_to_alignment): Return align_flags type instead of integer.
10261 (label_to_max_skip): Remove.
10262 (align_fuzz): Use align_flags type.
10263 (compute_alignments): Use align_flags type and use align_flags::max
10264 to combine multiple alignments.
10265 (grow_label_align): Grow vec instead of C array.
10266 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
10267 (shorten_branches): Use align_flags type and use align_flags::max
10268 to combine multiple alignments.
10269 (final_scan_insn_1): Remove usage of secondary alignment that comes
10270 from label alignment, but instead use proper secondary alignment
10271 which is computed in grow_label_align.
10272 * flags.h (struct align_flags_tuple): Move to align.h.
10273 (struct align_flags): Likewise.
10274 (state_align_loops): Rename to align_loops.
10275 (state_align_jumps): Rename to align_jumps.
10276 (state_align_labels): Rename to align_labels.
10277 (state_align_functions): Rename to align_functions.
10278 (align_loops_log): Remove.
10279 (align_jumps_log): Remove.
10280 (align_labels_log): Remove.
10281 (align_functions_log): Remove.
10282 (align_loops_max_skip): Remove.
10283 (align_jumps_max_skip): Remove.
10284 (align_labels_max_skip): Remove.
10285 (align_functions_max_skip): Remove.
10286 (align_loops_value): Remove.
10287 (align_jumps_value): Remove.
10288 (align_labels_value): Remove.
10289 (align_functions_value): Remove.
10290 * output.h (label_to_alignment): Change return type to align_flags.
10291 (label_to_max_skip): Remove.
10292 * target.def: Remove loop_align_max_skip, label_align_max_skip,
10293 jump_align_max_skip macros.
10294 * targhooks.h (default_loop_align_max_skip): Remove.
10295 (default_label_align_max_skip): Likewise.
10296 (default_jump_align_max_skip): Likewise.
10297 (default_label_align_after_barrier_max_skip): Remove.
10298 * toplev.c (read_log_maxskip): Use ::normalize function.
10299 (parse_N_M): Remove not used argument and also call ::normalize.
10300 (parse_alignment_opts): Do not pass unused arguments.
10301 * varasm.c (assemble_start_function): Use directly align_functions
10302 instead of removed macros.
10303 * system.h: Do not poison removed macros.
10305 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10307 PR middle-end/86539
10308 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
10309 and cond temporaries don't have reference type if iterator has
10310 pointer type. For init use &for_pre_body instead of pre_p if
10311 for_pre_body is non-empty.
10313 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10315 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
10316 double-double modes to SFmode directly directly.
10317 (trunc<mode>sf2_fprs): Delete.
10319 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10321 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
10322 for conversions between IFmode and the decimal floating point modes.
10323 (init_float128_ieee): Use the correct names for conversions between
10324 KFmode and the decimal floating point modes.
10326 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10328 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
10329 for the conversions between TDmode and IFmode.
10330 (init_float128_ieee): Use more correct names for the conversions
10331 between TDmode and KFmode.
10333 2018-07-16 Jakub Jelinek <jakub@redhat.com>
10335 PR tree-optimization/86526
10336 * builtins.c (expand_builtin_memcmp): Formatting fixes.
10337 (inline_expand_builtin_string_cmp): Likewise.
10338 (inline_string_cmp): Likewise. Use c_readstr instead of
10339 builtin_memcpy_read_str. Add unit_mode temporary.
10341 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10343 PR middle-end/86528
10344 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
10345 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
10347 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
10349 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
10350 Alphabetize prototypes of built-in functions, separating out
10351 built-in functions that are listed in this section but should be
10352 described elsewhere.
10354 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
10357 * expmed.c (emit_store_flag): Do not emit setcc followed by a
10358 conditional move when trapping comparison was split to a
10359 non-trapping one (and vice versa).
10361 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10363 * config/s390/s390.c (s390_function_profiler): Generate nops
10364 instead of profiler call sequences.
10365 * config/s390/s390.opt: Add the new option.
10367 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10369 * config/s390/s390.c (s390_function_profiler): Generate
10370 __mcount_loc section.
10371 * config/s390/s390.opt: Add the new option.
10373 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10375 * common.opt: Add the new warning.
10376 * config/s390/s390.c (s390_function_profiler): Emit "brasl
10377 %r0,__fentry__" when -mfentry is specified.
10378 (s390_option_override_internal): Disallow -mfentry for 31-bit
10380 * config/s390/s390.opt: Add the new option.
10382 2018-07-16 Richard Biener <rguenther@suse.de>
10385 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
10386 for function-local FUNCTION_DECL and RESULT_DECL immediately.
10388 2018-07-16 Martin Liska <mliska@suse.cz>
10391 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
10394 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10396 * config/arc/arcHS.md: Update ARCHS scheduling rules.
10398 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10400 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
10402 * config/arc/arc-cpus.def (hs4x): New cpu.
10404 * config/arc/arc-tables.opt: Regenerate.
10405 * config/arc/arc.c (arc_sched_issue_rate): New function.
10406 (TARGET_SCHED_ISSUE_RATE): Define.
10407 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
10408 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
10410 (attr tune): Add ARCHS4x tune values.
10411 (attr tune_dspmpy): Define.
10412 (*tst): Correct instruction type.
10413 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
10414 * config/arc/arcHS4x.md: New file.
10415 * config/arc/fpu.md: Update instruction type attributes.
10416 * config/arc/t-multilib: Regenerate.
10418 2018-07-16 Tom de Vries <tdevries@suse.de>
10421 * var-tracking.c (vt_initialize): Fix pre_dec handling.
10423 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10425 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
10428 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
10430 PR tree-optimization/86514
10431 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
10432 conversion to a boolean type from a type with greater precision.
10434 2018-07-16 Tom de Vries <tdevries@suse.de>
10436 * var-tracking.c (vt_initialize): Print adjusted insn slim if
10437 dump_flags request TDF_SLIM.
10439 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
10441 * fold-const.c (int_const_binop_1): Abstract...
10442 (wide_int_binop): ...wide int code here.
10443 (poly_int_binop): ...poly int code here.
10444 Abstract the rest of int_const_binop_1 into int_const_binop.
10445 * fold-const.h (wide_int_binop): New.
10446 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
10447 Remove useless PLUS/MINUS_EXPR case.
10448 (zero_nonzero_bits_from_vr): Move wide int code...
10449 (zero_nonzero_bits_from_bounds): ...here.
10450 (extract_range_from_binary_expr_1): Move mask optimization code...
10451 (range_easy_mask_min_max): ...here.
10452 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
10453 (range_easy_mask_min_max): New.
10455 2018-07-15 Jeff Law <law@redhat.com>
10458 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
10461 2018-07-14 Jim Wilson <jimw@sifive.com>
10463 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
10465 2018-07-14 Paul Koning <ni1d@arrl.net>
10467 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
10469 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
10471 * lto-streamer-out.c (copy_function_or_variable): Dump info about
10474 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
10475 Steve Munroe <munroesj52@gmail.com>
10477 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
10478 (_mm_andnot_si128): Likewise.
10479 (_mm_or_si128): Likewise.
10480 (_mm_xor_si128): Likewise.
10482 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
10484 PR middle-end/78809
10485 * builtins.c (expand_builtin_memcmp): Inline the calls first
10486 when result_eq is false.
10487 (expand_builtin_strcmp): Inline the calls first.
10488 (expand_builtin_strncmp): Likewise.
10489 (inline_string_cmp): New routine. Expand a string compare
10490 call by using a sequence of char comparison.
10491 (inline_expand_builtin_string_cmp): New routine. Inline expansion
10492 a call to str(n)cmp/memcmp.
10493 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
10495 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
10497 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
10499 * config/arm/driver-arm.c: Include arm-native.h.
10500 (host_detect_local_cpu): Use auto-generated data tables.
10501 (vendors, arm_cpu_table): Delete. Move part information to ...
10502 * config/arm/arm-cpus.in: ... here.
10503 * config/arm/parsecpu.awk (gen_native): New function.
10504 (vendor, part): New CPU fields.
10505 (END): Add support for building the native CPU detection tables.
10506 * config/arm/t-arm (arm-native.h): Add build rule.
10507 (driver-arm.o): Add dependency on arm-native.h.
10509 2018-07-13 Richard Biener <rguenther@suse.de>
10511 PR middle-end/85974
10512 * match.pd (addr1 - addr2): Allow either of the operand to
10515 2018-07-13 Tom de Vries <tdevries@suse.de>
10517 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
10520 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
10522 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
10523 arrays instead of numbered variables.
10525 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
10527 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
10528 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
10530 2018-07-13 Richard Biener <rguenther@suse.de>
10533 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
10534 instead of get_context_die.
10536 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10537 Richard Biener <rguenther@suse.de>
10539 PR middle-end/86489
10540 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
10541 that the loop latch destination where phi is defined.
10543 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
10545 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
10546 (riscv_expand_epilogue): Add assertion to check interrupt mode.
10547 (riscv_set_current_function): Extract getting interrupt type to new
10549 (riscv_get_interrupt_type): New function.
10550 (riscv_merge_decl_attributes): New function, checking interrupt type is
10552 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10554 2018-07-12 Paul Koning <ni1d@arrl.net>
10556 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10559 2018-07-12 Paul Koning <ni1d@arrl.net>
10561 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10562 zero reference, add doloop_end instead.
10563 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10564 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10565 detail for doloop_end.
10567 2018-07-12 Martin Sebor <msebor@redhat.com>
10570 * attribs.c (decl_attributes): Reject conflicting attributes before
10571 calling attribute handlers.
10573 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10575 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10577 (gcc::dump_manager::get_dump_file_name): likewise.
10578 (dump_begin): Likewise.
10579 * dumpfile.h (dump_begin): Update prototype.
10580 (gcc::dump_manager::get_dump_file_name,
10581 gcc::dump_manager::get_dump_file_name): Update prototype.
10583 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10585 * internal-fn.h (vectorizable_internal_fn_p): New function.
10586 * tree-vect-slp.c (compatible_calls_p): Likewise.
10587 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10588 to internal functions.
10589 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10591 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10593 * fold-const.h (inverse_conditions_p): Declare.
10594 * fold-const.c (inverse_conditions_p): New function.
10595 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10596 that test the inverse condition of a conditional internal function.
10597 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10598 * internal-fn.c (internal_fn_mask_index): Handle conditional
10599 internal functions.
10600 (vectorized_internal_fn_supported_p): New function.
10601 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10602 (any_pred_load_store): Replace with...
10603 (need_to_predicate): ...this new variable.
10604 (redundant_ssa_names): New variable.
10605 (ifcvt_can_use_mask_load_store): Move initial checks to...
10606 (ifcvt_can_predicate): ...this new function. Handle tree codes
10607 for which a conditional internal function exists.
10608 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10609 instead of ifcvt_can_use_mask_load_store. Update after variable
10611 (predicate_load_or_store): New function, split out from
10612 predicate_mem_writes.
10613 (check_redundant_cond_expr): New function.
10614 (value_available_p): Likewise.
10615 (predicate_rhs_code): Likewise.
10616 (predicate_mem_writes): Rename to...
10617 (predicate_statements): ...this. Use predicate_load_or_store
10618 and predicate_rhs_code.
10619 (combine_blocks, tree_if_conversion): Update after above name changes.
10620 (ifcvt_local_dce): Handle redundant_ssa_names.
10621 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10622 general conditional functions.
10623 * tree-vect-stmts.c (vectorizable_call): Likewise.
10625 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10626 Alan Hayward <alan.hayward@arm.com>
10627 David Sherwood <david.sherwood@arm.com>
10629 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10630 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10631 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10632 plus and minus and convert them into IFN_COND_FMA-based sequences.
10633 (convert_mult_to_fma): Handle conditional plus and minus.
10635 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10637 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10638 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10639 (cond_fnms_optab): New optabs.
10640 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10641 internal functions.
10642 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10643 * internal-fn.h (get_conditional_internal_fn): Declare.
10644 (get_unconditional_internal_fn): Likewise.
10645 * internal-fn.c (cond_ternary_direct): New macro.
10646 (expand_cond_ternary_optab_fn): Likewise.
10647 (direct_cond_ternary_optab_supported_p): Likewise.
10648 (FOR_EACH_COND_FN_PAIR): Likewise.
10649 (get_conditional_internal_fn): New function.
10650 (get_unconditional_internal_fn): Likewise.
10651 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10652 (gimple_match_op::gimple_match_op): Add a new overload for 5
10654 (gimple_match_op::set_op): Likewise.
10655 (gimple_resimplify5): Declare.
10656 * genmatch.c (decision_tree::gen): Generate simplifications for
10658 * gimple-match-head.c (gimple_simplify): Define an overload for
10659 5 operands. Handle calls with 5 arguments in the top-level overload.
10660 (convert_conditional_op): Handle conversions from unconditional
10661 internal functions to conditional ones.
10662 (gimple_resimplify5): New function.
10663 (build_call_internal): Pass a fifth operand.
10664 (maybe_push_res_to_seq): Likewise.
10665 (try_conditional_simplification): Try converting conditional
10666 internal functions to unconditional internal functions.
10667 Handle 3-operand unconditional forms.
10668 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10669 Define ternary equivalents of the current rules for binary conditional
10670 internal functions.
10671 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10672 ternary operations.
10673 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10674 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10675 (optab): Handle them.
10676 (SVE_COND_FP_TERNARY): New int iterator.
10677 (sve_fmla_op, sve_fmad_op): New int attributes.
10678 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10679 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10680 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10682 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10684 * target.def (preferred_else_value): New target hook.
10685 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10686 * doc/tm.texi: Regenerate.
10687 * targhooks.h (default_preferred_else_value): Declare.
10688 * targhooks.c (default_preferred_else_value): New function.
10689 * internal-fn.h (conditional_internal_fn_code): Declare.
10690 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10691 (get_conditional_internal_fn): Use it.
10692 (conditional_internal_fn_code): New function.
10693 * gimple-match.h (gimple_match_cond): New struct.
10694 (gimple_match_op): Add a cond member function.
10695 (gimple_match_op::gimple_match_op): Update all forms to take a
10697 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10698 for the suboperation, but don't specify a particular else_value.
10699 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10700 (visit_nary_op, visit_reference_op_load): Pass
10701 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10702 * gimple-match-head.c: Include tree-eh.h
10703 (convert_conditional_op): New function.
10704 (maybe_resimplify_conditional_op): Likewise.
10705 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10706 (gimple_resimplify2): Likewise.
10707 (gimple_resimplify3): Likewise.
10708 (gimple_resimplify4): Likewise.
10709 (maybe_push_res_to_seq): Return null for conditional operations.
10710 (try_conditional_simplification): New function.
10711 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10713 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10715 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10717 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10719 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10721 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10723 (hash_tree): Do not hash DECL_FCONTEXT
10724 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10725 Do not stream DECL_FCONTEXT.
10726 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10727 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10729 2018-07-12 Richard Biener <rguenther@suse.de>
10732 * dwarf2out.c (gen_block_die): Only output blocks when they have
10733 at least one !DECL_IGNORED_P variable.
10735 2018-07-12 Richard Biener <rguenther@suse.de>
10738 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10739 Remove -mieee-fp handling.
10741 2018-07-12 Richard Biener <rguenther@suse.de>
10743 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10744 left-over from last patch.
10746 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10748 PR tree-optimization/86492
10749 * gimple-ssa-store-merging.c
10750 (imm_store_chain_info::coalesce_immediate_stores): Call
10751 check_no_overlap even for the merge_overlapping case. Formatting fix.
10753 2018-07-12 Richard Biener <rguenther@suse.de>
10755 PR middle-end/86479
10756 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10757 move possibly trapping operations into the conditional.
10759 2018-07-12 Richard Biener <rguenther@suse.de>
10761 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10762 (vn_lookup_simplify_result): Remove recursion limit applied
10764 (vn_nary_build_or_lookup_1): Adjust.
10765 (try_to_simplify): Likewise.
10766 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10768 (gimple_resimplify2): Likewise.
10769 (gimple_resimplify3): Likewise.
10770 (gimple_resimplify4): Likewise.
10772 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10774 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10775 Use __mmask64 type instead of __mmask8 for __M argument.
10776 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10777 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10779 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10780 __mmask16 for __M argument.
10781 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10782 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10783 to __mmask16 instead of __mmask8.
10784 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10785 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10786 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10787 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10788 instead of __mmask16 for __U argument.
10789 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10790 __mmask16 instead of __mmask8 for __U argument.
10791 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10793 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10795 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10797 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10799 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10801 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10803 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10805 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10807 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10809 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10811 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10812 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10813 return type as well as __M argument type and all casts from __mmask8
10815 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10816 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10817 return type as well as __M argument type and all casts from __mmask8
10819 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10820 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10821 return type as well as __M argument type and all casts from __mmask8
10823 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10824 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10825 return type as well as __M argument type and all casts from __mmask8
10827 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10828 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10831 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10833 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10834 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10837 2018-07-11 Paul Koning <ni1d@arrl.net>
10839 * doc/md.texi (define_subst): Document how multiple occurrences of
10840 the same argument in the replacement pattern are handled.
10842 2018-07-11 Paul Koning <ni1d@arrl.net>
10844 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10845 alphabetical order.
10846 (Common Type Attributes): Add "mode" attribute.
10848 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10850 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10851 stream DECL_ORIGINAL_TYPE.
10852 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10853 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10854 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10855 Do not walk original type.
10856 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10858 (write_ts_decl_non_common_tree_pointers): Do not stream
10860 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10861 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10863 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10865 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10866 thread twice from the same starting edge.
10868 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10870 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10871 * gimple.c (gimple_call_nonnull_result_p): ...here...
10872 (gimple_call_nonnull_arg): ...and here.
10873 * gimple.h (gimple_call_nonnull_result_p): New.
10874 (gimple_call_nonnull_arg): New.
10876 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10878 * config/arm/arm-cpus.in: Move information from fpu field of each
10879 cpu definition to the isa field.
10880 * config/arm/parsecpu.awk (fpu): Delete match rule.
10881 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10883 2018-07-11 Richard Biener <rguenther@suse.de>
10886 * dwarf2out.c (init_sections_and_labels): Use
10887 output_asm_line_debug_info consistently.
10888 (dwarf2out_early_finish): Likewise.
10889 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10892 2018-07-11 Richard Biener <rguenther@suse.de>
10894 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10895 Rework father_bb setting in a way to avoid propagating constants
10896 multiple times on a loop body.
10898 2018-07-10 Mark Wielaard <mark@klomp.org>
10901 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10903 2018-07-10 Richard Biener <rguenther@suse.de>
10905 * hash-map.h (hash_map::iterator::operator*): Return
10906 references to key and value.
10908 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10911 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10912 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10913 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10914 (gimplify_omp_for): For composite loops, move outer
10915 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10916 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10917 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10918 TREE_LIST for both the original class iterator and the "last" helper
10919 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10920 loop, remember has_decl_expr from outer composite loops for the
10921 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10923 2018-07-09 Martin Sebor <msebor@redhat.com>
10925 PR middle-end/77357
10926 PR middle-end/86428
10927 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10928 accessing implicitly initialized array elements.
10929 * expr.c (string_constant): Handle string initializers of
10930 character arrays within aggregates.
10931 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10932 Store element offset. As a special case, handle zero size.
10933 (fold_nonarray_ctor_reference): Same.
10934 (fold_ctor_reference): Add argument. Store subobject offset.
10935 * gimple-fold.h (fold_ctor_reference): Add argument.
10937 2018-07-09 Paul Koning <ni1d@arrl.net>
10939 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10940 (pdp11_insn_cost): New function.
10941 (pdp11_md_asm_adjust): New function.
10942 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10943 (pdp11_rtx_costs): Update to match machine better.
10944 (output_addr_const_pdp11): Correct format mismatch warnings.
10945 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10946 * config/pdp11/pdp11.md: General change to add base_cost and/or
10947 length attributes for use by new pdp11_insn_cost function.
10948 (MIN_BRANCH): Correct definition.
10950 (doloop_end): Use standard pattern name for looping pattern.
10951 (doloop_end_nocc): New.
10952 (movsf): Add another constraint alternative.
10953 (zero_extendqihi2): Add constraint alternatives for not in place
10955 (zero_extendhisi2): Remove.
10956 (shift patterns): Add CC handling variants.
10960 (define_peephole2): New peephole to recognize mov that sets CC for
10963 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10965 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10966 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10969 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10972 * real.c (real_nextafter): Return true if result is denormal.
10974 2018-07-09 Martin Liska <mliska@suse.cz>
10976 * common.opt: Add back wrongly removed attribute.
10978 2018-07-09 Richard Biener <rguenther@suse.de>
10981 * dwarf2out.c (gen_block_die): For an early generated DIE
10982 always output high/low PC attributes.
10984 2018-07-09 Tom de Vries <tdevries@suse.de>
10986 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10987 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10988 onto VAR_DECL with abstract origin.
10990 2018-07-07 Jim Wilson <jimw@sifive.com>
10992 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10994 2018-07-07 Tom de Vries <tdevries@suse.de>
10996 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10998 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
11000 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
11002 (combine_bound): Use wide-int overflow calculation instead of
11004 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
11006 * fold-const.c (int_const_binop_2): Same.
11007 (extract_muldiv_1): Same.
11008 (fold_div_compare): Same.
11009 (fold_abs_const): Same.
11011 * poly-int.h (add): Same.
11015 * predict.c (predict_iv_comparison): Same.
11016 * profile-count.c (slow_safe_scale_64bit): Same.
11017 * simplify-rtx.c (simplify_const_binary_operation): Same.
11018 * tree-chrec.c (tree_fold_binomial): Same.
11019 * tree-data-ref.c (split_constant_offset_1): Same.
11020 * tree-if-conv.c (idx_within_array_bound): Same.
11021 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
11022 * tree-ssa-phiopt.c (minmax_replacement): Same.
11023 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
11024 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
11025 * vr-values.c (vr_values::adjust_range_with_scev): Same.
11026 * wide-int.cc (wi::add_large): Same.
11027 (wi::mul_internal): Same.
11028 (wi::sub_large): Same.
11029 (wi::divmod_internal): Same.
11030 * wide-int.h: Change overflow type to overflow_type for neg, add,
11031 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
11032 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
11033 mul_internal, divmod_internal.
11034 (overflow_type): New enum.
11035 (accumulate_overflow): New.
11037 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
11039 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
11040 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
11042 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
11044 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
11045 argument is checked for zero before entering loop, avoid checking again.
11047 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
11049 * gimplify.h (generic_expr_could_trap_p): Set as global function.
11050 * gimplify.h (generic_expr_could_trap_p): Likwise.
11051 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
11053 2018-07-06 Jakub Jelinek <jakub@redhat.com>
11055 PR tree-optimization/86401
11056 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
11057 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
11058 (fold_bit_and_mask): ... here. New helper function for match.pd.
11059 * fold-const.h (fold_bit_and_mask): Declare.
11060 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
11062 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
11065 * target.def (translate_mode_attribute): New hook.
11066 * targhooks.h (default_translate_mode_attribute): Declare.
11067 * targhooks.c (default_translate_mode_attribute): New function.
11068 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
11069 * doc/tm.texi: Regenerate.
11070 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
11071 (rs6000_translate_mode_attribute): New function.
11073 2018-07-06 Paul Koning <ni1d@arrl.net>
11075 * doc/md.texi (define_split): Document DONE and FAIL.
11076 (define_peephole2): Ditto.
11078 2018-07-05 Jeff Law <law@redhat.com>
11080 PR tree-optimization/86010
11081 * tree-ssa-dse.c (compute_trims): More aggressively trim at
11082 both the head and tail of mem* and str* calls.
11084 2018-07-05 Jim Wilson <jimw@sifive.com>
11086 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
11088 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
11090 * config/aarch64/aarch64-simd.md: correct flags text for
11091 MIN_EXPR replacement.
11093 2018-07-05 James Clarke <jrtc27@jrtc27.com>
11095 * configure: Regenerated.
11097 2018-07-05 Carl Love <cel@us.ibm.com>
11099 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
11100 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
11101 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
11102 VSX_BUILTIN_DOUBLEL_V4SF.
11104 2018-07-05 Martin Sebor <msebor@redhat.com>
11107 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
11108 than its domain to compute its the upper bound of a char array.
11110 2018-07-05 Nathan Sidwell <nathan@acm.org>
11112 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
11113 * doc/cpp.texi: Update comment.
11114 * doc/tm.texi: Rebuilt.
11115 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
11116 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
11117 * doc/extend.texi (Backwards Compatibility): Clarify it is system
11118 headers affected by extern "C".
11119 * system.h: Poison NO_IMPLICIT_EXTERN_C.
11120 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
11121 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
11122 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
11123 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
11124 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
11125 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
11126 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
11127 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
11128 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
11129 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
11130 config/sparc/sp64-elf.h, config/spu/spu.h,
11131 config/stormy16/stormy16.h, config/v850/v850.h,
11132 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
11133 define NO_IMPLICIT_EXTERN_C.
11134 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
11136 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11139 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
11140 instead of GET_MODE_SIZE when comparing Units.
11142 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11145 * rtlanal.c (set_noop_p): Constrain on mode change,
11146 include hard-reg-set.h
11148 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11150 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
11152 2018-07-05 Jakub Jelinek <jakub@redhat.com>
11155 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11158 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11160 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11162 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11165 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11167 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11169 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
11171 PR middle-end/86380
11172 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
11174 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
11176 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
11177 neg_*_op* variables.
11179 2018-07-04 Martin Liska <mliska@suse.cz>
11181 * tree-switch-conversion.c: Define
11182 max_ratio_for_speed and max_ratio_for_size constants.
11184 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11185 Martin Liska <mliska@suse.cz>
11187 PR middle-end/66240
11190 * common.opt: Rename align options with 'str_' prefix.
11191 * common/config/i386/i386-common.c (set_malign_value): New
11193 (ix86_handle_option): Use it to set -falign-* options/
11194 * config/aarch64/aarch64-protos.h (struct tune_params): Change
11195 type from int to string.
11196 * config/aarch64/aarch64.c: Update default values from int
11198 * config/alpha/alpha.c (alpha_override_options_after_change):
11200 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
11201 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11202 max skip conditionally.
11203 * config/i386/freebsd.h (SUBALIGN_LOG): New.
11204 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11205 max skip conditionally.
11206 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11207 max skip conditionally.
11208 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
11209 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11210 max skip conditionally.
11211 * config/i386/i386.c (struct ptt): Change type from int to
11213 (ix86_default_align): Set default values.
11214 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
11215 max skip conditionally.
11216 * config/i386/iamcu.h (SUBALIGN_LOG): New.
11217 (ASM_OUTPUT_MAX_SKIP_ALIGN):
11218 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
11219 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11220 max skip conditionally.
11221 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
11222 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
11223 * config/i386/x86-64.h (SUBALIGN_LOG): New.
11224 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11225 max skip conditionally.
11226 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11227 * config/ia64/ia64.c (ia64_option_override): Set default values
11228 for alignment options.
11229 * config/m68k/m68k.c: Handle new str_align_* options.
11230 * config/mips/mips.c (mips_set_compression_mode): Change
11232 (mips_option_override): Set default values for options.
11233 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
11235 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11237 * config/rx/rx.c (rx_option_override): Likewise.
11238 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
11239 (LABEL_ALIGN): Use align_labels_log.
11240 (LOOP_ALIGN): Use align_loops_align.
11241 * config/s390/s390.c (s390_asm_output_function_label): Use new
11243 * config/sh/sh.c (sh_override_options_after_change):
11244 Change type of constants.
11245 * config/spu/spu.c (spu_sched_init): Likewise.
11246 * config/sparc/sparc.c (sparc_option_override): Set default
11247 values for options.
11248 * config/visium/visium.c (visium_option_override): Likewise.
11249 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
11250 emit p2align format with last argument if it's not needed.
11251 * doc/invoke.texi: Document extended format of -falign-*.
11252 * final.c: Use align_labels alignment.
11253 * flags.h (struct target_flag_state): Change type to use
11255 (struct align_flags_tuple): New.
11256 (struct align_flags): Likewise.
11257 (align_loops_log): Redefine macro to use new types.
11258 (align_loops_max_skip): Redefine macro to use new types.
11259 (align_jumps_log): Redefine macro to use new types.
11260 (align_jumps_max_skip): Redefine macro to use new types.
11261 (align_labels_log): Redefine macro to use new types.
11262 (align_labels_max_skip): Redefine macro to use new types.
11263 (align_functions_log): Redefine macro to use new types.
11264 (align_loops): Redefine macro to use new types.
11265 (align_jumps): Redefine macro to use new types.
11266 (align_labels): Redefine macro to use new types.
11267 (align_functions): Redefine macro to use new types.
11268 (align_functions_max_skip): Redefine macro to use new types.
11269 (align_loops_value): New macro.
11270 (align_jumps_value): New macro.
11271 (align_labels_value): New macro.
11272 (align_functions_value): New macro.
11273 * function.c (invoke_set_current_function_hook): Propagate
11274 alignment values from flags to global variables default in
11276 * ipa-icf.c (sem_function::equals_wpa): Use
11277 cl_optimization_option_eq instead of memcmp.
11278 * lto-streamer.h (cl_optimization_stream_out): Support streaming
11280 (cl_optimization_stream_in): Likewise.
11281 * optc-save-gen.awk: Support strings in cl_optimization.
11282 * opth-gen.awk: Likewise.
11283 * opts.c (finish_options): Remove error checking of invalid
11285 (MAX_CODE_ALIGN): Remove.
11286 (MAX_CODE_ALIGN_VALUE): Likewise.
11287 (parse_and_check_align_values): New function.
11288 (check_alignment_argument): Likewise.
11289 (common_handle_option): Use check_alignment_argument.
11290 * opts.h (parse_and_check_align_values): Declare.
11291 * toplev.c (init_alignments): Remove.
11292 (read_log_maxskip): New.
11293 (parse_N_M): Likewise.
11294 (parse_alignment_opts): Likewise.
11295 (backend_init_target): Remove usage of init_alignments.
11296 * toplev.h (parse_alignment_opts): Declare.
11297 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
11299 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
11300 * tree.c (cl_option_hasher::equal): New.
11301 * varasm.c: Use new global macros.
11303 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11305 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
11306 Use a simpler align directive also if MAXSKIP = ALIGN-1.
11307 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11308 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11309 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11310 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11311 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
11312 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
11313 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
11314 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11315 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11316 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11317 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11319 2018-07-04 Martin Liska <mliska@suse.cz>
11320 Jonathan Wakely <jwakely@redhat.com>
11322 * coverage.c: Use correct type.
11323 * doc/invoke.texi: Language correction.
11325 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
11328 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
11329 ENDBRANCH for non-tail call which may return via indirect branch.
11330 * doc/extend.texi: Document indirect_return attribute.
11332 2018-07-03 Martin Sebor <msebor@redhat.com>
11334 PR tree-optimization/86274
11335 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
11337 (format_floating): Correct handling of infinities and NaNs.
11339 2018-07-03 Martin Sebor <msebor@redhat.com>
11341 * print-tree.c (print_real_cst): New function.
11342 (print_node_brief): Call it.
11343 (print_node): Ditto.
11345 2018-07-03 Jeff Law <law@redhat.com>
11347 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
11348 into a single pattern.
11350 * config/h8300/h8300.md (ors code_iterator): New.
11351 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
11352 a single pattern and single splitter.
11353 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
11354 (iorqi3_1, xorqi3_1): Likewise.
11355 (iorqi3, xorqi3 expanders): Similarly.
11357 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
11358 (movmd_internal) into a single pattern using the P mode iterator.
11359 (movmd splitters): Similarly.
11360 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
11361 (movsd splitters): Similarly.
11363 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
11364 ADDB, ADDW and ADDL into a single ADD attribute which selects the
11365 right table based on the size of the operand.
11366 * config/h8300/h8300.md (length_table): Corresponding changes. All
11367 references to "addb", "addw" and "addl" changed to "add".
11368 (btst patterns): Merge two variants into a single pattern.
11369 (tstqi, tsthi): Likewise.
11370 (addhi3_incdec, addsi3_incdec): Likewise.
11371 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
11372 (mulhi3, mulsi3): Likewise.
11373 (udivhi3, udivsi3): Likewise.
11374 (divhi3, divsi3): Likewise.
11375 (andorqi3, andorhi3, andorsi3): Likewise.
11377 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
11380 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
11381 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
11383 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11385 PR tree-optimization/85694
11386 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
11387 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
11389 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
11390 (<u>avg<mode>3_ceil): New patterns.
11392 2018-07-03 David Malcolm <dmalcolm@redhat.com>
11394 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
11395 scan-tree-dump directive.
11396 * gcc.dg/vect/slp-perm-2.c: Likewise.
11397 * gcc.dg/vect/slp-perm-3.c: Likewise.
11398 * gcc.dg/vect/slp-perm-5.c: Likewise.
11399 * gcc.dg/vect/slp-perm-6.c: Likewise.
11400 * gcc.dg/vect/slp-perm-7.c: Likewise.
11401 * gcc.dg/vect/slp-perm-8.c: Likewise.
11403 2018-07-03 Marek Polacek <polacek@redhat.com>
11405 PR middle-end/86202
11406 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
11408 2018-07-03 Richard Biener <rguenther@suse.de>
11411 * tree-ssa-structalias.c (find_func_clobbers): Properly
11412 handle indirect calls.
11414 2018-07-03 Jeff Law <law@redhat.com>
11416 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
11417 (shifts): New code iterator.
11418 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
11419 expander. Fix HImode handling on H8/SX.
11420 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
11421 (subqi3, subhi3, subsi3 expanders): Likewise.
11422 (andqi3, andhi3, andsi3 expanders): Likewise.
11423 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
11424 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
11425 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
11426 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
11427 (zero_extendqihi2, zero_extendqisi2): Likewise.
11428 (extendqihi2, extendqisi2): Likewise.
11429 (rotlqi3, rotlhi3, rotlsi3): Likewise.
11430 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
11431 (rotlqi3_1, rotlhi3_1): Likewise.
11432 (logicalhi3_sn, logicalsi3_sn): Likewise.
11433 (logicalhi3, logicalsi3): Likewise.
11435 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11437 * tree-vect-patterns.c (vect_recog_rotate_pattern)
11438 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11439 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
11440 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
11441 type to append_pattern_def_seq instead of creating a stmt_vec_info
11443 (build_mask_conversion): Likewise. Remove vinfo argument.
11444 (vect_add_conversion_to_patterm): Likewise, renaming to...
11445 (vect_add_conversion_to_pattern): ...this.
11446 (vect_recog_mask_conversion_pattern): Update call to
11447 build_mask_conversion. Pass the vector type to
11448 append_pattern_def_seq here too.
11449 (vect_recog_gather_scatter_pattern): Update call to
11450 vect_add_conversion_to_pattern.
11452 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11454 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
11455 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11456 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
11457 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
11458 STMT_VINFO_PATTERN_DEF_SEQ to null here.
11459 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
11460 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
11461 append_pattern_def_seq instead of new_pattern_def_seq.
11462 (vect_recog_divmod_pattern): Do both of the above.
11463 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
11466 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11468 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
11469 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
11470 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
11471 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
11472 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
11473 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11474 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
11475 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11476 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11477 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
11478 parameter with a single stmt_vec_info.
11479 (vect_recog_func_ptr): Likewise.
11480 (vect_recog_gather_scatter_pattern): Likewise, folding in...
11481 (vect_try_gather_scatter_pattern): ...this.
11482 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
11483 the stmt_vec_info of the statement to be matched. Don't clear
11484 STMT_VINFO_RELATED_STMT.
11485 (vect_pattern_recog): Update call accordingly.
11487 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11489 PR tree-optimization/85694
11490 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
11491 (uavgM3_ceil): Document new optabs.
11492 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
11493 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
11495 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
11496 (savg_ceil_optab): New optabs.
11497 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
11498 (vect_vect_recog_func_ptrs): Add it.
11499 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
11500 constant directly from the associated lhs.
11502 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11504 * tree-vect-patterns.c (vect_split_statement): New function.
11505 (vect_convert_input): Use it to try to split an existing cast.
11507 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11509 * poly-int.h (print_hex): New function.
11510 * dumpfile.h (dump_dec, dump_hex): Declare.
11511 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
11512 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
11513 min_input_precision, operation_precision and operation_sign.
11514 * tree-vect-patterns.c (vect_get_range_info): New function.
11515 (vect_same_loop_or_bb_p, vect_single_imm_use)
11516 (vect_operation_fits_smaller_type): Delete.
11517 (vect_look_through_possible_promotion): Add an optional
11518 single_use_p parameter.
11519 (vect_recog_over_widening_pattern): Rewrite to use new
11520 stmt_vec_info infomration. Handle one operation at a time.
11521 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
11522 (vect_truncatable_operation_p, vect_set_operation_type)
11523 (vect_set_min_input_precision): New functions.
11524 (vect_determine_min_output_precision_1): Likewise.
11525 (vect_determine_min_output_precision): Likewise.
11526 (vect_determine_precisions_from_range): Likewise.
11527 (vect_determine_precisions_from_users): Likewise.
11528 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
11529 (vect_vect_recog_func_ptrs): Put over_widening first.
11531 (vect_pattern_recog): Call vect_determine_precisions.
11533 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11535 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
11536 statements that have been replaced by further pattern statements.
11537 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
11539 2018-07-03 Richard Biener <rguenther@suse.de>
11541 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
11542 always set *dt. Dump vectype in vectype overload.
11543 * dumpfile.h (dump_gimple_expr): New function.
11544 (dump_gimple_expr_loc): Likewise.
11545 * dumpfile.c (dump_gimple_expr): New function.
11546 (dump_gimple_expr_loc): Likewise.
11548 2018-07-02 Jeff Law <law@redhat.com>
11550 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11551 the H8/300, H8/300H and H8/S variants into a single pattern.
11552 (movhi_h8300, movqi_h8300hs): Similarly.
11553 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11554 (QHI mode iterator): New.
11556 * config/h8300/h8300.md: Remove trailing whitespace.
11558 2018-07-02 Jim Wilson <jimw@sifive.com>
11560 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11561 instead of emit_insn for interrupt returns.
11562 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11563 (riscv_sret, riscv_uret): Likewise.
11565 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11567 * pretty-print.c (selftest::test_pp_format): Move save and restore
11568 of quotes to class auto_fix_quotes, and add an instance.
11569 * selftest.c: Include "intl.h".
11570 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11571 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11572 * selftest.h (selftest::auto_fix_quotes): New class.
11574 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11576 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11577 (aarch64_sve_prepare_conditional_op): Remove.
11578 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11579 Allow aarch64_simd_reg_or_zero as select operand; remove
11580 the aarch64_sve_prepare_conditional_op call.
11581 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11582 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11583 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11584 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11585 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11586 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11587 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11588 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11589 and a splitters to match all of the *_any patterns.
11590 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11592 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11593 (SVE_COND_FP_BINARY_REV): Remove.
11594 (sve_int_op_rev, sve_fp_op_rev): New.
11595 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11596 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11597 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11598 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11599 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11600 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11601 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11602 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11603 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11605 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11606 Remove match_dup 1 from the inner unspec.
11607 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11609 * config/aarch64/aarch64.md (movprfx): New attr.
11610 (length): Default movprfx to 8.
11611 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11612 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11613 (*<su>mul<SVE_I>3_highpart): Likewise.
11614 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11615 (*v<ASHIFT><SVE_I>3): Likewise.
11616 (*<su><MAXMIN><SVE_I>3): Likewise.
11617 (*<su><MAXMIN><SVE_F>3): Likewise.
11618 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11619 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11620 (*div<SVE_F>4): Likewise.
11622 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11624 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11627 2018-07-02 Richard Biener <rguenther@suse.de>
11629 PR tree-optimization/86363
11630 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11631 memset argument refers to a non-variable address.
11633 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11635 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11636 {PLUS,MINUS}_EXPR code to...
11637 (adjust_symbolic_bound): ...here,
11638 (combine_bound): ...here,
11639 (set_value_range_with_overflow): ...and here.
11641 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11643 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11645 (extract_range_from_abs_expr): ...here.
11647 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11649 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11650 -fno-omit-frame-pointer when not optimizing.
11652 2018-07-02 Martin Liska <mliska@suse.cz>
11655 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11656 (propagate_nothrow): Likewise.
11658 2018-07-02 Martin Liska <mliska@suse.cz>
11661 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11663 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11665 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11666 function in r262149, changing "loc" param from source_location to
11667 const dump_location_t &.
11668 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11669 declaration, as above.
11671 2018-07-01 Paul Koning <ni1d@arrl.net>
11673 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11674 -munit-asm, -mgnu-asm, -mdec-asm.
11675 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11676 (pdp11_output_labelref): New.
11677 (pdp11_output_def): New.
11678 (pdp11_output_addr_vec_elt): New.
11679 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11680 %# and %@ format codes.
11681 (pdp11_option_override): New.
11682 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11683 (pdp11_output_ident): New.
11684 (pdp11_asm_named_section): New.
11685 (pdp11_asm_init_sections): New.
11686 (pdp11_file_start): New.
11687 (pdp11_file_end): New.
11688 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11689 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11690 %o, like %c but octal.
11691 (pdp11_option_override): New.
11692 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11694 (DATA_SECTION_ASM_OP): Ditto.
11695 (READONLY_DATA_SECTION_ASM_OP): New.
11696 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11697 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11698 (ASM_OUTPUT_LABELREF): Ditto.
11699 (ASM_OUTPUT_DEF): Ditto.
11700 (ASM_OUTPUT_EXTERNAL): New.
11701 (ASM_OUTPUT_SOURCE_FILENAME): New.
11702 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11703 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11704 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11705 %# and %@ format codes.
11706 * config/pdp11/pdp11.opt (mgnu-asm): New.
11707 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11708 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11709 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11711 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11713 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11714 dereferencing path[] beyond its length.
11716 (debug_all_paths): New.
11717 (rewire_first_differing_edge): New.
11718 (adjust_paths_after_duplication): New.
11719 (duplicate_thread_path): Call adjust_paths_after_duplication.
11721 (thread_through_all_blocks): Add new argument to
11722 duplicate_thread_path.
11724 2018-06-30 Jim Wilson <jimw@sifive.com>
11726 * config/riscv/predicates.md (p2m1_shift_operand): New.
11727 (high_mask_shift_operand): New.
11728 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11729 pattern using p2m1_shift_operand.
11730 (lshsi3_zero_extend_3+2): New combiner pattern using
11731 high_mask_shift_operand.
11733 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11735 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11737 (vect_recog_rotate_pattern): ...here.
11738 (vect_convert_input): Try to insert casts of invariants in the
11740 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11741 preheader to be empty.
11743 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11745 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11746 vector type. If given, install it in the new statement's
11747 STMT_VINFO_VECTYPE.
11748 (vect_element_precision): New function.
11749 (vect_unpromoted_value): New struct.
11750 (vect_unpromoted_value::vect_unpromoted_value): New function.
11751 (vect_unpromoted_value::set_op): Likewise.
11752 (vect_look_through_possible_promotion): Likewise.
11753 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11754 (vect_widened_op_tree, vect_convert_input): Likewise.
11755 (vect_convert_inputs, vect_convert_output): Likewise.
11756 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11757 to handle the optional cast of the multiplication result and
11758 vect_widened_op_tree to detect the widened multiplication itself.
11759 Do not require the input and output of promotion casts to have
11760 the same sign, but base the signedness of the operation on the
11761 input rather than the result. If the pattern includes two
11762 promotions, check that those promotions have the same sign.
11763 Do not restrict the MULT_EXPR handling to a double-width result;
11764 handle quadruple-width results and wider. Use vect_convert_inputs
11765 to convert the inputs to the common type.
11766 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11767 to handle the optional cast of the ABS result. Also allow a sign
11768 change or a sign extension between the ABS and MINUS.
11769 Use vect_widened_op_tree to detect the widened subtraction and use
11770 vect_convert_inputs to convert the inputs to the common type.
11771 (vect_handle_widen_op_by_const): Delete.
11772 (vect_recog_widen_op_pattern): New function.
11773 (vect_recog_widen_mult_pattern): Use it.
11774 (vect_recog_widen_shift_pattern): Likewise.
11775 (vect_recog_widen_sum_pattern): Use
11776 vect_look_through_possible_promotion to handle the promoted
11779 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11781 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11782 the containing gimple_seq *.
11783 * gimple-iterator.h (gsi_for_stmt): Declare it.
11784 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11785 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11786 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11787 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11788 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11790 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11792 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11793 statement being replaced is part of an existing pattern
11794 definition sequence, inserting the new pattern statements before
11796 (vect_pattern_recog_1): Don't return a bool. If the statement
11797 is already part of a pattern, instead apply pattern matching
11798 to the pattern definition statements. Don't clear the
11799 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11800 (vect_pattern_recog): Don't break after the first match;
11801 continue processing the pattern definition statements instead.
11802 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11804 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11806 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11807 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11808 (vect_recog_widen_sum_pattern): Use it.
11810 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11812 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11813 phi is not a pattern statement and has not been replaced by
11814 a pattern statement.
11815 * tree-vect-patterns.c (type_conversion_p): Don't check
11816 STMT_VINFO_IN_PATTERN_P.
11817 (vect_recog_vector_vector_shift_pattern): Likewise.
11818 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11819 the pattern statement rather than the original statement; check
11820 directly for a WIDEN_MULT_EXPR here.
11821 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11822 vect_is_simple_use to return the pattern statement rather
11823 than the original statement; use is_pattern_stmt_p to check
11824 for such a pattern statement.
11825 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11826 to return the pattern statement rather than the original statement;
11827 don't do the same transformation here.
11828 (vect_is_simple_use): If the defining statement has been replaced
11829 by a pattern statement, return the pattern statement instead.
11830 Remove the corresponding (local) transformation from the vectype
11833 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11835 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11836 end and default to null.
11837 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11838 (vectorizable_reduction): Update calls accordingly, dropping the
11839 gimple ** argument if the passed-back statement isn't needed.
11840 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11841 (vect_recog_rotate_pattern): Likewise.
11842 (vect_recog_mask_conversion_pattern): Likewise.
11843 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11844 (vect_mask_constant_operand_p): Likewise.
11845 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11846 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11847 (get_group_load_store_type, get_load_store_type): Likewise.
11848 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11849 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11850 (vectorizable_conversion, vectorizable_assignment): Likewise.
11851 (vectorizable_shift, vectorizable_operation): Likewise.
11852 (vectorizable_store, vect_is_simple_cond): Likewise.
11853 (vectorizable_condition, vectorizable_comparison): Likewise.
11854 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11855 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11856 and move it to the end. Cope with null def_stmt_outs.
11858 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11860 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11862 2018-06-29 Jeff Law <law@redhat.com>
11864 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11865 displacements for TARGET_V850E2V3 and newer.
11866 (TARGET_LRA_P): Remove. Defaults to LRA now.
11867 * config/v850/v850.md (sign23byte_load): Remove.
11868 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11869 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11871 2018-06-29 Martin Liska <mliska@suse.cz>
11874 * coverage.c (coverage_init): Mangle full path name.
11875 * doc/invoke.texi: Document the change.
11876 * gcov-io.c (mangle_path): New.
11877 * gcov-io.h (mangle_path): Likewise.
11878 * gcov.c (mangle_name): Use mangle_path for path mangling.
11880 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11882 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11883 if starting source register is not even.
11885 2018-06-29 Martin Liska <mliska@suse.cz>
11887 PR tree-optimization/86263
11888 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11889 Make edge redirection.
11891 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11893 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11894 (dump_scope_depth): New variable.
11895 (get_dump_scope_depth): New function.
11896 (dump_begin_scope): New function.
11897 (dump_end_scope): New function.
11898 * dumpfile.h (get_dump_scope_depth): New declaration.
11899 (dump_begin_scope): New declaration.
11900 (dump_end_scope): New declaration.
11901 (class auto_dump_scope): New class.
11902 (AUTO_DUMP_SCOPE): New macro.
11903 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11906 2018-06-29 Richard Biener <rguenther@suse.de>
11908 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11909 compute_all_dependences succeeds.
11910 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11911 exceed --param loop-max-datarefs-for-datadeps.
11913 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11915 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11917 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11920 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11921 alternative 0 in preferred_for_speed attribute.
11923 2018-06-28 Paul Koning <ni1d@arrl.net>
11925 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11926 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11927 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11928 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11929 for shift insn patterns.
11931 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11933 * cgraph.c (cgraph_node::get_body): Replace assignments to
11934 "dump_file" with calls to set_dump_file.
11935 * dumpfile.c (alt_dump_file): Make static, and group with...
11936 (alt_flags): ...this definition.
11937 (dumps_are_enabled): New variable.
11938 (refresh_dumps_are_enabled): New function.
11939 (set_dump_file): New function.
11940 (set_alt_dump_file): New function.
11941 (gcc::dump_manager::dump_start): Replace assignments to
11942 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11944 (gcc::dump_manager::dump_finish): Likewise.
11945 * dumpfile.h (alt_dump_file): Delete decl.
11946 (dumps_are_enabled): New variable decl.
11947 (set_dump_file): New function decl.
11948 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11950 * tree-nested.c (lower_nested_functions): Replace assignments to
11951 "dump_file" with calls to set_dump_file.
11953 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11955 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11956 goto_locus of each outgoing edge of each basic block.
11958 2018-06-28 Richard Biener <rguenther@suse.de>
11960 * dwarf2out.c (decl_scope_table): Remove.
11961 (push_decl_scope): Likewise.
11962 (pop_decl_scope): Likewise.
11963 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11964 (gen_struct_or_union_type_die): Likewise.
11965 (gen_tagged_type_die): Likewise.
11966 (dwarf2out_init): Do not initialize decl_scope_table.
11967 (dwarf2out_c_finalize): Do not free it.
11969 2018-06-28 Richard Biener <rguenther@suse.de>
11971 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11972 deciding whether to not re-use a DIE.
11974 2018-06-28 Richard Biener <rguenther@suse.de>
11976 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11977 DW_AT_abstract_origin attribute.
11979 2018-06-28 Martin Liska <mliska@suse.cz>
11981 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11982 Use newly introduced constants.
11983 * tree-switch-conversion.h (struct jump_table_cluster):
11984 Define max_ratio_for_size and max_ratio_for_speed.
11986 2018-06-28 Martin Liska <mliska@suse.cz>
11988 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11989 Add new checking assert to catch invalid state.
11990 (jump_table_cluster::can_be_handled): Handle single case
11992 (jump_table_cluster::is_beneficial): Bail out for such case.
11993 (bit_test_cluster::find_bit_tests):
11994 Add new checking assert to catch invalid state.
11995 (bit_test_cluster::can_be_handled): Handle single case
11997 (bit_test_cluster::is_beneficial): Bail out for such case.
11998 (switch_decision_tree::analyze_switch_statement):
12001 2018-06-28 Martin Liska <mliska@suse.cz>
12003 * common.opt: Introduce -completion option.
12004 * gcc.c (driver_handle_option): Handle it.
12005 (driver::main): Print completions if completion
12007 * opt-suggestions.c (option_proposer::get_completions):
12009 (option_proposer::suggest_completion): Likewise.
12010 (option_proposer::find_param_completions): Likewise.
12011 (verify_autocompletions): Likewise.
12012 (test_completion_valid_options): Likewise.
12013 (test_completion_valid_params): Likewise.
12014 (in_completion_p): Likewise.
12015 (empty_completion_p): Likewise.
12016 (test_completion_partial_match): Likewise.
12017 (test_completion_garbage): Likewise.
12018 (opt_proposer_c_tests): Likewise.
12019 * opt-suggestions.h: Declare new functions.
12020 * opts.c (common_handle_option): Handle OPT__completion_.
12021 * selftest-run-tests.c (selftest::run_tests): Add
12022 opt_proposer_c_tests.
12023 * selftest.c (assert_str_startswith): New.
12024 * selftest.h (assert_str_startswith): Likewise.
12025 (opt_proposer_c_tests): New.
12026 (ASSERT_STR_STARTSWITH): Likewise.
12028 2018-06-28 Martin Liska <mliska@suse.cz>
12030 * Makefile.in: Add opt-suggestions.o.
12031 * gcc-main.c: Include opt-suggestions.h.
12032 * gcc.c (driver::driver): Likewise.
12033 (driver::~driver): Remove m_option_suggestions.
12034 (driver::build_option_suggestions): Moved to option_proposer.
12035 (driver::suggest_option): Likewise.
12036 (driver::handle_unrecognized_options): Use option_proposer.
12037 * gcc.h (class driver): Add new memver m_option_proposer.
12038 * opt-suggestions.c: New file.
12039 * opt-suggestions.h: New file.
12041 2018-06-28 Martin Liska <mliska@suse.cz>
12043 * vec.h (class auto_string_vec): New (moved from auto_argvec).
12044 (auto_string_vec::~auto_string_vec): Likewise.
12046 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
12048 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
12049 prevent_decl_creation_for_types fields up and add reset_location field.
12050 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
12051 statement if id->reset_location is true.
12052 (copy_edges_for_bb): Do not set goto_locus on the new edges if
12053 id->reset_location is true.
12054 (copy_phis_for_bb): Force input_location on the arguments if
12055 id->reset_location is true.
12056 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
12057 is set on the function to be inlined.
12059 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
12061 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
12063 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
12065 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
12066 registers for Pmode.
12067 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
12068 hard registers for the clobbered pseudo.
12070 2018-06-27 Paul Koning <ni1d@arrl.net>
12072 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
12073 mutually exclusive options.
12074 * config/pdp11/constraints.md (h): New constraint.
12075 (O): Update definition to match shift code generation.
12076 (D): New constraint.
12077 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
12079 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
12081 (output_jump): Change arguments.
12082 (pdp11_fixed_cc_regs): New function.
12083 (pdp11_cc_mode): Ditto.
12084 (pdp11_expand_shift): Ditto.
12085 (pdp11_assemble_shift): Ditto.
12086 (pdp11_small_shift): Ditto.
12087 (pdp11_branch_cost): Remove.
12088 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
12090 (pdp11_register_move_cost): Update for CC registers.
12091 (pdp11_rtx_costs): Add case for LSHIFTRT.
12092 (pdp11_output_jump): Add CCNZ mode conditional branches.
12093 (notice_update_cc_on_set): Remove.
12094 (pdp11_cc_mode): New function.
12095 (simple_memory_operand): Correct pre/post decrement case.
12096 (no_side_effect_operand): New function.
12097 (pdp11_regno_reg_class): Add CC_REGS class.
12098 (pdp11_fixed_cc_regs): New function.
12099 (pdp11_small_shift): New function.
12100 (pdp11_expand_shift): New function to expand shift insns.
12101 (pdp11_assemble_shift): New function to output shifts.
12102 (pdp11_branch_cost): Remove.
12103 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
12104 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
12105 (WCHAR_TYPE): Ditto.
12106 (PTRDIFF_TYPE): Ditto.
12107 (ADJUST_INSN_LENGTH): New macro.
12108 (FIXED_REGISTERS): Add CC registers.
12109 (CALL_USED_REGISTERS): Ditto.
12110 (reg_class): Ditto.
12111 (REG_CLASS_NAMES): Ditto.
12112 (REG_CLASS_CONTENTS): Ditto.
12113 (SELECT_CC_MODE): Use new function.
12114 (TARGET_FLAGS_REGNUM): New macro.
12115 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
12116 (cc0_reg_rtx): Remove.
12117 (CC_STATUS_MDEP): Remove.
12118 (CC_STATUS_MDEFP_INIT): Remove.
12119 (CC_IN_FPU): Remove.
12120 (NOTICE_UPDATE_CC): Remove.
12121 (REGISTER_NAMES): Add CC registers.
12122 (BRANCH_COST): Change to constant 1.
12123 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
12125 * config/pdp11/pdp11.opt (mbcopy): Remove.
12126 (mbcopy-builtin): Remove.
12127 (mbranch-cheap): Remove.
12128 (mbranch-expensive): Remove.
12129 * config/pdp11/predicates.md (expand_shift_operand): Update to
12130 match shift code generation.
12131 (ccnz_operator): New predicate.
12132 * doc/invoke.texi (PDP-11 Options): Remove deleted options
12133 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
12134 Remove non-existent option -mabshi, -mno-abshi. Document mutually
12136 * doc/md.texi (PDP-11): Document new D and h constraints. Update
12137 description of O constraint.
12139 2018-06-27 Jeff Law <law@redhat.com>
12140 Austin Law <austinklaw@gmail.com>
12142 * config/v850/v850.md (addsi3_set_flags): New pattern.
12143 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
12144 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
12145 (zero_extendhisi2_v850_set_flags): Likewise.
12146 (zero_extendqisi2_v850_set_flags): Likewise.
12147 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
12148 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
12149 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
12151 * config/v850/v850-protos.h (notice_update_cc): Remove.
12152 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
12153 (v850_print_operand): Handle 'D' and "d".
12154 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
12155 Add handling of arithmetic/logical operations compared against zero.
12156 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
12157 Do not look at v850_compare_op, instead get mode from last argument.
12158 (v850_gen_compare): Remove
12159 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
12160 after reload for prologue insns.
12161 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
12163 (construct_save_jarl): Likewise.
12164 (TARGET_FLAGS_REGNUM): Define.
12165 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
12166 (NOTICE_UPDATE_CC): Remove.
12167 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
12168 than cc0. Conditionalize on reload_completed.
12169 (cmpsi_insn, setfcc_insn): Likewise.
12170 (tst1 splitter): Turn into define_and_split which sets the flags
12172 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
12173 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
12174 (cstoresf4, cstoredf4): Clobber the flags.
12175 (cmpsi, cmpsf, cmpdf): Remove expanders.
12176 (setf_insn): Remove pattern.
12177 (addsi3): Turn into define_and_split which clobbers the flags after
12178 reload and a suitable pattern (addsi3_clobber_flags) for use after
12180 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
12181 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
12182 (ashrsi3, ashrsi3_v850e2): Likewise.
12183 (bins): Clobber the flags.
12184 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
12185 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
12186 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
12187 (call_value_internal_short, call_value_internal_long): Likewise.
12188 (callt_save_interrupt, callt_return_interrupt): Likewise.
12189 (save_interrupt, return_interrupt): Likewise.
12190 (callt_save_all_interrupt, save_all_interrupt): Likewise.
12191 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
12192 (restore_all_interrupt, _restore_all_interrupt): Likewise.
12193 (All FP comparisons): Only allow after reload has completed.
12195 (divh, divhu): Tweak output template.
12196 (branch_z_normal, branch_z_invert): Remove
12197 (branch_nz_normal, branch_nz_invert): Likewise.
12198 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
12200 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
12201 * config/v850/v850.c (notice_update_cc): Remove.
12202 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
12203 (CC_NO_CARRY): Likewise.
12204 (NOTICE_UPDATE_CC): Define to nothing.
12205 * config/v850/v850.md: Remove block comment on cc0 handling
12206 Remove "cc" attribute from all patterns. Remove cc_status handling
12207 from all patterns. Minor formatting fixes.
12209 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12211 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
12212 (cortex-a76.cortex-a55): Likewise.
12213 * config/aarch64/aarch64-tune.md: Regenerate.
12214 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
12215 cortex-a76.cortex-a55.
12217 2018-06-27 Jeff Law <law@redhat.com>
12219 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
12220 (MULTILIB_DIRNAMES): Similarly.
12222 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
12224 * gimple.h (gimple_return_retbnd): Delete.
12225 (gimple_return_set_retbnd): Likewise.
12226 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
12227 gimple_return_set_retbnd.
12228 * gimple-pretty-print.c (dump_gimple_return): Remove call to
12229 gimple_return_retbnd and adjust.
12230 * tree-inline.h (struct copy_body_data): Remove retbnd field.
12231 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
12232 Explicitly return NULL in a couple more cases. Move assertion
12233 on debug statements and remove unreachable code.
12234 (reset_debug_binding): Do not test id->retbnd.
12235 (expand_call_inline): Do not set it.
12237 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
12239 * configure.ac: Add --disable-gcov option.
12240 * configure: Regenerate.
12241 * Makefile.in: Honour @enable_gcov@.
12242 * doc/install.texi: Document --disable-gcov.
12244 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12246 * config/arm/arm-cpus.in (cortex-a76): New entry.
12247 (cortex-a76.cortex-a55): Likewise.
12248 * config/arm/arm-tables.opt: Regenerate.
12249 * config/arm/arm-tune.md: Likewise.
12250 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
12251 * doc/invoke.texi (ARM Options): Document cortex-a76 and
12252 cortex-a76.cortex-a55.
12254 2018-06-27 Tamar Christina <tamar.christina@arm.com>
12257 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
12259 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
12261 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
12263 (EPILOGUE_USES): Likewise.
12265 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
12267 * tree-inline.c (remap_location): New function extracted from...
12268 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
12269 (copy_phis_for_bb): ...here. Call remap_location.
12270 (copy_cfg_body): Adjust call to copy_edges_for_bb.
12272 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
12274 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
12275 unaligned vsx for 16B memset.
12277 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
12280 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
12281 ieee128_float_type_node to long_double_type_node unless
12282 TARGET_LONG_DOUBLE_128 is set.
12284 2018-06-26 David Malcolm <dmalcolm@redhat.com>
12286 * cfgloop.c (get_loop_location): Convert return type from
12287 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
12288 by implicit construction from rtx_insn *, and using
12289 dump_user_location_t::from_function_decl for the fallback case.
12290 * cfgloop.h (get_loop_location): Convert return type from
12291 location_t to dump_user_location_t.
12292 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
12293 dump_printf_loc to pass in a dump_location_t rather than a
12294 location_t, via the gimple stmt.
12295 * coverage.c (get_coverage_counts): Update calls to
12296 dump_printf_loc to pass in dump_location_t rather than a
12298 * doc/optinfo.texi (Dump types): Convert example of
12299 dump_printf_loc from taking "locus" to taking "insn". Update
12300 description of the "_loc" calls to cover dump_location_t.
12301 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
12303 (dump_user_location_t::dump_user_location_t): New constructors,
12304 from gimple *stmt and rtx_insn *.
12305 (dump_user_location_t::from_function_decl): New function.
12306 (dump_loc): Make static.
12307 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
12308 const dump_location_t &.
12309 (dump_generic_expr_loc): Delete.
12310 (dump_printf_loc): Convert param "loc" from location_t to
12311 const dump_location_t &.
12312 (selftest::test_impl_location): New function.
12313 (selftest::dumpfile_c_tests): New function.
12314 * dumpfile.h: Include "profile-count.h".
12315 (class dump_user_location_t): New class.
12316 (struct dump_impl_location_t): New struct.
12317 (class dump_location_t): New class.
12318 (dump_printf_loc): Convert 2nd param from source_location to
12319 const dump_location_t &.
12320 (dump_generic_expr_loc): Delete.
12321 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
12322 const dump_location_t &.
12323 * gimple-fold.c (fold_gimple_assign): Update call to
12324 dump_printf_loc to pass in a dump_location_t rather than a
12325 location_t, via the gimple stmt.
12326 (gimple_fold_call): Likewise.
12327 * gimple-loop-interchange.cc
12328 (loop_cand::analyze_iloop_reduction_var): Update for change to
12329 check_reduction_path.
12330 (tree_loop_interchange::interchange): Update for change to
12331 find_loop_location.
12332 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
12333 change in return-type of find_loop_location.
12334 (graphite_regenerate_ast_isl): Likewise.
12335 * graphite-optimize-isl.c (optimize_isl): Likewise.
12336 * graphite.c (graphite_transform_loops): Likewise.
12337 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
12338 pass in a dump_location_t rather than a location_t, via the
12340 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12341 * ipa.c (walk_polymorphic_call_targets): Likewise.
12342 * loop-unroll.c (report_unroll): Convert "locus" param from
12343 location_t to dump_location_t.
12344 (decide_unrolling): Update for change to get_loop_location's
12346 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
12347 location_t to dump_user_location_t.
12348 (grid_find_single_omp_among_assignments_1): Updates calls to
12349 dump_printf_loc to pass in a dump_location_t rather than a
12350 location_t, via the gimple stmt.
12351 (grid_parallel_clauses_gridifiable): Convert "tloc" from
12352 location_t to dump_location_t. Updates calls to dump_printf_loc
12353 to pass in a dump_location_t rather than a location_t, via the
12355 (grid_inner_loop_gridifiable_p): Likewise.
12356 (grid_dist_follows_simple_pattern): Likewise.
12357 (grid_gfor_follows_tiling_pattern): Likewise.
12358 (grid_target_follows_gridifiable_pattern): Likewise.
12359 (grid_attempt_target_gridification): Convert initialization
12360 of local "grid" from memset to zero-initialization; FIXME: does
12361 this require C++11? Update call to dump_printf_loc to pass in a
12362 optinfo_location rather than a location_t, via the gimple stmt.
12363 * profile.c (read_profile_edge_counts): Updates call to
12364 dump_printf_loc to pass in a dump_location_t rather than a
12366 (compute_branch_probabilities): Likewise.
12367 * selftest-run-tests.c (selftest::run_tests): Call
12369 * selftest.h (dumpfile_c_tests): New decl.
12370 * tree-loop-distribution.c (pass_loop_distribution::execute):
12371 Update for change in return type of find_loop_location.
12372 * tree-parloops.c (parallelize_loops): Likewise.
12373 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
12374 "locus" from location_t to dump_user_location_t.
12375 (canonicalize_loop_induction_variables): Likewise.
12376 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
12377 for change in return type of find_loop_location.
12378 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
12379 to dump_printf_loc to pass in a dump_location_t rather than a
12380 location_t, via the stmt.
12381 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
12383 * tree-vect-loop-manip.c (find_loop_location): Convert return
12384 type from source_location to dump_user_location_t.
12385 (vect_do_peeling): Update for above change.
12386 (vect_loop_versioning): Update for change in type of
12388 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
12389 from location_t to dump_user_location_t.
12390 (vect_estimate_min_profitable_iters): Update for change in type
12392 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
12393 location_t to dump_location_t.
12394 (vect_slp_bb): Update for change in type of vect_location.
12395 * tree-vectorizer.c (vect_location): Convert from source_location
12396 to dump_user_location_t.
12397 (try_vectorize_loop_1): Update for change in vect_location's type.
12398 (vectorize_loops): Likewise.
12399 (increase_alignment): Likewise.
12400 * tree-vectorizer.h (vect_location): Convert from source_location
12401 to dump_user_location_t.
12402 (find_loop_location): Convert return type from source_location to
12403 dump_user_location_t.
12404 (check_reduction_path): Convert 1st param from location_t to
12405 dump_user_location_t.
12406 * value-prof.c (check_counter): Update call to dump_printf_loc to
12407 pass in a dump_user_location_t rather than a location_t; update
12408 call to error_at for change in type of "locus".
12409 (check_ic_target): Update call to dump_printf_loc to
12410 pass in a dump_user_location_t rather than a location_t, via the
12413 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
12415 * config/s390/s390.h (enum processor_flags): Do not use
12416 default tune parameter when -march was specified.
12418 2018-06-26 Jakub Jelinek <jakub@redhat.com>
12421 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
12422 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
12425 2018-06-26 Richard Biener <rguenther@suse.de>
12427 PR tree-optimization/86287
12429 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
12430 (vect_analyze_loop): Initialize n_stmts.
12432 2018-06-26 Richard Biener <rguenther@suse.de>
12434 PR middle-end/86271
12435 * fold-const.c (fold_convertible_p): Pointer extension
12438 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
12441 * dwarf2out.c (loc_list_has_views): Adjust comments.
12442 (dw_loc_list): Split single cross-partition range with
12445 2018-06-25 Jeff Law <law@redhat.com>
12447 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
12448 on -mbig-switch by default.
12450 * config/v850/predicates.md (const_float_1_operand): Fix match_code
12452 (const_float_0_operand): Remove unused predicate.
12453 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
12454 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
12455 (recipsf2): New expander. Original pattern now called
12457 (recipdf2, recipdf2_insn): Similarly.
12458 (rsqrtsf2, rsqrtsf2_insn): Similarly
12459 (rsqrtdf2, rsqrtdf2_insn): Similarly
12461 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
12463 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
12464 Simplify logic for FreeBSD (twice).
12466 2018-06-25 Martin Sebor <msebor@redhat.com>
12468 PR tree-optimization/86204
12469 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
12470 a strnlen result if it's less than the length of the string.
12472 2018-06-25 Martin Sebor <msebor@redhat.com>
12474 PR tree-optimization/85700
12475 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
12476 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
12477 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
12479 2018-06-25 Martin Sebor <msebor@redhat.com>
12481 * doc/extend.texi (Zero-length arrays): Update and clarify.
12483 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
12485 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
12486 added IEEE/IBM long double multilib support on PowerPC little
12487 endian Linux systems.
12488 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
12489 (MULTILIB_DEFAULTS): Likewise.
12490 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12492 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
12493 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
12494 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
12496 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
12498 PR middle-end/86311
12499 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
12500 (REORDER_45): Likewise.
12502 2018-06-25 Jeff Law <law@redhat.com>
12504 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
12505 dividend to 32 bits. Adjust length.
12506 (udivmodhi4): Cleanup output template. Fix length.
12508 2018-06-25 Carl Love <cel@us.ibm.com>
12510 * config/rs6000/vsx.md: Change word selector to prefered location.
12512 2018-06-25 Richard Biener <rguenther@suse.de>
12514 PR tree-optimization/86304
12515 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
12516 epilogue-if-converted loops as well.
12518 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
12520 * lto-section-out.c (lto_begin_section): Do not print section
12521 name for noaddr and unnumbered dumps.
12523 2018-06-25 Richard Biener <rguenther@suse.de>
12525 * tree-vectorizer.h (struct vec_info_shared): New structure
12526 with parts split out from struct vec_info and loop_nest from
12527 struct _loop_vec_info.
12528 (struct vec_info): Adjust accordingly.
12529 (struct _loop_vec_info): Likewise.
12530 (LOOP_VINFO_LOOP_NEST): Adjust.
12531 (LOOP_VINFO_DATAREFS): Likewise.
12532 (LOOP_VINFO_DDRS): Likewise.
12533 (struct _bb_vec_info): Likewise.
12534 (BB_VINFO_DATAREFS): Likewise.
12535 (BB_VINFO_DDRS): Likewise.
12536 (struct _stmt_vec_info): Add dr_aux member.
12537 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
12538 (DR_MISALIGNMENT_UNINITIALIZED): New.
12539 (set_dr_misalignment): Adjust.
12540 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
12541 (vect_analyze_loop): Adjust prototype.
12542 (vect_analyze_loop_form): Likewise.
12543 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12544 Compute dependences lazily.
12545 (vect_record_base_alignments): Use shared datarefs/ddrs.
12546 (vect_verify_datarefs_alignment): Likewise.
12547 (vect_analyze_data_refs_alignment): Likewise.
12548 (vect_analyze_data_ref_accesses): Likewise.
12549 (vect_analyze_data_refs): Likewise.
12550 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12551 constructor parameter for shared part.
12552 (vect_analyze_loop_form): Pass in shared part and adjust.
12553 (vect_analyze_loop_2): Pass in storage for the number of
12554 stmts. Move loop nest finding to the caller. Compute
12556 (vect_analyze_loop): Pass in shared part.
12557 (vect_transform_loop): Verify shared datarefs are unchanged.
12558 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12559 constructor parameter for shared part.
12560 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12561 (vect_slp_bb): Verify shared datarefs are unchanged before
12563 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12565 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12566 DR_MISALIGNMENT_UNINITIALIZED.
12567 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12568 parameter for shared part.
12569 (vec_info::~vec_info): Adjust.
12570 (vec_info_shared::vec_info_shared): New.
12571 (vec_info_shared::~vec_info_shared): Likewise.
12572 (vec_info_shared::save_datarefs): Likewise.
12573 (vec_info_shared::check_datarefs): Likewise.
12574 (try_vectorize_loop_1): Construct shared part live for analyses
12575 of a single loop for multiple vector sizes.
12576 * tree-parloops.c (gather_scalar_reductions): Adjust.
12578 2018-06-25 Richard Biener <rguenther@suse.de>
12580 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12581 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12582 (vect_analyze_data_refs): Remove similar code from here and
12583 simplify accordingly.
12585 2018-06-25 Richard Biener <rguenther@suse.de>
12587 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12588 for reverse storage order accesses rather than asserting
12589 they cannot happen here.
12591 2018-06-25 Tom de Vries <tdevries@suse.de>
12594 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12595 Use data16 instead of .byte for insn prefix.
12597 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12600 * parser.c (make_char_string_pack): Pass this literal chars
12601 through cpp_interpret_string.
12602 (cp_parser_userdef_numeric_literal): Check the result of
12603 make_char_string_pack.
12605 2018-06-24 Maya Rashish <coypu@sdf.org>
12607 * ginclude/stddef.h: Simplify conditions around avoiding
12608 re-definition of __size_t.
12610 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12612 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12613 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12615 2018-06-22 Maya Rashish <coypu@sdf.org>
12617 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12618 made to OpenBSD and NetBSD through the years.
12620 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12622 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12623 behavior of vec_pack (vector double, vector double) to match
12624 behavior of vec_float2 (vector double, vector double).
12626 2018-06-22 Olivier Hainque <hainque@adacore.com>
12628 * gimplify.c (gimplify_function_tree): Prevent creation
12629 of a trampoline for the address of the current function
12630 passed to entry/exit instrumentation hooks.
12632 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12635 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12638 2018-06-22 Martin Liska <mliska@suse.cz>
12640 PR tree-optimization/86263
12641 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12642 Bail out if is_enabled is false.
12643 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12645 (jump_table_cluster::is_enabled): New function.
12647 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12649 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12650 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12651 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12652 (lto_input_ts_binfo_tree_pointers): Likewise.
12653 * tree-streamer-out.c (streamer_write_tree_bitfields,
12654 write_ts_binfo_tree_pointers): Likewise.
12655 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12657 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12659 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12661 2018-06-22 Martin Liska <mliska@suse.cz>
12663 * symbol-summary.h (get): Make it pure and inline move
12664 functionality from ::get function.
12665 (get): Remove and inline into ::get and ::get_create.
12666 (get_create): Move code from ::get function.
12668 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12671 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12672 -x assembler-with-cpp.
12674 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12676 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12677 _FILE_OFFSET_BITS=64 for C++.
12679 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12681 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12682 conversion insn that shows up when pr85657-3.c is compiled using
12683 IEEE 128-bit long double.
12684 (neg<mode>2_internal): Use the correct mode to check whether the
12685 mode is IBM extended.
12686 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12687 multiply and divide external functions from being created more
12690 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12692 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12694 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12695 the edge can be forwarded.
12696 (cfg_layout_merge_blocks): Likewise.
12698 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12700 * except.c (finish_eh_generation): Commit edge insertions only after
12701 the EH edges have been redirected from post-landing to landing pads.
12703 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12705 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12706 create_tmp_var_for to create the FRAME decl.
12707 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12709 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12711 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12712 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12713 debug statement when resetting its value.
12714 (expand_call_inline): Copy the locus of the call onto the assignment
12715 of the return value, if any. Use local variable in more cases.
12717 2018-06-21 Martin Liska <mliska@suse.cz>
12719 * ipa-pure-const.c (propagate_nothrow): Use
12720 funct_state_summaries->get.
12721 (dump_malloc_lattice): Likewise.
12722 (propagate_malloc): Likewise.
12724 2018-06-21 Richard Biener <rguenther@suse.de>
12726 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12727 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12728 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12730 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12731 BLOCK_ABSTRACT_ORIGIN unconditionally.
12733 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12735 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12737 * ipa-reference.c (ipa_reference_c_finalize): Delete
12738 ipa_ref_opt_sum_summaries and set it to NULL.
12740 2018-06-21 Tom de Vries <tdevries@suse.de>
12742 PR tree-optimization/85859
12743 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12744 test with comment from bb_no_side_effects_p.
12746 2018-06-21 Richard Biener <rguenther@suse.de>
12748 PR tree-optimization/86232
12749 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12750 max for constant niter.
12752 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12754 * config/aarch64/aarch64-simd.md
12755 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12757 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12759 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12760 Make opernads of the unspec commutative.
12762 2018-06-21 Richard Biener <rguenther@suse.de>
12764 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12765 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12766 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12767 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12769 (vect_record_base_alignments): Do not record base alignment
12770 for gathers or scatters.
12771 (vect_compute_data_ref_alignment): Drop return value that is always
12772 true. Bail out early for gathers or scatters.
12773 (vect_enhance_data_refs_alignment): Bail out early for gathers
12775 (vect_find_same_alignment_drs): Likewise.
12776 (vect_analyze_data_refs_alignment): Remove dead code.
12777 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12778 (vect_analyze_data_refs): For possible gathers or scatters do
12779 not create an alternate DR, just check their possible validity
12780 and mark them. Adjust DECL_NONALIASED handling to not rely
12781 on DR_BASE_ADDRESS.
12782 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12783 update inits of gathers or scatters.
12784 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12785 Also copy gather/scatter flag to pattern vinfo.
12787 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12789 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12790 behavior of vec_packsu (vector unsigned long long, vector unsigned
12791 long long) to match behavior of vec_packs with same signature.
12793 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12794 Thomas Schwinge <thomas@codesourcery.com>
12795 Cesar Philippidis <cesar@codesourcery.com>
12797 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12798 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12799 (gimplify_adjust_omp_clauses): Likewise.
12800 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12801 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12802 (gimplify_omp_target_update): Update handling of acc update and
12804 * omp-low.c (install_var_field): Remove unused parameter
12805 base_pointers_restrict.
12806 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12807 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12809 (omp_target_base_pointers_restrict_p): Delete.
12810 (scan_omp_target): Update call to scan_sharing_clauses.
12811 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12813 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12814 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12815 (convert_local_omp_clauses): Likewise.
12816 * tree-pretty-print.c (dump_omp_clause): Likewise.
12817 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12819 (omp_clause_code_name): Likewise.
12821 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12824 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12827 PR tree-optimization/86231
12828 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12829 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12831 2018-06-20 Tom de Vries <tdevries@suse.de>
12833 PR tree-optimization/86097
12834 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12835 iv type if signedness of iv type is not the same as that of *nit.
12837 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12839 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12840 EDGE_EH edges, verify they are all EDGE_EH.
12842 2018-06-20 Maya Rashish <coypu@sdf.org>
12844 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12846 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12848 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12849 * config/aarch64/aarch64.c (xgene1_tunings): Add
12850 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12851 (aarch64_mode_valid_for_sched_fusion_p):
12852 Allow 16-byte modes.
12853 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12854 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12856 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12858 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12859 * config/aarch64/iterators.md (VQ2): New mode iterator.
12861 2018-06-20 Martin Liska <mliska@suse.cz>
12863 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12864 Change default ratio from 10 to 8.
12866 2018-06-20 Martin Liska <mliska@suse.cz>
12868 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12870 (bit_test_cluster::find_bit_tests): Likewise.
12871 (switch_decision_tree::analyze_switch_statement): Find clusters.
12872 * tree-switch-conversion.h (struct jump_table_cluster): Document
12875 2018-06-20 Martin Liska <mliska@suse.cz>
12877 * tree-switch-conversion.c (switch_conversion::collect):
12878 Record m_uniq property.
12879 (switch_conversion::expand): Bail out for special conditions.
12880 (group_cluster::~group_cluster): New.
12881 (group_cluster::group_cluster): Likewise.
12882 (group_cluster::dump): Likewise.
12883 (jump_table_cluster::emit): New.
12884 (switch_decision_tree::fix_phi_operands_for_edges): New.
12885 (struct case_node): Remove struct.
12886 (jump_table_cluster::can_be_handled): New.
12887 (case_values_threshold): Moved to header.
12888 (reset_out_edges_aux): Likewise.
12889 (jump_table_cluster::is_beneficial): New.
12890 (bit_test_cluster::can_be_handled): Likewise.
12891 (add_case_node): Remove.
12892 (bit_test_cluster::is_beneficial): New.
12893 (case_bit_test::cmp): New.
12894 (bit_test_cluster::emit): New.
12895 (expand_switch_as_decision_tree_p): Remove.
12896 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12897 (fix_phi_operands_for_edge): Likewise.
12898 (switch_decision_tree::analyze_switch_statement): New.
12899 (compute_cases_per_edge): Move ...
12900 (switch_decision_tree::compute_cases_per_edge): ... here.
12901 (try_switch_expansion): Likewise.
12902 (switch_decision_tree::try_switch_expansion): Likewise.
12903 (record_phi_operand_mapping): Likewise.
12904 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12905 (emit_case_decision_tree): Likewise.
12906 (switch_decision_tree::emit): Likewise.
12907 (balance_case_nodes): Likewise.
12908 (switch_decision_tree::balance_case_nodes): Likewise.
12909 (dump_case_nodes): Likewise.
12910 (switch_decision_tree::dump_case_nodes): Likewise.
12911 (emit_jump): Likewise.
12912 (switch_decision_tree::emit_jump): Likewise.
12913 (emit_cmp_and_jump_insns): Likewise.
12914 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12915 (emit_case_nodes): Likewise.
12916 (switch_decision_tree::emit_case_nodes): Likewise.
12917 (conditional_probability): Remove.
12918 * tree-switch-conversion.h (enum cluster_type): New.
12920 (struct cluster): Likewise.
12921 (cluster::cluster): Likewise.
12922 (struct simple_cluster): Likewise.
12923 (simple_cluster::simple_cluster): Likewise.
12924 (struct group_cluster): Likewise.
12925 (struct jump_table_cluster): Likewise.
12926 (struct bit_test_cluster): Likewise.
12927 (struct min_cluster_item): Likewise.
12928 (struct case_tree_node): Likewise.
12929 (case_tree_node::case_tree_node): Likewise.
12930 (jump_table_cluster::case_values_threshold): Likewise.
12931 (struct case_bit_test): Likewise.
12932 (struct switch_decision_tree): Likewise.
12933 (struct switch_conversion): Likewise.
12934 (switch_decision_tree::reset_out_edges_aux): Likewise.
12936 2018-06-20 Martin Liska <mliska@suse.cz>
12938 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12939 (hoist_edge_and_branch_if_true): Likewise.
12940 (expand_switch_using_bit_tests_p): Likewise.
12941 (struct case_bit_test): Likewise.
12942 (case_bit_test_cmp): Likewise.
12943 (emit_case_bit_tests): Likewise.
12944 (switch_conversion::switch_conversion): New class.
12945 (struct switch_conv_info): Remove old struct.
12946 (collect_switch_conv_info): More to ...
12947 (switch_conversion::collect): ... this.
12948 (check_range): Likewise.
12949 (switch_conversion::check_range): Likewise.
12950 (check_all_empty_except_final): Likewise.
12951 (switch_conversion::check_all_empty_except_final): Likewise.
12952 (check_final_bb): Likewise.
12953 (switch_conversion::check_final_bb): Likewise.
12954 (create_temp_arrays): Likewise.
12955 (switch_conversion::create_temp_arrays): Likewise.
12956 (free_temp_arrays): Likewise.
12957 (gather_default_values): Likewise.
12958 (switch_conversion::gather_default_values): Likewise.
12959 (build_constructors): Likewise.
12960 (switch_conversion::build_constructors): Likewise.
12961 (constructor_contains_same_values_p): Likewise.
12962 (switch_conversion::contains_same_values_p): Likewise.
12963 (array_value_type): Likewise.
12964 (switch_conversion::array_value_type): Likewise.
12965 (build_one_array): Likewise.
12966 (switch_conversion::build_one_array): Likewise.
12967 (build_arrays): Likewise.
12968 (switch_conversion::build_arrays): Likewise.
12969 (gen_def_assigns): Likewise.
12970 (switch_conversion::gen_def_assigns): Likewise.
12971 (prune_bbs): Likewise.
12972 (switch_conversion::prune_bbs): Likewise.
12973 (fix_phi_nodes): Likewise.
12974 (switch_conversion::fix_phi_nodes): Likewise.
12975 (gen_inbound_check): Likewise.
12976 (switch_conversion::gen_inbound_check): Likewise.
12977 (process_switch): Use the newly created class.
12978 (switch_conversion::expand): New.
12979 (switch_conversion::~switch_conversion): New.
12980 * tree-switch-conversion.h: New file.
12982 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12984 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12985 tree-vect-patterns.c.
12986 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12987 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12988 (vect_recog_sad_pattern): Likewise.
12989 (vect_recog_widen_sum_pattern): Likewise.
12990 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12991 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12992 (vect_recog_rotate_pattern): Likewise.
12993 (vect_recog_mult_pattern): Likewise.
12994 (vect_recog_vector_vector_shift_pattern): Likewise.
12995 (vect_recog_divmod_pattern): Likewise.
12996 (vect_recog_mixed_size_cond_pattern): Likewise.
12997 (vect_recog_bool_pattern): Likewise.
12998 (vect_recog_mask_conversion_pattern): Likewise.
12999 (vect_try_gather_scatter_pattern): Likewise.
13000 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
13001 (vect_recog_over_widening_pattern): Likewise.
13002 (vect_recog_gather_scatter_pattern): Likewise.
13003 (vect_recog_func_ptr): Move from tree-vectorizer.h
13004 (vect_vect_recog_func_ptrs): Move further down the file.
13005 (vect_recog_func): Likewise. Remove the third argument.
13006 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
13007 (vect_pattern_recog_1): Expect the pattern function to do any
13008 necessary target tests. Also expect it to provide a vector type.
13009 Remove the type_in handling.
13011 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13013 * tree-vect-patterns.c (vect_pattern_detected): New function.
13014 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
13015 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
13016 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
13017 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
13018 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
13019 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
13020 (vect_recog_mask_conversion_pattern)
13021 (vect_try_gather_scatter_pattern): Likewise.
13023 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13025 * tree-vect-patterns.c (vect_get_internal_def): New function.
13026 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13027 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
13028 (search_type_for_mask_1): Use it.
13030 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13032 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
13033 redundant WIDEN_SUM_EXPR handling.
13034 (vect_recog_sad_pattern): Likewise.
13036 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13038 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
13039 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
13040 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
13041 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
13042 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
13044 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13046 * tree-vect-stmts.c (vectorizable_call): Make sure that we
13047 use the stmt_vec_info of the original bb statement for the
13048 new zero assignment, even if the call is part of a pattern.
13050 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13052 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
13053 that the sequence is attached to the original statement rather
13054 than the pattern statement.
13055 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
13056 PATTERN_DEF_SEQ from the original statement rather than
13057 the main pattern statement.
13058 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
13059 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
13060 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
13062 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
13064 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
13065 definition statements before the early exit for statements that aren't
13067 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
13069 (vect_transform_loop): ...here. Process pattern definition
13070 statements without first checking whether the main pattern
13071 statement is live or relevant.
13073 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13075 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
13076 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
13078 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
13080 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
13081 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
13082 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
13083 (expand_block_compare): Change select_block_compare_mode call.
13084 (expand_strncmp_align_check): Use new functions, fix comment.
13085 (emit_final_str_compare_gpr): New function.
13086 (expand_strn_compare): Refactor and clean up code.
13087 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
13089 2018-06-19 Tony Reix <tony.reix@atos.com>
13090 Damien Bergamini <damien.bergamini@atos.com>
13091 David Edelsohn <dje.gcc@gmail.com>
13093 * collect2.c (static_obj): New variable.
13094 (static_libs): New variable.
13095 (is_in_list): Uncomment declaration.
13096 (main): Track AIX libraries linked statically.
13097 (is_in_list): Uncomment definition.
13098 (scan_prog_file): Don't add AIX shared libraries initializer
13099 to constructor list if linking statically.
13101 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
13103 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
13105 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
13107 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
13109 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
13112 2018-06-19 Martin Liska <mliska@suse.cz>
13114 * config/i386/i386.c (ix86_can_inline_p): Do not use
13115 ipa_fn_summaries::get_create.
13116 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
13118 (devirtualization_time_bonus): Likewise.
13119 (ipcp_propagate_stage): Likewise.
13120 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
13121 (edge_set_predicate): Likewise.
13122 (evaluate_conditions_for_known_args): Likewise.
13123 (evaluate_properties_for_edge): Likewise.
13124 (ipa_call_summary::reset): Tranform to ...
13125 (ipa_call_summary::~ipa_call_summary): ... this.
13126 (ipa_fn_summary::reset): Transform to ...
13127 (ipa_fn_summary::~ipa_fn_summary): ... this.
13128 (ipa_fn_summary_t::remove): Rename to ...
13129 (ipa_fn_summary_t::remove_callees): ... this.
13130 (ipa_fn_summary_t::duplicate): Use placement new
13131 instead of memory copy.
13132 (ipa_call_summary_t::duplicate): Likewise.
13133 (ipa_call_summary_t::remove): Remove.
13134 (dump_ipa_call_summary): Change get_create to get.
13135 (ipa_dump_fn_summary): Dump only when summary exists.
13136 (analyze_function_body): Use symbol_summary::get instead
13138 (compute_fn_summary): Likewise.
13139 (estimate_edge_devirt_benefit): Likewise.
13140 (estimate_edge_size_and_time): Likewise.
13141 (inline_update_callee_summaries): Likewise.
13142 (remap_edge_change_prob): Likewise.
13143 (remap_edge_summaries): Likewise.
13144 (ipa_merge_fn_summary_after_inlining): Likewise.
13145 (write_ipa_call_summary): Likewise.
13146 (ipa_fn_summary_write): Likewise.
13147 (ipa_free_fn_summary): Likewise.
13148 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
13149 (struct ipa_call_summary): Likewise.
13150 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
13152 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
13153 (estimate_size_after_inlining): Likewise.
13154 (estimate_growth): Likewise.
13155 (growth_likely_positive): Likewise.
13156 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13157 (inline_call): Likewise.
13158 * ipa-inline.c (caller_growth_limits): Likewise.
13159 (can_inline_edge_p): Likewise.
13160 (can_inline_edge_by_limits_p): Likewise.
13161 (compute_uninlined_call_time): Likewise.
13162 (compute_inlined_call_time): Likewise.
13163 (want_inline_small_function_p): Likewise.
13164 (edge_badness): Likewise.
13165 (update_caller_keys): Likewise.
13166 (update_callee_keys): Likewise.
13167 (inline_small_functions): Likewise.
13168 (inline_to_all_callers_1): Likewise.
13169 (dump_overall_stats): Likewise.
13170 (early_inline_small_functions): Likewise.
13171 (early_inliner): Likewise.
13172 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13173 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13174 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13175 * ipa-split.c (execute_split_functions): Likewise.
13176 * symbol-summary.h: Likewise.
13177 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13179 2018-06-19 Richard Biener <rguenther@suse.de>
13181 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
13182 (vectorize_loops): ... here. Fix dbgcnt handling.
13183 (try_vectorize_loop): Wrap try_vectorize_loop_1.
13185 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
13188 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
13189 ieee128 argument takes up only one (vector) register, not two (floating
13192 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13194 * gimplify.c (gimplify_init_constructor): Really never clear for an
13195 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
13197 2018-06-19 Richard Biener <rguenther@suse.de>
13199 PR tree-optimization/86179
13200 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
13201 after failed recognition.
13203 2018-06-18 Martin Sebor <msebor@redhat.com>
13205 PR middle-end/85602
13206 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
13207 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
13208 Handle integer subtraction.
13209 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
13210 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
13212 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13214 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
13215 param from rtx to rtx_insn *.
13216 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
13218 (frv_ifcvt_modify_insn): Likwise.
13219 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
13220 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
13221 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
13222 as_a <rtx_insn *> cast to local "unprotected_region" once
13223 it's been established that it's not NULL or pc_rtx.
13224 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
13225 param "sethi" from rtx to rtx_insn *.
13226 (nds32_group_float_insns): Likewise for param "insn".
13227 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
13229 (vax_output_int_subtract): Likewise.
13230 * config/vax/vax.c (vax_output_int_add): Likewise for param
13232 (vax_output_int_subtract): Likewise.
13233 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
13234 (emit_pattern_after): Likewise for param "after".
13235 (emit_insn_after): Likewise.
13236 (emit_jump_insn_after): Likewise.
13237 (emit_call_insn_after): Likewise.
13238 (emit_debug_insn_after): Likewise.
13239 (emit_pattern_before): Likewise for param "before".
13240 (emit_insn_before): Likewise.
13241 (emit_jump_insn_before): Likewise.
13242 * final.c (get_insn_template): Likewise for param "insn", removing
13244 * output.h (get_insn_template): Likewise for 2nd param.
13245 * rtl.h (emit_insn_before): Likewise.
13246 (emit_jump_insn_before): Likewise.
13247 (emit_debug_insn_before_noloc): Likewise.
13248 (emit_insn_after): Likewise.
13249 (emit_jump_insn_after): Likewise.
13250 (emit_call_insn_after): Likewise.
13251 (emit_debug_insn_after): Likewise.
13252 (set_insn_deleted): Likewise for param.
13254 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
13257 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
13258 floating point modes, so that IFmode is numerically greater than
13259 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
13260 to declare the ordering. This prevents IFmode from being
13261 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
13262 machine. Include rs6000-modes.h to share the fractional values
13263 between genmodes* and the rest of the compiler.
13264 (IFmode): Likewise.
13265 (KFmode): Likewise.
13266 (TFmode): Likewise.
13267 * config/rs6000/rs6000-modes.h: New file.
13268 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
13269 meaning of rs6000_long_double_size so that 126..128 selects an
13270 appropriate 128-bit floating point type.
13271 (rs6000_option_override_internal): Likewise.
13272 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
13273 (TARGET_LONG_DOUBLE_128): Change the meaning of
13274 rs6000_long_double_size so that 126..128 selects an appropriate
13275 128-bit floating point type.
13276 (LONG_DOUBLE_TYPE_SIZE): Update comment.
13277 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
13278 source and destination to match the standard usage.
13279 (truncifkf2): Likewise.
13280 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
13281 ISA 2.07 to use an explicit clobber, instead of passing in a
13283 (copysign<mode>3_soft): Likewise.
13285 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13287 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13288 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
13289 (vect_slp_analyze_instance_dependence): Likewise.
13290 (vect_enhance_data_refs_alignment): Likewise.
13291 (vect_analyze_data_refs_alignment): Likewise.
13292 (vect_slp_analyze_and_verify_instance_alignment
13293 (vect_analyze_data_ref_accesses): Likewise.
13294 (vect_prune_runtime_alias_test_list): Likewise.
13295 (vect_analyze_data_refs): Likewise.
13296 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
13297 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
13298 (vect_analyze_scalar_cycles_1): Likewise.
13299 (vect_get_loop_niters): Likewise.
13300 (vect_analyze_loop_form_1): Likewise.
13301 (vect_update_vf_for_slp): Likewise.
13302 (vect_analyze_loop_operations): Likewise.
13303 (vect_analyze_loop): Likewise.
13304 (vectorizable_induction): Likewise.
13305 (vect_transform_loop): Likewise.
13306 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
13307 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13308 (vect_make_slp_decision): Likewise.
13309 (vect_detect_hybrid_slp): Likewise.
13310 (vect_slp_analyze_operations): Likewise.
13311 (vect_slp_bb): Likewise.
13312 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
13313 (vectorizable_bswap): Likewise.
13314 (vectorizable_call): Likewise.
13315 (vectorizable_simd_clone_call): Likewise.
13316 (vectorizable_conversion): Likewise.
13317 (vectorizable_assignment): Likewise.
13318 (vectorizable_shift): Likewise.
13319 (vectorizable_operation): Likewise.
13320 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
13322 2018-06-18 Martin Sebor <msebor@redhat.com>
13324 PR tree-optimization/81384
13325 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
13326 * builtins.c (expand_builtin_strnlen): New function.
13327 (expand_builtin): Call it.
13328 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
13329 * builtins.def (BUILT_IN_STRNLEN): New.
13330 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
13331 Warn for bounds in excess of maximum object size.
13332 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
13333 single-value ranges. Handle strnlen.
13334 (handle_builtin_strlen): Handle strnlen.
13335 (strlen_check_and_optimize_stmt): Same.
13336 * doc/extend.texi (Other Builtins): Document strnlen.
13338 2018-06-18 Maya Rashish <coypu@sdf.org>
13340 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
13341 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
13342 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
13344 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
13346 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
13348 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13350 * tree.c (escaped_string::escape): Replace cast to char * by
13351 const_cast<char *> (unescaped).
13353 2018-06-18 Nick Clifton <nickc@redhat.com>
13356 * tree.c (escaped_string): New class. Converts an unescaped
13357 string into its escaped equivalent.
13358 (warn_deprecated_use): Use the new class to convert the
13359 deprecation message, if present.
13360 (test_escaped_strings): New self test.
13361 (test_c_tests): Add test_escaped_strings.
13362 * doc/extend.texi (deprecated): Add a note that the
13363 deprecation message is affected by the -fmessage-length
13364 option, and that control characters will be escaped.
13365 (#pragma GCC error): Document this pragma.
13366 (#pragma GCC warning): Likewise.
13367 * doc/invoke.texi (-fmessage-length): Document this option's
13368 effect on the #warning and #error preprocessor directives and
13369 the deprecated attribute.
13371 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
13373 * tree.c (decl_value_expr_lookup): Revert latest change.
13374 (decl_value_expr_insert): Likewise.
13376 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
13378 * gimplify.c (nonlocal_vlas): Delete.
13379 (nonlocal_vla_vars): Likewise.
13380 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
13382 (gimplify_body): Do not create and destroy nonlocal_vlas.
13383 * tree-nested.c: Include diagnostic.h.
13384 (use_pointer_in_frame): Tweak.
13385 (lookup_field_for_decl): Add assertion and declare the transformation.
13386 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
13387 internal error when the reference is in a wrong context. Do not
13388 create a debug decl by default.
13389 (note_nonlocal_block_vlas): Delete.
13390 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
13391 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
13392 create a debug decl by default.
13393 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
13394 call is in a wrong context.
13395 (fixup_vla_decls): New function.
13396 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
13397 debug variables were created.
13398 * tree.c (decl_value_expr_lookup): Add checking assertion.
13399 (decl_value_expr_insert): Likewise.
13401 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13403 PR middle-end/82479
13404 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
13405 * tree-scalar-evolution.c (interpret_expr): Likewise.
13406 (expression_expensive_p): Likewise.
13407 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
13408 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
13409 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
13410 (ssa_defined_by_minus_one_stmt_p): New.
13412 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13414 PR middle-end/64946
13415 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
13416 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
13417 * dojump.c (do_jump): Likewise.
13418 * expr.c (expand_expr_real_2): Check operand type's sign.
13419 * fold-const.c (const_unop): Handle ABSU_EXPR.
13420 (fold_abs_const): Likewise.
13421 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
13422 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
13423 (strip_sign_op_1): Likesise.
13424 * match.pd: Add new pattern to generate ABSU_EXPR.
13425 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
13426 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
13427 * tree-eh.c (operation_could_trap_helper_p): Likewise.
13428 * tree-inline.c (estimate_operator_cost): Likewise.
13429 * tree-pretty-print.c (dump_generic_node): Likewise.
13430 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
13431 * tree.def (ABSU_EXPR): New.
13433 2018-06-16 Jakub Jelinek <jakub@redhat.com>
13435 PR middle-end/86095
13436 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
13437 documented as preserved for backward compatibility only.
13438 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
13440 PR rtl-optimization/86108
13441 * bb-reorder.c (create_forwarder_block): Renamed to ...
13442 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
13443 jump from new landing pad to the second part.
13444 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
13447 2018-06-15 Jakub Jelinek <jakub@redhat.com>
13449 PR middle-end/85878
13450 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
13451 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
13452 Only call store_expr for halves if the mode is the same.
13454 PR middle-end/86123
13455 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
13456 Fix up comment formatting.
13458 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13460 * typed-splay-tree.h (typed_splay_tree::remove): New function.
13461 (typed_splay_tree::closure,
13462 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
13463 (typed_splay_tree::typed_splay_tree,
13464 typed_splay_tree::operator =): Declared private.
13465 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
13466 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
13467 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
13468 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
13469 typed_splay_tree::splay_tree_splay,
13470 typed_splay_tree::splay_tree_foreach_helper,
13471 typed_splay_tree::splay_tree_insert,
13472 typed_splay_tree::splay_tree_remove,
13473 typed_splay_tree::splay_tree_lookup,
13474 typed_splay_tree::splay_tree_predecessor,
13475 typed_splay_tree::splay_tree_successor,
13476 typed_splay_tree::splay_tree_min,
13477 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
13478 (typed_splay_tree::root, typed_splay_tree::comp,
13479 typed_splay_tree::delete_key,
13480 typed_splay_tree::delete_value): New data members.
13481 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
13482 typed_splay_tree::remove.
13484 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
13486 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
13487 -mginv and -mno-ginv to the assembler.
13488 * config/mips/mips.opt (-mcrc): New option.
13489 (-mginv): Likewise.
13490 * doc/invoke.text (-mcrc): Document.
13491 (-mginv): Likewise.
13493 2018-06-15 Nick Clifton <nickc@redhat.com>
13496 * tree.c (escaped_string): New class. Converts an unescaped
13497 string into its escaped equivalent.
13498 (warn_deprecated_use): Use the new class to convert the
13499 deprecation message, if present.
13500 (test_escaped_strings): New self test.
13501 (test_c_tests): Add test_escaped_strings.
13502 * doc/extend.texi (deprecated): Add a note that the
13503 deprecation message is affected by the -fmessage-length
13504 option, and that control characters will be escaped.
13505 (#pragma GCC error): Document this pragma.
13506 (#pragma GCC warning): Likewise.
13507 * doc/invoke.texi (-fmessage-length): Document this option's
13508 effect on the #warning and #error preprocessor directives and
13509 the deprecated attribute.
13511 2018-06-15 Richard Biener <rguenther@suse.de>
13513 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
13514 here, also noting vector size used.
13515 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
13516 size used in MSG_OPTIMIZED_LOCATIONS dump.
13517 (pass_slp_vectorize::execute): Adjust.
13519 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
13522 * config/arc/arc.c (arc_return_address_register): Fix
13525 2018-06-15 Richard Biener <rguenther@suse.de>
13527 PR middle-end/86159
13528 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
13529 leave useless conversion stripping to force_gimple_operand_gsi.
13530 (gimplify_build2): Likewise.
13531 (gimplify_build1): Likewise.
13533 2018-06-15 Richard Biener <rguenther@suse.de>
13535 PR middle-end/86076
13536 * tree-cfg.c (move_stmt_op): unshare invariant addresses
13537 before adjusting their block.
13539 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
13541 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
13542 multilibs for *-*-rtems*.
13543 * config/riscv/t-rtems: New file.
13545 2018-06-14 Jakub Jelinek <jakub@redhat.com>
13547 PR middle-end/86122
13548 * match.pd ((A +- CST1) +- CST2): Punt if last resort
13549 unsigned_type_for returns NULL.
13552 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13553 subregs of multi-word pseudos unless the float mode has word size.
13555 2018-06-14 Richard Biener <rguenther@suse.de>
13557 PR middle-end/86139
13558 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13559 duplicate and harmful type_hash_canon.
13560 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13562 2018-06-14 Richard Biener <rguenther@suse.de>
13565 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13568 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13570 * config/rtems.h (STDINT_LONG32): Define.
13572 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13573 Prachi Godbole <prachi.godbole@imgtec.com>
13575 * config/mips/mips-cpus.def: Define P6600.
13576 * config/mips/mips-tables.opt: Regenerate.
13577 * config/mips/mips.c (mips_ucbranch_type): New enum.
13578 (mips_rtx_cost_data): Add support for P6600.
13579 (mips_issue_rate): Likewise.
13580 (mips_multipass_dfa_lookahead): Likewise.
13581 (mips_avoid_hazard): Likewise.
13582 (mips_reorg_process_insns): Likewise.
13583 (mips_classify_branch_p6600): New function.
13584 * config/mips/mips.h (TUNE_P6600): New define.
13585 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13586 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13587 * config/mips/mips.md: Include p6600.md.
13588 (processor): Add p6600.
13589 * config/mips/p6600.md: New file.
13590 * doc/invoke.texi: Add p6600 to supported architectures.
13592 2018-06-13 Martin Sebor <msebor@redhat.com>
13594 PR tree-optimization/86114
13595 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13597 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13599 2018-06-13 Richard Biener <rguenther@suse.de>
13601 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13602 Properly set vector type of the intermediate stmt.
13603 * tree-vect-stmts.c (vectorizable_operation): The destination
13604 var always has vectype_out type.
13606 2018-06-13 Jeff Law <law@redhat.com>
13608 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13609 integer 0 for argument to print_rtl_with_bb.
13610 (rl78_reorg): Likewise.
13612 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13614 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13615 from rtx to rtx_insn *.
13616 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13618 (add_sched_insns_for_speculation): Likewise for local "target",
13619 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13620 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13621 from rtx_insn ** to rtx_code_label **.
13622 (reorg_emit_nops): Likewise.
13623 (c6x_reorg): Likewise for local "call_labels".
13624 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13626 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13627 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13628 the loops over LABEL_REFS.
13629 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13631 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13632 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13633 (split_branches): Strengthen local "olabel" from rtx to
13634 rtx_insn *, adding a safe_as_a cast.
13635 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13637 (add_insn_after): Likewise for first two params.
13638 (add_insn_before): Likewise.
13639 (remove_insn): Likewise for param.
13640 (emit_pattern_before_noloc): Likewise for second and third params.
13641 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13642 (emit_call_insn_before_noloc): Likewise.
13643 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13645 (emit_barrier_before): Likewise.
13646 (emit_label_before): Strengthen "label" param from "rtx" to
13647 "rtx_code_label *". Strengthen "before" param from "rtx" to
13649 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13651 (emit_pattern_after_noloc): Likewise.
13652 (emit_insn_after_noloc): Likewise.
13653 (emit_jump_insn_after_noloc): Likewise.
13654 (emit_call_insn_after_noloc): Likewise.
13655 (emit_debug_insn_after_noloc): Likewise.
13656 (emit_barrier_after): Likewise.
13657 (emit_label_after): Likewise for both params.
13658 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13659 "loc" param from "int" to "location_t".
13660 (emit_insn_after_setloc): Likewise.
13661 (emit_jump_insn_after_setloc): Likewise.
13662 (emit_call_insn_after_setloc): Likewise.
13663 (emit_debug_insn_after_setloc): Likewise.
13664 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13665 "loc" param from "int" to "location_t".
13666 (emit_pattern_before): Convert NULL_RTX to NULL.
13667 (emit_insn_before_setloc): Convert "loc" param from "int" to
13669 (emit_jump_insn_before_setloc): Likewise.
13670 (emit_call_insn_before_setloc): Likewise.
13671 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13672 rtx_insn *. Convert "loc" param from "int" to "location_t".
13673 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13674 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13675 Convert 3rd param from "int" to "location_t".
13676 (emit_barrier_before, emit_barrier_after, next_real_insn):
13677 Strengthen param from rtx to rtx_insn *.
13678 (emit_label_before): Strengthen 1st param from "rtx" to
13679 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13681 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13682 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13683 Strengthen 2nd param from "rtx" to "rtx_insn *".
13684 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13685 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13686 Likewise. Convert 3rd param from "int" to "location_t".
13687 (emit_label_after): Strengthen 1st param from "rtx" to
13688 "rtx_code_label *".
13689 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13691 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13692 from "rtx" to "rtx_insn *".
13694 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13696 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13697 bodies streamed in with -Q.
13698 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13699 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13700 * gimple-streamer-out.c: Include gimple-pretty-print.h
13701 (output_bb): Dump stmts streamed.
13702 * lto-section-out.c: Include print-tree.h
13703 (lto_begin_section): Dump sections created.
13704 (lto_output_decl_index): Dump decl encoded.
13705 * lto-streamer-out.c: Include print-tree.h
13706 (create_output_block): Dump output block created.
13707 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13708 (output_function): Dump function output.
13709 (output_constructor): Dump constructor streamed.
13710 (write_global_stream): Output indexes encoded.
13711 (produce_asm_for_decls): Dump streams encoded.
13712 * lto-streamer.c (streamer_dump_file): New global var.
13713 * lto-streamer.h (streamer_dump_file): Declare.
13714 * passes.c (ipa_write_summaries): Initialize streamer dump.
13715 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13718 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13721 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13722 offsets for register save directives. Emit a second batch of save
13723 directives, if need be, when the function accesses prior frames.
13725 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13727 * config/arc/fpu.md (fmasf4): Force operand to register.
13728 (fnmasf4): Likewise.
13730 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13732 * config/arc/arc-protos.h (arc_pad_return): Remove.
13733 * config/arc/arc.c (machine_function): Remove force_short_suffix
13735 (arc_print_operand): Adjust printing of '&'.
13736 (arc_verify_short): Remove conditional printing of short suffix.
13737 (arc_final_prescan_insn): Remove reference to size_reason.
13738 (pad_return): New function.
13739 (arc_reorg): Call pad_return.
13740 (arc_pad_return): Remove.
13741 (arc_init_machine_status): Remove reference to force_short_suffix.
13742 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13743 (attr length): When attribute iscompact is true force to 2
13744 regardless; in the case of maybe check if we want to force the
13745 instruction to have 4 bytes length.
13746 (nopv): Change it to generate 4 byte long nop as well.
13747 (blockage): New pattern.
13748 (simple_return): Remove call to arc_pad_return.
13749 (p_return_i): Likewise.
13751 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13753 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13755 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13757 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13760 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13762 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13763 for ARC700 and ARCv2.
13765 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13768 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13769 operands[2] instead of operands[1].
13772 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13774 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13775 case, check whether the outer register overlaps an unallocatable
13776 register, not just whether it fits the required class.
13778 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13780 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13782 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13783 (memrefs_conflict_p): Likewise.
13784 (init_alias_analysis): Likewise.
13785 * cfgexpand.c (expand_debug_expr): Likewise.
13786 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13787 * cse.c (fold_rtx): Likewise.
13788 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13789 * expr.c (emit_block_move_hints): Likewise.
13790 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13791 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13792 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13793 (emit_group_store): Likewise.
13794 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13795 to read the PRE/POST_MODIFY increment.
13796 * calls.c (store_one_arg): Use strip_offset.
13797 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13799 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13801 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13802 (simplify_binary_operation_1): Extend CONST_INT handling to
13804 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13805 than a HOST_WIDE_INT.
13806 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13808 (adjust_mems, add_stores): Update accodingly.
13809 (vt_canonicalize_addr): Track polynomial offsets.
13810 (emit_note_insn_var_location): Likewise.
13811 (vt_add_function_parameter): Likewise.
13812 (vt_initialize): Likewise.
13814 2018-06-12 Jeff Law <law@redhat.com>
13816 * config.gcc (alpha*-*-freebsd*): Remove.
13817 * config/alpha/freebsd.h: Remove.
13819 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13822 * spellcheck-tree.c (levenshtein_distance): Rename to...
13823 (get_edit_distance): ...this, and update for underlying renaming.
13824 * spellcheck-tree.h (levenshtein_distance): Rename to...
13825 (get_edit_distance): ...this.
13826 * spellcheck.c (levenshtein_distance): Rename to...
13827 (get_edit_distance): ...this. Convert from Levenshtein distance
13828 to Damerau-Levenshtein distance by supporting transpositions of
13829 adjacent characters. Rename "v1" to "v_next" and "v0" to
13831 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13832 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13833 update for underlying renaming.
13834 (selftest::levenshtein_distance_unit_test): Rename to...
13835 (selftest::test_get_edit_distance_unit): ...this, and update for
13836 underlying renaming.
13837 (selftest::test_find_closest_string): Add example from PR 69968
13838 where transposition helps
13839 (selftest::test_metric_conditions): Update for renaming.
13840 (selftest::test_metric_conditions): Likewise.
13841 (selftest::spellcheck_c_tests): Likewise.
13842 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13843 (get_edit_distance): ...this.
13844 (best_match::consider): Update for renaming.
13846 2018-06-12 Martin Sebor <msebor@redhat.com>
13848 PR tree-optimization/85259
13849 * builtins.c (compute_objsize): Handle constant offsets.
13850 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13851 true iff a warning has been issued.
13852 * gimple.h (gimple_nonartificial_location): New function.
13853 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13854 gimple_nonartificial_location and handle -Wno-system-headers.
13855 (handle_builtin_stxncpy): Same.
13857 2018-06-12 Martin Sebor <msebor@redhat.com>
13860 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13862 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13864 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13865 BUILTIN_VEC_XST entries for pointer to double and long long.
13867 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13870 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13872 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13875 2018-06-12 Martin Liska <mliska@suse.cz>
13877 * doc/options.texi: Document IntegerRange.
13879 2018-06-12 Martin Liska <mliska@suse.cz>
13881 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13882 * opt-functions.awk: Handle Deprecated flag.
13883 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13885 (read_cmdline_option): Report warning for a deprecated option.
13886 * opts.h (struct cl_option): Add new field cl_deprecated.
13887 (CL_ERR_DEPRECATED): New.
13889 2018-06-12 Martin Liska <mliska@suse.cz>
13891 * doc/options.texi: Document Deprecated option flag.
13893 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13895 * config/arc/arc-arch.h (arc_extras): New enum.
13896 (arc_cpu_t):Add field extra.
13897 (arc_cpu_types): Consider the extras.
13898 * config/arc/arc-cpus.def: Add extras info.
13899 * config/arc/arc-opts.h (processor_type): Consider extra field.
13900 * config/arc/arc.c (arc_override_options): Handle extra field.
13902 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13904 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13905 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13907 * config/arc/arc.c (arc_init): Update pic warning.
13908 (irq_range): Update irq range parsing warnings.
13909 (arc_override_options): Update various warning messages.
13910 (arc_handle_aux_attribute): Likewise.
13912 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13914 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13916 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13918 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13919 compiler messages directives.
13921 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13923 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13924 * config/mips/mips-tables.opt: Regenerate.
13925 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13927 * doc/invoke.texi: Document -march=i6500.
13929 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13931 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13932 (i6400_gpmul): Add cpu_unit.
13933 (i6400_gpdiv): Likewise.
13934 (i6400_msa_add_d): Update reservations.
13935 (i6400_msa_int_add) Likewise.
13936 (i6400_msa_short_logic3) Likewise.
13937 (i6400_msa_short_logic2) Likewise.
13938 (i6400_msa_short_logic) Likewise.
13939 (i6400_msa_move) Likewise.
13940 (i6400_msa_cmp) Likewise.
13941 (i6400_msa_short_float2) Likewise.
13942 (i6400_msa_div_d) Likewise.
13943 (i6400_msa_long_logic1) Likewise.
13944 (i6400_msa_long_logic2) Likewise.
13945 (i6400_msa_mult) Likewise.
13946 (i6400_msa_long_float2) Likewise.
13947 (i6400_msa_long_float4) Likewise.
13948 (i6400_msa_long_float5) Likewise.
13949 (i6400_msa_long_float8) Likewise.
13950 (i6400_fpu_fadd): Include frint type.
13951 (i6400_fpu_store): New define_insn_reservation.
13952 (i6400_fpu_load): Likewise.
13953 (i6400_fpu_move): Likewise.
13954 (i6400_fpu_fcmp): Likewise.
13955 (i6400_fpu_fmadd): Likewise.
13956 (i6400_int_mult): Include imul3nc type and update reservation.
13957 (i6400_int_div): Include idiv3 type and update reservation.
13958 (i6400_int_load): Update to check type not move_type.
13959 (i6400_int_store): Likewise.
13960 (i6400_int_prefetch): Set zero latency.
13962 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13964 * gcc.c: Document new %@{...} sequence.
13965 (LINK_COMMAND_SPEC): Use it for the -L switches.
13966 (cpp_unique_options): Use it for the -I switches.
13967 (at_file_argbuf): New global variable.
13968 (in_at_file): Likewise.
13969 (alloc_args): Create at_file_argbuf.
13970 (clear_args): Truncate at_file_argbuf.
13971 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13972 (open_at_file): New function.
13973 (close_at_file): Likewise.
13974 (create_at_file): Delete.
13975 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13978 (validate_switches_from_spec): Deal with %@{...} sequence.
13979 (validate_switches): Likewise.
13980 (driver::finalize): Call clear_args.
13982 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13984 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13986 2018-06-11 Martin Sebor <msebor@redhat.com>
13988 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13989 (Wno-class-memaccess): Add @opindex.
13990 (Wno-templates, Wno-multiple-inheritance): Same.
13991 (Wno-virtual-inheritance, Wno-namespaces): Same.
13992 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13993 (Wno-format-overflow, Wno-format-truncation): Same.
13994 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13995 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13996 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13997 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13998 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13999 (Wno-misspelled-isr): Same.
14001 2018-06-11 Martin Sebor <msebor@redhat.com>
14003 * PR tree-optimization/86083
14004 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
14006 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
14008 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
14010 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
14013 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
14014 on the correct operand.
14015 (*movdi_internal64): Ditto.
14017 2018-06-11 Martin Liska <mliska@suse.cz>
14019 PR tree-optimization/86089
14020 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
14022 2018-06-11 Julia Koval <julia.koval@intel.com>
14024 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
14025 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
14026 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
14028 2018-06-11 Olivier Hainque <hainque@adacore.com>
14030 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
14031 for Ada with strict dwarf2.
14033 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
14036 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
14039 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
14041 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
14043 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
14045 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
14048 2018-06-08 Martin Liska <mliska@suse.cz>
14050 * tree-cfg.h (debug_function): Fix argument type to match
14053 2018-06-08 Martin Liska <mliska@suse.cz>
14055 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
14056 Remove usage of MPX-related (and removed) fields.
14057 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
14059 2018-06-08 David Malcolm <dmalcolm@redhat.com>
14061 * cfg.c (debug): Use TDF_NONE rather than 0.
14062 * cfghooks.c (debug): Likewise.
14063 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
14064 (struct dump_option_value_info): Convert to...
14065 (struct kv_pair): ...this template type.
14066 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
14068 (optinfo_verbosity_options): Likewise.
14069 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
14071 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
14072 than int for "optgroup_flags" param.
14073 (dump_generic_expr_loc): Use dump_flags_t rather than int for
14075 (dump_dec): Likewise.
14076 (dump_finish): Use TDF_NONE rather than 0.
14077 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
14078 rather than int for "optgroup_flags" param. Use TDF_NONE rather
14079 than 0. Update for change to option_ptr.
14080 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
14081 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
14082 0. Update for changes to optinfo_verbosity_options and
14084 (opt_info_switch_p): Convert optgroup_flags from int to
14086 (dump_basic_block): Use dump_flags_t rather than int
14087 for "dump_kind" param.
14088 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
14089 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
14090 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
14091 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
14092 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
14093 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
14094 TDF_NONE): Convert from macros to...
14095 (enum dump_flag): ...this new enum.
14096 (dump_flags_t): Update to use enum.
14097 (operator|, operator&, operator~, operator|=, operator&=):
14098 Implement for dump_flags_t.
14099 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
14100 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
14101 Convert from macros to...
14102 (enum optgroup_flag): ...this new enum.
14103 (optgroup_flags_t): New typedef.
14104 (operator|, operator|=): Implement for optgroup_flags_t.
14105 (struct dump_file_info): Convert field "alt_flags" to
14106 dump_flags_t. Convert field "optgroup_flags" to
14108 (dump_basic_block): Use dump_flags_t rather than int for param.
14109 (dump_generic_expr_loc): Likewise.
14110 (dump_dec): Likewise.
14111 (dump_register): Convert param "optgroup_flags" to
14113 (opt_info_enable_passes): Likewise.
14114 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
14116 * gimple-pretty-print.c (debug): Likewise.
14117 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
14118 (merged_store_group::apply_stores): Likewise.
14119 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
14120 * gimple.c (verify_gimple_pp): Likewise.
14121 * graphite-poly.c (print_pbb_body): Likewise.
14122 * passes.c (pass_manager::register_one_dump_file): Convert
14123 local "optgroup_flags" to optgroup_flags_t.
14124 * print-tree.c (print_node): Use TDF_NONE rather than 0.
14126 (debug_body): Likewise.
14127 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
14128 to optgroup_flags_t.
14129 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
14131 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
14132 (convert_mult_to_fma): Likewise.
14133 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
14134 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
14135 * tree-vect-data-refs.c (dump_lower_bound): Convert param
14136 "dump_kind" to dump_flags_t.
14138 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14140 * config/rs6000/rs6000.c (min, max): Delete.
14142 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14144 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
14145 -mabi=spe and -mabi=no-spe.
14147 2018-06-08 Martin Liska <mliska@suse.cz>
14149 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
14150 where we expect an existing summary.
14152 2018-06-08 Martin Liska <mliska@suse.cz>
14154 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
14155 * ipa-inline.h (estimate_edge_growth): Likewise.
14157 2018-06-08 Martin Liska <mliska@suse.cz>
14159 * cgraph.c (function_version_hasher::hash): Use
14160 cgraph_node::get_uid ().
14161 (function_version_hasher::equal):
14162 * cgraph.h (cgraph_node::get_uid): New method.
14163 * ipa-inline.c (update_caller_keys): Use
14164 cgraph_node::get_uid ().
14165 (update_callee_keys): Likewise.
14166 * ipa-utils.c (searchc): Likewise.
14167 (ipa_reduced_postorder): Likewise.
14168 * lto-cgraph.c (input_node): Likewise.
14169 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
14170 * symbol-summary.h (symtab_insertion): Likewise.
14171 (symtab_removal): Likewise.
14172 (symtab_duplication): Likewise.
14173 * tree-pretty-print.c (dump_function_header): Likewise.
14174 * tree-sra.c (convert_callers_for_node): Likewise.
14176 2018-06-08 Martin Liska <mliska@suse.cz>
14178 * cgraph.c (symbol_table::create_edge): Always assign a new
14180 (symbol_table::free_edge): Do not recycle numbers.
14181 * cgraph.h (cgraph_edge::get): New method.
14182 * symbol-summary.h (symtab_removal): Use it.
14183 (symtab_duplication): Likewise.
14184 (call_summary::hashable_uid): Remove.
14186 2018-06-08 Martin Liska <mliska@suse.cz>
14188 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
14189 (initialize_growth_caches): Remove.
14190 (free_growth_caches): Likewise.
14191 (do_estimate_edge_time): Use edge_growth_cache.
14192 (do_estimate_edge_size): Likewise.
14193 (do_estimate_edge_hints): Likewise.
14194 * ipa-inline.c (reset_edge_caches): Likewise.
14195 (recursive_inlining): Likewise.
14196 (inline_small_functions): Likewise.
14197 * ipa-inline.h (initialize_growth_caches): Remove.
14198 (estimate_edge_size): Likewise.
14199 (estimate_edge_time): Likewise.
14200 (estimate_edge_hints): Likewise.
14201 (reset_edge_growth_cache): Likewise.
14202 * symbol-summary.h (call_summary::remove): New method.
14204 2018-06-08 Martin Liska <mliska@suse.cz>
14206 * ipa-cp.c (class edge_clone_summary): New summary.
14207 (grow_edge_clone_vectors): Remove.
14208 (ipcp_edge_duplication_hook): Remove.
14209 (class edge_clone_summary_t): New call_summary class.
14210 (ipcp_edge_removal_hook): Remove.
14211 (edge_clone_summary_t::duplicate): New function.
14212 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
14213 (create_specialized_node): Likewise.
14214 (ipcp_driver): Initialize edge_clone_summaries and do not
14217 2018-06-08 Martin Liska <mliska@suse.cz>
14219 * symbol-summary.h (get): New function.
14220 (call_summary::m_initialize_when_cloning): New class member.
14222 2018-06-08 Martin Liska <mliska@suse.cz>
14224 * cgraph.c (cgraph_node::remove): Do not recycle uid.
14225 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
14226 (symbol_table::allocate_cgraph_symbol): Do not set uid.
14227 * passes.c (uid_hash_t): Record removed_nodes by their uids.
14228 (remove_cgraph_node_from_order): Use the removed_nodes set.
14229 (do_per_function_toporder): Likwise.
14230 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
14231 instead of summary_uid.
14232 (symtab_removal): Likewise.
14233 (symtab_duplication): Likewise.
14235 2018-06-08 Martin Liska <mliska@suse.cz>
14237 * ipa-cp.c (ipcp_store_bits_results): Use
14238 ipcp_transformation_sum.
14239 (ipcp_store_vr_results): Likewise.
14240 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
14242 (ipcp_transformation_initialize): ... this.
14243 (ipa_set_node_agg_value_chain):
14244 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
14245 (write_ipcp_transformation_info): Likewise.
14246 (read_ipcp_transformation_info): Likewise.
14247 (ipcp_update_bits): Likewise.
14248 (ipcp_update_vr): Likewise.
14249 (ipcp_transform_function): Likewise.
14250 * ipa-prop.h: Rename ipcp_transformation_summary to
14251 ipcp_transformation.
14252 (class ipcp_transformation_t): New function summary.
14253 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
14254 (ipa_get_agg_replacements_for_node): Likewise.
14256 2018-06-08 Martin Liska <mliska@suse.cz>
14258 * ipa-pure-const.c (struct funct_state_d): Do it class instead
14260 (class funct_state_summary_t): New function_summary class.
14261 (has_function_state): Remove.
14262 (get_function_state): Likewise.
14263 (set_function_state): Likewise.
14264 (add_new_function): Likewise.
14265 (funct_state_summary_t::insert): New function.
14266 (duplicate_node_data): Remove.
14267 (remove_node_data): Remove.
14268 (funct_state_summary_t::duplicate): New function.
14269 (register_hooks): Create new funct_state_summaries.
14270 (pure_const_generate_summary): Use it.
14271 (pure_const_write_summary): Likewise.
14272 (pure_const_read_summary): Likewise.
14273 (propagate_pure_const): Likewise.
14274 (propagate_nothrow): Likewise.
14275 (dump_malloc_lattice): Likewise.
14276 (propagate_malloc): Likewise.
14277 (execute): Do not register hooks, just remove summary
14279 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
14282 2018-06-08 Martin Liska <mliska@suse.cz>
14284 * ipa-reference.c (remove_node_data): Remove.
14285 (duplicate_node_data): Likewise.
14286 (class ipa_ref_var_info_summary_t): New class.
14287 (class ipa_ref_opt_summary_t): Likewise.
14288 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
14289 (get_reference_optimization_summary): Use
14290 ipa_ref_opt_sum_summaries.
14291 (set_reference_vars_info): Remove.
14292 (set_reference_optimization_summary): Likewise.
14293 (ipa_init): Create summaries.
14294 (init_function_info): Use function summary.
14295 (ipa_ref_opt_summary_t::duplicate): New function.
14296 (ipa_ref_opt_summary_t::remove): New function.
14297 (get_read_write_all_from_node): Fix GNU coding style.
14298 (propagate): Use function summary.
14299 (write_node_summary_p): Fix GNU coding style.
14300 (stream_out_bitmap): Likewise.
14301 (ipa_reference_read_optimization_summary): Use function summary.
14302 (ipa_reference_c_finalize): Do not release hooks.
14304 2018-06-08 Martin Liska <mliska@suse.cz>
14306 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
14307 (analyze_function_body): Extract multiple calls of get_create.
14308 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14309 * ipa-inline.c (recursive_inlining): Use ::get method.
14310 * ipa-inline.h (estimate_edge_growth): Likewise.
14312 2018-06-08 Martin Liska <mliska@suse.cz>
14314 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
14316 (hsa_function_summary::hsa_function_summary): Use the new enum
14318 (hsa_gpu_implementation_p): Use hsa_summaries::get.
14319 * hsa-gen.c (hsa_get_host_function): Likewise.
14320 (get_brig_function_name): Likewise.
14321 * ipa-hsa.c (process_hsa_functions): Likewise.
14322 (ipa_hsa_write_summary): Likewise.
14323 * symbol-summary.h (symtab_duplication): Use ::get function/
14324 (get): New function.
14326 2018-06-08 Martin Liska <mliska@suse.cz>
14328 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
14330 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
14331 (hsa_register_kernel): Likewise.
14332 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
14333 * hsa-gen.c (hsa_get_host_function): Likewise.
14334 (get_brig_function_name): Likewise.
14335 (generate_hsa): Likewise.
14336 (pass_gen_hsail::execute): Likewise.
14337 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
14338 (devirtualization_time_bonus): Likewise.
14339 (ipcp_propagate_stage): Likewise.
14340 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
14341 (edge_set_predicate): Likewise.
14342 (evaluate_conditions_for_known_args): Likewise.
14343 (evaluate_properties_for_edge): Likewise.
14344 (ipa_fn_summary::reset): Likewise.
14345 (ipa_fn_summary_t::duplicate): Likewise.
14346 (dump_ipa_call_summary): Likewise.
14347 (ipa_dump_fn_summary): Likewise.
14348 (analyze_function_body): Likewise.
14349 (compute_fn_summary): Likewise.
14350 (estimate_edge_devirt_benefit): Likewise.
14351 (estimate_edge_size_and_time): Likewise.
14352 (estimate_calls_size_and_time): Likewise.
14353 (estimate_node_size_and_time): Likewise.
14354 (inline_update_callee_summaries): Likewise.
14355 (remap_edge_change_prob): Likewise.
14356 (remap_edge_summaries): Likewise.
14357 (ipa_merge_fn_summary_after_inlining): Likewise.
14358 (ipa_update_overall_fn_summary): Likewise.
14359 (read_ipa_call_summary): Likewise.
14360 (inline_read_section): Likewise.
14361 (write_ipa_call_summary): Likewise.
14362 (ipa_fn_summary_write): Likewise.
14363 (ipa_free_fn_summary): Likewise.
14364 * ipa-hsa.c (process_hsa_functions): Likewise.
14365 (ipa_hsa_write_summary): Likewise.
14366 (ipa_hsa_read_section): Likewise.
14367 * ipa-icf.c (sem_function::merge): Likewise.
14368 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14369 (do_estimate_edge_time): Likewise.
14370 (estimate_size_after_inlining): Likewise.
14371 (estimate_growth): Likewise.
14372 (growth_likely_positive): Likewise.
14373 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
14374 (inline_call): Likewise.
14375 * ipa-inline.c (caller_growth_limits): Likewise.
14376 (can_inline_edge_p): Likewise.
14377 (can_inline_edge_by_limits_p): Likewise.
14378 (compute_uninlined_call_time): Likewise.
14379 (compute_inlined_call_time): Likewise.
14380 (want_inline_small_function_p): Likewise.
14381 (edge_badness): Likewise.
14382 (update_caller_keys): Likewise.
14383 (update_callee_keys): Likewise.
14384 (recursive_inlining): Likewise.
14385 (inline_small_functions): Likewise.
14386 (inline_to_all_callers_1): Likewise.
14387 (dump_overall_stats): Likewise.
14388 (early_inline_small_functions): Likewise.
14389 (early_inliner): Likewise.
14390 * ipa-inline.h (estimate_edge_growth): Likewise.
14391 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14392 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14393 * ipa-prop.h (IPA_NODE_REF): Likewise.
14394 (IPA_EDGE_REF): Likewise.
14395 * ipa-pure-const.c (malloc_candidate_p): Likewise.
14396 (propagate_malloc): Likewise.
14397 * ipa-split.c (execute_split_functions): Likewise.
14398 * symbol-summary.h: Rename get to get_create.
14400 (get_create): Likewise.
14401 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
14403 2018-06-08 Martin Liska <mliska@suse.cz>
14405 * symbol-summary.h (release): Move definition out of class
14407 (symtab_removal): Likewise.
14408 (symtab_duplication): Likewise.
14410 2018-06-08 Martin Liska <mliska@suse.cz>
14412 * symbol-summary.h (function_summary): Move constructor
14413 implementation out of class declaration.
14414 (release): Likewise.
14415 (symtab_insertion): Likewise.
14416 (symtab_removal): Likewise.
14417 (symtab_duplication): Likewise.
14420 2018-06-08 Martin Liska <mliska@suse.cz>
14422 * Makefile.in: Remove support for MPX (macros, related functions,
14423 fields in cgraph_node, ...).
14424 * builtin-types.def (BT_BND): Likewise.
14425 (BT_FN_BND_CONST_PTR): Likewise.
14426 (BT_FN_CONST_PTR_BND): Likewise.
14427 (BT_FN_VOID_PTR_BND): Likewise.
14428 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
14429 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
14430 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
14431 (expand_builtin_mempcpy_with_bounds): Likewise.
14432 (expand_builtin_memset_with_bounds): Likewise.
14433 (expand_builtin_memset_args): Likewise.
14434 (std_expand_builtin_va_start): Likewise.
14435 (expand_builtin): Likewise.
14436 (expand_builtin_with_bounds): Likewise.
14437 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
14438 (DEF_LIB_BUILTIN_CHKP): Likewise.
14439 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
14440 (DEF_CHKP_BUILTIN): Likewise.
14441 (BUILT_IN_MEMCPY): Likewise.
14442 (BUILT_IN_MEMMOVE): Likewise.
14443 (BUILT_IN_MEMPCPY): Likewise.
14444 (BUILT_IN_MEMSET): Likewise.
14445 (BUILT_IN_STPCPY): Likewise.
14446 (BUILT_IN_STRCAT): Likewise.
14447 (BUILT_IN_STRCHR): Likewise.
14448 (BUILT_IN_STRCPY): Likewise.
14449 (BUILT_IN_STRLEN): Likewise.
14450 (BUILT_IN_MEMCPY_CHK): Likewise.
14451 (BUILT_IN_MEMMOVE_CHK): Likewise.
14452 (BUILT_IN_MEMPCPY_CHK): Likewise.
14453 (BUILT_IN_MEMSET_CHK): Likewise.
14454 (BUILT_IN_STPCPY_CHK): Likewise.
14455 (BUILT_IN_STRCAT_CHK): Likewise.
14456 (BUILT_IN_STRCPY_CHK): Likewise.
14457 * calls.c (store_bounds): Likewise.
14458 (emit_call_1): Likewise.
14459 (special_function_p): Likewise.
14460 (maybe_warn_nonstring_arg): Likewise.
14461 (initialize_argument_information): Likewise.
14462 (finalize_must_preallocate): Likewise.
14463 (compute_argument_addresses): Likewise.
14464 (expand_call): Likewise.
14465 * cfgexpand.c (expand_call_stmt): Likewise.
14466 (expand_return): Likewise.
14467 (expand_gimple_stmt_1): Likewise.
14468 (pass_expand::execute): Likewise.
14469 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14470 (cgraph_node::remove): Likewise.
14471 (cgraph_node::dump): Likewise.
14472 (cgraph_node::verify_node): Likewise.
14473 * cgraph.h (chkp_function_instrumented_p): Likewise.
14474 (symtab_node::get_alias_target): Likewise.
14475 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
14476 (cgraph_local_p): Likewise.
14477 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
14478 (cgraph_edge::rebuild_references): Likewise.
14479 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
14480 (walk_polymorphic_call_targets): Likewise.
14481 (cgraph_node::expand_thunk): Likewise.
14482 (symbol_table::output_weakrefs): Likewise.
14483 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
14484 (ix86_handle_option): Likewise.
14485 * config/i386/constraints.md: Likewise.
14486 * config/i386/i386-builtin-types.def (BND): Likewise.
14490 * config/i386/i386-builtin.def (BDESC_END): Likewise.
14491 (BDESC_FIRST): Likewise.
14493 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
14494 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
14495 * config/i386/i386.c (enum reg_class): Likewise.
14496 (ix86_target_string): Likewise.
14497 (ix86_option_override_internal): Likewise.
14498 (ix86_conditional_register_usage): Likewise.
14499 (ix86_valid_target_attribute_inner_p): Likewise.
14500 (ix86_set_indirect_branch_type): Likewise.
14501 (ix86_set_current_function): Likewise.
14502 (ix86_function_arg_regno_p): Likewise.
14503 (init_cumulative_args): Likewise.
14504 (ix86_function_arg_advance): Likewise.
14505 (ix86_function_arg): Likewise.
14506 (ix86_pass_by_reference): Likewise.
14507 (ix86_function_value_regno_p): Likewise.
14508 (ix86_function_value_1): Likewise.
14509 (ix86_function_value_bounds): Likewise.
14510 (ix86_return_in_memory): Likewise.
14511 (ix86_setup_incoming_vararg_bounds): Likewise.
14512 (ix86_va_start): Likewise.
14513 (indirect_thunk_need_prefix): Likewise.
14514 (print_reg): Likewise.
14515 (ix86_print_operand): Likewise.
14516 (ix86_expand_call): Likewise.
14517 (ix86_output_function_return): Likewise.
14518 (reg_encoded_number): Likewise.
14519 (BDESC_VERIFYS): Likewise.
14520 (ix86_init_mpx_builtins): Likewise.
14521 (ix86_init_builtins): Likewise.
14522 (ix86_emit_cmove): Likewise.
14523 (ix86_emit_move_max): Likewise.
14524 (ix86_expand_builtin): Likewise.
14525 (ix86_builtin_mpx_function): Likewise.
14526 (ix86_get_arg_address_for_bt): Likewise.
14527 (ix86_load_bounds): Likewise.
14528 (ix86_store_bounds): Likewise.
14529 (ix86_load_returned_bounds): Likewise.
14530 (ix86_store_returned_bounds): Likewise.
14531 (ix86_class_likely_spilled_p): Likewise.
14532 (ix86_hard_regno_mode_ok): Likewise.
14533 (x86_order_regs_for_local_alloc): Likewise.
14534 (ix86_mitigate_rop): Likewise.
14535 (ix86_bnd_prefixed_insn_p): Likewise.
14536 (ix86_mpx_bound_mode): Likewise.
14537 (ix86_make_bounds_constant): Likewise.
14538 (ix86_initialize_bounds): Likewise.
14539 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
14540 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
14541 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
14542 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
14543 (TARGET_CHKP_BOUND_MODE): Likewise.
14544 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
14545 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
14546 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
14547 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
14548 * config/i386/i386.h (TARGET_MPX): Likewise.
14549 (TARGET_MPX_P): Likewise.
14550 (VALID_BND_REG_MODE): Likewise.
14551 (FIRST_BND_REG): Likewise.
14552 (LAST_BND_REG): Likewise.
14553 (enum reg_class): Likewise.
14554 (BND_REG_P): Likewise.
14555 (BND_REGNO_P): Likewise.
14556 (BNDmode): Likewise.
14557 (ADJUST_INSN_LENGTH): Likewise.
14558 * config/i386/i386.md: Likewise.
14559 * config/i386/i386.opt: Likewise.
14560 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14561 (defined): Likewise.
14562 (LINK_MPX): Likewise.
14563 (MPX_SPEC): Likewise.
14564 (LIBMPX_SPEC): Likewise.
14565 (LIBMPXWRAPPERS_SPEC): Likewise.
14566 (CHKP_SPEC): Likewise.
14567 * config/i386/predicates.md: Likewise.
14568 * dbxout.c (dbxout_type): Likewise.
14569 * doc/extend.texi: Likewise.
14570 * doc/invoke.texi: Likewise.
14571 * doc/md.texi: Likewise.
14572 * doc/tm.texi: Likewise.
14573 * doc/tm.texi.in: Likewise.
14574 * dwarf2out.c (is_base_type): Likewise.
14575 (gen_formal_types_die): Likewise.
14576 (gen_subprogram_die): Likewise.
14577 (gen_type_die_with_usage): Likewise.
14578 (gen_decl_die): Likewise.
14579 (dwarf2out_late_global_decl): Likewise.
14580 * expr.c (expand_assignment): Likewise.
14581 (emit_storent_insn): Likewise.
14582 (store_expr_with_bounds): Likewise.
14583 (store_expr): Likewise.
14584 (expand_expr_real_1): Likewise.
14585 * expr.h (store_expr_with_bounds): Likewise.
14586 * function.c (use_register_for_decl): Likewise.
14587 (struct bounds_parm_data): Likewise.
14588 (assign_parms_augmented_arg_list): Likewise.
14589 (assign_parm_find_entry_rtl): Likewise.
14590 (assign_parm_is_stack_parm): Likewise.
14591 (assign_parm_load_bounds): Likewise.
14592 (assign_bounds): Likewise.
14593 (assign_parms): Likewise.
14594 (expand_function_start): Likewise.
14595 * gcc.c (CHKP_SPEC): Likewise.
14596 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14597 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14598 (wrestrict_dom_walker::check_call): Likewise.
14599 * gimple.c (gimple_build_call_from_tree): Likewise.
14600 * gimple.h (enum gf_mask): Likewise.
14601 (gimple_call_with_bounds_p): Likewise.
14602 (gimple_call_set_with_bounds): Likewise.
14603 * gimplify.c (gimplify_init_constructor): Likewise.
14604 * ipa-cp.c (initialize_node_lattices): Likewise.
14605 (propagate_constants_across_call): Likewise.
14606 (find_more_scalar_values_for_callers_subset): Likewise.
14607 * ipa-hsa.c (process_hsa_functions): Likewise.
14608 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14609 * ipa-icf.c (sem_function::merge): Likewise.
14610 * ipa-inline.c (early_inliner): Likewise.
14611 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14612 (warn_function_cold): Likewise.
14613 (propagate_pure_const): Likewise.
14614 * ipa-ref.h (enum GTY): Likewise.
14615 * ipa-split.c (find_retbnd): Likewise.
14616 (consider_split): Likewise.
14617 (split_function): Likewise.
14618 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14619 * ipa.c (walk_polymorphic_call_targets): Likewise.
14620 (symbol_table::remove_unreachable_nodes): Likewise.
14621 (process_references): Likewise.
14622 (cgraph_build_static_cdtor_1): Likewise.
14623 * lto-cgraph.c (lto_output_node): Likewise.
14624 (output_refs): Likewise.
14625 (compute_ltrans_boundary): Likewise.
14626 (input_overwrite_node): Likewise.
14627 (input_node): Likewise.
14628 (input_cgraph_1): Likewise.
14629 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14630 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14631 (class pass_chkp_instrumentation_passes): Likewise.
14632 (make_pass_chkp_instrumentation_passes): Likewise.
14633 * passes.def: Likewise.
14634 * rtl.h (struct GTY): Likewise.
14635 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14636 * stor-layout.c (layout_type): Likewise.
14637 * symtab.c: Likewise.
14638 * target.def: Likewise.
14639 * targhooks.c (default_chkp_bound_type): Likewise.
14640 (default_chkp_bound_mode): Likewise.
14641 (default_builtin_chkp_function): Likewise.
14642 (default_chkp_function_value_bounds): Likewise.
14643 (default_chkp_make_bounds_constant): Likewise.
14644 (default_chkp_initialize_bounds): Likewise.
14645 * targhooks.h (default_chkp_bound_type): Likewise.
14646 (default_chkp_bound_mode): Likewise.
14647 (default_builtin_chkp_function): Likewise.
14648 (default_chkp_function_value_bounds): Likewise.
14649 (default_chkp_make_bounds_constant): Likewise.
14650 (default_chkp_initialize_bounds): Likewise.
14651 * toplev.c (compile_file): Likewise.
14652 (process_options): Likewise.
14653 * tree-core.h (DEF_BUILTIN): Likewise.
14654 (DEF_BUILTIN_CHKP): Likewise.
14655 * tree-inline.c (declare_return_variable): Likewise.
14656 (remap_gimple_stmt): Likewise.
14657 (copy_bb): Likewise.
14658 (initialize_inlined_parameters): Likewise.
14659 (expand_call_inline): Likewise.
14660 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14661 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14662 (make_pass_ipa_chkp_produce_thunks): Likewise.
14663 (make_pass_chkp): Likewise.
14664 (make_pass_chkp_opt): Likewise.
14665 (make_pass_chkp_instrumentation_passes): Likewise.
14666 * tree-pretty-print.c (dump_generic_node): Likewise.
14667 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14668 * tree-ssa-dce.c (propagate_necessity): Likewise.
14669 (eliminate_unnecessary_stmts): Likewise.
14670 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14671 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14672 * tree-ssa-sccvn.h: Likewise.
14673 * tree-ssa-strlen.c (get_string_length): Likewise.
14674 (valid_builtin_call): Likewise.
14675 (adjust_last_stmt): Likewise.
14676 (handle_builtin_strchr): Likewise.
14677 (handle_builtin_strcpy): Likewise.
14678 (handle_builtin_stxncpy): Likewise.
14679 (handle_builtin_memcpy): Likewise.
14680 (handle_builtin_strcat): Likewise.
14681 (strlen_check_and_optimize_stmt): Likewise.
14682 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14683 * tree-streamer-in.c: Likewise.
14684 * tree-streamer.c (record_common_node): Likewise.
14685 * tree.c (tree_code_size): Likewise.
14686 (wide_int_to_tree_1): Likewise.
14687 (type_contains_placeholder_1): Likewise.
14688 (build_common_tree_nodes): Likewise.
14689 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14690 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14691 (POINTER_BOUNDS_P): Likewise.
14692 (BOUNDED_TYPE_P): Likewise.
14693 (BOUNDED_P): Likewise.
14694 (CALL_WITH_BOUNDS_P): Likewise.
14695 (pointer_bounds_type_node): Likewise.
14696 * value-prof.c (gimple_ic): Likewise.
14697 * var-tracking.c (vt_add_function_parameters): Likewise.
14698 * varasm.c (make_decl_rtl): Likewise.
14699 (assemble_start_function): Likewise.
14700 (output_constant): Likewise.
14701 (maybe_assemble_visibility): Likewise.
14702 * varpool.c (ctor_for_folding): Likewise.
14703 * chkp-builtins.def: Remove.
14704 * ipa-chkp.c: Remove.
14705 * ipa-chkp.h: Remove.
14706 * rtl-chkp.c: Remove.
14707 * rtl-chkp.h: Remove.
14708 * tree-chkp-opt.c: Remove.
14709 * tree-chkp.c: Remove.
14710 * tree-chkp.h: Remove.
14712 2018-06-07 Carl Love <cel@us.ibm.com>
14714 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14715 vextract_fp_from_shortl): Add BE support.
14717 2018-06-07 Paul Koning <ni1d@arrl.net>
14719 * compare-elim.c (try_merge_compare): Don't merge compare if
14720 address contains a side effect.
14721 (try_eliminate_compare): Likewise.
14723 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14725 * config.gcc: Support "tremont".
14726 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14727 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14729 * config/i386/i386.c (m_TREMONT): Define.
14730 (processor_target_table): Add "tremont".
14731 (PTA_TREMONT): Define.
14732 (ix86_lea_outperforms): Add TARGET_TREMONT.
14733 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14734 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14735 and M_INTEL_GOLDMONT_PLUS.
14736 (fold_builtin_cpu): Add "tremont".
14737 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14738 (ix86_option_override_internal): Add "tremont".
14739 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14740 (processor_type): Add PROCESSOR_TREMONT.
14741 * config/i386/x86-tune.def: Add m_TREMONT.
14742 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14744 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14746 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14747 symbol defined for msp430i* devices to be lower case.
14749 2018-06-07 Richard Biener <rguenther@suse.de>
14751 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14752 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14753 Properly wrap signed arithmetic if overflow wraps.
14755 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14757 PR tree-optimization/69615
14758 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14759 of a cast from a same precision integral SSA_NAME in a bb dominated
14760 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14761 cast to utype if rhs2 has already a compatible type.
14763 2018-06-07 Richard Biener <rguenther@suse.de>
14765 PR tree-optimization/85935
14766 * graphite-scop-detection.c (find_params_in_bb): Analyze
14767 condition operands with respect to the correct loop. Assert
14768 the analysis doesn't fail.
14770 2018-06-04 Carl Love <cel@us.ibm.com>
14772 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14773 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14774 as it is slightly cheaper.
14775 (first_match_or_eos_index_<mode>):
14776 Calculate index using natural element order.
14777 (first_match_index_<mode>):
14778 Calculate index using natural element order.
14779 (first_match_or_eos_index_<mode>):
14780 Calculate index using natural order.
14781 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14782 for BE and LE modes.
14783 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14784 P9V_BUILTIN_VCLZLSBB_V16QI.
14785 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14788 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14790 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14791 indentation and line wrap for many prototypes. Add missing
14792 @smallexample directives around block of prototypes for vec_xl and
14795 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14797 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14798 track if we pass or return IEEE 128-bit floating point.
14799 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14800 C++ mangling that is compatible with GCC 8.1.
14801 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14802 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14803 floating point types.
14804 (rs6000_function_arg_advance_1): Likewise.
14805 (rs6000_mangle_type): Optionally generate mangled names that match
14806 what GCC 8.1 generated for IEEE 128-bit floating point types.
14807 (rs6000_globalize_decl_name): If we have an external function that
14808 passes or returns IEEE 128-bit types, generate a weak reference
14809 from the mangled name used in GCC 8.1 to the current mangled
14811 (rs6000_init_builtins): Make __ibm128 use the long double type if
14812 long double is IBM extended double. Make __float128 use the long
14813 double type if long double is IEEE 128-bit.
14816 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14817 macro for __ibm128 built-in functions.
14818 (PACK_IF): Add __ibm128 pack/unpack functions.
14819 (UNPACK_IF): Likewise.
14820 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14821 enable long double built-in functions if long double is IEEE
14822 128-bit floating point.
14823 (rs6000_invalid_builtin): Update long double built-in function
14825 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14826 functions, adjust the built-in function to use the long double
14827 built-in function if __ibm128 and long double are the same type.
14828 * doc/extend.texi (PowerPC builtins): Update documention for
14829 __builtin_{,un}pack_longdouble. Add documentation for
14830 __builtin_{,un}pack_ibm128.
14832 2018-06-06 Jim Wilson <jimw@sifive.com>
14834 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14835 (struct machine_function): New field interrupt_mode.
14836 (riscv_handle_type_attribute): New function. Add forward declaration.
14837 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14838 (riscv_expand_epilogue): Check interrupt_mode field.
14839 (riscv_set_current_function): Check interrupt attribute args and
14840 set interrupt_mode field.
14841 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14842 (riscv_sret, riscv_uret): New.
14843 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14844 new arguments to interrupt attribute.
14846 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14849 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14850 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14852 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14854 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14855 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14856 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14857 several redundant entries.
14859 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14861 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14862 type from "rtx" to "rtx_insn *".
14863 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14864 for local "call_insn", removing cast.
14865 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14867 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14869 PR tree-optimization/86066
14870 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14871 for BIT_INSERT_EXPR stores.
14873 2018-06-06 Richard Biener <rguenther@suse.de>
14875 PR tree-optimization/86062
14876 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14877 component refs ontop
14878 of to be offsetted base.
14880 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14882 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14883 to be static and remove check on interrupt attribute name.
14885 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14887 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14888 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14890 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14893 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14895 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14896 Remove second argument, change how error is called.
14897 (aarch64_layout_arg): Remove second argument from
14898 aarch64_err_no_fpadvsimd call.
14899 (aarch64_init_cumulative_args): Ditto.
14900 (aarch64_gimplify_va_arg_expr): Ditto.
14901 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14903 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14905 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14906 (*simple_return_indirect_internal<mode>): Rename from
14907 simple_return_indirect_internal. Use W mode iterator.
14908 (rstorssp): New expander.
14909 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14910 (clrssbsy): New expander.
14911 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14913 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14915 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14917 (cmse_check_pointed_object): Likewise.
14919 2018-06-05 Martin Liska <mliska@suse.cz>
14921 PR gcov-profile/47618
14922 * doc/invoke.texi: Document how -fprofile-dir format
14925 2018-06-05 Richard Biener <rguenther@suse.de>
14927 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14928 removal pretend DOM info isn't available so we do not update
14929 it and only remove edges, not dominated blocks. Actually free
14930 DOM info in case we removed something. Remove unreachable blocks.
14931 (mfb_keep_latches): Work with either DOM info or marked backedges.
14932 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14933 first. Mark backedges if DOM info isn't available.
14934 (Re-)compute DOM info after cleanup_control_flow_pre.
14936 2018-06-05 Richard Biener <rguenther@suse.de>
14938 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14939 (locus_discrim_hasher::hash): Adjust.
14940 (locus_discrim_hasher::equal): Likewise.
14941 (next_discriminator_for_locus): Work on line directly.
14942 (same_line_p): Pass in expanded locus1 as well.
14943 (assign_discriminators): Avoid redundant location expansions.
14945 2018-06-05 Richard Biener <rguenther@suse.de>
14947 PR tree-optimization/86046
14948 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14949 if required after clearing TREE_ADDRESSABLE.
14951 2018-06-05 Richard Biener <rguenther@suse.de>
14953 PR tree-optimization/86047
14954 * tree-ssa-loop.c (for_each_index): Glob handling of all
14955 decls and constants and really handle all of them.
14957 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14960 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14961 qualifier_void_pointer and qualifier_const_void_pointer.
14962 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14963 (arm_init_builtins): Handle the above.
14964 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14965 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14966 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14969 2018-06-05 Martin Liska <mliska@suse.cz>
14971 * auto-profile.c (read_autofdo_file): Do not use
14972 gcov_ctr_summary struct.
14973 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14974 * coverage.c (struct counts_entry): Likewise.
14975 (read_counts_file): Read just single summary entry.
14976 (get_coverage_counts): Use gcov_summary struct.
14977 * coverage.h (get_coverage_counts): Likewise.
14978 * gcov-dump.c (dump_working_sets): Likewise.
14979 (tag_summary): Dump just single summary.
14980 * gcov-io.c (gcov_write_summary): Write just histogram
14982 (gcov_read_summary): Read just single summary.
14983 (compute_working_sets): Use gcov_summary struct.
14984 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14985 of GCOV_COUNTERS_SUMMABLE.
14986 (GCOV_COUNTERS_SUMMABLE): Remove.
14987 (GCOV_FIRST_VALUE_COUNTER): Replace with
14988 GCOV_COUNTER_V_INTERVAL.
14989 (struct gcov_ctr_summary): Remove.
14990 (struct gcov_summary): Directly use fields of former
14992 (compute_working_sets): Use gcov_summary struct.
14993 * gcov.c (read_count_file): Do not use ctrs fields.
14994 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14996 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14998 * profile.c: Likewise.
14999 * profile.h: Likewise.
15001 2018-06-05 Martin Liska <mliska@suse.cz>
15003 PR gcov-profile/84846
15004 * gcov.c (output_lines): Print working directory only
15005 in intermediate format.
15007 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
15009 * config/s390/s390-builtin-types.def: Add void function type.
15010 * config/s390/s390-builtins.def: Use the function type for the
15013 2018-06-04 Jim Wilson <jimw@sifive.com>
15015 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
15017 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
15018 and maybe_eh_return. Change regno to unsigned int. Use new args to
15019 handle EH_RETURN_DATA_REGNO registers properly.
15020 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
15021 (riscv_expand_epilogue): Update comment. Change argument name and
15022 type. Update code to use new name and type. Pass new args to
15023 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
15025 * config/riscv/riscv.md (NORMAL_RETURN): New.
15026 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
15027 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
15028 (eh_return): Call gen_eh_return_internal and emit barrier.
15029 (eh_return_internal): Call riscv_expand_epilogue.
15031 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
15033 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
15034 bit_insertion field and declare can_be_merged_into method.
15035 (merged_store_group::can_be_merged_into): New method.
15036 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
15037 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
15038 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
15040 2018-06-04 Richard Biener <rguenther@suse.de>
15042 PR tree-optimization/85955
15043 * builtins.c (fold_builtin_sincos): Convert pointers to
15044 destination to appropriate type before dereferencing.
15046 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
15048 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
15050 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
15052 * expr.c (expand_expr_real_1): Force the operand into memory if
15053 its TYPE_MODE is BLKmode and if there is no integer mode for
15054 the number of bits being extracted.
15056 2018-06-04 Jakub Jelinek <jakub@redhat.com>
15060 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15061 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
15063 2018-06-04 Richard Biener <rguenther@suse.de>
15065 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
15066 (cleanup_tree_cfg_noloop): ... single caller. Do
15067 start_recording_case_labels later.
15069 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
15071 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
15072 to _IMMINTRIN_H_INCLUDED.
15073 * config/i386/pconfigintrin.h: Ditto.
15074 * config/i386/waitpkgintrin.h: Ditto.
15075 * config/i386/immintrin.h: Add includes for sgxintrin.h,
15076 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
15077 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
15078 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
15079 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
15080 waitpkgintrin.h and cldemoteintrin.h.
15082 2018-06-04 Richard Biener <rguenther@suse.de>
15084 PR tree-optimization/86038
15085 * tracer.c (find_best_successor): Check probability for
15086 being initialized, bail out if not.
15088 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
15091 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
15092 of bits to ignore when comparing architectures.
15094 2018-06-04 Jakub Jelinek <jakub@redhat.com>
15096 PR tree-optimization/69615
15097 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
15098 maximum or minimum of the type, try to merge it also as if
15099 range1 is + [-, x - 1] or + [x + 1, -].
15102 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
15104 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
15106 PR tree-optimization/86034
15107 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
15108 the unsigned bitfield type in a bit insertion sequence if it does not
15109 have a larger precision than the bitfield size.
15110 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
15112 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
15114 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
15116 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
15118 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
15119 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
15120 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
15121 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
15123 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
15125 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
15126 Disable -fdelete-null-pointer-checks for ELF toolchain.
15128 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15129 Kito Cheng <kito.cheng@gmail.com>
15131 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
15132 (nds32le-*-*, nds32be-*-*): Integrate checking process.
15133 (nds32*-*-*): Add glibc and uclibc conditions.
15134 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
15135 (TARGET_EXCEPT_UNWIND_INFO): Define.
15136 * config/nds32/elf.h: New file.
15137 * config/nds32/linux.h: New file.
15138 * config/nds32/nds32-elf.opt: New file.
15139 * config/nds32/nds32-linux.opt: New file.
15140 * config/nds32/nds32-fp-as-gp.c
15141 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
15142 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
15144 (nds32_asm_file_end): Ditto.
15145 (nds32_print_operand): Ditto.
15146 (nds32_insert_attributes): Ditto.
15147 (nds32_init_libfuncs): New function.
15148 (TARGET_HAVE_TLS): Define.
15149 (TARGET_INIT_LIBFUNCS): Define.
15150 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
15152 (TARGET_ELF): Apply different mcmodel setting.
15153 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
15154 been migrated into elf.h and linux.h files.
15155 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
15156 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
15157 (mcmodel): The content has been migrated into nds32-elf.opt and
15158 nds32-linux.opt files.
15159 * config/nds32/t-elf: New file.
15160 * config/nds32/t-linux: New file.
15162 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15163 Shiva Chen <shiva0217@gmail.com>
15165 * config/nds32/constants.md (unspec_volatile_element): Add
15166 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
15167 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
15169 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
15170 (make_pass_nds32_fp_as_gp): Declare.
15171 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
15173 (nds32_asm_function_end_prologue): Remove unused asm output.
15174 (nds32_asm_function_begin_epilogue): Remove unused asm output.
15175 (nds32_asm_file_start): Output necessary fp_as_gp information.
15176 (nds32_option_override): Adjust register usage.
15177 (nds32_expand_prologue): Consider fp_as_gp situation.
15178 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
15179 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
15182 (simple_return): Ditto.
15183 (omit_fp_begin): Output special directive for fp_as_gp.
15184 (omit_fp_end): Output special directive for fp_as_gp.
15185 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
15186 mforbid-fp-as-gp): New options.
15188 2018-06-01 Mark Wielaard <mark@klomp.org>
15190 * dwarf2out.c (dwarf2out_finish): Remove generation of
15191 DW_AT_loclists_base.
15193 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
15195 * gimple-ssa-store-merging.c: Include gimple-fold.h.
15196 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
15197 (struct merged_store_group): Add bit_insertion field.
15198 (dump_char_array): Use standard hexadecimal format.
15199 (merged_store_group::merged_store_group): Set bit_insertion to false.
15200 (merged_store_group::apply_stores): Use optimal buffer size. Deal
15201 with BIT_INSERT_EXPR stores. Move up code updating the mask and
15202 also print the mask in the dump file.
15203 (pass_store_merging::gate): Minor tweak.
15204 (imm_store_chain_info::coalesce_immediate): Fix wrong association
15205 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
15206 stores with INTEGER_CST stores.
15207 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
15208 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
15209 and use it throughout. Generate bit insertion sequences if need be.
15210 (pass_store_merging::process_store): Remove redundant condition.
15211 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
15213 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
15215 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
15216 the 128-bit floating point types. Fix function comment.
15218 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15220 * config/aarch64/aarch64-simd.md
15221 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
15223 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
15226 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15228 PR tree-optimization/85989
15229 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
15231 (backprop::intersect_uses): Check it when deciding whether this
15232 is a backedge reference.
15233 (backprop::process_block): Add each phi to m_visited_phis
15234 after visiting it, then clear it at the end.
15236 2018-06-01 Richard Biener <rguenther@suse.de>
15238 * tree-vectorizer.h (vect_dr_stmt): New function.
15239 (vect_get_load_cost): Adjust.
15240 (vect_get_store_cost): Likewise.
15241 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15242 Use vect_dr_stmt instead of DR_SMTT.
15243 (vect_record_base_alignments): Likewise.
15244 (vect_calculate_target_alignment): Likewise.
15245 (vect_compute_data_ref_alignment): Likewise and make static.
15246 (vect_update_misalignment_for_peel): Likewise.
15247 (vect_verify_datarefs_alignment): Likewise.
15248 (vector_alignment_reachable_p): Likewise.
15249 (vect_get_data_access_cost): Likewise. Pass down
15250 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
15251 (vect_get_peeling_costs_all_drs): Likewise.
15252 (vect_peeling_hash_get_lowest_cost): Likewise.
15253 (vect_enhance_data_refs_alignment): Likewise.
15254 (vect_find_same_alignment_drs): Likewise.
15255 (vect_analyze_data_refs_alignment): Likewise.
15256 (vect_analyze_group_access_1): Likewise.
15257 (vect_analyze_group_access): Likewise.
15258 (vect_analyze_data_ref_access): Likewise.
15259 (vect_analyze_data_ref_accesses): Likewise.
15260 (vect_vfa_segment_size): Likewise.
15261 (vect_small_gap_p): Likewise.
15262 (vectorizable_with_step_bound_p): Likewise.
15263 (vect_prune_runtime_alias_test_list): Likewise.
15264 (vect_analyze_data_refs): Likewise.
15265 (vect_supportable_dr_alignment): Likewise.
15266 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
15267 (vect_gen_prolog_loop_niters): Likewise.
15268 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
15269 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
15271 (vect_recog_mask_conversion_pattern): Likewise.
15272 (vect_try_gather_scatter_pattern): Likewise.
15273 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
15274 to vect_get_store_cost.
15275 (vect_get_store_cost): Get stmt_info instead of DR.
15276 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
15277 (vect_get_load_cost): Get stmt_info instead of DR.
15279 2018-06-01 Richard Biener <rguenther@suse.de>
15281 PR middle-end/86017
15282 * gimple-fold.c (var_decl_component_p): Also allow offsetted
15283 vars wrapped in MEM_REFs.
15285 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15287 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15288 Fix subreg tests so that we only return a choice between
15289 GENERAL_REGS and FP_REGS if the original classes included both.
15291 2018-06-01 Richard Biener <rguenther@suse.de>
15294 * tree-ssa-structalias.c (get_function_part_constraint):
15295 Handle NULL fi->decl.
15296 (find_func_aliases_for_call): Properly handle indirect
15297 fi from direct call.
15298 (find_func_clobbers): Likewise.
15299 (ipa_pta_execute): Likewise.
15300 (create_variable_info_for): For functions that are ifunc_resolver
15301 resolve to a varinfo that contains the result of the resolver call.
15302 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
15305 2018-05-31 Michael Collison <michael.collison@arm.com>
15307 * config/aarch64/aarch64.md:
15308 (*fix_to_zero_extenddfdi2): New pattern.
15309 * gcc.target/aarch64/fix_extend1.c: New testcase.
15311 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
15313 PR middle-end/78809
15314 PR middle-end/83026
15315 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
15316 and BUILT_IN_STRNCMP_EQ.
15317 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
15318 BUILT_IN_STRNCMP_EQ.
15319 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
15320 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15321 (gimple_fold_builtin): Likewise.
15322 * tree-ssa-strlen.c (compute_string_length): New function.
15323 (determine_min_obsize): New function.
15324 (handle_builtin_string_cmp): New function to handle calls to
15325 string compare functions.
15326 (strlen_optimize_stmt): Add handling to builtin string compare
15328 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15329 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15330 * tree.c (build_common_builtin_nodes): Add new defines of
15331 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
15333 2018-05-31 Jakub Jelinek <jakub@redhat.com>
15336 * bb-reorder.c (pass_partition_blocks::gate): Return false for
15337 functions with naked attribute.
15339 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
15341 * config/i386/sse.md (avx_vec_concat<mode>):
15342 Substitute concat_tg_mode mode attribute with xtg_mode.
15343 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
15344 (concat_tg_mode): Remove mode attribute.
15346 2018-05-31 Martin Sebor <msebor@redhat.com>
15349 * calls.c (alloc_max_size): Correct a logic error/typo.
15350 Treat excessive arguments as infinite. Warn for invalid arguments.
15351 * doc/invoke.texi (-Walloc-size-larger-than): Update.
15353 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
15356 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
15357 and movx for Haswell.
15359 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
15360 Cesar Philippidis <cesar@codesourcery.com>
15362 PR middle-end/85879
15363 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
15364 when emitting error on private/firstprivate reductions.
15365 * omp-low.c (lower_omp_target): Avoid reference-type processing
15366 on pointers for firstprivate clause.
15368 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15370 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
15373 * config/aarch64/aarch64-simd.md
15374 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
15375 (aarch64_ld1_x3_<mode>): Likewise
15376 (aarch64_st1x2<VALLDIF:mode>): Likewise
15377 (aarch64_st1_x2_<mode>): Likewise
15378 (aarch64_st1x3<VALLDIF:mode>): Likewise
15379 (aarch64_st1_x3_<mode>): Likewise
15380 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
15381 (vld1_s8_x3): Likewise.
15382 (vld1_u16_x3): Likewise.
15383 (vld1_s16_x3): Likewise.
15384 (vld1_u32_x3): Likewise.
15385 (vld1_s32_x3): Likewise.
15386 (vld1_u64_x3): Likewise.
15387 (vld1_s64_x3): Likewise.
15388 (vld1_f16_x3): Likewise.
15389 (vld1_f32_x3): Likewise.
15390 (vld1_f64_x3): Likewise.
15391 (vld1_p8_x3): Likewise.
15392 (vld1_p16_x3): Likewise.
15393 (vld1_p64_x3): Likewise.
15394 (vld1q_u8_x3): Likewise.
15395 (vld1q_s8_x3): Likewise.
15396 (vld1q_u16_x3): Likewise.
15397 (vld1q_s16_x3): Likewise.
15398 (vld1q_u32_x3): Likewise.
15399 (vld1q_s32_x3): Likewise.
15400 (vld1q_u64_x3): Likewise.
15401 (vld1q_s64_x3): Likewise.
15402 (vld1q_f16_x3): Likewise.
15403 (vld1q_f32_x3): Likewise.
15404 (vld1q_f64_x3): Likewise.
15405 (vld1q_p8_x3): Likewise.
15406 (vld1q_p16_x3): Likewise.
15407 (vld1q_p64_x3): Likewise.
15408 (vst1_s64_x2): Likewise.
15409 (vst1_u64_x2): Likewise.
15410 (vst1_f64_x2): Likewise.
15411 (vst1_s8_x2): Likewise.
15412 (vst1_p8_x2): Likewise.
15413 (vst1_s16_x2): Likewise.
15414 (vst1_p16_x2): Likewise.
15415 (vst1_s32_x2): Likewise.
15416 (vst1_u8_x2): Likewise.
15417 (vst1_u16_x2): Likewise.
15418 (vst1_u32_x2): Likewise.
15419 (vst1_f16_x2): Likewise.
15420 (vst1_f32_x2): Likewise.
15421 (vst1_p64_x2): Likewise.
15422 (vst1q_s8_x2): Likewise.
15423 (vst1q_p8_x2): Likewise.
15424 (vst1q_s16_x2): Likewise.
15425 (vst1q_p16_x2): Likewise.
15426 (vst1q_s32_x2): Likewise.
15427 (vst1q_s64_x2): Likewise.
15428 (vst1q_u8_x2): Likewise.
15429 (vst1q_u16_x2): Likewise.
15430 (vst1q_u32_x2): Likewise.
15431 (vst1q_u64_x2): Likewise.
15432 (vst1q_f16_x2): Likewise.
15433 (vst1q_f32_x2): Likewise.
15434 (vst1q_f64_x2): Likewise.
15435 (vst1q_p64_x2): Likewise.
15436 (vst1_s64_x3): Likewise.
15437 (vst1_u64_x3): Likewise.
15438 (vst1_f64_x3): Likewise.
15439 (vst1_s8_x3): Likewise.
15440 (vst1_p8_x3): Likewise.
15441 (vst1_s16_x3): Likewise.
15442 (vst1_p16_x3): Likewise.
15443 (vst1_s32_x3): Likewise.
15444 (vst1_u8_x3): Likewise.
15445 (vst1_u16_x3): Likewise.
15446 (vst1_u32_x3): Likewise.
15447 (vst1_f16_x3): Likewise.
15448 (vst1_f32_x3): Likewise.
15449 (vst1_p64_x3): Likewise.
15450 (vst1q_s8_x3): Likewise.
15451 (vst1q_p8_x3): Likewise.
15452 (vst1q_s16_x3): Likewise.
15453 (vst1q_p16_x3): Likewise.
15454 (vst1q_s32_x3): Likewise.
15455 (vst1q_s64_x3): Likewise.
15456 (vst1q_u8_x3): Likewise.
15457 (vst1q_u16_x3): Likewise.
15458 (vst1q_u32_x3): Likewise.
15459 (vst1q_u64_x3): Likewise.
15460 (vst1q_f16_x3): Likewise.
15461 (vst1q_f32_x3): Likewise.
15462 (vst1q_f64_x3): Likewise.
15463 (vst1q_p64_x3): Likewise.
15465 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15467 * config/msp430/msp430.c (msp430_output_labelref): Prepend
15468 user_label_prefix to name.
15470 * tree-core.h: Update comment about the format of NAME string
15471 passed to handler in attribute_spec.
15473 * config/msp430/msp430.md: Remove erroneous subreg expression from
15474 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
15475 zero_extend{q,h}isi2.
15477 2018-05-30 Borislav Petkov <bp@suse.de>
15479 * doc/extend.texi: Document some architecture specific
15480 constraints and sort entries.
15482 2018-05-30 Martin Sebor <msebor@redhat.com>
15484 PR middle-end/85369
15485 * builtins.c (expand_builtin_stpcpy_1): New function.
15486 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
15487 only if the former succeeds.
15489 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15491 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
15494 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15496 * doc/invoke.texi (-flinker-output): Document
15498 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15500 * passes.c (ipa_write_summaries): Only modify statements if body
15502 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
15503 incrementally linking.
15504 (ipa_passes): Likewise.
15505 * lto-cgraph.c (lto_output_node): When incrementally linking do not
15506 pass down resolution info.
15507 * common.opt (flag_incremental_link): Update info.
15508 * gcc.c (plugin specs): Turn flinker-output=* to
15509 -plugin-opt=-linker-output-known
15510 * toplev.c (compile_file): Also cut compilation when doing incremental
15512 * flag-types. (enum lto_partition_model): Add
15513 LTO_LINKER_OUTPUT_NOLTOREL.
15514 (invoke.texi): Add -flinker-output docs.
15515 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
15516 link same way as WPA; do not stream in dead initializers.
15518 * dwarf2out.c (dwarf2out_die_ref_for_decl,
15519 darf2out_register_external_decl): Support incremental link.
15521 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15523 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
15525 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15527 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
15528 it down to simple_object_copy_lto_debug_sections.
15529 (run_gcc): Determine incremental LTO link time and configure
15530 lto1 into non-wpa mode, disable renaming of debug sections.
15532 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
15534 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
15535 descriptions of various incorrectly documented functions.
15537 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
15540 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15541 address check not strict.
15543 2018-05-30 Richard Biener <rguenther@suse.de>
15545 PR tree-optimization/85964
15546 * tracer.c (better_p): Drop initialized count check, we only
15547 call the function with initialized counts now.
15548 (find_best_successor): Do find a best edge if one
15549 has uninitialized count.
15550 (find_best_predecessor): Likewise. Do BB frequency check only
15551 if count is initialized.
15553 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15555 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15556 (aarch64_ldrstr_offset_compare): New.
15557 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15558 load/store orderings.
15559 (aarch64_gen_adjusted_ldpstp): Likewise.
15561 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15563 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15564 Check for subset of GENERAL_REGS and FP_REGS.
15565 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15568 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15570 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15571 and wi::to_poly_offset. Add the current offset and then check
15572 whether the sum fits, rather than using an unchecked addition of
15573 a checked term. Check for a shwi rather than a uhwi.
15574 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15575 (store_constructor): Use poly_int_tree_p.
15576 (expand_expr_real_1): Likewise.
15577 * function.c (assign_temp): Likewise.
15578 * fold-const.c (const_binop): Use poly_int_tree_p and
15579 wi::to_poly_offset.
15580 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15582 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15583 to_poly_offset for MEM offsets.
15584 * ipa-icf.c (sem_variable::equals): Likewise.
15585 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15586 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15587 wi::to_poly_offset for BIT_FIELD_REF offsets.
15588 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15589 wi::to_poly_offset.
15590 * var-tracking.c (emit_note_insn_var_location): Use
15591 tree_to_poly_uint64.
15593 2018-05-29 Jim Wilson <jimw@sifive.com>
15595 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15597 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15600 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15601 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15603 (sse4_1_round<mode>2): Use nonimmediate_operand
15604 for operand 1 predicate.
15606 2018-05-29 Martin Sebor <msebor@redhat.com>
15607 Richard Biener <rguenther@suse.de>
15610 * calls.c (get_size_range): Call determine_value_range instead
15611 of get_value_range..
15612 * tree-vrp.h (determine_value_range): Declared new function.
15613 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15615 2018-05-29 Richard Biener <rguenther@suse.de>
15617 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15618 sure to use non-pattern stmts for get_earlier_stmt arguments.
15619 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15620 called on pattern stmts.
15621 (get_later_stmt): Likewise.
15623 2018-05-29 Martin Liska <mliska@suse.cz>
15625 PR gcov-profile/85759
15626 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15629 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15631 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15633 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15637 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15638 VEC_PACK_FLOAT_EXPR): New tree codes.
15639 * tree-pretty-print.c (op_code_prio): Handle
15640 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15641 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15642 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15643 * tree-inline.c (estimate_operator_cost): Likewise.
15644 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15645 * fold-const.c (const_binop): Likewise.
15646 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15647 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15648 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15649 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15650 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15651 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15652 * expr.c (expand_expr_real_2): Likewise.
15653 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15654 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15655 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15657 * optabs.c (expand_widen_pattern_expr): For
15658 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15659 sign from result type rather than operand's type.
15660 (expand_binop_directly): For vec_packu_float_optab and
15661 vec_packs_float_optab allow result type to be different from operand's
15663 * optabs-tree.c (optab_for_tree_code): Handle
15664 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15665 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15666 * tree-vect-generic.c (expand_vector_operations_1): Handle
15667 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15668 VEC_PACK_FLOAT_EXPR.
15669 * tree-vect-stmts.c (supportable_widening_operation): Handle
15671 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15672 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15673 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15674 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15675 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15677 (vec_pack<floatprefix>_float_<mode>): New expander.
15678 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15680 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15681 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15682 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15683 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15684 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15686 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15687 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15688 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15689 VEC_PACK_FLOAT_EXPR): Document.
15691 2018-05-29 Richard Biener <rguenther@suse.de>
15693 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15695 (stmt_vec_info_vec): Make pointer.
15696 (init_stmt_vec_info_vec): Remove.
15697 (free_stmt_vec_info_vec): Likewise.
15698 (set_stmt_vec_info_vec): New function.
15699 (free_stmt_vec_infos): Likewise.
15700 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15701 (set_vinfo_for_stmt): Likewise.
15702 (get_earlier_stmt): Likewise.
15703 (get_later_stmt): Likewise.
15704 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15705 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15706 (vec_info::~vec_info): Free stmt_vec_infos.
15707 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15708 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15709 (pass_slp_vectorize::execute): Likewise.
15710 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15711 (free_stmt_vec_info_vec): Likewise.
15712 (set_stmt_vec_info_vec): New function.
15713 (free_stmt_vec_infos): Likewise.
15714 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15715 the global stmt_vec_info_vec.
15716 * tree-parloops.c (gather_scalar_reductions): Use
15717 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15720 2018-05-29 Richard Biener <rguenther@suse.de>
15722 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15724 2018-05-29 Martin Liska <mliska@suse.cz>
15725 David Malcolm <dmalcolm@redhat.com>
15727 * vec.c (test_reverse): New.
15728 (vec_c_tests): Add new test.
15729 * vec.h (vl_ptr>::reverse): New function.
15731 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15733 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15735 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15738 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15740 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15742 2018-05-28 Richard Biener <rguenther@suse.de>
15744 PR tree-optimization/85933
15745 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15746 look at stmts marked as vectorizable.
15748 2018-05-28 Richard Biener <rguenther@suse.de>
15750 PR tree-optimization/85934
15751 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15752 vector boolean check before scalar optimization.
15754 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15756 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15759 2018-05-28 Mark Wielaard <mark@klomp.org>
15761 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15762 if it is an expression containing a minus sign.
15764 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15766 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15768 2018-05-27 Paul Koning <ni1d@arrl.net>
15770 * config/pdp11/pdp11.md (truncsihi2): Remove.
15772 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15773 Chung-Ju Wu <jasonwucj@gmail.com>
15775 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15777 (unaligned_store_dw): Ditto.
15778 * config/nds32/nds32-memory-manipulation.c
15779 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15780 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15781 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15782 (emit_setmem_word_loop): Rename to ...
15783 (emit_setmem_doubleword_loop): ... this.
15784 (nds32_gen_dup_4_byte_to_word_value): New function.
15785 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15786 (nds32_expand_setmem_loop): Refine implementation.
15787 (nds32_expand_setmem_loop_v3m): Ditto.
15788 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15791 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15793 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15795 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15797 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15798 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15799 machine->attr_no_prologue_p.
15800 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15801 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15802 (nds32_expand_epilogue): Consider attr_naked_p.
15803 (nds32_expand_epilogue_v3pop): Likewise.
15804 (nds32_can_use_return_insn): Likewise.
15805 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15806 attr_no_prologue_p fields.
15807 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15809 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15812 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15814 * config/i386/sse.md
15815 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15817 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15819 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15821 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15823 (*<floatsuffix>floatv2div2sf2): Rename to ...
15824 (*float<floatunssuffix>v2div2sf2): ... this.
15825 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15826 (float<floatunssuffix>v2div2sf2_mask): ... this.
15827 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15828 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15829 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15831 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15833 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15835 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15837 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15839 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15841 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15842 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15843 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15844 gen_ufix_truncv8dfv8si2.
15845 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15846 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15847 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15848 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15849 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15850 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15851 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15852 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15854 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15858 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15860 2018-05-25 Jim Wilson <jimw@sifive.com>
15862 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15863 * config/riscv/riscv.c (struct machine_function): Add
15864 interrupt_handler_p and attribute_checked_p fields.
15865 (riscv_attribute_table): Add interrupt.
15866 (riscv_interrupt_type_p): New.
15867 (riscv_save_reg_p): Save extra regs for interrupt handler.
15868 (riscv_use_save_libcall): Return false for interrupt handler.
15869 (riscv_first_stack_step): Add forward declaration.
15870 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15871 for interrupt handler with large frame. Use it for saved reg list.
15872 (riscv_expand_prologue): Move flag_stack_usage_info support to
15873 eliminate duplication.
15874 (riscv_expand_epilogue): Generate mret for interrupt handler.
15875 (riscv_epilogue_uses): New.
15876 (riscv_can_use_return_insn): Return false for interrupt handler.
15877 (riscv_function_ok_for_sibcall): Likewise.
15878 (riscv_set_current_function): Add interrupt handler support.
15879 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15880 * config/riscv/riscv.md (UNSPECV_MRET): New.
15882 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15884 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15886 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15888 PR tree-optimization/85712
15889 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15890 this candidate has already been replaced in-situ by a copy.
15892 2018-05-25 Jason Merrill <jason@redhat.com>
15894 PR c++/80485 - inline function non-zero address.
15895 * symtab.c (nonzero_address): Check DECL_COMDAT.
15897 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15900 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15901 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15902 extension of SImode operation. Use const123_operand predicate.
15903 (*saddsi_1): Remove.
15904 (*saddl_se_1): Ditto.
15905 (*ssubsi_1): Ditto.
15906 (*ssubl_se_1): Ditto.
15907 * config/alpha/predicates.md (const123_operand): New predicate.
15908 * config/alpha/constraints.md (P): Use IN_RANGE.
15910 2018-05-25 Richard Biener <rguenther@suse.de>
15912 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15914 (ref_maybe_used_by_stmt_p): Likewise.
15915 (stmt_may_clobber_ref_p): Likewise.
15916 (stmt_may_clobber_ref_p_1): Likewise.
15917 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15919 (ref_maybe_used_by_stmt_p): Likewise.
15920 (stmt_may_clobber_ref_p): Likewise.
15921 (stmt_may_clobber_ref_p_1): Likewise.
15922 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15923 the alias oracle to disambiguate DRs with stmts DR analysis
15925 (vect_analyze_data_refs): Do not give up on not analyzable
15926 DRs for BB vectorization. Remove code truncating the dataref
15929 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15932 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15933 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15934 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15936 2018-05-25 Richard Biener <rguenther@suse.de>
15938 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15939 function, combining stmt data ref gathering and fatal analysis
15941 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15942 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15943 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15944 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15945 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15947 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15949 PR tree-optimization/85720
15950 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15951 SCC if all partitions are builtins.
15952 (version_loop_by_alias_check): New parameter. Generate cancelable
15953 runtime alias check if all partitions are builtins.
15954 (distribute_loop): Update call to above function.
15956 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15958 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15959 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15960 (parm_default_def_partition_arg): Ditto.
15961 (set_parm_default_def_partition): Ditto.
15962 (get_parm_default_def_partitions): Ditto and make it static.
15963 (get_undefined_value_partitions): Ditto and make it static.
15964 (remove_ssa_form): Refactor call to init_var_map here.
15965 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15966 computation for loop region.
15967 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15968 (register_default_def): Delete.
15969 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15970 (parm_default_def_partition_arg): Ditto.
15971 (set_parm_default_def_partition): Ditto.
15972 (get_parm_default_def_partitions): Ditto and make it static.
15973 (get_undefined_value_partitions): Ditto and make it static.
15974 (coalesce_with_default, coalesce_with_default): Update comment.
15975 (create_coalesce_list_for_region): New func factored out from
15976 create_outofssa_var_map.
15977 (populate_coalesce_list_for_outofssa): New func factored out from
15978 create_outofssa_var_map and coalesce_ssa_name.
15979 (create_outofssa_var_map): Delete.
15980 (coalesce_ssa_name): Refactor to support live range computation.
15981 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15982 (get_parm_default_def_partitions): Delete.
15983 (get_undefined_value_partitions): Ditto.
15984 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15985 computation for loop region.
15986 (new_tree_live_info, loe_visit_block): Ditto.
15987 (live_worklist, set_var_live_on_entry): Ditto.
15988 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15989 * tree-ssa-live.h (struct _var_map): New fields.
15990 (init_var_map): Change decl.
15991 (region_contains_p): New.
15993 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15995 * tree-ssa-live.h (live_merge_and_clear): Delete.
15997 2018-05-25 Richard Biener <rguenther@suse.de>
16000 * tree-dump.c (dequeue_and_dump): Remove access to removed
16001 operand 2 of a SWITCH_EXPR.
16003 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16005 * doc/sourcebuild.texi (vect_double_cond_arith): Include
16006 multiplication and division.
16007 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
16008 (cond_udiv@var{m}, cond_umod@var{m}): Document.
16009 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
16010 (cond_udiv_optab, cond_umod_optab): New optabs.
16011 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
16012 (IFN_COND_RDIV): New internal functions.
16013 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
16014 TRUNC_MOD_EXPR and RDIV_EXPR.
16015 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
16016 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
16018 (SVE_INT_BINARY): Include mult.
16019 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
16020 (optab, sve_int_op): Handle mult.
16021 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
16023 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
16024 for SVE_INT_BINARY_SD.
16026 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16028 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
16029 (optab, sve_int_op): Handle div and udiv.
16030 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
16031 for SVE_INT_BINARY_SD.
16032 (*<optab><mode>3): New insn for the same.
16034 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16036 * tree-vect-patterns.c: Include predict.h.
16037 (vect_recog_divmod_pattern): Restrict check for division support
16038 to when optimizing for size.
16040 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16042 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
16043 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
16044 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
16045 (gimple_match_op::set_op): Likewise.
16046 (gimple_resimplify4): Declare.
16047 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
16048 (expr::gen_transform): Likewise.
16049 (decision_tree::gen): Generate a simplification routine for 4 operands.
16050 * gimple-match-head.c (gimple_simplify): Add an overload for
16051 4 operands. In the top-level function, handle up to 4 call
16052 arguments and call gimple_resimplify4.
16053 (gimple_resimplify4): New function.
16054 (build_call_internal): Pass a fourth operand.
16055 (maybe_push_to_seq): Likewise.
16056 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
16057 Fold VEC_COND_EXPRs of an operation and a default value into
16058 an IFN_COND_* function if possible.
16059 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
16061 (SVE_COND_FP_BINARY): Include them.
16062 (optab, sve_fp_op): Handle them.
16063 (SVE_INT_BINARY_REV): New code iterator.
16064 (SVE_COND_FP_BINARY_REV): New int iterator.
16065 (commutative): New int attribute.
16066 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
16068 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
16070 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
16071 (*cond_<optab><mode>): New patterns for reversed operands.
16073 2018-05-25 Richard Biener <rguenther@suse.de>
16075 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
16076 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
16077 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
16078 (STMT_VINFO_GROUPED_ACCESS): Adjust.
16079 * tree-vect-data-refs.c (everywhere): Adjust users.
16080 * tree-vect-loop.c (everywhere): Likewise.
16081 * tree-vect-slp.c (everywhere): Likewise.
16082 * tree-vect-stmts.c (everywhere): Likewise.
16083 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
16085 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16087 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
16089 (gcc_cv_as_section_exclude): ... this.
16090 Try Solaris as #exclude syntax.
16091 * configure: Regenerate.
16092 * config.in: Regenerate.
16093 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
16095 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
16096 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
16098 * varasm.c (default_elf_asm_named_section): Don't check if
16099 HAVE_GAS_SECTION_EXCLUDE is defined.
16101 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16103 * doc/md.texi: Update the documentation of the cond_* optabs
16104 to mention the new final operand. Fix GET_MODE_NUNITS call.
16105 Describe the scalar case too.
16106 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
16107 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
16109 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
16110 (get_conditional_internal_fn): Update comment.
16111 * tree-vect-loop.c (vectorizable_reduction): Pass the original
16112 accumulator value as a final argument to conditional functions.
16113 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
16114 a define_expand and add an "else" operand. Assert for now that
16115 the else operand is equal to operand 2. Use SVE_INT_BINARY and
16116 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
16117 (*cond_<optab><mode>): New patterns.
16118 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
16119 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
16120 (UNSPEC_COND_EOR): Delete.
16121 (optab): Remove associated mappings.
16122 (SVE_INT_BINARY): New code iterator.
16123 (sve_int_op): Remove int attribute and add "minus" to the code
16125 (SVE_COND_INT_OP): Delete.
16126 (SVE_COND_FP_OP): Rename to...
16127 (SVE_COND_FP_BINARY): ...this.
16129 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16131 * optabs.c (can_reuse_operands_p): New function.
16132 (maybe_legitimize_operands): Try to reuse the results for
16135 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16137 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
16138 Add {q} suffix to insn mnemonic.
16140 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16142 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
16143 (msp430_warn_func_return): New.
16145 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
16147 * fold-const.c (tree_nonzero_bits): New function.
16148 * fold-const.h (tree_nonzero_bits): Likewise.
16149 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
16150 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
16152 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
16156 * varasm.c (assemble_alias): Check ifunc_resolver only on
16159 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16162 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
16163 when memory input operand is handled.
16165 2018-05-24 Luis Machado <luis.machado@linaro.org>
16167 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
16169 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
16171 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16173 * match.pd: Delay FMA folds until after vectorization.
16175 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
16178 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
16179 address check not strict.
16181 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16183 * gimple-match.h (gimple_match_op): New class.
16184 (mprts_hook): Replace parameters with a gimple_match_op *.
16185 (maybe_build_generic_op): Likewise.
16186 (gimple_simplified_result_is_gimple_val): Replace parameters with
16187 a const gimple_match_op *.
16188 (gimple_simplify): Replace code_helper * and tree * parameters with
16189 a gimple_match_op * parameter.
16190 (gimple_resimplify1): Replace code_helper *, tree and tree *
16191 parameters with a gimple_match_op * parameter.
16192 (gimple_resimplify2): Likewise.
16193 (gimple_resimplify3): Likewise.
16194 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
16195 parameters with a gimple_match_op * parameter.
16196 * gimple-match-head.c (gimple_simplify): Change prototypes of
16197 auto-generated functions to take a gimple_match_op * instead of
16198 separate code_helper * and tree * parameters. Make the same
16199 change in the top-level overload and update calls to the
16200 gimple_resimplify routines. Update calls to the auto-generated
16201 functions and to maybe_push_res_to_seq in the publicly-facing
16202 operation-specific gimple_simplify overloads.
16203 (gimple_match_op::MAX_NUM_OPS): Define.
16204 (gimple_resimplify1): Replace rcode and ops with a single res_op
16205 parameter. Update call to gimple_simplify.
16206 (gimple_resimplify2): Likewise.
16207 (gimple_resimplify3): Likewise.
16208 (mprts_hook): Replace parameters with a gimple_match_op *.
16209 (maybe_build_generic_op): Likewise.
16210 (build_call_internal): Replace type, nargs and ops with
16211 a gimple_match_op *.
16212 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
16213 with a single gimple_match_op *. Update calls to mprts_hook,
16214 build_call_internal and gimple_simplified_result_is_gimple_val.
16215 Factor out code that is common to the tree_code and combined_fn cases.
16216 * genmatch.c (expr::gen_transform): Replace tem_code and
16217 tem_ops with a gimple_match_op called tem_op. Update calls
16218 to the gimple_resimplify functions and maybe_push_res_to_seq.
16219 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
16220 res_ops. Update call to the gimple_resimplify functions.
16221 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
16222 (decision_tree::gen): Make the functions take a gimple_match_op *
16223 called res_op instead of separate res_code and res_ops parameters.
16224 Update call accordingly.
16225 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
16226 and ops with a single res_op parameter. Update calls to
16227 maybe_build_generic_op and maybe_push_res_to_seq.
16228 (fold_stmt_1): Update calls to gimple_simplify and
16229 replace_stmt_with_simplification.
16230 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
16231 and gimple_simplified_result_is_gimple_val.
16232 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
16234 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
16235 with a gimple_match_op *.
16236 (vn_nary_build_or_lookup): Likewise. Update call to
16237 vn_nary_build_or_lookup_1.
16238 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
16239 gimple_match_op *. Update calls to the gimple_resimplify routines
16240 and to gimple_simplified_result_is_gimple_val.
16241 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
16242 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
16243 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
16244 (visit_nary_op): Likewise.
16245 (visit_reference_op_load): Likewise.
16247 2018-05-23 Luis Machado <luis.machado@linaro.org>
16249 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
16250 modifier for printing the step amount.
16252 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16255 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
16258 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
16260 * doc/sourcebuild.texi (Endianness): New subsubsection.
16262 2018-05-23 Luis Machado <luis.machado@linaro.org>
16264 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16265 <prefetch_dynamic_strides>: New const bool field.
16266 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16267 prefetch_dynamic_strides.
16268 (exynosm1_prefetch_tune): Likewise.
16269 (thunderxt88_prefetch_tune): Likewise.
16270 (thunderx_prefetch_tune): Likewise.
16271 (thunderx2t99_prefetch_tune): Likewise.
16272 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
16274 (aarch64_override_options_internal): Update to set
16275 PARAM_PREFETCH_DYNAMIC_STRIDES.
16276 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16277 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16278 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16279 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16280 prefetch-dynamic-strides setting.
16282 2018-05-23 Luis Machado <luis.machado@linaro.org>
16284 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16285 <minimum_stride>: New const int field.
16286 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16287 minimum_stride field defaulting to -1.
16288 (exynosm1_prefetch_tune): Likewise.
16289 (thunderxt88_prefetch_tune): Likewise.
16290 (thunderx_prefetch_tune): Likewise.
16291 (thunderx2t99_prefetch_tune): Likewise.
16292 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
16293 <default_opt_level>: Set to 3.
16294 (aarch64_override_options_internal): Update to set
16295 PARAM_PREFETCH_MINIMUM_STRIDE.
16296 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16297 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16298 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16299 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16300 stride is constant and is below the minimum stride threshold.
16302 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16304 * config/arm/arm-cpus.in (mode26): Delete.
16305 (armv4): Delete mode26 reference.
16306 * config/arm/arm.c (arm_configure_build_target): Delete use of
16309 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16311 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
16313 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
16314 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
16316 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
16317 (floatunsdidf2): Ditto.
16319 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16321 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
16322 (fixuns_trunc<mode>si2_avx512f): Ditto.
16323 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
16324 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
16325 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
16327 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16329 PR rtl-optimization/79985
16330 * df-scan.c (df_insn_refs_collect): Remove special case for
16331 global registers and asm statements.
16333 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16335 * extend.texi (Global Register Variables): Rewrite the bullet list.
16336 Note that the register is available for allocation. Note that access
16337 via inline asm must use constraints. Add note about async-signal
16338 handlers. Remove paragraph about automagic register selection.
16340 2018-05-23 Richard Biener <rguenther@suse.de>
16342 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
16343 of fixed offset from memset VN.
16345 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
16347 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
16348 first_interp field.
16349 (alloc_cand_and_find_basis): Initialize first_interp field.
16350 (slsr_process_mul): Modify first_interp field.
16351 (slsr_process_add): Likewise.
16352 (slsr_process_cast): Modify first_interp field for each new
16354 (slsr_process_copy): Likewise.
16355 (dump_candidate): Dump first_interp field.
16356 (replace_mult_candidate): Process all interpretations, not just
16358 (replace_rhs_if_not_dup): Likewise.
16359 (replace_one_candidate): Likewise.
16361 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
16363 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
16365 (aarch64_needs_frame_chain): New function.
16366 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
16368 2018-05-23 Sudakshina Das <sudi.das@arm.com>
16371 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
16372 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
16373 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
16374 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
16375 as true for strict-align.
16376 (aarch64_can_inline_p): Perform checks even when callee has no
16377 attributes to check for strict alignment.
16378 * doc/extend.texi (AArch64 Function Attributes): Document
16380 * doc/invoke.texi: (AArch64 Options): Likewise.
16382 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
16384 PR tree-optimization/85853
16385 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
16386 the handling of the root of the node to...
16387 (vect_slp_analyze_node_operations_1): ...this new function,
16388 and run the whole thing with the child nodes' def types
16389 set according to their SLP node's def type.
16391 2018-05-23 Richard Biener <rguenther@suse.de>
16393 PR middle-end/85874
16394 * tree-data-ref.c (create_runtime_alias_checks): Defer
16395 and ignore overflow warnings.
16397 2018-05-23 Yury Gribov <tetra2005@gmail.com>
16399 PR tree-optimization/85822
16400 * tree-vrp.c (is_masked_range_test): Fix handling of negative
16403 2018-05-23 Richard Biener <rguenther@suse.de>
16405 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
16406 memset constants via native_interpret_expr.
16408 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
16411 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
16413 (cgraph_node::create_alias): Likewise.
16414 (cgraph_node::get_availability): Check ifunc_resolver instead
16415 of looking up ifunc attribute.
16416 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
16417 * varasm.c (do_assemble_alias): Likewise.
16418 (assemble_alias): Likewise.
16419 (default_binds_local_p_3): Likewise.
16420 * cgraph.h (cgraph_node): Add ifunc_resolver.
16421 (cgraph_node::only_called_directly_or_aliased_p): Return false
16422 for IFUNC resolver.
16423 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
16425 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
16426 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
16427 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
16428 instead of looking up ifunc attribute.
16430 2018-05-22 Luis Machado <luis.machado@linaro.org>
16432 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
16434 2018-05-22 Martin Sebor <msebor@redhat.com>
16436 PR middle-end/85359
16437 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
16438 only when expasion succeeds.
16439 (expand_builtin_strcmp): Same.
16440 (expand_builtin_strncmp): Same.
16442 2018-05-22 Martin Sebor <msebor@redhat.com>
16444 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
16446 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16447 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16449 * config/aarch64/aarch64-ldpstp.md: Replace uses of
16450 aarch64_mem_pair_operand with memory_operand and delete operand swapping
16452 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16453 Add check for legitimate_address.
16454 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
16455 (aarch64_swap_ldrstr_operands): New.
16456 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
16459 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16460 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16462 * config/aarch64/aarch64.md: New patterns to generate stp
16464 (store_pair_sw, store_pair_dw): New patterns to generate stp for
16465 single words and double words.
16466 (load_pair_sw, load_pair_dw): Likewise.
16467 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
16469 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
16471 * config/aarch64/aarch64-ldpstp.md: Modify peephole
16472 for different mode ldpstp and add peephole for merged zero stores.
16473 Likewise for loads.
16474 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16476 (aarch64_gen_store_pair): Rename calls to match new patterns.
16477 (aarch64_gen_load_pair): Rename calls to match new patterns.
16478 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
16479 (load_pair<DREG:mode><DREG2:mode>): ... This.
16480 (store_pair<mode>): Rename to...
16481 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
16482 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
16483 New mode iterators.
16484 (V_INT_EQUIV): Handle SImode.
16485 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
16488 2018-05-22 Martin Sebor <msebor@redhat.com>
16491 * calls.c (maybe_warn_nonstring_arg): Use string length to set
16492 or ajust the presumed bound on an operation to avoid unnecessary
16495 2018-05-22 Martin Sebor <msebor@redhat.com>
16497 PR tree-optimization/85826
16498 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
16499 assuming that a DECL necesarily has a constant size.
16501 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
16503 PR middle-end/85862
16504 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
16506 2018-05-22 Richard Biener <rguenther@suse.de>
16508 PR tree-optimization/85834
16509 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
16510 non-constant and non-zero memset arguments.
16512 2018-05-22 Martin Liska <mliska@suse.cz>
16515 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
16517 2018-05-22 Richard Biener <rguenther@suse.de>
16519 PR tree-optimization/85863
16520 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
16521 comparisons when vectype is specified.
16522 (vectorizable_condition): Do not specify vectype for
16523 vect_is_simple_cond when SLP vectorizing.
16525 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
16528 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
16529 define __ibm128 as long double.
16530 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
16531 as a distinct type when IEEE 128-bit support is enabled.
16532 (init_float128_ieee): Fix up conversions between IFmode and IEEE
16533 128-bit types to use the correct functions.
16534 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
16535 convert between 128-bit floating point types that have different
16536 modes but the same representation, instead of using gen_lowpart to
16538 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
16540 (IFKF_reg): New attributes to give the register constraints for
16542 (extend<mode>tf2_internal): New insns to mark an explicit
16543 conversion between 128-bit floating point types that have a
16544 different mode but share the same representation.
16546 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
16548 PR tree-optimization/85814
16549 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
16550 a null return from get_strinfo when unsharing the next
16551 strinfo in the chain.
16553 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16556 * varasm.c (weak_finish): Clean up weak_decls.
16558 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16560 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16561 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16562 UNSPEC_UADALP values.
16563 * config/aarch64/iterators.md (ABAL): New int iterator.
16566 (sur): Add mappings for the above.
16567 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16569 (aarch64_<sur>abal<mode>_4): Likewise.
16570 (aarch64_<sur>adalp<mode>_3): Likewise.
16571 (<sur>sadv16qi): New define_expand.
16573 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16575 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16576 (*movdf_internal): Ditto.
16577 (*rcpsf2_sse): Ditto.
16578 (*rsqrtsf2_sse): Ditto.
16579 (*sqrt<mode>2_sse): Ditto.
16581 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16583 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16585 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16586 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16587 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16588 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16590 * config/aarch64/arm_neon.h: Likewise.
16591 * config/aarch64/iterators.md (VQ_I): New.
16593 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16595 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16597 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16600 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16602 * config/nds32/constraints.md (S): New constraint.
16603 * config/nds32/nds32.md (call_internal): Use constraint S.
16604 (call_value_internal): Likewise.
16605 (sibcall_internal): Likewise.
16606 (sibcall_value_internal): Likewise.
16608 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16609 Chung-Ju Wu <jasonwucj@gmail.com>
16611 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16612 into consideration.
16614 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16615 Chung-Ju Wu <jasonwucj@gmail.com>
16617 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16618 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16619 (nds32_rtx_costs_impl): Simplify.
16620 (nds32_address_cost_impl): Simplify.
16621 (nds32_init_rtx_costs): New function.
16622 (nds32_rtx_costs_speed_prefer): Likewise.
16623 (nds32_rtx_costs_size_prefer): Likewise.
16624 (nds32_address_cost_speed_prefer): Likewise.
16625 (nds32_address_cost_speed_fwprop): Likewise.
16626 (nds32_address_cost_size_prefer): Likewise.
16627 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16628 * config/nds32/nds32.c (nds32_option_override): Use
16629 nds32_init_rtx_costs function.
16631 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16633 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16634 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16635 (TARGET_PIPELINE_N8): Likewise.
16636 (TARGET_PIPELINE_N10): Likewise.
16637 (TARGET_PIPELINE_N13): Likewise.
16638 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16640 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16642 * config/nds32/nds32-fpu.md: Update copyright year.
16644 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16646 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16648 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16650 * config/nds32/nds32.c
16651 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16652 * config/nds32/nds32.opt (minline-asm-r15): New option.
16654 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16656 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16658 * config/nds32/nds32.md (abssi2): New pattern.
16660 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16662 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16663 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16664 Merge insn pattern from sse_cvtsi2ss<round_name> and
16665 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16666 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16667 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16668 using SWI48 mode iterator.
16669 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16670 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16671 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16672 pattern from sse_cvttss2si<round_saeonly_name>
16673 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16674 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16675 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16676 using SWI48 mode iterator.
16677 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16678 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16679 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16680 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16681 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16682 using SWI48 mode iterator.
16683 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16684 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16685 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16686 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16687 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16688 SWI48 mode iterator.
16689 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16690 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16691 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16692 pattern from sse_cvttsd2si<round_saeonly_name>
16693 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16695 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16697 * config/nds32/nds32-md-auxiliary.c
16698 (nds32_valid_smw_lwm_base_p): Refine.
16699 (nds32_output_smw_single_word): Refine.
16700 (nds32_output_smw_double_word): New.
16701 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16703 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16705 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16706 (nds32_output_stack_pop): Refine.
16707 (nds32_expand_unaligned_load): Refine.
16708 (nds32_expand_unaligned_store): Refine.
16710 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16711 Chung-Ju Wu <jasonwucj@gmail.com>
16713 * config/nds32/constants.md: Add TP_REGNUM constant.
16714 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16715 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16717 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16718 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16719 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16720 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16721 and PIC code generation.
16722 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16724 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16726 * config/nds32/nds32.md: Support TLS and PIC.
16727 * config/nds32/nds32.c: Support TLS and PIC.
16728 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16729 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16732 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16734 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16735 mode with E_ prefix.
16737 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16738 Chung-Ju Wu <jasonwucj@gmail.com>
16740 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16741 * config/nds32/nds32-md-auxiliary.c
16742 (symbolic_reference_mentioned_p): New.
16743 (nds32_legitimize_ict_address): New.
16744 (nds32_expand_ict_move): New.
16745 (nds32_indirect_call_referenced_p): New.
16746 (nds32_symbol_binds_local_p): Delete.
16747 (nds32_long_call_p): Modify.
16748 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16749 * config/nds32/nds32-protos.h
16750 (symbolic_reference_mentioned_p): Declare.
16751 (nds32_legitimize_ict_address): Declare.
16752 (nds32_expand_ict_move): Declare.
16753 (nds32_indirect_call_referenced_p): Declare.
16754 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16755 (nds32_relax_group): Use nds32_ict_const_p as condition.
16756 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16757 (nds32_asm_file_start): Output ict_model directive in asm code.
16758 (nds32_legitimate_address_p): Consider indirect call.
16759 (nds32_print_operand): Consider indirect call.
16760 (nds32_print_operand_address): Consider indirect call.
16761 (nds32_insert_attributes): Handle "indirect_call" attribute.
16762 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16763 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16764 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16765 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16766 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16767 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16768 (TARGET_ICT_MODEL_SMALL): Define.
16769 (TARGET_ICT_MODEL_LARGE): Define.
16770 * config/nds32/nds32.md (movsi): Consider ict model.
16771 (call, call_value): Consider ict model.
16772 (sibcall, sibcall_value): Consider ict model.
16773 * config/nds32/nds32.opt (mict-model): New option.
16774 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16777 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16778 Monk Chiang <sh.chiang04@gmail.com>
16779 Jim Wilson <jimw@sifive.com>
16781 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16782 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16783 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16784 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16785 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16786 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16787 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16788 compute save_libcall_adjustment properly.
16789 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16790 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16791 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16792 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16793 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16794 (ABI_SPEC): Handle mabi=ilp32e.
16795 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16796 (RVE): Add RVE mask.
16797 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16798 <-march>: Add rv32e as an example.
16800 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16803 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16804 (intra_create_variable_infos): Handle C++ constructors.
16806 2018-05-18 Martin Liska <mliska@suse.cz>
16808 * passes.def: Remove a redundant pass.
16810 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16813 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16815 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16817 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16819 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16821 (armv2, armv2a, armv3, armv3m): Delete architectures.
16822 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16823 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16824 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16826 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16827 (*mulsidi3adddi): Likewise.
16828 (mulsidi3): Likewise.
16829 (*mulsidi3_nov6): Likewise.
16830 (umulsidi3): Likewise.
16831 (umulsidi3_nov6): Likewise.
16832 (umaddsidi4): Likewise.
16833 (*umulsidi3adddi): Likewise.
16834 (smulsi3_highpart): Likewise.
16835 (*smulsi3_highpart_nov6): Likewise.
16836 (umulsi3_highpart): Likewise.
16837 (*umulsi3_highpart_nov6): Likewise.
16838 * config/arm/arm.h (arm_arch3m): Delete.
16839 * config/arm/arm.c (arm_arch3m): Delete.
16840 (arm_option_override_internal): Update armv3-related comment.
16841 (arm_configure_build_target): Delete use of isa_bit_mode32.
16842 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16843 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16844 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16845 (mulsa3): Likewise.
16846 (mulusa3): Likewise.
16847 * config/arm/arm-protos.h (arm_arch3m): Delete.
16848 * config/arm/arm-tables.opt: Regenerate.
16849 * config/arm/arm-tune.md: Likewise.
16850 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16851 deleted architectures.
16853 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16855 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16856 (armv5t, armv5te): New features.
16857 (ARMv5, ARMv5e): Delete fgroups.
16858 (ARMv5t, ARMv5te): Adjust for above changes.
16859 (ARMv6m): Likewise.
16860 (armv5, armv5e): Delete arches.
16861 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16863 (*call_reg_arm): Likewise.
16864 (*call_value_reg_armv5): Likewise.
16865 (*call_value_reg_arm): Likewise.
16866 (*call_symbol): Likewise.
16867 (*call_value_symbol): Likewise.
16868 (*sibcall_insn): Likewise.
16869 (*sibcall_value_insn): Likewise.
16870 (clzsi2): Likewise.
16871 (prefetch): Likewise.
16872 (define_split and define_peephole2 dependent on arm_arch5):
16874 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16876 (TARGET_ARM_QBIT): Likewise.
16877 (TARGET_DSP_MULTIPLY): Likewise.
16878 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16879 (arm_arch5, arm_arch5e): Delete.
16880 (arm_arch5t, arm_arch5te): Declare.
16881 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16882 (arm_arch5t): Declare.
16883 (arm_option_reconfigure_globals): Update for the above.
16884 (arm_options_perform_arch_sanity_checks): Update comment, replace
16885 use of arm_arch5 with arm_arch5t.
16886 (use_return_insn): Likewise.
16887 (arm_emit_call_insn): Likewise.
16888 (output_return_instruction): Likewise.
16889 (arm_final_prescan_insn): Likewise.
16890 (arm_coproc_builtin_available): Likewise.
16891 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16892 arm_arch5e with arm_arch5t and arm_arch5te.
16893 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16894 (arm_arch5t, arm_arch5te): Declare.
16895 * config/arm/arm-tables.opt: Regenerate.
16896 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16897 * config/arm/t-multilib: Likewise.
16898 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16899 instead of arm_arch5.
16900 (*call_reg_thumb1): Likewise.
16901 (*call_value_reg_thumb1_v5): Likewise.
16902 (*call_value_reg_thumb1): Likewise.
16903 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16905 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16907 2018-05-18 Martin Liska <mliska@suse.cz>
16909 PR gcov-profile/84846
16910 * doc/gcov.texi: Document -t option of gcov tool.
16912 2018-05-18 Martin Liska <mliska@suse.cz>
16914 PR gcov-profile/84846
16915 * gcov.c (print_usage): Add new -t option.
16916 (process_args): Handle the option.
16917 (generate_results): Use stdout as output when requested by
16920 2018-05-18 Martin Liska <mliska@suse.cz>
16922 PR gcov-profile/84846
16923 * coverage.c (coverage_init): Write PWD to .gcno file.
16924 * doc/gcov.texi: Document how working directory is printed.
16925 * gcov-dump.c (dump_gcov_file): Print PWD.
16926 * gcov.c (output_intermediate_file): Likewise.
16927 (read_graph_file): Read PWD string.
16928 (output_lines): Print PWD.
16930 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16932 PR middle-end/85817
16933 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16934 for retval and return false if all args to phi are zero.
16936 2018-05-18 Richard Biener <rguenther@suse.de>
16938 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16940 (evrp_dom_walker::before_dom_children): Call it.
16942 2018-05-18 Richard Biener <rguenther@suse.de>
16944 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16945 results when processing array refs with variable index.
16947 2018-05-18 Toon Moene <toon@moene.org>
16949 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16950 directly after that of -floop-interchange. Indicate that both
16951 options are enabled by default when specifying -O3.
16953 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16955 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16956 iterator. Delete separate integer-mode vec_set<mode> expander.
16957 (aarch64_simd_vec_setv2di): Delete.
16958 (vec_setv2di): Delete.
16959 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16960 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16961 the "w, r" alternative.
16963 2018-05-18 Martin Liska <mliska@suse.cz>
16965 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16966 * tree-pass.h (make_pass_lower_switch_O0): New function.
16967 * tree-switch-conversion.c (node_has_low_bound): Remove.
16968 (node_has_high_bound): Likewise.
16969 (node_is_bounded): Likewise.
16970 (class pass_lower_switch): Make it a template type and create
16972 (pass_lower_switch::execute): Add template argument.
16973 (make_pass_lower_switch): New function.
16974 (make_pass_lower_switch_O0): New function.
16975 (do_jump_if_equal): Remove.
16976 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16977 all the hard work to tree optimization passes.
16979 2018-05-18 Martin Liska <mliska@suse.cz>
16981 * dbgcnt.c (limit_low): Renamed from limit.
16982 (limit_high): New variable.
16983 (dbg_cnt_is_enabled): Check for upper limit.
16984 (dbg_cnt): Adjust dumping.
16985 (dbg_cnt_set_limit_by_index): Add new argument for high
16987 (dbg_cnt_set_limit_by_name): Likewise.
16988 (dbg_cnt_process_single_pair): Parse new format.
16989 (dbg_cnt_process_opt): Use strtok.
16990 (dbg_cnt_list_all_counters): Remove 'value' and add
16992 * doc/invoke.texi: Document changes.
16994 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16996 * doc/sourcebuild.texi (scalar_all_fma): Document.
16997 * tree.def (FMA_EXPR): Delete.
16998 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16999 * internal-fn.c (ternary_direct): New macro.
17000 (expand_ternary_optab_fn): Likewise.
17001 (direct_ternary_optab_supported_p): Likewise.
17002 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
17003 * builtins.c (fold_builtin_fma): Delete.
17004 (fold_builtin_3): Don't call it.
17005 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
17006 * expr.c (expand_expr_real_2): Likewise.
17007 * fold-const.c (operand_equal_p): Likewise.
17008 (fold_ternary_loc): Likewise.
17009 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
17010 * gimple.c (DEFTREECODE): Likewise.
17011 * gimplify.c (gimplify_expr): Likewise.
17012 * optabs-tree.c (optab_for_tree_code): Likewise.
17013 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17014 * tree-eh.c (operation_could_trap_p): Likewise.
17015 (stmt_could_throw_1_p): Likewise.
17016 * tree-inline.c (estimate_operator_cost): Likewise.
17017 * tree-pretty-print.c (dump_generic_node): Likewise.
17018 (op_code_prio): Likewise.
17019 * tree-ssa-loop-im.c (stmt_cost): Likewise.
17020 * tree-ssa-operands.c (get_expr_operands): Likewise.
17021 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
17022 * fold-const-call.h (fold_fma): Delete.
17023 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
17024 CFN_FNMA and CFN_FNMS.
17025 (fold_fma): Delete.
17026 * genmatch.c (combined_fn): New enum.
17027 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
17028 (commutative_op): New function.
17029 (commutate): Use it. Handle more than 2 operands.
17030 (dt_operand::gen_gimple_expr): Use commutative_op.
17031 (parser::parse_expr): Allow :c to be used with non-binary
17032 operators if the commutative operand is known.
17033 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
17034 CFN_FMS, CFN_FNMA and CFN_FNMS.
17035 (backprop::process_assign_use): Remove FMA_EXPR handling.
17036 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
17037 (gen_hsa_fma): New function.
17038 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
17039 IFN_FNMA and IFN_FNMS.
17040 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
17041 * gimple-fold.h (follow_all_ssa_edges): Declare.
17042 * gimple-fold.c (follow_all_ssa_edges): New function.
17043 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
17044 gimple_build interface and use follow_all_ssa_edges to fold the result.
17045 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
17046 instead of checking for optabs directly.
17047 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
17048 rather than FMA_EXPRs.
17049 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
17050 call to IFN_FMA instead of an FMA_EXPR.
17052 2018-05-17 Jim Wilson <jimw@sifive.com>
17054 * expr.c (do_tablejump): When converting index to Pmode, if we have a
17055 sign extended promoted subreg, and the range does not have the sign bit
17056 set, then do a sign extend.
17058 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
17059 test, check for sign extended subreg and/or constant operands, and
17060 do a sign extend in that case.
17062 2018-05-17 Steve Ellcey <sellcey@cavium.com>
17064 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
17065 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
17067 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
17068 Change logics_shift_reg to logics_shift_imm.
17069 (thunderx2t99_fp_loadpair_basic): Delete.
17070 (thunderx2t99_fp_storepair_basic): Delete.
17071 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
17072 (thunderx2t99_asimd_polynomial): Delete.
17073 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
17074 and neon_fp_mul_d_scalar_q.
17075 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
17076 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
17077 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
17078 (thunderx2t99_asimd_lut): Add missing tbl types.
17079 (thunderx2t99_asimd_ext): Delete.
17080 (thunderx2t99_asimd_load1_1_mult): Delete.
17081 (thunderx2t99_asimd_load1_2_mult): Delete.
17082 (thunderx2t99_asimd_load1_ldp): New.
17083 (thunderx2t99_asimd_load1): New.
17084 (thunderx2t99_asimd_load2): Add missing *load2* types.
17085 (thunderx2t99_asimd_load3): New.
17086 (thunderx2t99_asimd_load4): New.
17087 (thunderx2t99_asimd_store1_1_mult): Delete.
17088 (thunderx2t99_asimd_store1_2_mult): Delete.
17089 (thunderx2t99_asimd_store2_mult): Delete.
17090 (thunderx2t99_asimd_store2_onelane): Delete.
17091 (thunderx2t99_asimd_store_stp): New.
17092 (thunderx2t99_asimd_store1): New.
17093 (thunderx2t99_asimd_store2): New.
17094 (thunderx2t99_asimd_store3): New.
17095 (thunderx2t99_asimd_store4): New.
17097 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
17099 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
17100 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
17102 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
17103 Segher Boessenkool <segher@kernel.crashing.org>
17106 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
17109 2018-05-17 Richard Biener <rguenther@suse.de>
17111 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
17112 for pruning loop and prune defs feeding only already visited PHIs.
17114 2018-05-17 Richard Biener <rguenther@suse.de>
17116 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
17118 2018-05-17 Bin Cheng <bin.cheng@arm.com>
17119 Richard Biener <rguenther@suse.de>
17121 PR tree-optimization/85793
17122 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
17123 for VMAT_ELEMENTWISE.
17125 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17127 * internal-fn.h (lookup_internal_fn): Declare
17128 * internal-fn.c (lookup_internal_fn): New function.
17129 * gimple.c (gimple_build_call_from_tree): Handle calls to
17130 internal functions.
17131 * gimple-pretty-print.c (dump_gimple_call): Print "." before
17132 internal function names.
17133 * tree-pretty-print.c (dump_generic_node): Likewise.
17134 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
17136 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17138 * gimple-fold.h (gimple_build): Make the function forms take
17139 combined_fn rather than built_in_function.
17140 (gimple_simplify): Likewise.
17141 * gimple-match-head.c (gimple_simplify): Likewise.
17142 * gimple-fold.c (gimple_build): Likewise.
17143 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
17144 rather than gimple_build_call_internal.
17145 (get_initial_defs_for_reduction): Likewise.
17146 (vect_create_epilog_for_reduction): Likewise.
17147 (vectorizable_live_operation): Likewise.
17149 2018-05-17 Martin Liska <mliska@suse.cz>
17151 * gimple-ssa-sprintf.c (format_directive): Do not use
17152 space in between 'G_' and '('.
17154 2018-05-17 Jakub Jelinek <jakub@redhat.com>
17157 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
17158 even if the mask is not all ones.
17161 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
17163 (ix86_gimple_fold_builtin): Likewise.
17166 * config/i386/i386.c: Include tree-vector-builder.h.
17167 (ix86_vector_shift_count): New function.
17168 (ix86_fold_builtin): Fold shift builtins by scalar count.
17169 (ix86_gimple_fold_builtin): Likewise.
17171 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
17172 _mm512_setzero): New intrinsics.
17174 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
17175 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17177 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
17178 code generation for cases where splatting a value is not useful.
17179 * simplify-rtx.c (simplify_ternary_operation): Simplify
17180 vec_merge across a vec_duplicate and a paradoxical subreg forming
17181 a vector mode to a vec_concat.
17183 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
17185 * config.gcc: Support "goldmont-plus".
17186 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17188 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17189 PROCESSOR_GOLDMONT_PLUS.
17190 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
17191 (processor_target_table): Add "goldmont-plus".
17192 (PTA_GOLDMONT_PLUS): Define.
17193 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
17194 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
17195 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
17196 (fold_builtin_cpu): Add "goldmont-plus".
17197 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
17198 (ix86_option_override_internal): Add "goldmont-plus".
17199 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
17200 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
17201 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
17202 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
17204 2018-05-17 Richard Biener <rguenther@suse.de>
17206 PR tree-optimization/85757
17207 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
17208 remove defs that only feed that PHI from further processing.
17210 2018-05-16 Jim Wilson <jimw@sifive.com>
17212 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
17214 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
17216 2018-05-16 Mark Wielaard <mark@klomp.org>
17218 * dwarf2out.c (count_index_strings): New function.
17219 (output_indirect_strings): Call count_index_strings and generate
17220 header for dwarf_version >= 5.
17222 2018-05-16 Mark Wielaard <mark@klomp.org>
17224 * dwarf2out.c (dwarf_FORM): New function.
17225 (set_indirect_string): Use dwarf_FORM.
17226 (reset_indirect_string): Likewise.
17227 (size_of_die): Likewise.
17228 (value_format): Likewise.
17229 (output_die): Likewise.
17230 (add_skeleton_AT_string): Likewise.
17231 (output_macinfo_op): Likewise.
17232 (index_string): Likewise.
17233 (output_index_string_offset): Likewise.
17234 (output_index_string): Likewise.
17235 (count_index_strings): Likewise.
17237 2018-05-16 Carl Love <cel@us.ibm.com>
17239 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17240 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
17242 2018-05-16 Martin Jambor <mjambor@suse.cz>
17244 * ipa-prop.c (ipa_free_all_edge_args): Remove.
17245 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
17247 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
17249 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
17250 (fnma<mode>4): Likewise.
17251 (fms<mode>4): Likewise.
17252 (fnms<mode>4): Likewise.
17253 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
17254 (aarch64_fnma<mode>4): Likewise.
17255 (aarch64_fms<mode>4): Likewise.
17256 (aarch64_fnms<mode>4): Likewise.
17257 (aarch64_fnmadd<mode>4): Likewise.
17259 2018-05-16 Jason Merrill <jason@redhat.com>
17261 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
17263 2018-05-16 Richard Biener <rguenther@suse.de>
17265 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
17266 (dump_stmt_cost): Declare.
17267 (add_stmt_cost): Dump cost we add.
17268 (add_stmt_costs): New function.
17269 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
17270 No longer exported.
17271 (vect_analyze_stmt): Adjust prototype.
17272 (vectorizable_condition): Likewise.
17273 (vectorizable_live_operation): Likewise.
17274 (vectorizable_reduction): Likewise.
17275 (vectorizable_induction): Likewise.
17276 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
17277 cost vector to pass to vectorizable_ and record afterwards.
17278 (vect_model_reduction_cost): Take cost vector argument and adjust.
17279 (vect_model_induction_cost): Likewise.
17280 (vectorizable_reduction): Likewise.
17281 (vectorizable_induction): Likewise.
17282 (vectorizable_live_operation): Likewise.
17283 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
17284 SLP_TREE_NUMBER_OF_VEC_STMTS.
17285 (vect_analyze_slp_cost_1): Remove.
17286 (vect_analyze_slp_cost): Likewise.
17287 (vect_slp_analyze_node_operations): Take visited args and
17288 a target cost vector. Avoid processing already visited stmt sets.
17289 (vect_slp_analyze_operations): Use a local cost vector to gather
17290 costs and register those of non-discarded instances.
17291 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
17292 (vect_schedule_slp_instance): Remove copying of
17293 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
17295 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
17296 adding cost. Record cost entry location.
17297 (vect_prologue_cost_for_slp_op): Function to compute cost of
17298 a constant or invariant generated for SLP vect in the prologue,
17299 split out from vect_analyze_slp_cost_1.
17300 (vect_model_simple_cost): Make static. Adjust for SLP costing.
17301 (vect_model_promotion_demotion_cost): Likewise.
17302 (vect_model_store_cost): Likewise, make static.
17303 (vect_model_load_cost): Likewise.
17304 (vectorizable_bswap): Add cost vector arg and adjust.
17305 (vectorizable_call): Likewise.
17306 (vectorizable_simd_clone_call): Likewise.
17307 (vectorizable_conversion): Likewise.
17308 (vectorizable_assignment): Likewise.
17309 (vectorizable_shift): Likewise.
17310 (vectorizable_operation): Likewise.
17311 (vectorizable_store): Likewise.
17312 (vectorizable_load): Likewise.
17313 (vectorizable_condition): Likewise.
17314 (vectorizable_comparison): Likewise.
17315 (can_vectorize_live_stmts): Likewise.
17316 (vect_analyze_stmt): Likewise.
17317 (vect_transform_stmt): Adjust calls to vectorizable_*.
17318 * tree-vectorizer.c: Include gimple-pretty-print.h.
17319 (dump_stmt_cost): New function.
17321 2018-05-16 Richard Biener <rguenther@suse.de>
17323 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
17324 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
17325 * tree-ssa-dse.c: Include tree-ssa-loop.h.
17326 (check_name): New callback.
17327 (dse_classify_store): Track cycles via a visited bitmap of PHI
17328 defs and simplify handling of in-loop and across loop dead stores
17329 and properly fail for loop-variant refs. Handle byte-tracking with
17330 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
17333 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
17335 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
17336 (vect_get_mask_type_for_stmt): Likewise.
17337 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
17339 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
17340 to determine the statement's vector type and the vector type that
17341 should be used for calculating nunits. Deal with cases in which
17342 the type has to be deferred.
17343 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
17344 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
17345 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
17346 (vect_determine_vf_for_stmt): New functions, split out from...
17347 (vect_determine_vectorization_factor): ...here.
17348 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
17349 (vect_get_mask_type_for_stmt): New functions, split out from
17350 vect_determine_vectorization_factor.
17352 2018-05-16 Richard Biener <rguenther@suse.de>
17354 * tree-cfg.c (verify_gimple_assign_ternary): Properly
17355 verify the [VEC_]COND_EXPR embedded comparison.
17357 2018-05-15 Martin Sebor <msebor@redhat.com>
17359 PR tree-optimization/85753
17360 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
17361 RECORD_TYPE in addition to ARRAY_TYPE.
17363 2018-05-15 Martin Sebor <msebor@redhat.com>
17365 PR middle-end/85643
17366 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
17368 2018-05-15 Richard Biener <rguenther@suse.de>
17370 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
17371 add by_clobber_p one. Change algorithm to collect all defs
17372 representing uses we need to walk and try reducing them to
17373 a single one before failing.
17374 (dse_dom_walker::dse_optimize_stmt): Adjust.
17376 2018-05-13 Mark Wielaard <mark@klomp.org>
17378 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
17379 (size_of_loc_descr): Likewise.
17380 (output_loc_operands): Likewise.
17381 (output_loc_operands_raw): Likewise.
17382 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
17383 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
17384 (hash_loc_operands): Likewise.
17385 (compare_loc_operands): Likewise.
17387 2018-05-14 Mark Wielaard <mark@klomp.org>
17389 * dwarf2out.c (count_index_addrs): New function.
17390 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
17392 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17394 PR tree-optimization/83648
17395 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
17396 return value as malloc candidate.
17398 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17401 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
17402 param as true in call to suggest_attribute.
17404 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
17406 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
17407 -mreadonly-in-sdata.
17409 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17411 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
17413 (aarch64_crypto_aesd_fused): Likewise.
17415 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
17417 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
17418 (movsi_aarch64): Likewise.
17419 (load_pairsi): Likewise.
17420 (load_pairdi): Likewise.
17421 (store_pairsi): Likewise.
17422 (store_pairdi): Likewise.
17423 (load_pairsf): Likewise.
17424 (load_pairdf): Likewise.
17425 (store_pairsf): Likewise.
17426 (store_pairdf): Likewise.
17427 (zero_extend): Likewise.
17428 (trunc): Swap alternatives.
17429 (fcvt_target): Add '?' to prefer w over r.
17431 2018-05-14 Jakub Jelinek <jakub@redhat.com>
17434 * config/i386/i386.md: Disallow non-commutative arithmetics in
17435 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
17436 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
17437 in the peephole2 before it.
17439 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
17441 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
17442 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
17443 (ix86_handle_option): Handle -mcldemote.
17444 * config.gcc: New header.
17445 * config/i386/cldemoteintrin.h: New file.
17446 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
17447 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17449 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17450 OPTION_MASK_ISA_CLDEMOTE.
17451 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
17452 (ix86_valid_target_attribute_inner_p): Ditto.
17453 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
17454 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
17455 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
17456 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
17457 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
17459 * config/i386/i386.opt: Add -mcldemote.
17460 * config/i386/x86intrin.h: New header.
17461 * doc/invoke.texi: Add -mcldemote.
17463 2018-05-14 Richard Biener <rguenther@suse.de>
17465 * doc/match-and-simplify.texi: Adjust :s documentation.
17467 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
17469 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
17470 intended memcpy size.
17471 (REORDER_45): Likewise.
17473 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
17475 * sort.cc: New file.
17476 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
17477 * vec.c (qsort_chk): Use gcc_qsort.
17478 * Makefile.in (OBJS-libcommon): Add sort.o.
17479 (build/sort.o): New target. Use it...
17480 (BUILD_RTL): ... here, and...
17481 (build/gencfn-macros): ... here, and...
17482 (build/genmatch): ... here.
17484 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17485 Chung-Ju Wu <jasonwucj@gmail.com>
17487 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
17488 * config/nds32/nds32-graywolf.md: New file.
17489 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
17490 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
17492 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
17493 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
17494 * config/nds32/nds32.md (pipeline_model): Add graywolf.
17495 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
17496 * config/nds32/pipelines.md: Include n15 settings.
17498 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17499 Chung-Ju Wu <jasonwucj@gmail.com>
17501 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
17502 * config/nds32/nds32-n13.md: New file.
17503 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
17504 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
17506 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
17507 * config/nds32/nds32.md (pipeline_model): Add n13.
17508 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
17509 * config/nds32/pipelines.md: Include n13 settings.
17511 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17512 Chung-Ju Wu <jasonwucj@gmail.com>
17514 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
17515 * config/nds32/nds32-n10.md: New file.
17516 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
17517 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
17519 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
17520 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
17521 * config/nds32/nds32.md (pipeline_model): Add n10.
17522 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
17523 * config/nds32/pipelines.md: Include n10 settings.
17525 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
17526 Kito Cheng <kito.cheng@gmail.com>
17527 Chung-Ju Wu <jasonwucj@gmail.com>
17529 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
17530 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17531 Add enum values for DSP extension instructions.
17532 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
17534 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
17535 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
17536 New code iterators.
17537 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
17538 * config/nds32/nds32-dspext.md: New file for DSP implementation.
17539 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
17540 * config/nds32/nds32-intrinsic.md: Likewise.
17541 * config/nds32/nds32_intrinsic.h: Likewise.
17542 * config/nds32/nds32-md-auxiliary.c: Likewise.
17543 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
17544 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
17545 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
17546 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
17547 * config/nds32/nds32-protos.h: New declarations for DSP extension.
17548 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
17549 TYPE_DMAC in switch statement.
17550 * config/nds32/nds32.c: New checking and implementation for DSP
17551 extension instructions.
17552 * config/nds32/nds32.h: Likewise.
17553 * config/nds32/nds32.md: Likewise.
17554 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17555 * config/nds32/predicates.md: Implement new predicates for DSP
17558 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17560 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17561 Reformat alternatives and attributes so it is easier to identify
17562 which constraints/attributes go with which instruction.
17563 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17564 (mov<mode>_softfloat32, FMOVE64): Likewise.
17565 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17566 (mov<mode>_softfloat64, FMOVE64): Likewise.
17568 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17570 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17572 (Basic PowerPC Built-in Functions): The new name of the
17573 subsection previously known as "PowerPC Built-in Functions".
17574 (Basic PowerPC Built-in Functions Available on all Configurations):
17576 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17577 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17578 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17579 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17581 2018-05-11 Martin Jambor <mjambor@suse.cz>
17584 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17587 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17590 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17592 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17594 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17595 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17596 (ix86_handle_option): Handle -mwaitpkg.
17597 * config.gcc: New header.
17598 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17599 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17600 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17602 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17603 OPTION_MASK_ISA_WAITPKG.
17604 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17605 (ix86_option_override_internal): Add PTA_WAITPKG.
17606 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17607 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17608 IX86_BUILTIN_TPAUSE.
17609 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17610 __builtin_ia32_umwait and __builtin_ia32_tpause.
17611 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17612 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17613 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17614 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17615 UNSPECV_TPAUSE): New.
17616 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17617 * config/i386/i386.opt: Add -mwaitpkg.
17618 * config/i386/waitpkgintrin.h: New file.
17619 * config/i386/x86intrin.h: New header.
17620 * doc/invoke.texi: Add -mwaitpkg.
17622 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17625 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17627 (cortex-m0): Use armv6s-m isa.
17628 (cortex-m0plus): Likewise.
17629 (cortex-m1): Likewise.
17630 (cortex-m0.small-multiply): Likewise.
17631 (cortex-m0plus.small-multiply): Likewise.
17632 (cortex-m1.small-multiply): Likewise.
17634 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17635 Jakub Jelinek <jakub@redhat.com>
17637 PR tree-optimization/85692
17638 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17639 source permute as well.
17641 2018-05-11 Martin Liska <mliska@suse.cz>
17644 * doc/extend.texi: Document LLVM style format for no_sanitize
17647 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17649 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17650 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17651 (mode_supports_vsx_dform_quad): Likewise.
17652 (mode_supports_vmx_dform): Move these functions to be next to the
17653 other mode_supports functions.
17654 (mode_supports_dq_form): Likewise.
17655 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17656 mode_supports_dq_form.
17657 (reg_offset_addressing_ok_p): Likewise.
17658 (offsettable_ok_by_alignment): Likewise.
17659 (rs6000_legitimate_offset_address_p): Likewise.
17660 (legitimate_lo_sum_address_p): Likewise.
17661 (rs6000_legitimize_address): Likewise.
17662 (rs6000_legitimize_reload_address): Likewise.
17663 (rs6000_secondary_reload_inner): Likewise.
17664 (rs6000_preferred_reload_class): Likewise.
17665 (rs6000_output_move_128bit): Likewise.
17667 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17669 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17670 Generate SImode target register for null target.
17671 <case IX86_BUILTIN_XGETBV>: Ditto.
17672 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17673 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17675 2018-05-10 Carl Love <cel@us.ibm.com>
17677 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17678 dcbtt and dcbtstt if operands[2] is 0.
17680 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17683 * config/i386/sse.md (usadv64qi): New expander.
17685 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17687 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17688 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17689 -maltivec=be support.
17690 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17691 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17692 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17693 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17694 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17695 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17696 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17697 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17698 altivec_vsumsws): Adjust.
17699 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17700 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17701 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17703 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17704 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17705 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17706 (altivec_lve<VI_char>x): Delete expand.
17707 (*altivec_lve<VI_char>x_internal): Rename to...
17708 (altivec_lve<VI_char>x): ... this.
17709 (altivec_lvxl_<mode>): Delete expand.
17710 (*altivec_lvxl_<mode>_internal): Rename to ...
17711 (altivec_lvxl_<mode>): ... this.
17712 (altivec_stvxl_<mode>): Delete expand.
17713 (*altivec_stvxl_<mode>_internal): Rename to ...
17714 (altivec_stvxl_<mode>): ... this.
17715 (altivec_stve<VI_char>x): Delete expand.
17716 (*altivec_stve<VI_char>x_internal): Rename to ...
17717 (altivec_stve<VI_char>x): ... this.
17718 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17719 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17720 reduc_plus_scal_<mode>): Adjust.
17721 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17723 (rs6000_cpu_cpp_builtins): Adjust.
17724 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17725 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17726 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17728 -maltivec=be support.
17729 (rs6000_split_vec_extract_var): Adjust.
17730 (rs6000_split_v4si_init): Adjust.
17731 (swap_selector_for_mode): Delete.
17732 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17733 altivec_expand_stvex_be): Delete.
17734 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17735 -maltivec=be support.
17736 (rs6000_gimple_fold_builtin): Ditto.
17737 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17739 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17740 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17741 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17742 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17743 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17744 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17745 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17746 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17747 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17748 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17749 anonymous split): Adjust.
17750 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17751 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17753 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17755 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17756 when --with-gxx-include-dir is also specified.
17757 * configure: Regenerate.
17759 2018-05-09 Jim Wilson <jimw@sifive.com>
17762 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17763 * config/riscv/t-withmultilib: New.
17764 * config/riscv/withmultilib.h: New.
17765 * doc/install.texi: Document RISC-V --with-multilib-list support.
17767 2018-05-09 Richard Biener <rguenther@suse.de>
17769 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17771 (vect_bb_vectorization_profitable_p): Adjust. Compute
17772 actual scalar cost using the cost vector and the add_stmt_cost
17775 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17777 PR rtl-optimization/85645
17778 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17779 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17781 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17783 PR rtl-optimization/85645
17784 * shrink-wrap.c (spread_components): Return a boolean saying if
17785 anything was changed.
17786 (try_shrink_wrapping_separate): Iterate spread_components until
17787 nothing changes anymore.
17789 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17791 PR rtl-optimization/85645
17792 * regrename.c (build_def_use): Also kill the chains that include the
17793 destination of a REG_CFA_REGISTER note.
17795 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17797 PR rtl-optimization/85645
17798 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17799 insn that has a REG_CFA_REGISTER note.
17801 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17803 * cfgexpand.c (expand_clobber): New function.
17804 (expand_gimple_stmt_1): Use it.
17805 * tree-vect-stmts.c (vect_clobber_variable): New function,
17807 (vectorizable_simd_clone_call): ...here.
17808 (vectorizable_store): Emit a clobber either side of an
17809 IFN_STORE_LANES sequence.
17810 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17812 2018-05-09 Tom de Vries <tom@codesourcery.com>
17815 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17816 (define_insn "trap_if_false"): Add exit after trap.
17818 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17820 PR rtl-optimization/85638
17821 * bb-reorder.c: Include common/common-target.h.
17822 (create_forwarder_block): New function extracted from...
17823 (fix_up_crossing_landing_pad): ...here. Rename into...
17824 (dw2_fix_up_crossing_landing_pad): ...this.
17825 (sjlj_fix_up_crossing_landing_pad): New function.
17826 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17827 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17828 from both partitions and exit the loop after one iteration.
17830 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17833 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17835 (Basic PowerPC Built-in Functions): The new name of the
17836 subsection previously known as "PowerPC Built-in Functions".
17837 (Basic PowerPC Built-in Functions Available on all Configurations):
17839 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17841 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17843 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17845 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17848 2018-05-08 Jim Wilson <jimw@sifive.com>
17850 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17851 (LD_EMUL_SUFFIX): New.
17852 (LINK_SPEC): Use it.
17854 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17856 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17858 (Basic PowerPC Built-in Functions): The new name of the
17859 subsection previously known as "PowerPC Built-in Functions".
17860 (Basic PowerPC Built-in Functions Available on all Configurations):
17862 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17864 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17866 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17868 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17871 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17874 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17875 after cmpelim optimization.
17877 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17879 * config.gcc: Support "goldmont".
17880 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17881 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17882 PROCESSOR_GOLDMONT.
17883 * config/i386/i386.c (m_GOLDMONT): Define.
17884 (processor_target_table): Add "goldmont".
17885 (PTA_GOLDMONT): Define.
17886 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17887 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17888 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17889 (fold_builtin_cpu): Add "goldmont".
17890 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17891 (ix86_option_override_internal): Add "goldmont".
17892 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17893 (processor_type): Add PROCESSOR_GOLDMONT.
17894 * config/i386/i386.md: Add CPU "glm".
17895 * config/i386/glm.md: New file.
17896 * config/i386/x86-tune.def: Add m_GOLDMONT.
17897 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17899 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17902 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17904 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17905 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17906 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17909 * config/i386/i386.c (ix86_fold_builtin): Handle
17910 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17913 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17914 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17916 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17919 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17920 (check_arch): Likewise.
17921 (check_fpu): Return the result rather than printing it.
17922 (end arch): Fix operator precedence.
17923 (end cpu): Likewise.
17924 (END): Print the result from check_fpu.
17926 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17927 Alan Hayward <alan.hayward@arm.com>
17928 David Sherwood <david.sherwood@arm.com>
17930 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17931 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17932 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17933 (*fcmuo<mode>_and): New patterns.
17935 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17937 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17938 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17939 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17940 (cmp_op, sve_imm_con): New code attributes.
17941 (SVE_COND_INT_CMP, imm_con): Delete.
17942 (cmp_op): Remove above unspecs from int attribute.
17943 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17945 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17946 comparison-specific unspecs.
17947 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17948 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17949 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17950 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17951 (*vec_fcm<cmp_op><mode>): Rename to...
17952 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17953 (*vec_fcmuo<mode>): Rename to...
17954 (*fcmuo<mode>): ...this and adjust likewise.
17955 (*pred_fcm<cmp_op><mode>): New pattern.
17956 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17957 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17959 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17961 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17962 (aarch64_emit_sve_predicated_cond): New function.
17963 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17964 (aarch64_emit_unspec_cond_or): Replace with...
17965 (aarch64_emit_sve_or_conds): ...this new function. Use
17966 aarch64_emit_sve_ptrue_op for the individual comparisons and
17967 aarch64_emit_binop to OR them together.
17968 (aarch64_emit_inverted_unspec_cond): Replace with...
17969 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17970 aarch64_emit_sve_ptrue_op for the comparison and
17971 aarch64_emit_unop to invert the result.
17972 (aarch64_expand_sve_vec_cmp_float): Update after the above
17973 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17975 2018-05-07 Nathan Sidwell <nathan@acm.org>
17977 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17978 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17979 (Backwards Compatibility): Likewise.
17981 2018-05-07 Luis Machado <luis.machado@linaro.org>
17985 2018-05-07 Luis Machado <luis.machado@linaro.org>
17987 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17988 <prefetch_dynamic_strides>: New const bool field.
17989 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17990 prefetch_dynamic_strides.
17991 (exynosm1_prefetch_tune): Likewise.
17992 (thunderxt88_prefetch_tune): Likewise.
17993 (thunderx_prefetch_tune): Likewise.
17994 (thunderx2t99_prefetch_tune): Likewise.
17995 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17997 (aarch64_override_options_internal): Update to set
17998 PARAM_PREFETCH_DYNAMIC_STRIDES.
17999 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18000 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18001 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18002 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18003 prefetch-dynamic-strides setting.
18005 2018-05-07 Luis Machado <luis.machado@linaro.org>
18007 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18008 <minimum_stride>: New const int field.
18009 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18010 minimum_stride field.
18011 (exynosm1_prefetch_tune): Likewise.
18012 (thunderxt88_prefetch_tune): Likewise.
18013 (thunderx_prefetch_tune): Likewise.
18014 (thunderx2t99_prefetch_tune): Likewise.
18015 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
18016 (aarch64_override_options_internal): Update to set
18017 PARAM_PREFETCH_MINIMUM_STRIDE.
18018 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18019 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18020 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18021 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18022 stride is constant and is below the minimum stride threshold.
18024 2018-05-07 Luis Machado <luis.machado@linaro.org>
18026 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
18029 2018-05-07 Luis Machado <luis.machado@linaro.org>
18031 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18032 <prefetch_dynamic_strides>: New const bool field.
18033 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18034 prefetch_dynamic_strides.
18035 (exynosm1_prefetch_tune): Likewise.
18036 (thunderxt88_prefetch_tune): Likewise.
18037 (thunderx_prefetch_tune): Likewise.
18038 (thunderx2t99_prefetch_tune): Likewise.
18039 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
18041 (aarch64_override_options_internal): Update to set
18042 PARAM_PREFETCH_DYNAMIC_STRIDES.
18043 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18044 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18045 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18046 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18047 prefetch-dynamic-strides setting.
18049 2018-05-07 Luis Machado <luis.machado@linaro.org>
18051 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18052 <minimum_stride>: New const int field.
18053 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18054 minimum_stride field.
18055 (exynosm1_prefetch_tune): Likewise.
18056 (thunderxt88_prefetch_tune): Likewise.
18057 (thunderx_prefetch_tune): Likewise.
18058 (thunderx2t99_prefetch_tune): Likewise.
18059 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
18060 (aarch64_override_options_internal): Update to set
18061 PARAM_PREFETCH_MINIMUM_STRIDE.
18062 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18063 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18064 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18065 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18066 stride is constant and is below the minimum stride threshold.
18068 2018-05-06 Jakub Jelinek <jakub@redhat.com>
18071 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
18072 the type is addressable. Don't force op into register if it has
18075 2018-05-05 Roland McGrath <mcgrathr@google.com>
18078 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
18079 any section for which we don't know a specific type it should have,
18080 regardless of name. Previously this was done only for the exact
18081 names ".init_array", ".fini_array", and ".preinit_array".
18082 (default_elf_asm_named_section): Add comment about
18083 relationship with default_section_type_flags and SECTION_NOTYPE.
18084 (get_section): Don't consider it a type conflict if one side has
18085 SECTION_NOTYPE and the other doesn't, as long as neither has the
18086 SECTION_BSS et al used in the default_section_type_flags logic.
18088 2018-05-05 Tom de Vries <tom@codesourcery.com>
18091 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
18092 (workaround_barsyncs): New function.
18093 (nvptx_reorg): Use workaround_barsyncs.
18094 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
18095 (define_expand "nvptx_membar_cta"): New define_expand.
18096 (define_insn "*nvptx_membar_cta"): New insn.
18098 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
18100 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
18101 To improve optimization opportunities.
18102 * builtin-types.def: The new needed builtin types for the above.
18104 2018-05-04 Richard Biener <rguenther@suse.de>
18106 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
18107 * gimple-ssa-store-merging.c
18108 (imm_store_chain_info::output_merged_store): Remove redundant create,
18109 release split_store vector contents on failure.
18110 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
18111 scalar stmt vector on cache hit.
18113 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
18115 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
18117 * config.gcc (powerpc-xilinx-eabi*): Remove.
18118 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
18120 (fusion_addis_mem_combo_load): Ditto.
18121 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
18123 (rs6000_cpu_cpp_builtins): Ditto.
18124 * config/rs6000/rs6000-linux.c
18125 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
18126 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
18127 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
18129 (rs6000_setup_reg_addr_masks): Ditto.
18130 (rs6000_init_hard_regno_mode_ok): Ditto.
18131 (rs6000_option_override_internal): Ditto.
18132 (legitimate_lo_sum_address_p): Ditto.
18133 (rs6000_legitimize_address): Ditto.
18134 (rs6000_legitimize_reload_address): Ditto.
18135 (rs6000_legitimate_address_p): Ditto.
18136 (abi_v4_pass_in_fpr): Ditto.
18137 (setup_incoming_varargs): Ditto.
18138 (rs6000_gimplify_va_arg): Ditto.
18139 (rs6000_split_multireg_move): Ditto.
18140 (rs6000_savres_strategy): Ditto.
18141 (rs6000_emit_prologue_components): Ditto.
18142 (rs6000_emit_epilogue_components): Ditto.
18143 (rs6000_emit_prologue): Ditto.
18144 (rs6000_emit_epilogue): Ditto.
18145 (rs6000_elf_file_end): Ditto.
18146 (rs6000_function_value): Ditto.
18147 (rs6000_libcall_value): Ditto.
18148 * config/rs6000/rs6000.h: Ditto.
18149 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
18150 (TARGET_MINMAX): ... this. New.
18151 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
18152 * config/rs6000/rs6000.md: Remove Xilinx FP support.
18153 (*movsi_internal1_single): Delete.
18154 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
18155 mfpu=, mxilinx-fpu): Delete.
18156 * config/rs6000/singlefp.h: Delete.
18157 * config/rs6000/sysv4.h: Remove Xilinx FP support.
18158 * config/rs6000/t-rs6000: Ditto.
18159 * config/rs6000/t-xilinx: Delete.
18160 * config/rs6000/titan.md: Adjust for fp_type removal.
18161 * config/rs6000/vsx.md: Remove Xilinx FP support.
18162 (VStype_simple): Delete.
18163 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
18164 * config/rs6000/xfpu.h: Delete.
18165 * config/rs6000/xfpu.md: Delete.
18166 * config/rs6000/xilinx.h: Delete.
18167 * config/rs6000/xilinx.opt: Delete.
18168 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
18169 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
18171 2018-05-04 Tom de Vries <tom@codesourcery.com>
18174 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
18177 2018-05-04 Richard Biener <rguenther@suse.de>
18179 PR middle-end/85627
18180 * tree-complex.c (update_complex_assignment): We are always in SSA form.
18181 (expand_complex_div_wide): Likewise.
18182 (expand_complex_operations_1): Likewise.
18183 (expand_complex_libcall): Preserve EH info of the original stmt.
18184 (tree_lower_complex): Handle removed blocks.
18185 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
18186 on complex multiplication and division libcall builtins.
18188 2018-05-04 Richard Biener <rguenther@suse.de>
18190 PR middle-end/85574
18191 * fold-const.c (negate_expr_p): Restrict negation of operand
18192 zero of a division to when we know that can happen without
18194 (fold_negate_expr_1): Likewise.
18196 2018-05-04 Jakub Jelinek <jakub@redhat.com>
18199 * real.h (real_nextafter): Declare.
18200 * real.c (real_nextafter): New function.
18201 * fold-const-call.c (fold_const_nextafter): New function.
18202 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
18203 CASE_CFN_NEXTTOWARD.
18204 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
18205 even when arg1_mode is different from arg0_mode.
18207 2018-05-03 Nathan Sidwell <nathan@acm.org>
18209 * doc/extend.texi (Deprecated Features): Remove
18210 -ffriend-injection.
18211 (Backwards Compatibility): Likewise.
18212 * doc/invoke.texi (C++ Language Options): Likewise.
18213 (C++ Dialect Options): Likewise.
18215 2018-05-03 Jakub Jelinek <jakub@redhat.com>
18218 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
18219 _mm512_mask_mullox_epi64): New intrinsics.
18221 2018-05-03 Tom de Vries <tom@codesourcery.com>
18224 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18225 dump files): Add offload-tree.
18227 2018-05-03 Richard Biener <rguenther@suse.de>
18229 PR tree-optimization/85615
18230 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
18231 to loops not nested in BBs loop father to avoid creating multi-entry
18234 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18236 PR tree-optimization/70291
18237 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
18238 arguments. Change return type to tree. Emit libcall as a new
18239 statement rather than replacing existing one when inplace_p is true.
18240 (expand_complex_multiplication_components): New function.
18241 (expand_complex_multiplication): Expand floating-point complex
18242 multiplication using the above.
18243 (expand_complex_division): Rename inner_type parameter to type.
18244 Update expand_complex_libcall call-site.
18245 (expand_complex_operations_1): Update expand_complex_multiplication
18246 and expand_complex_division call-sites.
18248 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18251 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18252 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18253 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
18254 the highest significant bit of the shift count mask is clear. In
18255 check whether and[sq]i3 is needed verify that all significant bits
18256 of the shift count other than the highest are set.
18258 2018-05-02 Tom de Vries <tom@codesourcery.com>
18261 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
18262 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
18263 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
18264 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
18265 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
18266 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
18267 __builtin_goacc_parlevel_size.
18269 2018-05-02 Richard Biener <rguenther@suse.de>
18271 PR tree-optimization/85597
18272 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
18273 do not use split vect_get_vec_defs call but call vect_get_slp_defs
18276 2018-05-02 Tom de Vries <tom@codesourcery.com>
18279 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18280 dump files): Add ltrans-tree.
18282 2018-05-02 Tom de Vries <tom@codesourcery.com>
18285 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18286 dump files): Add wpa-ipa.
18288 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
18290 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
18291 powerpc*-*-linux*paired* target.
18292 * config/rs6000/750cl.h: Delete.
18293 * config/rs6000/paired.h: Delete.
18294 * config/rs6000/paired.md: Delete.
18295 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
18297 * config/rs6000/rs6000-builtin.def: Remove paired float support.
18298 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18299 comment. Remove paired float support.
18300 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
18301 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
18303 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
18304 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
18306 * config/rs6000/rs6000.c: Remove paired float support.
18307 (paired_expand_vector_init, paired_expand_vector_move,
18308 paired_emit_vector_compare, paired_emit_vector_cond_expr,
18309 (paired_expand_lv_builtin, paired_expand_stv_builtin,
18310 paired_expand_builtin, paired_expand_predicate_builtin,
18311 paired_init_builtins): Delete.
18312 * config/rs6000/rs6000.h: Remove paired float support.
18313 * config/rs6000/rs6000.md: Remove paired float support.
18314 (move_from_CR_ov_bit): Delete.
18315 * config/rs6000/rs6000.opt (mpaired): Delete.
18316 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
18317 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
18319 2018-05-02 Richard Biener <rguenther@suse.de>
18321 PR middle-end/85567
18322 * gimplify.c (gimplify_save_expr): When in SSA form allow
18323 SAVE_EXPRs to compute to SSA vars.
18325 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18328 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18329 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18330 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
18331 clobber operands[2], instead use a new pseudo. Formatting fixes.
18333 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
18335 PR tree-optimization/85586
18336 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
18337 exit early for statements in the same group if the accesses are
18340 2018-05-02 Tom de Vries <tom@codesourcery.com>
18343 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
18346 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
18348 PR tree-optimization/85143
18349 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
18351 2018-05-01 Tom de Vries <tom@codesourcery.com>
18354 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
18355 not found" error message.
18357 2018-05-01 Tom de Vries <tom@codesourcery.com>
18360 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
18361 * vec.c (test_ordered_remove_if): New function.
18362 (vec_c_tests): Call test_ordered_remove_if.
18363 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
18364 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
18365 * tree-vect-patterns.c (vect_pattern_recog_1): Use
18366 VEC_ORDERED_REMOVE_IF.
18368 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18370 PR tree-optimization/82665
18371 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
18372 pointer subtraction where arguments come from a memchr call.
18374 2018-05-01 Jakub Jelinek <jakub@redhat.com>
18376 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
18377 --push-state --as-needed and --pop-state instead of --as-needed and
18378 --no-as-needed if ld supports it.
18379 * configure: Regenerated.
18382 * doc/install.texi2html: Replace _002d with - and _002a with * in
18383 generated html files using sed.
18385 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18388 * gcc-rich-location.c (blank_line_before_p): New function.
18389 (use_new_line): New function.
18390 (gcc_rich_location::add_fixit_insert_formatted): New function.
18391 * gcc-rich-location.h
18392 (gcc_rich_location::add_fixit_insert_formatted): New function.
18394 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18396 * selftest.c (assert_streq): Rename "expected" and "actual" to
18397 "val1" and "val2". Extend NULL-handling to cover both inputs
18398 symmetrically, while still requiring both to be non-NULL for a pass.
18399 * selftest.h (assert_streq): Rename "expected" and "actual" to
18401 (ASSERT_EQ): Likewise.
18402 (ASSERT_EQ_AT): Likewise.
18403 (ASSERT_KNOWN_EQ): Likewise.
18404 (ASSERT_KNOWN_EQ_AT): Likewise.
18405 (ASSERT_NE): Likewise.
18406 (ASSERT_MAYBE_NE): Likewise.
18407 (ASSERT_MAYBE_NE_AT): Likewise.
18408 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
18409 the assertion to pass.
18410 (ASSERT_STREQ_AT): Likewise.
18412 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18414 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
18415 interaction with -pie.
18417 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18419 * selftest.h: Fix alphabetization of per-source-file selftest
18422 2018-04-30 Jason Merrill <jason@redhat.com>
18424 PR c++/61982 - dead stores to destroyed objects.
18425 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
18428 2018-04-30 Jason Merrill <jason@redhat.com>
18430 * tree.c (build_clobber): New.
18431 * tree.h: Declare it.
18432 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
18434 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18436 * diagnostic-show-locus.c (layout::layout): Update for
18437 location_get_source_line returning a char_span.
18438 (struct char_span): Move to input.h.
18439 (struct correction): Update for fields in char_span becoming
18441 (struct source_line): Update for location_get_source_line
18442 returning a char_span.
18443 (layout::print_line): Likewise.
18444 * edit-context.c (edited_file::print_content): Likewise.
18445 (edited_file::print_diff_hunk): Likewise.
18446 (edited_file::print_run_of_changed_lines): Likewise.
18447 (edited_file::get_num_lines): Likewise.
18448 (edited_line::edited_line): Likewise.
18449 * final.c (asm_show_source): Likewise.
18450 * input.c (location_get_source_line): Convert return type
18451 from const char * to char_span, losing the final "line_len"
18453 (dump_location_info): Update for the above.
18454 (get_substring_ranges_for_loc): Likewise. Use a char_span
18455 when handling the literal within the line.
18456 (test_reading_source_line): Update for location_get_source_line
18457 returning a char_span.
18458 * input.h (class char_span): Move here from
18459 diagnostic-show-locus.c, converting from a struct to a class.
18460 Make data members private.
18461 (char_span::operator bool): New.
18462 (char_span::length): New.
18463 (char_span::get_buffer): New.
18464 (char_span::operator[]): New.
18465 (char_span::subspan): Make const.
18466 (char_span::xstrdup): New.
18467 (location_get_source_line): Convert return type from const char *
18468 to char_span, losing the final "line_size" param.
18470 2018-04-30 Jan Hubicka <jh@suse.cz>
18472 * lto-wrapper.c (ltrans_priorities): New static var.
18473 (cmp_priority): New.
18474 (run_gcc): Read priorities and if doing parallel build order
18475 the Makefile by them.
18477 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18479 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
18481 2018-04-30 Richard Biener <rguenther@suse.de>
18483 * tree-cfg.c (verify_address): Remove base argument, add
18484 flag whether to check TREE_ADDRESSABLE and do that.
18485 (verify_expr): Remove.
18486 (verify_types_in_gimple_reference): Add pieces from verify_expr.
18487 (verify_gimple_assign_single): Likewise.
18488 (verify_gimple_switch): Likewise.
18489 (verify_expr_location_1): Dereference tp once. Add (disabled)
18490 piece from verify_expr.
18491 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
18493 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18495 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
18497 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18499 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
18500 (small_data_pattern): Likewise.
18501 (arc_rewrite_small_data): Likewise.
18502 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
18503 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
18504 (get_symbol_alignment): New function.
18505 (legitimate_small_data_address_p): Likewise.
18506 (legitimate_scaled_address): Update, call
18507 legitimate_small_data_address_p.
18508 (output_sdata): New static variable.
18509 (arc_print_operand): Update how we handle small data operands.
18510 (arc_print_operand_address): Likewise.
18511 (arc_legitimate_address_p): Update, use
18512 legitimate_small_data_address_p.
18513 (arc_rewrite_small_data_p): Remove.
18514 (arc_rewrite_small_data_1): Likewise.
18515 (arc_rewrite_small_data): Likewise.
18516 (small_data_pattern): Likewise.
18517 (compact_sda_memory_operand): Update to use
18518 legitimate_small_data_address_p and get_symbol_alignment.
18519 (prepare_move_operands): Don't rewite sdata pattern.
18520 (prepare_extend_operands): Remove.
18521 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
18523 (zero_extendqisi2): Likewise.
18524 (zero_extendhisi2): Likewise.
18525 (extendqihi2): Likewise.
18526 (extendqisi2): Likewise.
18527 (extendhisi2): Likewise.
18528 (addsi3): Likewise.
18529 (subsi3): Likewise.
18530 (andsi3): Likewise.
18531 * config/arc/constraints.md (Usd): Change it to memory constraint.
18533 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18535 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
18536 as source of std instructions.
18537 * config/arc/arc.md (movsi_insn): Update pattern predicate to
18538 allow 6-bit constants as source for store instructions.
18539 (movdi_insn): Update instruction pattern to allow 6-bit constants
18540 as source for store instructions.
18542 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18544 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
18546 2018-04-30 Nathan Sidwell <nathan@acm.org>
18547 Sandra Loosemore <sandra@codesourcery.com>
18549 * dumpfile.c (dump_open): Allow '-' for stdout.
18550 * doc/invoke.texi (Developer Options): Document dump filename
18551 determination early. Document stdin/stdout selection.
18553 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18555 Microblaze Target: PIC data text relative
18557 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18558 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18560 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18561 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18562 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18563 New addressing mode for data-text relative position indepenedent code.
18564 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18565 'ADDRESS_SYMBOLIC_TXT_REL'.
18566 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18567 (microblaze_legitimate_pic_operand): Exclude function calls from
18568 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18569 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18571 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18572 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18573 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18574 for 'address + offset'.
18575 (microblaze_expand_prologue): Add new function prologue call for
18577 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18578 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18579 table in case of TARGET_PIC_DATA_TEXT_REL.
18580 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18581 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18582 Add new macros 'UNSPEC_TEXT',
18583 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18584 + exclude function calls from 'UNSPEC_PLT' in case of data text
18586 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18587 new target hook for generating address diff vector tables in case of
18589 * doc/tm.texi : Regenerate.
18590 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18591 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18592 of addr diff vector generation.
18593 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18594 target hook definition.
18595 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18596 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18597 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18598 option for data text relative.
18600 2018-04-30 Richard Biener <rguenther@suse.de>
18602 * tree-chrec.h (evolution_function_is_constant_p): Remove
18604 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18606 2018-04-30 Richard Biener <rguenther@suse.de>
18609 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18611 2018-04-30 Richard Biener <rguenther@suse.de>
18613 PR tree-optimization/28364
18614 PR tree-optimization/85275
18615 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18616 copying first exit test.
18618 2018-04-28 Mark Wielaard <mark@klomp.org>
18620 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18621 dwarf_version >= 5.
18622 (dwarf_AT): Handle DW_AT_addr_base.
18623 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18625 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18628 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18629 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18630 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18631 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18633 2018-04-28 Richard Biener <rguenther@suse.de>
18635 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18636 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18637 to reflect use. Only add interesting stmts.
18639 2018-04-27 Martin Jambor <mjambor@suse.cz>
18642 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18643 the jump function allows for passing through aggregate values.
18645 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18647 * input.h (in_system_header_at): Convert from macro to inline
18649 (from_macro_expansion_at): Likewise.
18650 (from_macro_definition_at): Likewise.
18652 2018-04-27 Jeff Law <law@redhat.com>
18654 * config.gcc: Mark tile* targets as deprecated/obsolete.
18656 2018-04-27 Richard Biener <rguenther@suse.de>
18658 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18661 2018-04-27 Richard Biener <rguenther@suse.de>
18663 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18665 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18667 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18668 with Yd constraint. Set "preferred_for_speed" attribute from
18669 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18670 with Yd constraint.
18671 (*movdi_internal): Ditto.
18672 (movti_interunit splitters): Remove
18673 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18674 (movdi_interunit splitters): Ditto.
18675 * config/i386/constraints.md (Ye): Remove.
18676 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18678 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18681 * config/aarch64/constraints.md (Usg): Limit to 31.
18682 (Usj): Limit to 63.
18684 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18686 PR tree-optimization/85529
18687 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18688 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18689 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18690 zero extension or masking of the MSB bit.
18691 (optimize_range_tests): Add FIRST_BB argument, pass it through
18692 to optimize_range_tests_var_bound.
18693 (maybe_optimize_range_tests, reassociate_bb): Adjust
18694 optimize_range_tests callers.
18696 2018-04-26 Richard Biener <rguenther@suse.de>
18697 Jakub Jelinek <jakub@redhat.com>
18699 * cgraph.h (symbol_table): Just declare debug method here.
18700 * symtab.c (symbol_table::debug): Define.
18702 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18704 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18706 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18708 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18709 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18710 (*movdi_internal): Substitute Yi and Yj constraint with x
18711 and Ym and Yn constraint with y constraint. Update "isa"
18712 attribute and set "preferred_for_speed" attribute from
18713 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18714 (*movsi_internal): Ditto.
18715 (*movdf_internal): Ditto.
18716 (*movsf_internal): Ditto.
18717 (*zero_extendsidi2): Ditto.
18718 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18719 (sse2_loadld): Ditto.
18720 (*vec_extract<ssevecmodelower>_0): Ditto.
18721 (*vec_extractv4si_0_zext_sse4): Ditto.
18722 (vec_concatv2di): Ditto.
18723 (*vec_dup<mode>): Ditto.
18724 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18725 * config/i386/constraints.md (Yi): Remove.
18730 2018-04-26 Nathan Sidwell <nathan@acm.org>
18732 * dumpfile.c (dump_open): New.
18733 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18734 (dump_finish): Detect stdio/stderr by value not name.
18736 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18738 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18740 2018-04-26 Tom de Vries <tom@codesourcery.com>
18743 * config/nvptx/nvptx.c (verify_neutering_jumps)
18744 (verify_neutering_labels): New function
18745 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18747 2018-04-26 Tom de Vries <tom@codesourcery.com>
18750 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18751 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18754 2018-04-26 Richard Biener <rguenther@suse.de>
18755 Tom de Vries <tom@codesourcery.com>
18758 * lto-streamer-out.c (output_function): Fixup loops if required to match
18759 discovery done in the reader.
18761 2018-04-26 Richard Biener <rguenther@suse.de>
18763 PR tree-optimization/85116
18764 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18765 have a loop exit from the single latch predecessor. Remove
18766 case of header with just condition.
18767 (ch_base::copy_headers): Exclude infinite loops from any
18769 (pass_ch::execute): Record exits.
18771 2018-04-26 Richard Biener <rguenther@suse.de>
18773 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18774 prologue cost vector and pass it to vect_get_load_cost.
18775 (vect_get_peeling_costs_all_drs): Likewise.
18776 (vect_peeling_hash_get_lowest_cost): Likewise.
18777 (vect_enhance_data_refs_alignment): Likewise.
18779 2018-04-26 Richard Biener <rguenther@suse.de>
18781 PR middle-end/85450
18782 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18783 checking of integer<->pointer conversions.
18784 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18785 sign-/zero-extending pointer types.
18786 (expand_omp_for_static_chunk): Likewise.
18788 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18789 Jean Lee <xiaoyur347@gmail.com>
18791 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18792 (TARGET_ASAN_SHADOW_OFFSET): Define.
18793 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18794 true for -fsanitize=address.
18796 2018-04-25 Mark Wielaard <mark@klomp.org>
18798 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18801 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18803 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18804 than "alu", remove explicit "memory" and "imm_disp" attributes.
18805 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18807 PR middle-end/85414
18808 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18809 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18810 gen_lowpart_no_emit.
18812 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18815 * config/i386/i386.c (ix86_expand_builtin): Change memory
18816 operand to XI, extend p0 to Pmode.
18817 * config/i386/i386.md: Change unspec volatile and operand
18818 1 mode to XI, change operand 0 mode to P.
18820 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18822 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18823 GET_MODE_MASK before any checking.
18824 (nds32_can_use_bset_p): Likewise.
18825 (nds32_can_use_btgl_p): Likewise.
18827 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18829 * config/nds32/nds32-doubleword.md: New define_split pattern for
18830 illegal register number.
18832 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18834 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18836 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18838 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18840 2018-04-25 Richard Biener <rguenther@suse.de>
18842 * lto-streamer.h (LTO_major_version): Bump to 8.
18844 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18846 * BASE-VER: Set to 9.0.0.
18848 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18850 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18851 in __abskf2 and __powikf2.
18853 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18856 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18857 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18858 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18859 Use the above on operand 2. Reindent.
18860 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18862 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18865 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18867 * config/i386/i386.opt (mcet): Removed.
18868 * doc/install.texi: Remove -mcet documentation.
18869 * doc/invoke.texi: Likewise.
18871 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18874 * doc/install.texi: Remove -mcet from bootstrap-cet.
18876 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18879 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18880 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18884 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18885 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18886 containing a CONST_VECTOR.
18888 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18890 * doc/install.texi: Update newlib dependency for nvptx.
18892 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18895 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18896 instead of INTVAL when shifting x left.
18898 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18900 PR tree-optimization/85478
18901 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18902 vect_grouped_store_supported for single element vectors.
18904 2018-04-24 Richard Biener <rguenther@suse.de>
18907 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18908 load cost increase to the case of non-constant step.
18910 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18913 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18914 destination if any_malformed_asm.
18916 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18918 PR middle-end/85496
18919 * expr.c (store_field): In the bitfield case, if the value comes from
18920 a function call and is returned in registers by means of a PARALLEL,
18921 do not change the mode of the temporary unless BLKmode and VOIDmode.
18923 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18925 PR rtl-optimization/85423
18926 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18927 dependencies to debug insns when the previous insn is non-debug.
18929 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18931 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18932 enums into a single definition.
18933 (fls): Fix predicates and printing.
18936 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18938 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18939 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18940 and short u6 immediate.
18941 (check_if_valid_sleep_operand): Remove.
18942 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18944 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18946 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18947 flag_always_save_lp condition.
18948 * config/nds32/nds32.opt (malways-save-lp): New option.
18950 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18952 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18953 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18954 * config/nds32/nds32.h
18955 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18956 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18958 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18960 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18961 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18963 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18964 Chung-Ju Wu <jasonwucj@gmail.com>
18966 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18968 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18969 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18971 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18973 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18975 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18977 * config/nds32/nds32-protos.h (nds32_data_alignment,
18978 nds32_local_alignment): Declare.
18979 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18980 nds32_local_alignment): New functions.
18981 (TARGET_CONSTANT_ALIGNMENT): Define.
18982 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18984 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18986 * config/nds32/nds32.c
18987 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18988 (TARGET_MODES_TIEABLE_P): Likewise.
18990 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18992 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18993 level Ofast and Og.
18995 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18996 Chung-Ju Wu <jasonwucj@gmail.com>
18998 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18999 for unaligned access.
19000 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19002 * config/nds32/nds32-intrinsic.md: Likewise.
19003 * config/nds32/nds32_intrinsic.h: Likewise.
19004 * config/nds32/nds32.h (nds32_builtins): Likewise.
19005 * config/nds32/nds32.opt (munaligned-access): New option.
19006 * config/nds32/nds32.c (nds32_asm_file_start): Display
19007 flag_unaligned_access status.
19009 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
19011 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
19012 -mno-relax is present.
19013 * config/riscv/linux.h (LINK_SPEC): Ditto.
19015 2018-04-20 Martin Sebor <msebor@redhat.com>
19018 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
19020 (gimple_fold_builtin_stxcpy_chk): Same.
19021 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
19023 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
19026 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
19027 __powikf2 when long double is IEEE 128-bit.
19029 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
19031 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
19032 step to make sure stack always aligned.
19034 2018-04-20 Carl Love <cel@us.ibm.com>
19037 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
19038 size check for arg0.
19040 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
19041 Tom de Vries <tom@codesourcery.com>
19044 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
19045 Emit insns for calls too.
19046 (nvptx_find_par): Always look for worker-level predecessor insn.
19047 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
19049 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
19050 (nvptx_process_pars): Propagate frames for calls.
19052 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
19055 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
19057 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
19058 (ix86_handle_option): Don't handle OPT_mibt.
19059 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
19061 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
19063 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
19064 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
19065 (ix86_target_macros): Define __CET__ with flag_cf_protection
19066 for -fcf-protection.
19067 * config/i386/i386.c (isa2_opts): Remove -mibt.
19068 * config/i386/i386.h (TARGET_IBT): Removed.
19069 (TARGET_IBT_P): Likewise.
19070 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
19071 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
19072 * config/i386/i386.opt (mcet): Update help message.
19073 (mshstk): Likewise.
19075 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
19076 -mcet as an alias for -mshstk.
19078 2018-04-20 Richard Biener <rguenther@suse.de>
19080 PR middle-end/85475
19081 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
19082 complexity by forcing a single use of the multiply operand.
19084 2018-04-20 Martin Jambor <mjambor@suse.cz>
19087 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
19088 recursion dependency to only apply to non-clones.
19090 2018-04-20 Martin Jambor <mjambor@suse.cz>
19093 * ipa-cp.c (create_specialized_node): Check that clones of
19094 self-recursive edges exist during IPA-CP.
19096 2018-04-19 Toon Moene <toon@moene.org>
19098 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
19101 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19103 PR tree-optimization/85467
19104 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
19105 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
19106 VECTOR_CST element to type.
19108 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19111 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
19112 * config/i386/i386.md (builtin_setjmp_setup): Removed.
19113 (builtin_longjmp): Likewise.
19114 (save_stack_nonlocal): New pattern.
19115 (restore_stack_nonlocal): Likewise.
19117 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19120 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
19121 Replace ASM_OUTPUT_LABEL with fprintf.
19123 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
19126 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
19127 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
19128 * config/i386/i386-c.c (ix86_target_macros_internal): Also
19129 define __IBT__ and __SHSTK__ for -fcf-protection.
19130 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
19132 (ix86_trampoline_init): Likewise.
19133 (x86_output_mi_thunk): Likewise.
19134 (ix86_notrack_prefixed_insn_p): Likewise.
19135 (ix86_option_override_internal): Don't disallow -fcf-protection.
19136 * config/i386/i386.md (rdssp<mode>): Also enable for
19138 (incssp<mode>): Likewise.
19139 (nop_endbr): Likewise.
19140 * config/i386/i386.opt (mcet): Change help message to built-in
19143 (mshstk): Likewise.
19144 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
19145 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
19146 enable CET built-in functions.
19148 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
19150 * common/config/i386/i386-common.c
19151 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
19152 OPTION_MASK_ISA_MOVDIRI_UNSET,
19153 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
19154 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
19155 * config.gcc (movdirintrin.h): New header.
19156 * config/i386/cpuid.h (bit_MOVDIRI,
19157 bit_MOVDIR64B): New bits.
19158 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
19160 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
19161 (VOID, PVOID, PCVOID)): New function types.
19162 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
19163 __builtin_ia32_directstoreu_u64,
19164 __builtin_ia32_movdir64b): New builtins.
19165 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
19166 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
19168 (ix86_valid_target_attribute_inner_p): Ditto.
19169 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
19170 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
19171 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
19172 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
19173 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
19174 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
19175 (movdiri<mode>, movdir64b_<mode>): New.
19176 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
19177 * config/i386/immintrin.h: Include movdirintrin.h.
19178 * config/i386/movdirintrin.h: New file.
19179 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
19181 2018-04-19 Richard Biener <rguenther@suse.de>
19183 PR middle-end/85455
19184 * cfg.c (clear_bb_flags): When loop state says we have
19185 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
19187 2018-04-19 Richard Biener <rguenther@suse.de>
19189 PR tree-optimization/84737
19190 * tree-vect-data-refs.c (vect_copy_ref_info): New function
19191 copying restrict info.
19192 (vect_setup_realignment): Use it.
19193 * tree-vectorizer.h (vect_copy_ref_info): Declare.
19194 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
19195 the first DR to all generated stores.
19196 (vectorizable_load): Likewise for loads.
19198 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19200 PR tree-optimization/85446
19201 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
19202 the integral and pointer types to have the same precision.
19204 * doc/install.texi: Document --disable-cet being the default and
19207 2018-04-18 Martin Liska <mliska@suse.cz>
19209 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
19212 2018-04-18 Martin Liska <mliska@suse.cz>
19215 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
19218 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
19219 arguments if they are comparable.
19221 2018-04-18 Martin Liska <mliska@suse.cz>
19224 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
19227 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
19230 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
19233 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
19234 ENDBR after calling __morestack.
19236 2018-04-18 David Malcolm <dmalcolm@redhat.com>
19239 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
19240 by using gcc_base_ver to generate a gcc_driver_version, and use
19241 it when generating GCC_DRIVER_NAME.
19242 * configure: Regenerate.
19244 2018-04-18 Jakub Jelinek <jakub@redhat.com>
19247 * config.gcc: Obsolete powerpc*-*-*spe*.
19249 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19252 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
19253 (stabstr_D): Change type of unum from unsigned int to
19254 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
19257 2018-04-17 Jim Wilson <jimw@sifive.com>
19260 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
19261 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
19262 Set arg_pointer_offset after using pretend_args_size.
19264 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19266 PR rtl-optimization/85431
19267 * dse.c (record_store): Ignore zero width stores.
19270 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
19271 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
19272 __builtin_stack_restore rather than after it.
19273 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
19274 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
19275 argument instead of virtual_dynamic_stack_rtx.
19277 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
19279 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
19281 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19282 Add note to error message to explain internal mapping of overloaded
19283 built-in function name to non-overloaded built-in function name.
19284 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
19287 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
19290 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
19291 where the inputs overlap with the output.
19293 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19295 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
19296 (=v, v) alternative and explicit "memory" attribute.
19297 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
19298 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19300 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
19301 "sselog1" type instead of "sselog".
19302 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
19303 "sselog". Remove explicit "memory" attribute.
19304 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
19305 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19307 (vec_extract_hi_v32hi): Merge all alternatives into one, use
19308 "sselog1" type instead of "sselog". Remove explicit "memory"
19310 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
19311 use "sselog1" type instead of "sselog". Remove explicit "memory"
19313 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
19314 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19316 (vec_extract_hi_v64qi): Merge all alternatives into one, use
19317 "sselog1" type instead of "sselog". Remove explicit "memory"
19319 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
19320 use "sselog1" type instead of "sselog". Remove explicit "memory"
19324 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
19326 PR middle-end/85414
19327 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
19330 2018-04-17 Martin Jambor <mjambor@suse.cz>
19333 * ipa-cp.c (create_specialized_node): Call
19334 expand_all_artificial_thunks if necessary.
19336 2018-04-17 Martin Liska <mliska@suse.cz>
19339 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
19340 in message, remote space in between '_G' and '('.
19342 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19345 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
19346 avx512f_vmcmp<mode>3<round_saeonly_name>,
19347 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
19348 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
19349 avx512f_rndscale<mode><round_saeonly_name>,
19350 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
19351 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
19352 Use %<iptr>2 instead of %2 for -masm=intel.
19353 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
19354 avx512f_vcvttss2usi<round_saeonly_name>,
19355 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
19357 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
19358 avx512f_vcvttsd2usi<round_saeonly_name>,
19359 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
19360 Use %q1 instead of %1 for -masm=intel.
19361 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
19362 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
19363 of %3 for -masm=intel.
19364 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
19366 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
19368 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
19370 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
19371 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
19373 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
19375 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
19376 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
19377 %g1 and one with %0 and %1.
19378 (avx512er_vmrcp28<mode><round_saeonly_name>,
19379 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
19380 %1 for -masm=intel.
19381 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
19382 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
19383 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
19384 of %0 and %{%4%} for -masm=intel.
19385 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
19386 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
19387 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
19388 order of %0 and %{%5%}%{z%} for -masm=intel.
19390 2018-04-17 Jan Hubicka <jh@suse.cz>
19393 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
19395 2018-04-17 Martin Liska <mliska@suse.cz>
19398 * multiple_target.c (create_dispatcher_calls): Set apostrophes
19399 for target_clone error message. Make default implementation
19400 clone to be a local declaration.
19401 (separate_attrs): Add new argument and check for an empty
19403 (expand_target_clones): Handle it.
19404 (ipa_target_clone): Make redirection just for target_clones
19407 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
19408 Tom de Vries <tom@codesourcery.com>
19410 PR middle-end/84955
19411 * omp-expand.c (expand_oacc_for): Add dummy false branch for
19412 tiled basic blocks without omp continue statements.
19414 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
19417 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
19418 vec_extract expression as having side effects to make sure it gets
19421 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
19424 * config/i386/i386.c (get_builtin_code_for_version): Check
19427 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
19430 * config.gcc: Support "skylake".
19431 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19433 * config/i386/i386.c (m_SKYLAKE): Define.
19434 (processor_target_table): Add "skylake".
19435 (ix86_option_override_internal): Add "skylake".
19436 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
19437 PROCESSOR_CANNONLAKE.
19438 (get_builtin_code_for_version): Fix priority for
19439 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
19440 PROCESSOR_SKYLAKE-AVX512.
19441 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
19442 (processor_type): Add PROCESSOR_SKYLAKE.
19444 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
19445 Jason Merrill <jason@redhat.com>
19448 * convert.c (convert_to_integer_1): Use direct recursion for
19449 enumeral types and types with a precision less than the number
19450 of bits in their mode.
19452 2018-04-16 Julia Koval <julia.koval@intel.com>
19455 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
19456 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
19458 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
19461 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
19462 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
19463 and -mno-direct-move.
19465 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
19468 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
19469 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
19470 Ensure negative shifts result in {0}.
19472 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
19474 PR rtl-optimization/79916
19475 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
19476 regs (if any) to define how to gnerate SD moves when LRA is in
19479 2018-04-13 Jakub Jelinek <jakub@redhat.com>
19481 PR rtl-optimization/85393
19482 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
19483 * except.c (expand_dw2_landing_pad_for_region): Make static.
19484 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
19485 a label and unconditional jump to old_bb, rather than
19486 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
19489 PR rtl-optimization/85376
19490 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
19491 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
19492 instead of a specific value.
19494 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19495 Bin Cheng <bin.cheng@arm.com>
19497 PR tree-optimization/82965
19498 PR tree-optimization/83991
19499 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
19500 by_profile_only parameter.
19501 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
19502 information if the loop was predicted to iterate too many times.
19503 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
19505 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19508 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
19511 2018-04-13 Martin Liska <mliska@suse.cz>
19512 Jakub Jelinek <jakub@redhat.com>
19514 PR middle-end/81657
19515 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
19516 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
19517 * builtins.c (expand_builtin_memory_copy_args): Use
19518 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
19519 handle dest_addr == pc_rtx.
19521 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19524 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
19525 asked to not generate direct moves.
19526 (fix_trunc<mode>si2_stfiwx): Similar.
19527 (fix_trunc<mode>si2_internal): Similar.
19529 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19532 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
19533 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
19534 lookup if dest in some wider mode is known to be const0_rtx and
19535 if so, record permanent equivalence for it to be ZERO_EXTEND of
19536 the narrower mode destination.
19538 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19540 * lto-streamer-out.c (output_function): Revert 259346.
19541 * omp-expand.c (expand_oacc_for): Likewise.
19543 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
19545 PR rtl-optimization/85354
19546 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
19547 * sel-sched.c (sel_global_init): ... here.
19549 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19552 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19553 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19554 mode for PE-COFF targets.
19555 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19556 (i386_pe_asm_lto_end): Likewise.
19557 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19558 (TARGET_ASM_LTO_END): Likewise.
19559 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19560 (i386_pe_asm_lto_start): New function.
19561 (i386_pe_asm_lto_end): Likewise.
19563 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19564 Richard Biener <rguenther@suse.de>
19566 PR middle-end/84955
19567 * lto-streamer-out.c (output_function): Fix CFG loop state before
19569 * omp-expand.c (expand_oacc_for): Handle calls to internal
19570 functions like regular functions.
19572 2018-04-12 Richard Biener <rguenther@suse.de>
19575 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19576 for the early LTO debug to properly generate references to it
19577 during DIE emission. Do not re-use that for the skeleton for
19579 (dwarf2out_early_finish): Likewise.
19581 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19584 * config/i386/sse.md
19585 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19586 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19587 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19588 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19589 and output is a reg, avoid creating invalid lowpart subreg, but
19590 instead split into a 512-bit move. Don't split if not AVX512VL,
19591 input is xmm16+ reg and output is a mem.
19592 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19593 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19594 xmm16+ reg and output is a mem.
19596 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19598 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19599 also for flag_dwarf2_cfi_asm.
19601 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19603 PR rtl-optimization/85342
19604 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19605 a bool scalar var inside of the loop instead. Don't try to update
19606 recog_data.operand after failed apply_change_group.
19608 2018-04-12 Tom de Vries <tom@codesourcery.com>
19611 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19612 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19613 array with flexible array member as array without given dimension.
19614 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19615 argument for undefined param to true.
19617 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19620 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19621 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19622 from PowerPC section.
19623 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19624 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19625 help text that is too long.
19626 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19627 help text that is too long.
19628 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19629 help text that is too long.
19631 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19633 * config/alpha/alpha.md (stack_probe_internal): Rename
19634 from "probe_stack". Update all callers.
19636 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19638 PR rtl-optimization/84566
19639 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19640 sched_macro_fuse_insns.
19642 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19645 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19646 (compute_block_dependences): ... from here.
19648 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19650 PR tree-optimization/85331
19651 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19652 from int to HOST_WIDE_INT.
19654 2018-04-11 Martin Jambor <mjambor@suse.cz>
19657 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19658 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19659 not the same as the source val.
19660 (cgraph_edge_brings_value_p): New parameter.
19661 (gather_edges_for_value): Pass destination value to
19662 cgraph_edge_brings_value_p.
19663 (perhaps_add_new_callers): Likewise.
19664 (get_info_about_necessary_edges): Likewise and exclude values brought
19665 only by self-recursive edges.
19666 (create_specialized_node): Redirect only clones of self-calling edges.
19667 (+self_recursive_pass_through_p): New function.
19668 (find_more_scalar_values_for_callers_subset): Use it.
19669 (find_aggregate_values_for_callers_subset): Likewise.
19670 (known_aggs_to_agg_replacement_list): Removed.
19671 (decide_whether_version_node): Re-calculate known constants for all
19672 remaining context clones.
19674 2018-04-11 Richard Biener <rguenther@suse.de>
19677 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19678 from early DWARF output.
19679 (dwarf2out_early_finish): Output line info unconditionally into
19680 early DWARF and add reference to it.
19682 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19685 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19686 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19687 other than V2DFmode using iptr mode attribute.
19688 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19690 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19692 PR rtl-optimization/84659
19693 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19695 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19698 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19700 (output_loc_list): Pass address of a dummy size variable even in the
19701 locview handling loop.
19702 (index_location_lists): Add comment on why skip_loc_list_entry can't
19705 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19708 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19711 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19714 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19715 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19716 and -mstring-compare-inline-limit.
19718 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19721 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19722 for stack clash protection in a register whenever we need it to be in
19725 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19727 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19728 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19730 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19733 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19735 (mlong-double-): Ditto.
19736 * config/rs6000/sysv4.opt (msdata=): Ditto.
19737 (mtls-size=): Ditto.
19739 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19741 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19742 erroneous entries for
19743 "vector int vec_ldl (int, long int *)", and
19744 "vector unsigned int vec_ldl (int, unsigned long int *)".
19745 Add comments and entries for
19746 "vector bool char vec_ldl (int, bool char *)",
19747 "vector bool short vec_ldl (int, bool short *)",
19748 "vector bool int vec_ldl (int, bool int *)",
19749 "vector bool long long vec_ldl (int, bool long long *)",
19750 "vector pixel vec_ldl (int, pixel *)",
19751 "vector long long vec_ldl (int, long long *)",
19752 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19753 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19754 type tree bool_long_long_type_node and correct definition of
19755 bool_V2DI_type_node to make reference to this new type tree.
19756 (rs6000_mangle_type): Replace erroneous reference to
19757 bool_long_type_node with bool_long_long_type_node.
19758 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19759 comments to emphasize sign distinctions for char and int types and
19760 replace RS6000_BTI_bool_long constant with
19761 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19762 use of RS6000_BTI_pixel.
19763 (bool_long_type_node): Remove this macro definition.
19764 (bool_long_long_type_node): New macro definition
19766 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19768 PR rtl-optimization/85300
19769 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19770 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19771 simplify_unary_operation fails.
19773 2018-04-10 Martin Liska <mliska@suse.cz>
19775 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19776 cgraph_edge and ipa_ref.
19778 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19782 * config/i386/sse.md
19783 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19784 computation of the VEC_MERGE selector from mask.
19785 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19786 Fix decoding of the VEC_MERGE selector into mask.
19788 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19790 PR tree-optimization/85286
19791 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19793 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19795 * final.c (final_1): Set insn_last_address as well as
19796 insn_current_address.
19798 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19801 * explow.c (emit_stack_probe): Call validize_mem on memory location
19802 before passing it to gen_probe_stack. Create address operand and
19803 legitimize it for the probe_stack_address case.
19805 2018-04-09 Jan Hubicka <jh@suse.cz>
19808 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19809 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19810 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19812 (free_lang_data): Rebuild type inheritance graph.
19814 2018-04-09 Martin Sebor <msebor@redhat.com>
19816 * invoke.texi (-finline-small-functions): Mention other optimization
19818 (-findirect-inlining, -fpartial-inlining): Same.
19819 (-finline-functions-called-once): Same.
19820 (-freorder-blocks-and-partition): Same.
19822 2018-04-09 Jan Hubicka <jh@suse.cz>
19825 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19826 jumps; choose last target that matches the criteria (i.e.
19827 no partition changes for non-crossing jumps).
19828 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19829 support for redirecting crossing jumps to non-crossing.
19831 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19833 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19834 also for naked functions.
19836 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19838 * config/arc/arc.md (add_shift): New pattern.
19839 (add_shift2): Likewise.
19840 (sub_shift): Likewise.
19841 (sub_shift_cmp0_noout): Likewise.
19842 (compare_si_ashiftsi): Likewise.
19843 (xbfu_cmp0_noout): New combine pattern.
19844 (xbfu_cmp0"): Likewise.
19845 (movsi_set_cc_insn): Place the predicable variant first.
19846 (commutative_binary_cmp0_noout): Remove clobber.
19847 (commutative_binary_cmp0): New pattern.
19848 (noncommutative_binary_cmp0): Likewise.
19849 (noncommutative_binary_cmp0_noout): Likewise.
19850 (noncommutative_binary_comparison_result_used): Removed.
19851 (rsub_cmp0): New pattern.
19852 (rsub_cmp0_noout): Likewise.
19853 (extzvsi): Changed, keep only meaningful variants.
19854 (SQH, SEZ): New iterators.
19855 (SQH_postfix): New mode attribute.
19856 (SEZ_prefix): New code attribute.
19857 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19858 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19859 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19860 of numerical value.
19861 (noncommutative_operator): Check the availability of barrel
19864 2018-04-09 Richard Biener <rguenther@suse.de>
19866 PR tree-optimization/85284
19867 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19868 Only use the niter constraining form of simple_iv when the exit
19869 is always executed.
19871 2018-04-09 Tom de Vries <tom@codesourcery.com>
19874 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19875 (define_expand "*memory_barrier"): New define_expand.
19876 (define_insn "memory_barrier"): New insn.
19878 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19880 PR rtl-optimization/80463
19881 PR rtl-optimization/83972
19882 PR rtl-optimization/83480
19884 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19885 correct producer for the insn.
19886 (tidy_control_flow): Fixup seqnos in case of debug insns.
19888 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19890 PR rtl-optimization/83913
19892 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19893 different sched-times when merging exprs.
19895 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19897 PR rtl-optimization/83962
19899 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19900 tidy_fallthru_edge and tidy_control_flow.
19902 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19904 PR rtl-optimization/83530
19906 * sel-sched.c (force_next_insn): New global variable.
19907 (remove_insn_for_debug): When force_next_insn is true, also leave only
19908 next insn in the ready list.
19909 (sel_sched_region): When the region wasn't scheduled, make another pass
19910 over it with force_next_insn set to 1.
19912 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19914 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19916 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19917 for interrupt control.
19918 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19919 functions for interrupt control.
19920 * config/nds32/nds32-intrinsic.md: Likewise.
19921 * config/nds32/nds32_intrinsic.h: Likewise.
19922 * config/nds32/nds32.h (nds32_builtins): Likewise.
19924 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19926 * config/nds32/nds32.c (nds32_init_machine_status,
19927 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19928 strict_aligned_p field.
19929 (nds32_expand_to_rtl_hook): New function.
19930 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19931 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19933 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19934 Chung-Ju Wu <jasonwucj@gmail.com>
19936 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19937 * config/nds32/nds32-n7.md: New file.
19938 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19939 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19941 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19942 * config/nds32/nds32.md (pipeline_model): Add n7.
19943 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19944 * config/nds32/pipelines.md: Include n7 settings.
19946 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19947 Chung-Ju Wu <jasonwucj@gmail.com>
19949 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19950 * config/nds32/nds32-e8.md: New file.
19951 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19952 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19954 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19955 * config/nds32/nds32.md (pipeline_model): Add e8.
19956 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19957 * config/nds32/pipelines.md: Include e8 settings.
19959 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19960 Chung-Ju Wu <jasonwucj@gmail.com>
19962 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19963 * config/nds32/nds32-n8.md: New file.
19964 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19965 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19967 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19968 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19969 * config/nds32/nds32.md (pipeline_model): Add n8.
19970 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19971 * config/nds32/pipelines.md: Include n8 settings.
19973 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19974 Chung-Ju Wu <jasonwucj@gmail.com>
19976 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19977 * config/nds32/nds32-n9-2r1w.md: New file.
19978 * config/nds32/nds32-n9-3r2w.md: New file.
19979 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19980 nds32_register_ports): New or modify for cpu n9.
19981 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19983 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19984 * config/nds32/nds32-utils.c: New file.
19985 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19986 TARGET_MUL_SLOW): Define.
19987 * config/nds32/nds32.md (pipeline_model): New attribute.
19988 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19989 New options that support cpu n9.
19990 * config/nds32/pipelines.md: Include n9 settings.
19991 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19993 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19995 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19996 information if necessary.
19997 (output_cond_branch_compare_zero): Likewise.
19998 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19999 (nds32_target_alignment): Refine for alignment.
20000 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
20001 (FUNCTION_BOUNDARY): Modify.
20002 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
20004 * config/nds32/nds32.opt (malways-align, malign-functions): New.
20006 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
20008 * config/nds32/constants.md (unspec_volatile_element): Add values for
20009 TLB operation and data prefetch.
20010 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
20011 functions for TLB operation and data prefetch.
20012 * config/nds32/nds32-intrinsic.md: Likewise.
20013 * config/nds32/nds32_intrinsic.h: Likewise.
20014 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
20015 (nds32_print_operand): Likewise.
20016 * config/nds32/nds32.h (nds32_builtins): Likewise.
20018 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20019 Andrew Pinski <pinsika@gcc.gnu.org>
20021 PR middle-end/82976
20022 * match.pd: Use constant_boolean_node of correct type instead of
20023 boolean_true_node or boolean_false_node for simplifying
20024 pointer comparisons to zero.
20026 2018-04-07 Jakub Jelinek <jakub@redhat.com>
20028 PR tree-optimization/80021
20029 * tree.c (verify_type_variant): Make error call in verify_variant_match
20030 translatable and remove final full stop.
20032 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20034 * config/nds32/constants.md (unspec_volatile_element): Add
20035 UNSPEC_VOLATILE_EH_RETURN.
20036 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
20037 nds32_output_stack_pop): Support dwarf exception handling process.
20038 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
20039 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
20040 exception handling process.
20041 (nds32_compute_stack_frame): Likewise.
20042 (nds32_return_addr_rtx): Likewise.
20043 (nds32_initial_elimination_offset): Likewise.
20044 (nds32_expand_prologue): Likewise.
20045 (nds32_expand_epilogue): Likewise.
20046 (nds32_dynamic_chain_address): New function.
20047 * config/nds32/nds32.h (machine_function): Add fields for dwarf
20048 exception handling.
20049 (DYNAMIC_CHAIN_ADDRESS): Define.
20050 (EH_RETURN_DATA_REGNO): Define.
20051 (EH_RETURN_STACKADJ_RTX): Define.
20052 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
20053 patterns for dwarf exception handling.
20055 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20057 * config/nds32/nds32.h: Clean up obsolete macros.
20059 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20061 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20062 Add enum values for particular instructions.
20063 * config/nds32/nds32-intrinsic.c: Implementation of expanding
20064 particular intrinsic functions.
20065 * config/nds32/nds32-intrinsic.md: Likewise.
20066 * config/nds32/nds32_intrinsic.h: Likewise.
20067 * config/nds32/nds32.h (nds32_builtins): Likewise.
20068 * config/nds32/nds32.md (type): Add pbsad and pbsada.
20069 (btst, ave): New patterns for particular instructions.
20071 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20073 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
20074 Add enum values for atomic load/store and memory sync.
20075 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
20077 * config/nds32/nds32-intrinsic.md: Likewise.
20078 * config/nds32/nds32_intrinsic.h: Likewise.
20079 * config/nds32/nds32.h (nds32_builtins): Likewise.
20081 2018-04-07 Jakub Jelinek <jakub@redhat.com>
20083 PR tree-optimization/85257
20084 * fold-const.c (native_encode_vector): If not all elts could fit
20085 and off is -1, return 0 rather than offset.
20086 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
20087 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
20088 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
20089 adjust buffer in native_interpret_expr call.
20091 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20093 * config/nds32/constants.md (unspec_volatile_element): Add cache
20094 control enum values.
20095 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
20096 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
20097 * config/nds32/nds32.c (nds32_cctl_names): New.
20098 (nds32_print_operand): Handle cache control register names.
20099 * config/nds32/nds32.h (nds32_builtins): New enum values.
20100 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
20102 * config/nds32/nds32.md (type): Add mmu.
20103 * config/nds32/pipelines.md (simple_insn): Add mmu.
20105 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20107 * config/nds32/nds32.md (type): Remove call.
20108 * config/nds32/pipelines.md (simple_insn): Likewise.
20110 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20112 * config/nds32/constants.md (unspec_volatile_element): Add
20113 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
20114 UNSPEC_VOLATILE_FMFCFG.
20115 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
20116 description for fmfcfg and fmfcsr.
20117 (bdesc_1arg): Add fmtcsr.
20118 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
20119 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
20120 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
20121 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
20122 unspec_fmfcfg): New patterns.
20123 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
20124 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
20125 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
20126 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
20127 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
20128 __nds32__fmfcfg): Define.
20130 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20132 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
20133 intrinsic register names.
20134 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
20135 intrinsic register enum values and macros.
20137 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20139 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
20140 for load/store addressing form.
20141 (nds32_print_operand_address): Likewise.
20143 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
20146 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
20147 based on LABEL_REF. Remove useless assertion.
20148 (pic_address_needs_scratch): Fix formatting.
20149 (sparc_legitimize_pic_address): Minor tweaks.
20150 (sparc_delegitimize_address): Adjust assertion accordingly.
20151 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
20152 into symbolic_operand.
20153 (movsi_high_pic_label_ref): Likewise.
20154 (movsi_lo_sum_pic_label_ref): Likewise.
20155 (movdi_pic_label_ref): Likewise.
20156 (movdi_high_pic_label_ref): Likewise.
20157 (movdi_lo_sum_pic_label_ref): Likewise.
20159 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
20161 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
20162 custom LIB_SPEC setup.
20164 2018-04-06 Ruslan Bukin <br@bsdpad.com>
20165 Kito Cheng <kito.cheng@gmail.com>
20167 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
20168 * config/riscv/freebsd.h: New.
20170 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20172 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
20173 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
20176 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20177 Kito Cheng <kito.cheng@gmail.com>
20179 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
20180 nds32_output_call, nds32_symbol_binds_local_p): New functions.
20181 * config/nds32/nds32-protos.h (nds32_output_call,
20182 nds32_output_return): Declare.
20183 * config/nds32/nds32.md: Refine all the call and return patterns.
20185 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20188 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
20189 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
20191 PR rtl-optimization/84872
20192 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
20193 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
20194 EDGE_CROSSING edge.
20196 2018-04-06 Tamar Christina <tamar.christina@arm.com>
20198 * expr.c (copy_blkmode_to_reg): Revert 254862.
20199 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
20201 2018-04-06 Richard Biener <rguenther@suse.de>
20203 PR middle-end/85244
20204 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
20205 after seeing a component reference with an adjacent field. Treat
20206 refs to arrays at struct end of external decls similar to
20207 refs to unconstrained commons.
20209 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20212 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
20213 look through SAVE_EXPRs with non-side-effects argument. Adjust
20215 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
20218 2018-04-06 Richard Biener <rguenther@suse.de>
20220 PR middle-end/85180
20221 * alias.c (find_base_term): New wrapper around find_base_term
20222 unwinding CSELIB_VAL_PTR changes.
20223 (find_base_term): Do not restore CSELIB_VAL_PTR during the
20226 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20228 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
20230 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
20231 constant definitions.
20232 ("nop"): lr 0,0 -> nopr r0
20233 ("nop_lr0", "nop_lr1"): New insn definitions.
20235 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20237 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
20238 NDS32_V3PUSH_AVAILABLE_P macro.
20240 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
20241 Chung-Ju Wu <jasonwucj@gmail.com>
20243 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
20244 (nds32*-*-*): Add float and fpu_config into supported_defaults.
20245 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
20246 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
20247 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
20248 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
20249 * config/nds32/constraints.md: New constraints and checking for hard
20250 float configuration.
20251 * config/nds32/iterators.md: New mode iterator and attribute for hard
20252 float configuration.
20253 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
20255 * config/nds32/nds32-fpu.md: New file.
20256 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
20257 deal with hard float code generation.
20258 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
20260 (abi_type, float_reg_number): New enum type.
20261 * config/nds32/nds32-predicates.c: New predicates for hard float.
20262 * config/nds32/nds32-protos.h: Declare functions for hard float.
20263 * config/nds32/nds32.c: Implementation for hard float configuration.
20264 * config/nds32/nds32.h: Definitions for hard float configuration.
20265 * config/nds32/nds32.md: Include hard float machine description and
20266 modify patterns for hard float configuration.
20267 * config/nds32/nds32.opt: New options for hard float configuration.
20268 * config/nds32/predicates.md: New predicates for hard float
20271 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20273 * common/config/nds32/nds32-common.c
20274 (nds32_option_optimization_table): Enable -mreleax-hint by default.
20276 2018-04-05 Jakub Jelinek <jakub@redhat.com>
20278 PR middle-end/85195
20279 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
20280 CONSTRUCTOR_ELT (ctor, ...)->value.
20282 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
20285 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
20287 2018-04-05 Tom de Vries <tom@codesourcery.com>
20290 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
20293 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20294 Kito Cheng <kito.cheng@gmail.com>
20296 * config/nds32/constraints.md (U33): Fine-tune checking condition.
20297 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
20298 * config/nds32/nds32.h (nds32_16bit_address_type): Add
20299 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
20301 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20302 Kito Cheng <kito.cheng@gmail.com>
20304 * config/nds32/constraints.md (Ufe): New memory constraint.
20305 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
20306 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
20307 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
20309 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
20310 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
20312 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20314 * config/nds32/nds32.md: Use optimize_size in the condition for
20315 alu-shift instructions.
20317 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20319 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
20321 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20323 * config/nds32/nds32.md (negsi2): Refine pattern.
20325 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20326 Chung-Ju Wu <jasonwucj@gmail.com>
20328 * config/nds32/iterators.md (shift_rotate): New code iterator.
20329 (shift): New code attribute.
20330 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
20331 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
20332 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
20333 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
20334 bit-wise operations.
20335 (andsi3, *andsi3): Ditto.
20336 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
20337 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
20338 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
20339 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
20340 nds32_ior_operand, nds32_xor_operand): New predicates.
20342 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20344 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
20345 (addsi3, subsi3): ... this.
20347 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20349 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
20351 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20353 * config/nds32/nds32.md: Adjust indention.
20355 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20357 * config/nds32/nds32.md (feature): New attribute.
20359 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20361 * config/nds32/nds32.md (subtype): New attribute.
20363 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20366 * config/arm/arm-builtins.c (arm_expand_builtin): Change
20367 expansion to perform a bitwise AND of the argument followed by a
20368 boolean negation of the result.
20370 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
20372 PR rtl-optimization/84878
20373 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
20374 the basic block. Assert the use reference is not artificial and that
20375 it has an associated insn.
20377 2018-04-04 Michael Matz <matz@suse.de>
20379 * builtins.c (compute_objsize): Pass correct operand
20380 to array_at_struct_end_p.
20382 2018-04-04 Richard Biener <rguenther@suse.de>
20385 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
20386 from contexts for DINFO_LEVEL_TERSE and below.
20388 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20390 * config/nds32/nds32-doubleword.md (move_<mode>): Require
20391 resiter_operand condition.
20392 * config/nds32/nds32.md (*move<mode>): Ditto.
20394 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20395 Monk Chiang <sh.chiang04@gmail.com>
20397 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
20399 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20401 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
20403 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20404 Kito Cheng <kito.cheng@gmail.com>
20406 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
20407 nds32_cond_code_str, output_cond_branch,
20408 output_cond_branch_compare_zero, nds32_expand_cbranch,
20409 nds32_expand_cstore, nds32_expand_movcc,
20410 nds32_output_cbranchsi4_equality_zero,
20411 nds32_output_cbranchsi4_equality_reg,
20412 nds32_output_cbranchsi4_equality_reg_or_const_int,
20413 nds32_output_cbranchsi4_greater_less_zero: New functions.
20414 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
20415 nds32_expand_cstore, nds32_expand_movcc,
20416 nds32_output_cbranchsi4_equality_zero,
20417 nds32_output_cbranchsi4_equality_reg,
20418 nds32_output_cbranchsi4_equality_reg_or_const_int,
20419 nds32_output_cbranchsi4_greater_less_zero): Declare.
20420 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
20421 nds32_rimm11s_operand): New predicates.
20422 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
20423 * config/nds32/nds32.md: Rewrite all the branch and conditional move
20426 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20428 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
20429 * config/nds32/nds32.md: Ditto.
20430 * config/nds32/pipelines.md: Ditto.
20432 2018-04-04 Richard Biener <rguenther@suse.de>
20434 PR tree-optimization/85168
20435 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
20436 propagating abnormals.
20438 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20440 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
20442 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20443 Kito Cheng <kito.cheng@gmail.com>
20445 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
20446 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
20447 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
20448 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
20449 * config/nds32/nds32.md (sibcall_internal): New.
20450 (sibcall_register): Remove.
20451 (sibcall_immediate): Remove.
20452 (sibcall_value_internal): New.
20453 (sibcall_value_register): Remove.
20454 (sibcall_value_immediate): Remove.
20455 * config/nds32/predicates.md (nds32_general_register_operand): New.
20456 (nds32_call_address_operand): New.
20458 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20460 PR rtl-optimization/85167
20461 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
20462 bb_defs if *split_p, instead preinitialize it to NULL.
20464 PR tree-optimization/85156
20465 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
20466 evaluating the argument multiple times.
20468 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
20470 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
20472 (_mm_cvtpd_ps): Likewise.
20473 (_mm_cvttpd_epi32): Likewise.
20474 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
20475 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
20476 vector, pixel, and bool following altivec.h include.
20478 2018-04-03 Martin Sebor <msebor@redhat.com>
20480 * doc/extend.texi (Common Function Attributes): Clarify.
20481 (const attribute): Likewise.
20482 (pure attribute): Likewise.
20484 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20487 * config/i386/i386.c (ix86_expand_vector_set): Use
20488 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
20490 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
20492 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
20493 instructions when changing rounding bits to preserve precision bits
20494 in the x87 control word.
20496 2018-04-03 Martin Liska <mliska@suse.cz>
20498 PR tree-optimization/82491
20499 * rtl.h (strip_offset_and_add): Replace += suboffset with
20500 poly_uint64 () + suboffset.
20502 2018-03-29 Martin Liska <mliska@suse.cz>
20503 Martin Jambor <mjambor@suse.cz>
20506 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
20507 param_type is not an integral or pointer type.
20509 2018-04-03 Richard Biener <rguenther@suse.de>
20511 * sese.h (recompute_all_dominators): Remove.
20513 2018-04-02 Martin Sebor <msebor@redhat.com>
20515 * doc/invoke.texi (-Wrestrict): Fix typos.
20517 2018-04-02 Jim Wilson <jimw@sifive.com>
20519 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
20520 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
20521 (<optab>di3, <optab>si3_extend): Likewise.
20522 (<optab>si3_mask, <optab>si3_mask_1): New.
20523 (<optab>di3_mask, <optab>di3_mask_1): New.
20524 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
20525 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
20526 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
20528 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
20530 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
20533 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
20535 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
20536 (nds32_canonicalize_comparison): New function.
20538 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20539 Kito Cheng <kito.cheng@gmail.com>
20540 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20542 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
20543 * config/nds32/constants.md (unspec_volatile_element): Add
20544 UNSPEC_VOLATILE_RELAX_GROUP.
20545 * config/nds32/nds32-relax-opt.c: New file.
20546 * config/nds32/nds32-predicates.c
20547 (nds32_symbol_load_store_p): New function.
20548 * config/nds32/nds32-protos.h
20549 (nds32_symbol_load_store_p): Declare function.
20550 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20551 * config/nds32/nds32.c
20552 (nds32_register_pass): New function to register pass.
20553 (nds32_register_passes): New function to register passes.
20554 * config/nds32/nds32.md (relax_group): New pattern.
20555 * config/nds32/nds32.opt (mrelax-hint): New option.
20556 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20558 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20560 * config/nds32/t-nds32: Modify files dependency.
20562 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20564 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20565 (PROFILE_HOOK): Define its implementation.
20567 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20569 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20570 type and 32-bit size.
20572 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20574 PR middle-end/85090
20575 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20576 (V_128_256): New mode iterator.
20577 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20578 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20579 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20581 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20582 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20584 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20587 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20588 NaN inputs correctly.
20590 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20593 * config/rs6000/vsx.md (??r): New mode attribute.
20594 (*vsx_mov<mode>_64bit): Use it.
20595 (*vsx_mov<mode>_32bit): Likewise.
20597 2018-03-30 Martin Sebor <msebor@redhat.com>
20599 PR tree-optimization/84818
20600 * builtins.c (check_access): Use warning_n.
20602 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20605 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20607 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20610 2018-03-30 Julia Koval <julia.koval@intel.com>
20613 * x86-tune.def (movx, partial_reg_dependency): Enable for
20616 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20618 PR inline-asm/84985
20619 * lra-constraints.c (process_alt_operands): Move setting
20620 this_alternative_matches below.
20622 2018-03-29 Martin Liska <mliska@suse.cz>
20625 * doc/invoke.texi: Document how LTO works with debug info.
20626 Describe auto-load support of binutils. Mention 'x86-64'
20627 as valid option value of -march option.
20629 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20631 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20634 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20635 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20636 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20639 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20642 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20643 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20644 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20645 for RS6000_BTM_POWERPC64.
20646 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20647 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20648 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20651 (DIVDEU): Likewise.
20653 2018-03-28 Carl Love <cel@us.ibm.com>
20656 2017-09-27 Carl Love <cel@us.ibm.com>
20658 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20659 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20660 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20663 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20665 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20666 instead of __vector bool.
20667 (_mm_max_pu8): Likewise.
20668 (_mm_min_pi16): Likewise.
20670 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20673 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20674 (DIVWEUO): Likewise.
20675 (DIVDEO): Likewise.
20676 (DIVDEUO): Likewise.
20677 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20678 DIVWEUO and DIVDEUO.
20679 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20680 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20681 (div_extend): Likewise.
20682 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20684 (__builtin_divweuo): Likewise.
20685 (__builtin_divdeo): Likewise.
20686 (__builtin_divdeuo): Likewise.
20688 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20691 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20692 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20694 PR tree-optimization/82004
20695 * gimple-match-head.c (optimize_pow_to_exp): New function.
20696 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20697 Don't fold to exp if optimize_pow_to_exp is false.
20699 2018-03-28 Martin Liska <mliska@suse.cz>
20702 * calls.c (initialize_argument_information): Fix trailing space.
20703 * common.opt: Fix typo and provide better explanation for
20704 -fsanitize-coverage option.
20705 * config/i386/i386.opt: Fix typo.
20707 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20708 Martin Liska <mliska@suse.cz>
20711 * gimplify.c (asan_poison_variable): Don't do the check for
20712 gimplify_omp_ctxp here.
20713 (gimplify_decl_expr): Do it here.
20714 (gimplify_target_expr): Likewise.
20716 2018-03-28 Martin Liska <mliska@suse.cz>
20719 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20720 chkp_type_bounds_count if MPX is not enabled.
20722 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20724 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20726 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20729 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20730 function to create the function decl for complex long double
20731 multiply and divide for -mabi=ieeelongdouble.
20732 (init_float128_ieee): Call it.
20734 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20737 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20738 -fcf-protection=branch -mibt.
20739 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20741 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20744 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20745 arm_word_relocations.
20747 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20750 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20751 extern array declarations.
20753 2018-03-27 Richard Biener <rguenther@suse.de>
20755 PR middle-end/84067
20756 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20757 explicit single_use checks.
20759 2018-03-27 Richard Biener <rguenther@suse.de>
20761 PR tree-optimization/85082
20762 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20765 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20767 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20768 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20769 Turn on fasynchronous-unwind-tables and funwind-tables.
20771 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20774 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20775 (*bmi_blsr_<mode>_ccz): Ditto.
20777 2018-03-26 Tom de Vries <tom@codesourcery.com>
20779 PR tree-optimization/85063
20780 * omp-general.c (offloading_function_p): New function. Factor out
20782 * omp-offload.c (pass_omp_target_link::gate): ... here.
20783 * omp-general.h (offloading_function_p): Declare.
20784 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20785 with attribute omp declare target for offloading functions.
20787 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20789 PR tree-optimization/84005
20790 * tree-data-ref.h (get_base_for_alignment): Declare.
20791 * tree-data-ref.c (get_base_for_alignment_1): New function.
20792 (get_base_for_alignment): Likewise.
20793 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20794 get_base_for_alignment to find a suitable base object, instead
20795 of always using drb->base_address.
20797 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20799 PR inline-asm/85022
20800 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20801 known size by default.
20803 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20805 PR inline-asm/85030
20806 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20807 and non BLKmode operands.
20809 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20812 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20813 Clean up attributes.
20815 2018-03-23 Richard Biener <rguenther@suse.de>
20818 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20819 we are going to emit early debug for LTO.
20821 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20823 PR inline-asm/85034
20824 * function.c (match_asm_constraints_1): Don't optimize if input
20825 doesn't satisfy general_operand predicate for output's mode.
20827 PR inline-asm/85022
20828 * alias.c (write_dependence_p): Don't require for x_canonicalized
20829 non-VOIDmode if x has VOIDmode.
20832 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20833 just don't try to optimize it rather than assert it never happens.
20835 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20837 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20838 macro expansions for definition of ST_INTERNAL_<mode> and
20839 LD_INTERNAL_<mode> builtins.
20840 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20842 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20844 (altivec_expand_st_builtin): Likewise.
20845 (altivec_expand_builtin): Remove calls to deleted functions.
20846 (rs6000_address_for_altivec): Delete this function.
20847 * config/rs6000/vector.md: Remove expands for
20848 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20850 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20853 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20854 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20855 re-computing once computed.
20856 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20857 (arm_init_machine_status): Initialize
20858 machine->static_chain_stack_bytes.
20860 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20863 * doc/extend.texi: Add four new prototypes for vec_ld.
20864 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20865 definitions for more logical presentation.
20866 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20867 entries for V1TI variants of __builtin_altivec_ld builtin.
20868 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20869 handling of V1TI variant of LVX icode pattern.
20870 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20871 (rs6000_gimple_fold_builtin): Likewise.
20872 (altivec_init_builtins): Add code to define
20873 __builtin_altivec_lvx_v1ti function.
20875 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20877 PR inline-asm/84941
20878 * function.c (match_asm_constraints_1): Don't do the optimization
20879 if input isn't a REG, SUBREG, MEM or constant.
20881 2018-03-22 Tom de Vries <tom@codesourcery.com>
20883 PR tree-optimization/84956
20884 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20885 bb_has_abnormal_pred.
20887 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20890 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20891 DECL_INITIAL (decl) to decl at the end.
20892 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20893 adjust the comment.
20895 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20897 * doc/extend.texi (__builtin_tgmath): Document when complex
20898 integer types are treated as _Complex _Float64.
20900 2018-03-21 Tom de Vries <tom@codesourcery.com>
20902 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20904 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20906 PR tree-optimization/84960
20907 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20908 if it is ENTRY block, move them into single succ of ENTRY in that case.
20910 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20912 PR tree-optimization/84811
20913 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20914 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20915 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20916 (known_subrange_p): Update accordingly. Cast each value involved
20917 in the size comparison, rather than casting the result of the
20920 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20922 PR tree-optimization/84982
20923 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20924 by flipping the least significant bit rather than all bits from
20925 bitpos to bitpos + bitsize - 1.
20927 2018-03-21 Nathan Sidwell <nathan@acm.org>
20929 * doc/extend.texi (Deprecated Features): Remove mention of
20930 long-deleted deprecations.
20932 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20935 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20936 * configure: Regenerate.
20938 2018-03-21 Tom de Vries <tom@codesourcery.com>
20940 PR tree-optimization/83126
20941 * tree-parloops.c (num_phis): New function.
20942 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20944 2018-03-21 Nathan Sidwell <nathan@acm.org>
20946 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20947 mention anon-struct/union members and trailing attributes.
20949 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20951 PR tree-optimization/84969
20952 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20953 builtin memset partitions if they set different rhs values.
20955 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20957 PR rtl-optimization/84989
20958 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20959 VEC_DUPLICATE with scalar result mode.
20961 2018-03-21 Martin Liska <mliska@suse.cz>
20964 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20965 not intended return statement.
20967 2018-03-21 Martin Liska <mliska@suse.cz>
20970 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20971 (chkp_find_bound_slots_1): Limit number of iterations.
20973 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20976 * Minor grammar fixes for x86 options.
20978 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20981 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20982 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20984 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20987 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20988 (altivec_lvx_<mode>_1op): Likewise.
20989 (altivec_stvx_<mode>_2op): Likewise.
20990 (altivec_stvx_<mode>_1op): Likewise.
20991 (altivec_lvx_<VM2:mode>): New define_expand.
20992 (altivec_stvx_<VM2:mode>): Likewise.
20993 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20994 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20995 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20996 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20997 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20998 (rs6000_gen_lvx): Likewise.
20999 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
21000 (altivec_expand_stv_builtin): Likewise.
21001 (altivec_expand_builtin): Likewise.
21002 * config/rs6000/vector.md: Likewise.
21004 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21007 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
21010 2018-03-20 Richard Biener <rguenther@suse.de>
21013 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
21014 sign-conversions as zero, fall back to standard scalar_stmt
21017 2018-03-20 Martin Liska <mliska@suse.cz>
21020 * predict.c (rebuild_frequencies): Handle case when we have
21021 PROFILE_ABSENT, but flag_guess_branch_prob is false.
21023 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21026 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
21027 flag_section_anchors.
21028 * varasm.c (use_blocks_for_decl_p): Remove hack for
21029 dw2_force_const_mem.
21032 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
21034 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
21035 be created, use lowpart_subreg of operands[0] rather than operands[0]
21037 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
21038 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
21039 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
21040 and n constraint instead of aarch64_shift_imm_di and Usd.
21041 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
21042 (*aarch64_<optab>_reg_minus<mode>3): ... this.
21044 2018-03-20 Sudakshina Das <sudi.das@arm.com>
21047 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
21048 to favor GPR over NEON registers.
21049 (<shift>di3_neon): Likewise.
21051 2018-03-20 Tom de Vries <tom@codesourcery.com>
21054 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
21055 (nvptx_process_pars): Emit bar.sync asap and alap.
21057 2018-03-20 Tom de Vries <tom@codesourcery.com>
21060 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
21061 seen_label if seen_label is already set.
21063 2018-03-20 Jakub Jelinek <jakub@redhat.com>
21066 * config/i386/i386.c (fold_builtin_cpu): For features above 31
21067 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
21068 Use 1U instead of 1. Formatting fixes.
21071 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
21072 instead of TREE_TYPE (s1) for the return value.
21074 2018-03-19 Jakub Jelinek <jakub@redhat.com>
21076 PR tree-optimization/84946
21077 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
21078 bitsize + bitsize in poly_uint64 rather than poly_int64.
21081 * dwarf2asm.c: Include fold-const.c.
21082 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
21083 of decl rather than decl itself.
21085 PR rtl-optimization/84643
21086 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
21088 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
21091 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
21092 calling assemble_variable.
21094 2018-03-19 Sudakshina Das <sudi.das@arm.com>
21097 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
21098 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
21100 2018-03-19 Jim Wilson <jimw@sifive.com>
21103 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
21104 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
21105 (riscv_first_stack_step): Likewise.
21106 (riscv_option_override): Use STACK_BOUNDARY instead of
21107 MIN_STACK_BOUNDARY.
21108 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
21109 MIN_STACK_BOUNDARY.
21110 (BIGGEST_ALIGNMENT): Set to 128.
21111 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
21112 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
21115 2018-03-19 Richard Biener <rguenther@suse.de>
21117 PR tree-optimization/84933
21118 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
21119 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
21121 2018-03-19 Richard Biener <rguenther@suse.de>
21123 PR tree-optimization/84859
21124 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
21125 (cond_if_else_store_replacement): Perform sinking operation on
21126 single-store BBs regardless of MAX_STORES_TO_SINK setting.
21127 Generalize what a BB with a single eligible store is.
21129 2018-03-19 Richard Biener <rguenther@suse.de>
21131 PR tree-optimization/84929
21132 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
21133 chrec_is_positive against non-chrec arg.
21135 2018-03-19 Tamar Christina <tamar.christina@arm.com>
21138 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
21140 2018-03-18 Martin Liska <mliska@suse.cz>
21142 PR rtl-optimization/84635
21143 * regrename.c (build_def_use): Use matches_mode only when
21146 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
21148 PR tree-optimization/84913
21149 * tree-vect-loop.c (vectorizable_reduction): Don't try to
21150 vectorize chains of COND_EXPRs.
21152 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21154 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
21156 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21158 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
21160 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21162 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
21164 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21165 Kito Cheng <kito.cheng@gmail.com>
21167 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
21168 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
21169 (nds32_adjust_reg_alloc_order): New function.
21170 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
21172 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
21174 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
21175 nds32_print_operand, nds32_print_operand_address): Use
21176 HOST_WIDE_INT_PRINT_DEC instead.
21178 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21180 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
21182 2018-03-17 Jakub Jelinek <jakub@redhat.com>
21185 * config/i386/i386.c (initial_ix86_tune_features,
21186 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
21187 unsigned long long.
21188 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
21189 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
21190 rather than 1u << ix86_tune. Formatting fix.
21191 (ix86_option_override_internal): Change ix86_arch_mask from
21192 unsigned int to unsigned HOST_WIDE_INT, initialize to
21193 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
21194 (ix86_function_specific_restore): Likewise.
21196 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21199 * postreload.c (reload_combine_recognize_pattern): Perform
21200 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
21201 truncate_int_for_mode the result for the destination's mode.
21204 * hsa-gen.c (mem_type_for_type): Fix comment typo.
21205 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
21207 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21210 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
21213 * lra-assigns.c (lra_split_hard_reg_for): Don't use
21214 regno_allocno_class_array and sorted_pseudos.
21215 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
21216 insns where regno is used.
21218 2018-03-16 Martin Liska <mliska@suse.cz>
21221 * multiple_target.c (create_dispatcher_calls): Redirect
21222 reference in the symbol table.
21224 2018-03-16 Martin Liska <mliska@suse.cz>
21227 * multiple_target.c (create_dispatcher_calls): Redirect also
21230 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21234 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
21235 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
21236 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
21238 2018-03-16 Julia Koval <julia.koval@intel.com>
21240 * doc/invoke.texi (Skylake Server): Add CLWB.
21241 Cannonlake): Remove CLWB.
21243 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21245 PR tree-optimization/84841
21246 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
21248 (FLOAT_ONE_CONST_TYPE): Define.
21249 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
21250 (sort_by_operand_rank): Put entries with higher constant_type last
21251 rather than first to match comments.
21253 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
21255 * config/nios2/nios2.md (movsi_internal): Fix thinko in
21258 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21261 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
21262 check and use address of target always.
21264 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
21267 * config/i386/i386.c (indirect_thunk_needed): Update comments.
21268 (indirect_thunk_bnd_needed): Likewise.
21269 (indirect_thunks_used): Likewise.
21270 (indirect_thunks_bnd_used): Likewise.
21271 (indirect_return_needed): New.
21272 (indirect_return_bnd_needed): Likewise.
21273 (output_indirect_thunk_function): Add a bool argument for
21275 (output_indirect_thunk_function): Don't generate alias for
21276 function return thunk.
21277 (ix86_code_end): Call output_indirect_thunk_function to generate
21278 function return thunks.
21279 (ix86_output_function_return): Set indirect_return_bnd_needed
21280 and indirect_return_needed instead of indirect_thunk_bnd_needed
21281 and indirect_thunk_needed.
21283 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
21285 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
21286 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
21287 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
21289 2018-03-15 David Malcolm <dmalcolm@redhat.com>
21290 Paul Hua <paul.hua.gm@gmail.com>
21293 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
21295 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
21297 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
21298 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
21299 resp. SFmode cases.
21301 2018-03-15 Tamar Christina <tamar.christina@arm.com>
21304 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
21305 instead of GET_MODE_SIZE when comparing Units.
21307 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21310 * varasm.c (hash_section): Return an unchangeble hash value
21311 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
21312 Return !aarch64_can_use_per_function_literal_pools_p ().
21314 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21317 * optabs.c (emit_conditional_move): Pass address of cmode's copy
21318 rather than address of cmode as last argument to prepare_cmp_insn.
21320 2018-03-15 Julia Koval <julia.koval@intel.com>
21322 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
21323 F_AVX512VNNI, F_AVX512BITALG): New.
21325 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
21328 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
21329 insn for floating-point loads and stores.
21331 2018-03-14 Carl Love <cel@us.ibm.com>
21333 * config/rs6000/rs6000-c.c: Add macro definitions for
21334 ALTIVEC_BUILTIN_VEC_PERMXOR.
21335 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
21336 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
21337 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
21338 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
21340 * config/doc/extend.texi: Add prototypes for vec_permxor.
21342 2018-03-14 David Malcolm <dmalcolm@redhat.com>
21345 * diagnostic-show-locus.c (class layout_point): Convert m_line
21346 from int to linenum_type.
21347 (line_span::comparator): Use linenum "compare" function when
21348 comparing line numbers.
21349 (test_line_span): New function.
21350 (layout_range::contains_point): Convert param "row" from int to
21352 (layout_range::intersects_line_p): Likewise.
21353 (layout::will_show_line_p): Likewise.
21354 (layout::print_source_line): Likewise.
21355 (layout::should_print_annotation_line_p): Likewise.
21356 (layout::print_annotation_line): Likewise.
21357 (layout::print_leading_fixits): Likewise.
21358 (layout::annotation_line_showed_range_p): Likewise.
21359 (struct line_corrections): Likewise for field m_row.
21360 (line_corrections::line_corrections): Likewise for param "row".
21361 (layout::print_trailing_fixits): Likewise.
21362 (layout::get_state_at_point): Likewise.
21363 (layout::get_x_bound_for_row): Likewise.
21364 (layout::print_line): Likewise.
21365 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
21366 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
21367 * input.c (selftest::test_linenum_comparisons): New function.
21368 (selftest::input_c_tests): Call it.
21369 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
21370 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
21371 * selftest.h (ASSERT_GT): New macro.
21372 (ASSERT_GT_AT): New macro.
21373 (ASSERT_LT): New macro.
21374 (ASSERT_LT_AT): New macro.
21376 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
21378 PR rtl-optimization/84780
21379 * combine.c (distribute_links): Don't make a link based on pc_rtx.
21381 2018-03-14 Martin Liska <mliska@suse.cz>
21383 * tree.c (record_node_allocation_statistics): Use
21384 get_stats_node_kind.
21385 (get_stats_node_kind): New function extracted from
21386 record_node_allocation_statistics.
21387 (free_node): Use get_stats_node_kind.
21389 2018-03-14 Richard Biener <rguenther@suse.de>
21391 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
21392 that the value-set of ANTIC_IN doesn't grow.
21395 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21397 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21398 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21400 2018-03-14 Julia Koval <julia.koval@intel.com>
21402 * config.gcc (icelake-client, icelake-server): New.
21404 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
21405 (initial_ix86_arch_features): Ditto.
21406 (PTA_SKYLAKE): Add SGX.
21407 (PTA_ICELAKE): Remove.
21408 (PTA_ICELAKE_CLIENT): New.
21409 (PTA_ICELAKE_SERVER): New.
21410 (ix86_option_override_internal): Split up icelake on icelake client and
21412 (get_builtin_code_for_version): Ditto.
21413 (fold_builtin_cpu): Ditto.
21414 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
21415 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
21416 * config/i386/i386.h (processor_type): Ditto.
21417 * doc/invoke.texi: Ditto.
21419 2018-03-14 Jakub Jelinek <jakub@redhat.com>
21422 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
21423 INTEGER_CST offset, add it together with bitpos / 8 and
21424 sign extend based on POINTER_SIZE.
21428 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21431 * config/i386/constraints.md (Yc): New register constraint.
21432 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21433 Use Yc constraint for alternative 2 of operand 0. Remove
21434 preferred_for_speed attribute.
21436 2018-03-14 Richard Biener <rguenther@suse.de>
21438 PR tree-optimization/84830
21439 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
21440 with the old one to avoid oscillations.
21442 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
21445 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
21447 (assign_by_spills): Return a flag of reload assignment failure.
21448 Do not process the reload assignment failures. Do not spill other
21449 reload pseudos if they has the same reg class. Update n if
21451 (lra_assign): Add a return arg. Set up from the result of
21452 assign_by_spills call.
21453 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21454 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21455 usage_insns if it is not NULL.
21456 (spill_hard_reg_in_range): New function.
21457 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21458 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21459 function prototypes.
21460 (lra_assign): Change prototype.
21461 * lra.c (lra): Add code to deal with fails by splitting hard reg
21464 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
21466 * config/riscv/riscv.opt (mrelax): New option.
21467 * config/riscv/riscv.c (riscv_file_start): Emit ".option
21468 "norelax" when riscv_mrelax is disabled.
21469 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
21471 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21474 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
21475 reassociation for int modes.
21477 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21479 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
21480 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
21482 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
21483 * config/aarch64/aarch64-sve.md
21484 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
21485 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
21486 (*extend<mode><Vwide>2): Rename to...
21487 (aarch64_sve_extend<mode><Vwide>2): ...this.
21488 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
21489 renaming the old pattern to...
21490 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
21492 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
21493 define_expand, renaming the old pattern to...
21494 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
21495 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
21496 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
21497 account when deciding which SVE instruction the optab should use.
21498 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
21500 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21502 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
21503 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
21504 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
21505 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
21506 (tlsdesc_small_<mode>): Turn a define_expand and use
21507 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
21508 (tlsdesc_small_advsimd_<mode>): ...this.
21509 (tlsdesc_small_sve_<mode>): New pattern.
21511 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21513 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
21514 (UNSPEC_UMUL_HIGHPART): New constants.
21515 (MUL_HIGHPART): New int iteraor.
21516 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
21517 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
21519 (*<su>mul<mode>3_highpart): New define_insn.
21521 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21524 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21527 2018-03-13 Martin Liska <mliska@suse.cz>
21530 * (sem_item_optimizer::sem_item_optimizer): Initialize new
21532 (sem_item_optimizer::~sem_item_optimizer): Release it.
21533 (sem_item_optimizer::merge_classes): Register variable aliases.
21534 (sem_item_optimizer::fixup_pt_set): New function.
21535 (sem_item_optimizer::fixup_points_to_sets): Likewise.
21536 * ipa-icf.h: Declare new variables and functions.
21538 2018-03-13 Jakub Jelinek <jakub@redhat.com>
21540 PR middle-end/84834
21541 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
21542 integer_pow2p@2 and test integer_pow2p in condition.
21543 (A < 0 ? C : 0): Similarly for @1.
21545 PR middle-end/84831
21546 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
21547 characters starting at p contain '\0' character, don't look beyond
21551 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21552 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21555 * reg-stack.c (change_stack): Change update_end var from int to
21556 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21557 also call set_block_for_insn on the newly added insns and rescan.
21560 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21561 on the last operand.
21564 * tree.c (stabilize_reference_1): Return save_expr (e) for
21565 STATEMENT_LIST even if it doesn't have side-effects.
21567 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21569 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21571 2018-03-12 Renlin Li <renlin.li@arm.com>
21573 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21574 aarch64_output_scalar_simd_mov_immediate.
21576 2018-03-12 Martin Sebor <msebor@redhat.com>
21578 PR tree-optimization/83456
21579 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21580 for perfectly overlapping calls to memcpy.
21581 (gimple_fold_builtin_memory_chk): Same.
21582 (gimple_fold_builtin_strcpy): Handle no-warning.
21583 (gimple_fold_builtin_stxcpy_chk): Same.
21584 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21586 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21588 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21589 parameter. Use it for SFmode.
21590 (rs6000_function_arg_advance_1): Adjust.
21591 (rs6000_function_arg): Adjust.
21592 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21594 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21596 PR rtl-optimization/84169
21597 PR rtl-optimization/84780
21598 * combine.c (can_combine_p): Check for a 2-insn combination whether
21599 the destination register is used between the two insns, too.
21601 2018-03-12 Richard Biener <rguenther@suse.de>
21603 PR tree-optimization/84803
21604 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21605 for refs DR analysis didn't process.
21607 2018-03-12 Richard Biener <rguenther@suse.de>
21609 PR tree-optimization/84777
21610 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21611 force-vectorize loops ignore whether we are optimizing for size.
21613 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21615 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21616 (TARGET_MD_ASM_ADJUST): Define.
21618 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21619 Kito Cheng <kito.cheng@gmail.com>
21620 Chung-Ju Wu <jasonwucj@gmail.com>
21622 * config/nds32/nds32.c (nds32_compute_stack_frame,
21623 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21624 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21625 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21626 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21627 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21628 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21629 * config/nds32/nds32.md (prologue, epilogue): Use macro
21630 NDS32_V3PUSH_AVAILABLE_P to do checking.
21632 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21635 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21636 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21637 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21638 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21639 addition of most attributes on !orig_type_die or the attribute not
21640 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21642 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21643 Chung-Ju Wu <jasonwucj@gmail.com>
21645 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21646 __NDS32_VH__ macro.
21647 * config/nds32/nds32.opt (mvh): New option.
21649 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21650 Chung-Ju Wu <jasonwucj@gmail.com>
21652 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21654 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21655 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21658 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21659 Chung-Ju Wu <jasonwucj@gmail.com>
21661 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21663 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21664 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21666 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21667 Kito Cheng <kito.cheng@gmail.com>
21668 Chung-Ju Wu <jasonwucj@gmail.com>
21670 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21671 UNSPEC_FFMISM and UNSPEC_FLMISM.
21672 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21673 for ffb, ffmism and flmism.
21674 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21675 (unspec_ffmism): Ditto.
21676 (unspec_flmism): Ditto.
21677 (nds32_expand_builtin_impl): Check if string extension is available.
21678 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21679 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21681 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21684 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21687 * lra-assigns.c (assign_by_spills): Return a flag of reload
21688 assignment failure. Do not process the reload assignment
21689 failures. Do not spill other reload pseudos if they has the same
21691 (lra_assign): Add a return arg. Set up from the result of
21692 assign_by_spills call.
21693 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21694 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21695 usage_insns if it is not NULL.
21696 (spill_hard_reg_in_range): New function.
21697 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21698 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21699 function prototypes.
21700 (lra_assign): Change prototype.
21701 * lra.c (lra): Add code to deal with fails by splitting hard reg
21704 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21707 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21709 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21712 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21713 (dw_val_node): Add val_symbolic_view.
21714 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21715 (symview_upper_bound): New.
21716 (new_line_info_table): Initialize symviews_since_reset.
21717 (dwarf2out_source_line): Count symviews_since_reset and set
21718 symview_upper_bound.
21719 (dw_val_equal_p): Handle symview.
21720 (add_AT_symview): New.
21721 (print_dw_val): Handle symview.
21722 (attr_checksum, attr_checksum_ordered): Likewise.
21723 (same_dw_val_p, size_of_die): Likewise.
21724 (value_format, output_die): Likewise.
21725 (add_high_low_attributes): Use add_AT_symview for entry_view.
21726 (dwarf2out_finish): Reset symview_upper_bound, clear
21729 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21732 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21733 Add strict argument and use it.
21734 (rs6000_split_multireg_move): Update for new strict argument.
21735 (mem_operand_gpr): Disallow all non-offsettable addresses.
21736 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21738 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21741 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21742 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21743 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21746 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21747 decl, use remap_type if we want to use the type.
21749 2018-03-09 Martin Sebor <msebor@redhat.com>
21751 PR tree-optimization/84526
21752 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21754 (builtin_access::generic_overlap): Be prepared to handle non-array
21757 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21759 PR rtl-optimization/84682
21760 * lra-constraints.c (process_address_1): Check is_address flag
21761 for address constraints.
21762 (process_alt_operands): Likewise.
21763 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21764 preprocess_constraints.
21765 * recog.h (preprocess_constraints): Add oploc parameter.
21767 * recog.c (preprocess_constraints): Test address_operand for
21768 CT_ADDRESS constraints.
21770 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21773 * lra-assigns.c (assign_by_spills): Return a flag of reload
21774 assignment failure. Do not process the reload assignment
21775 failures. Do not spill other reload pseudos if they has the same
21777 (lra_assign): Add a return arg. Set up from the result of
21778 assign_by_spills call.
21779 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21780 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21781 usage_insns if it is not NULL.
21782 (spill_hard_reg_in_range): New function.
21783 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21784 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21785 function prototypes.
21786 (lra_assign): Change prototype.
21787 * lra.c (lra): Add code to deal with fails by splitting hard reg
21790 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21793 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21794 Accept complain bool parameter. Only emit errors if it is true.
21795 (arm_parse_cpu_option_name): Likewise.
21796 (arm_target_thumb_only): Adjust callers of the above.
21797 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21798 prototype to take a default true bool parameter.
21799 (arm_parse_arch_option_name): Likewise.
21801 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21802 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21806 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21807 * configure: Regenerate.
21808 * configure.ac ("linker --version-script option"): New.
21809 ("linker soname option"): New.
21811 2018-03-09 Richard Biener <rguenther@suse.de>
21813 PR tree-optimization/84775
21814 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21815 immediate uses of predicate stmts and mark them modified.
21818 PR tree-optimization/84178
21819 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21821 (version_loop_for_if_conversion): Delay update_ssa call.
21822 (tree_if_conversion): Delay update_ssa until after predicate
21825 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21828 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21829 when the function accesses prior frames.
21831 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21834 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21835 gen_llsym, otherwise call maybe_gen_llsym.
21837 PR inline-asm/84742
21838 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21839 has ',' character inside of it.
21841 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21844 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21845 as clobbering CC_REGNUM.
21847 2018-03-08 Richard Biener <rguenther@suse.de>
21849 PR middle-end/84552
21850 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21851 (follow_copies_to_constant): Do not follow SSA names registered
21854 2018-03-08 Richard Biener <rguenther@suse.de>
21856 PR tree-optimization/84178
21857 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21859 (version_loop_for_if_conversion): Delay update_ssa call.
21860 (tree_if_conversion): Delay update_ssa until after predicate
21863 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21865 PR tree-optimization/84178
21866 * tree-if-conv.c (release_bb_predicate): Remove the
21867 the assertion that the stmts have NULL use_ops.
21868 Discard the statements, asserting that they haven't
21869 yet been added to a BB.
21871 2018-03-08 Richard Biener <rguenther@suse.de>
21873 PR tree-optimization/84746
21874 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21875 (phi_translate): Pass in destination ANTIC_OUT set.
21876 (phi_translate_1): Likewise. For a simplified result lookup
21877 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21878 (phi_translate_set): Adjust.
21879 (do_pre_regular_insertion): Likewise.
21880 (do_pre_partial_partial_insertion): Likewise.
21882 2018-03-08 Martin Liska <mliska@suse.cz>
21884 PR gcov-profile/84735
21885 * doc/gcov.texi: Document usage of profile files.
21886 * gcov-io.h: Document changes in the format.
21888 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21892 * dwarf2out.c (struct dw_line_info_table): Update comments for
21894 (FORCE_RESET_NEXT_VIEW): New.
21895 (FORCE_RESETTING_VIEW_P): New.
21896 (RESETTING_VIEW_P): Check for -1 too.
21897 (ZERO_VIEW_P): Likewise.
21898 (new_line_info_table): Force-reset next view.
21899 (dwarf2out_begin_function): Likewise.
21900 (dwarf2out_source_line): Simplify zero_view_p initialization.
21901 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21902 view directly. Omit view when omitting .loc at line 0.
21904 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21906 PR tree-optimization/84740
21907 * tree-switch-conversion.c (process_switch): Call build_constructors
21908 only if info.phi_count is non-zero.
21910 PR tree-optimization/84739
21911 * tree-tailcall.c (find_tail_calls): Check call arguments against
21912 DECL_ARGUMENTS (current_function_decl) rather than
21913 DECL_ARGUMENTS (func) when checking for tail recursion.
21915 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21917 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21918 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21919 Volker Reichelt's entry and add entries for people that perform
21920 GCC fuzzy testing and report numerous bugs.
21922 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21925 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21926 readonly data in sdata, if that is disabled.
21927 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21928 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21929 -mreadonly-in-sdata option.
21931 2018-03-07 Martin Sebor <msebor@redhat.com>
21933 PR tree-optimization/84468
21934 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21935 basic block when looking for nul assignment.
21937 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21940 * except.h (output_function_exception_table): Adjust prototype.
21941 * except.c (output_function_exception_table): Remove FNNAME parameter
21942 and add SECTION parameter. Ouput one part of the table at a time.
21943 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21944 the first part of the exception table and emit unwind directives.
21945 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21946 (i386_pe_seh_cold_init): Likewise.
21947 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21948 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21949 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21950 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21951 * config/i386/winnt.c: Include except.h.
21952 (struct seh_frame_state): Add reg_offset, after_prologue and
21953 in_cold_section fields.
21954 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21955 (i386_pe_seh_cold_init): New function.
21956 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21957 to seh->in_cold_section.
21958 (seh_emit_push): Record the offset of the push.
21959 (seh_emit_save): Record the offet of the save.
21960 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21961 Test seh->after_prologue to disregard the epilogue.
21962 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21963 (i386_pe_end_cold_function): New function.
21965 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21968 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21969 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21972 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21974 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21975 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21977 PR middle-end/84723
21978 * multiple_target.c: Include tree-inline.h and intl.h.
21979 (expand_target_clones): Diagnose and fail if node->definition and
21980 !tree_versionable_function_p (node->decl).
21982 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21984 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21986 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21987 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21988 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21990 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21993 * combine.c (try_combine): Use reg_or_subregno instead of handling
21994 just paradoxical SUBREGs and REGs.
21996 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21998 * config/arc/arc.c (arc_finalize_pic): Remove function.
21999 (arc_must_save_register): We use single base PIC register, remove
22000 checks to save/restore the PIC register.
22001 (arc_expand_prologue): Likewise.
22002 * config/arc/arc-protos.h (arc_set_default_type_attributes):
22004 (arc_verify_short): Likewise.
22005 (arc_attr_type): Likewise.
22006 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
22007 (walk_stores): Likewise.
22008 (arc_address_cost): Make it static.
22009 (arc_verify_short): Likewise.
22010 (branch_dest): Likewise.
22011 (arc_attr_type): Likewise.
22012 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
22013 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
22014 (arc_final_prescan_insn): Remove inserting the nops due to
22015 hardware hazards. It is done in reorg step.
22016 (insn_length_variant_t): Remove.
22017 (insn_length_parameters_t): Likewise.
22018 (arc_insn_length_parameters): Likewise.
22019 (arc_get_insn_variants): Likewise.
22020 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
22022 2018-03-06 Jakub Jelinek <jakub@redhat.com>
22024 PR inline-asm/84683
22025 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
22028 PR tree-optimization/84687
22029 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
22031 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
22033 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22035 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
22036 Rename to ppc_speculation_barrier.
22037 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
22038 __builtin_ppc_speculation_barrier.
22040 2018-03-05 Jakub Jelinek <jakub@redhat.com>
22043 * combine.c (combine_simplify_rtx): Don't try to simplify if
22044 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
22047 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
22049 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
22050 to 32 bytes when compiling for POWER9.
22052 2018-03-05 Jakub Jelinek <jakub@redhat.com>
22055 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
22056 regparm >= 3 with no arg reg available also for calls with
22057 flag_force_indirect_call. Pass decl to ix86_function_regparm.
22060 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
22062 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
22064 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
22067 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
22069 2018-03-05 Richard Biener <rguenther@suse.de>
22071 PR tree-optimization/84486
22072 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
22073 When inserting a __builtin_assume_aligned call set the LHS
22074 SSA name alignment info accordingly.
22076 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
22078 PR tree-optimization/84114
22079 * config/aarch64/aarch64.c (aarch64_reassociation_width)
22080 Avoid reassociation of FLOAT_MODE addition.
22082 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
22084 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
22085 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
22086 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
22087 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
22088 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
22089 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
22090 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
22092 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
22093 __builtin_ia32_wbinvd): New builtins.
22094 (SPECIAL_ARGS2): New.
22095 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
22096 (SPECIAL_ARGS2): New.
22097 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
22098 (ix86_valid_target_attribute_inner_p): Ditto.
22099 (ix86_init_mmx_sse_builtins): Add special_args2.
22100 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
22101 TARGET_WBNOINVD_P): New.
22102 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
22103 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
22104 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
22105 * config/i386/immintrin.h (_wbinvd): New intrinsic.
22106 * config/i386/pconfigintrin.h: New file.
22107 * config/i386/wbnoinvdintrin.h: Ditto.
22108 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
22110 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
22112 2018-03-05 Richard Biener <rguenther@suse.de>
22114 PR tree-optimization/84670
22115 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
22117 (BB_VISITED_WITH_VISITED_SUCCS): New define.
22118 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
22119 (compute_antic_aux): Only assert the number of values in ANTIC_IN
22120 doesn't grow if all successors (recursively) were visited at least
22123 2018-03-05 Richard Biener <rguenther@suse.de>
22125 PR tree-optimization/84650
22126 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
22127 if executed in the loop pipeline.
22129 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
22131 * doc/configfiles.texi (Configuration Files): Move info about
22132 conditionalizing $target-protos.h to...
22133 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
22134 differs from $target-protos.h.
22136 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
22137 Chung-Ju Wu <jasonwucj@gmail.com>
22139 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
22140 * config/nds32/nds32-multiple.md (setmemsi): Define.
22141 * config/nds32/nds32-memory-manipulation.c
22142 (nds32_gen_dup_4_byte_to_word_value): New.
22143 (emit_setmem_word_loop): New.
22144 (emit_setmem_byte_loop): New.
22145 (nds32_expand_setmem_loop): New.
22146 (nds32_expand_setmem_loop_v3m): New.
22147 (nds32_expand_setmem_unroll): New.
22148 (nds32_expand_setmem): New.
22150 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22151 Chung-Ju Wu <jasonwucj@gmail.com>
22153 * config/nds32/nds32-memory-manipulation.c
22154 (nds32_emit_load_store): New.
22155 (nds32_emit_post_inc_load_store): New.
22156 (nds32_emit_mem_move): New.
22157 (nds32_emit_mem_move_block): New.
22158 (nds32_expand_movmemsi_loop_unknown_size): New.
22159 (nds32_expand_movmemsi_loop_known_size): New.
22160 (nds32_expand_movmemsi_loop): New.
22161 (nds32_expand_movmemsi_unroll): New.
22162 (nds32_expand_movmemqi): Rename ...
22163 (nds32_expand_movmemsi): ... to this.
22164 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
22165 (movmemsi): ... to this.
22166 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
22167 (nds32_expand_movmemsi): ... to this.
22169 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22170 Monk Chiang <sh.chiang04@gmail.com>
22171 Chung-Ju Wu <jasonwucj@gmail.com>
22173 * config/nds32/nds32-protos.h
22174 (nds32_expand_load_multiple): New arguments.
22175 (nds32_expand_store_multiple): Ditto.
22176 (nds32_valid_multiple_load_store): Rename ...
22177 (nds32_valid_multiple_load_store_p): ... to this.
22178 * config/nds32/nds32-memory-manipulation.c
22179 (nds32_expand_load_multiple): Refine implementation.
22180 (nds32_expand_store_multiple): Ditto.
22181 * config/nds32/nds32-multiple.md
22182 (load_multiple): Update nds32_expand_load_multiple interface.
22183 (store_multiple): Update nds32_expand_store_multiple interface.
22184 * config/nds32/nds32-predicates.c
22185 (nds32_valid_multiple_load_store): Rename ...
22186 (nds32_valid_multiple_load_store_p): ... to this and refine
22188 * config/nds32/predicates.md
22189 (nds32_load_multiple_and_update_address_operation): New predicate.
22190 (nds32_store_multiple_and_update_address_operation): New predicate.
22192 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22193 Chung-Ju Wu <jasonwucj@gmail.com>
22195 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
22196 (combo): New attribute.
22197 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
22199 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
22201 * config/nds32/nds32.opt: Change -mcmodel= default value.
22203 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
22204 Monk Chiang <sh.chiang04@gmail.com>
22205 Chung-Ju Wu <jasonwucj@gmail.com>
22207 * config/nds32/constants.md (unspec_element): New enum.
22208 * config/nds32/constraints.md (Umw): New constraint.
22209 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
22210 * config/nds32/nds32-intrinsic.md: Likewise.
22211 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
22212 (nds32_valid_smw_lwm_base_p): New.
22213 (nds32_output_smw_single_word): New.
22214 (nds32_output_lmw_single_word): New.
22215 (nds32_expand_unaligned_load): New.
22216 (nds32_expand_unaligned_store): New.
22217 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
22218 (nds32_output_smw_single_word): Declare.
22219 (nds32_output_lmw_single_word): Declare.
22220 (nds32_expand_unaligned_load): Declare.
22221 (nds32_expand_unaligned_store): Declare.
22222 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
22223 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
22224 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
22225 NDS32_BUILTIN_UASTORE_DW.
22226 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
22229 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
22230 Kito Cheng <kito.cheng@gmail.com>
22231 Chung-Ju Wu <jasonwucj@gmail.com>
22233 * config/nds32/nds32-intrinsic.c
22234 (nds32_expand_builtin_null_ftype_reg): Delete.
22235 (nds32_expand_builtin_reg_ftype_imm): Ditto.
22236 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
22237 (nds32_read_argument): New.
22238 (nds32_legitimize_target): Ditto.
22239 (nds32_legitimize_argument): Ditto.
22240 (nds32_check_constant_argument): Ditto.
22241 (nds32_expand_unop_builtin): Ditto.
22242 (nds32_expand_unopimm_builtin): Ditto.
22243 (nds32_expand_binop_builtin): Ditto.
22244 (nds32_builtin_decl_impl): Ditto.
22245 (builtin_description): Ditto.
22246 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
22247 (nds32_init_builtins_impl): Ditto.
22248 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
22249 (nds32_builtin_decl): New.
22250 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
22251 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
22253 2018-03-02 Jeff Law <law@redhat.com>
22255 * reorg.c (stop_search_p): Handle DEBUG_INSN.
22256 (redundant_insn, fill_simple_delay_slots): Likewise.
22257 (fill_slots_from_thread): Likewise.
22258 * resource.c (mark_referenced_resources): Likewise.
22259 (mark_set_resources, find_dead_or_set_registers): Likewise.
22261 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22263 * substring-locations.h (format_warning_va): Formatting fix for
22264 ATTRIBUTE_GCC_DIAG.
22265 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
22267 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
22268 * substring-locations.c: Include intl.h.
22269 (format_warning_va): Turned into small wrapper around
22270 format_warning_n_va, renamed to ...
22271 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
22272 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
22274 (format_warning_at_substring_n): New function.
22275 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
22276 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
22277 format_warning_at_substring with just a shorter name instead of
22278 const function pointer.
22279 (fmtwarn_n): New function.
22280 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
22281 appropriate, get rid of all the fmtstr temporaries, move conditionals
22282 with G_() wrapped string literals directly into fmtwarn arguments,
22283 cast dir.len to (int), formatting fixes.
22285 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
22287 * doc/invoke.texi: Remove "Cilk Plus" references.
22289 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22290 Richard Biener <rguenther@suse.de>
22293 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
22294 for error or warning attributes if CALL_FROM_THUNK_P is set.
22297 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22300 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
22301 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
22304 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
22305 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
22307 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
22308 instead of -1U in last predictors element's probability member.
22310 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22313 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22314 arguments if they are comparable.
22316 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22318 PR tree-optimization/84634
22319 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
22320 masks and masked_loop_p with a single loop_masks, making sure it's
22321 null for bb vectorization.
22323 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22325 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
22326 (vect_analyze_data_ref_access): Use loop->safe_len rather than
22327 loop->force_vectorize to check whether there is no alias.
22329 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22332 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
22334 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
22336 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
22337 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
22338 instead of a loop around prev_real_insn.
22339 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
22342 PR inline-asm/84625
22343 * config/i386/i386.c (ix86_print_operand): Use conditional
22344 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
22347 2018-03-02 Richard Biener <rguenther@suse.de>
22349 PR tree-optimization/84427
22350 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
22351 (bitmap_set_subtract_values): Rewrite to handle multiple
22354 (prune_clobbered_mems): Likewise.
22355 (phi_translate): Take edge instead of pred/phiblock.
22356 (phi_translate_1): Likewise.
22357 (phi_translate_set): Likewise. Insert all translated
22358 exprs for a value into the set, keeping possibly multiple
22359 expressions per value.
22360 (compute_antic_aux): Adjust for phi_translate changes.
22361 When intersecting union the expressions and prune those
22362 not in the final value set, keeping possibly multiple
22363 expressions per value. Do not use value-insertion
22364 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
22365 all expressions. Add verification that the value-sets
22366 only shrink during iteration.
22367 (compute_partial_antic_aux): Adjust for the phi_translate changes.
22368 (do_pre_regular_insertion): Likewise.
22369 (do_pre_partial_partial_insertion): Likewise.
22371 2018-03-02 Richard Biener <rguenther@suse.de>
22374 * config/darwin.c (saved_debug_info_level): New static global.
22375 (darwin_asm_lto_start): Disable debug info generation for LTO out.
22376 (darwin_asm_lto_end): Restore debug info generation settings.
22378 2018-03-01 Martin Liska <mliska@suse.cz>
22381 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
22382 volatile arguments.
22384 2018-03-01 Richard Biener <rguenther@suse.de>
22387 * dwarf2out.c (gen_variable_die): Properly handle late VLA
22388 type annotation with LTO when debug was disabled at compile-time.
22390 2018-03-01 Matthew Fortune <mfortune@gmail.com>
22392 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
22394 (mips_final_postscan_insn): Likewise.
22396 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
22398 PR rtl-optimization/84528
22399 * alias.c (init_alias_target): Add commentary.
22400 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
22401 a unique base value if the frame pointer is not eliminated
22402 to the stack pointer.
22404 2018-03-01 Tom de Vries <tom@codesourcery.com>
22406 PR rtl-optimization/83327
22407 * lra-int.h (hard_regs_spilled_into): Declare.
22408 * lra.c (hard_regs_spilled_into): Define.
22409 (init_reg_info): Init hard_regs_spilled_into.
22410 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
22411 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
22412 (process_bb_lives): Handle hard_regs_spilled_into.
22413 (lra_create_live_ranges_1): Before doing liveness propagation, clear
22414 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
22416 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
22418 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
22419 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
22420 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
22421 * config/rs6000/aix72.h: New file.
22423 2018-02-28 Jakub Jelinek <jakub@redhat.com>
22425 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
22426 instead of warning_at with conditional singular and plural messages
22430 * stor-layout.c (update_alignment_for_field): For
22431 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
22432 && !DECL_PACKED (field), do the alignment update, just use
22433 only desired_align instead of MAX (type_align, desired_align)
22435 (place_field): Don't do known_align < desired_align handling
22436 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
22437 is non-NULL, instead do it after rli->prev_field handling and
22438 only if not within a bitfield word. For DECL_PACKED (field)
22439 use type_align of BITS_PER_UNIT.
22441 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
22443 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
22444 superfluous parentheses and trailing spaces.
22446 2018-02-28 Richard Biener <rguenther@suse.de>
22448 PR tree-optimization/84584
22449 * graphite-scop-detection.c (scop_detection::add_scop): Discard
22450 SCoPs with fake exit edge.
22452 2018-02-28 Martin Liska <mliska@suse.cz>
22455 * timevar.c (timer::print): Fix format to properly print 100%
22458 2018-02-28 Richard Biener <rguenther@suse.de>
22460 PR middle-end/84607
22461 * genmatch.c (capture_info::walk_match): Do not mark
22462 captured expressions without operands as expr_p given
22463 they act more like predicates and should be subject to
22464 "lost tail" side-effect preserving.
22466 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
22468 PR rtl-optimization/81611
22469 * auto-inc-dec.c (attempt_change): Move dead note from
22470 mem_insn if it's the next use of regno
22471 (find_address): Take address use of reg holding
22472 non-incremented value. Add parm to limit search to the named
22474 (merge_in_block): Attempt to use a mem insn that is the next
22475 use of the original regno.
22477 2018-02-27 Martin Sebor <msebor@redhat.com>
22480 * doc/invoke.texi (-Wmissing-attributes): New option.
22481 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
22483 2018-02-27 Martin Sebor <msebor@redhat.com>
22485 PR translation/84207
22486 * diagnostic-core.h (warning_n, error_n, inform_n): Change
22487 n argument to unsigned HOST_WIDE_INT.
22488 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
22489 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
22490 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
22491 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
22493 2018-02-27 Richard Biener <rguenther@suse.de>
22495 PR tree-optimization/84512
22496 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
22497 Do not use the estimate returned from record_stmt_cost for
22498 the scalar iteration cost but sum properly using add_stmt_cost.
22500 2018-02-27 Richard Biener <rguenther@suse.de>
22502 PR tree-optimization/84466
22503 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22504 Adjust last change to less strictly validate use operands.
22506 2018-02-27 Martin Liska <mliska@suse.cz>
22508 PR gcov-profile/84548
22509 * gcov.c (process_file): Allow partial overlap and consider it
22510 also as group functions.
22511 (output_lines): Properly calculate range of lines for a group.
22513 2018-02-27 Martin Liska <mliska@suse.cz>
22515 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
22516 'ggc' suffixes. Change first column width.
22517 (timer::print): Fix formatting of the column.
22519 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
22521 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
22522 preserve inline entry blocks for the sake of debug inline
22523 entry point markers alone.
22524 (remove_unused_locals): Suggest in comments a better place to
22525 force the preservation of inline entry blocks that are
22526 otherwise unused, but do not preserve them.
22528 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22530 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
22532 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22535 * config/i386/constraints.md (Bs): Replace
22536 ix86_indirect_branch_register with
22537 TARGET_INDIRECT_BRANCH_REGISTER.
22539 * config/i386/i386.md (indirect_jump): Likewise.
22540 (tablejump): Likewise.
22541 (*sibcall_memory): Likewise.
22542 (*sibcall_value_memory): Likewise.
22543 Peepholes of indirect call and jump via memory: Likewise.
22544 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
22545 (*sibcall_value_GOT_32): Likewise.
22546 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22547 (GOT_memory_operand): Likewise.
22548 (call_insn_operand): Likewise.
22549 (sibcall_insn_operand): Likewise.
22550 (GOT32_symbol_operand): Likewise.
22551 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22553 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22555 PR rtl-optimization/83496
22556 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22557 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22558 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22559 redundant insn, if any.
22560 (relax_delay_slots): Likewise.
22561 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22563 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22565 PR tree-optimization/83965
22566 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22567 that grouped statements are part of a reduction chain. Return
22568 true if the statement is not marked as a reduction itself but
22569 is part of a group.
22570 (vect_recog_dot_prod_pattern): Don't check whether the statement
22571 is part of a group here.
22572 (vect_recog_sad_pattern): Likewise.
22573 (vect_recog_widen_sum_pattern): Likewise.
22575 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22578 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22580 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22583 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22585 (ix86_output_indirect_function_return): New prototype.
22586 (ix86_split_simple_return_pop_internal): Likewise.
22587 * config/i386/i386.c (indirect_return_via_cx): New.
22588 (indirect_return_via_cx_bnd): Likewise.
22589 (indirect_thunk_name): Handle return va CX_REG.
22590 (output_indirect_thunk_function): Create alias for
22591 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22592 (ix86_output_indirect_jmp): Remove the bool argument.
22593 (ix86_output_indirect_function_return): New function.
22594 (ix86_split_simple_return_pop_internal): Likewise.
22595 * config/i386/i386.md (*indirect_jump): Don't pass false
22596 to ix86_output_indirect_jmp.
22597 (*tablejump_1): Likewise.
22598 (simple_return_pop_internal): Change it to define_insn_and_split.
22599 Call ix86_split_simple_return_pop_internal to split it for
22600 -mfunction-return=.
22601 (simple_return_indirect_internal): Call
22602 ix86_output_indirect_function_return instead of
22603 ix86_output_indirect_jmp.
22605 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22608 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22609 memset and value initialization afterwards.
22611 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22613 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22615 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22618 * common/config/aarch64/aarch64-common.c
22619 (aarch_option_optimization_table[]): Switch
22620 off fomit-frame-pointer
22622 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22623 Chung-Ju Wu <jasonwucj@gmail.com>
22625 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22627 (store_multiple): Ditto.
22629 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22631 * config.gcc: Add --with-cpu support for nds32 target.
22632 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22633 * config/nds32/nds32.opt: Add -mcpu= option.
22635 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22637 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22638 isel=yes): Warn for these deprecated options.
22640 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22642 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22643 ISA_2_5_MASKS_EMBEDDED.
22645 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22647 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22648 p->max as pointers rather than using iterative_hash_expr.
22650 2018-02-23 Carl Love <cel@us.ibm.com>
22652 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22653 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22655 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22656 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22657 P8V_BUILTIN_VEC_VUNSIGNED2.
22659 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22662 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22663 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22665 * lra-constraints.c (curr_insn_transform): Set up
22666 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22668 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22669 * lra-eliminations.c (spill_pseudos): Ditto.
22670 (process_insn_for_elimination): Ditto.
22671 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22672 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22673 LRA_NON_CLOBBERED_ALT.
22675 2018-02-22 Martin Sebor <msebor@redhat.com>
22677 PR tree-optimization/84480
22678 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22679 to maybe_diag_stxncpy_trunc. Call it.
22680 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22681 from gimple_fold_builtin_strcpy. Print inlining stack.
22682 (handle_builtin_stxncpy): Print inlining stack.
22683 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22685 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22688 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22689 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22690 and -fcheck-pointer-bounds are used together.
22691 (indirect_thunk_prefix): New enum.
22692 (indirect_thunk_need_prefix): New function.
22693 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22694 "_nt" instead of "_bnd" for NOTRACK prefix.
22695 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22696 (output_indirect_thunk_function): Likewise.
22698 (ix86_code_end): Update output_indirect_thunk_function calls.
22699 (ix86_output_indirect_branch_via_reg): Replace
22700 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22701 (ix86_output_indirect_branch_via_push): Likewise.
22702 (ix86_output_function_return): Likewise.
22703 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22704 incompatible with -fcf-protection=branch and
22705 -fcheck-pointer-bounds.
22707 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22710 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22711 Change gcc_assert call to output_operand_lossage.
22713 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22715 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22717 2018-02-22 DJ Delorie <dj@redhat.com>
22718 Sebastian Perta <sebastian.perta@renesas.com>
22719 Oleg Endo <olegendo@gcc.gnu.org>
22721 * config/rx/rx.c (rx_rtx_costs): New function.
22722 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22724 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22726 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22728 2018-02-22 Martin Liska <mliska@suse.cz>
22731 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22732 Add "native" as a possible value.
22734 2018-02-22 Martin Liska <mliska@suse.cz>
22737 * config/i386/i386.c (ix86_option_override_internal):
22738 Add "native" as a possible value for -march and -mtune.
22740 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22743 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22744 to all type variants.
22746 PR tree-optimization/84503
22747 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22748 width as info->bitpos + info->bitsize - start.
22749 (merged_store_group::merge_overlapping): Simplify width computation.
22750 (check_no_overlap): New function.
22751 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22752 start + width and last_order of the group, fail if check_no_overlap
22754 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22755 to group if check_no_overlap fails.
22757 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22759 * config/rs6000/altivec.md: Delete contraint arguments to
22760 define_expand, define_split, and define_peephole2, and in
22761 define_insn_and_split if always unused.
22762 * config/rs6000/darwin.md: Ditto.
22763 * config/rs6000/dfp.md: Ditto.
22764 * config/rs6000/rs6000.md: Ditto.
22765 * config/rs6000/sync.md: Ditto.
22766 * config/rs6000/vector.md: Ditto.
22767 * config/rs6000/vsx.md: Ditto.
22769 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22771 * config/rs6000/altivec.md: Write output control strings as braced
22772 blocks instead of double-quoted strings.
22773 * config/rs6000/darwin.md: Ditto.
22774 * config/rs6000/rs6000.md: Ditto.
22775 * config/rs6000/vector.md: Ditto.
22776 * config/rs6000/vsx.md: Ditto.
22778 2018-02-21 Jason Merrill <jason@redhat.com>
22780 PR c++/84314 - ICE with templates and fastcall attribute.
22781 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22783 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22785 * ipa-cp.c (determine_versionability): Fix comment typos.
22787 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22790 * ipa-cp.c (determine_versionability): Do not version functions caling
22793 2018-02-21 Martin Liska <mliska@suse.cz>
22796 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22797 Add "native" as a possible value.
22798 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22799 the macro when native cpu detection is available.
22801 2018-02-21 Martin Liska <mliska@suse.cz>
22804 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22805 Add "native" as a possible value.
22806 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22807 when native cpu detection is available.
22809 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22810 Martin Sebor <msebor@redhat.com>
22812 PR tree-optimization/84478
22813 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22815 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22816 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22817 support which is conservatively correct, for 2 only stay conservative
22818 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22819 argument to the 2 argument get_range_strlen, adjust 6 arg
22820 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22822 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22823 (gimple_fold_builtin_strlen): Pass true as last argument to
22826 2018-02-20 Martin Sebor <msebor@redhat.com>
22828 PR middle-end/84095
22829 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22830 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22831 (builtin_memref::builtin_memref): Factor out parts into
22832 set_base_and_offset and call it.
22834 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22836 PR middle-end/84406
22837 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22838 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22839 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22840 search at the associated MODE_INT.
22842 2018-02-20 Jeff Law <law@redhat.com>
22844 PR middle-end/82123
22845 PR tree-optimization/81592
22846 PR middle-end/79257
22847 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22848 for range data rather than using global data.
22849 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22850 range data rather than using global data.
22851 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22852 pass it to children as needed.
22853 (struct directive::fmtresult): Similarly.
22854 (struct directive::set_width): Similarly.
22855 (struct directive::set_precision): Similarly.
22856 (format_integer, format_directive, parse_directive): Similarly.
22857 (format_none): Accept unnamed vr_values parameter.
22858 (format_percent, format_floating, format_character): Similarly.
22859 (format_string, format_plain): Similarly.
22860 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22861 the EVRP range analyzer for range data rather than using global data.
22862 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22863 gimple-ssa-evrp-analyze.h
22864 (class sprintf_dom_walker): Add after_dom_children member function.
22865 Add evrp_range_analyzer member.
22866 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22867 range analyzer as needed.
22868 (sprintf_dom_walker::after_dom_children): New member function.
22869 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22871 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22872 (evrp_range_analyzer::pop_to_marker): Likewise.
22874 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22876 PR tree-optimization/84419
22877 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22878 with the required type if its current type is compatible but
22881 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22883 PR middle-end/82004
22884 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22885 after vectorization.
22887 2018-02-20 Martin Liska <mliska@suse.cz>
22890 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22891 possible values if we don't have a hint.
22893 2018-02-20 Martin Liska <mliska@suse.cz>
22897 * final.c (shorten_branches): Build align_tab array with one
22899 * opts.c (finish_options): Add alignment option limit check.
22900 (MAX_CODE_ALIGN): Likewise.
22901 (MAX_CODE_ALIGN_VALUE): Likewise.
22902 * doc/invoke.texi: Document maximum allowed option value for
22903 all -falign-* options.
22905 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22908 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22909 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22910 * var-tracking.c (emit_note_insn_var_location): Remove all references
22911 to NOTE_INSN_CALL_ARG_LOCATION.
22912 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22913 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22914 Use copy_rtx_if_shared.
22915 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22916 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22917 (dwarf2out_var_location): Remove handling of
22918 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22920 * final.c (final_scan_insn): Remove all references to
22921 NOTE_INSN_CALL_ARG_LOCATION.
22922 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22923 before dumping final insns.
22924 * except.c (emit_note_eh_region_end): Remove all references to
22925 NOTE_INSN_CALL_ARG_LOCATION.
22926 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22927 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22928 * config/arc/arc.c (hwloop_optimize): Likewise.
22929 * config/arm/arm.c (create_fix_barrier): Likewise.
22930 * config/s390/s390.c (s390_chunkify_start): Likewise.
22931 * config/sh/sh.c (find_barrier): Likewise.
22932 * config/i386/i386.c (rest_of_insert_endbranch,
22933 ix86_seh_fixup_eh_fallthru): Likewise.
22934 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22935 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22936 * config/frv/frv.c (frv_function_prologue): Likewise.
22937 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22939 (note_outside_basic_block_p): Remove all references to
22940 NOTE_INSN_CALL_ARG_LOCATION.
22941 * gengtype.c (adjust_field_rtx_def): Likewise.
22942 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22944 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22945 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22948 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22951 PR tree-optimization/84452
22952 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22953 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22956 2018-02-19 Martin Liska <mliska@suse.cz>
22959 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22961 2018-02-19 Martin Liska <mliska@suse.cz>
22962 Richard Sandiford <richard.sandiford@linaro.org>
22964 PR tree-optimization/82491
22965 * gimple-fold.c (get_base_constructor): Make earlier bail out
22968 2018-02-19 Carl Love <cel@us.ibm.com>
22970 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22971 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22973 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22974 P8V_BUILTIN_VEC_NEG.
22976 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22978 * config/rl78/rl78.md (movdf): New define expand.
22980 2018-02-19 Martin Liska <mliska@suse.cz>
22983 * doc/invoke.texi: Fix typo.
22984 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22986 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22988 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22989 handle rs6000_single_float and rs6000_double_float specially for
22992 2018-02-16 Jeff Law <law@redhat.com>
22994 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22995 (pop_regs): Use it.
22997 2018-02-16 Jakub Jelinek <jakub@redhat.com>
23000 * ipa-inline.c (inline_small_functions): Fix a typo.
23002 2018-02-16 Nathan Sidwell <nathan@acm.org>
23004 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
23006 2018-02-16 Carl Love <cel@us.ibm.com>
23008 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
23009 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
23010 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
23011 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
23012 expansion to P8V_BUILTIN_VEC_FLOAT2.
23014 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
23016 PR rtl-optimization/70023
23017 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
23018 src_regno into account.
23020 2018-02-16 Carl Love <cel@us.ibm.com>
23022 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
23023 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
23024 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
23025 * config/rs6000/rs6000.c: Remove case statements for
23026 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
23027 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
23028 and P9V_BUILTIN_VEC_VINSERT4B.
23029 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
23030 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
23031 * config/rs6000/vsx.md:
23032 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
23035 2018-02-16 Carl Love <cel@us.ibm.com>
23037 * config/rs6000/altivec.h: Add builtin names vec_extract4b
23039 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
23041 * config/rs6000/rs6000-c.c: Add the definitions for
23042 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
23043 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
23044 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
23045 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
23046 definition for insert4b and define insn *insert3b_internal.
23047 * doc/extend.texi: Add documentation for vec_extract4b.
23049 2018-02-16 Nathan Sidwell <nathan@acm.org>
23051 * doc/extend.texi (Backwards Compatibility): Mention friend
23052 injection. Note for-scope is deprecated.
23053 * doc/invoke.texi (-ffriend-injection): Deprecate.
23055 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
23057 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
23058 that moved to I2, also allow destinations that are a paradoxical
23059 subreg (instead of a normal reg).
23061 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
23064 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
23067 2018-02-16 Richard Biener <rguenther@suse.de>
23069 PR tree-optimization/84037
23070 PR tree-optimization/84016
23072 * config/i386/i386.c (ix86_builtin_vectorization_cost):
23073 Adjust vec_construct for the fact we need additional higher latency
23074 128bit inserts for AVX256 and AVX512 vector builds.
23075 (ix86_add_stmt_cost): Scale vector construction cost for
23078 2018-02-16 Richard Biener <rguenther@suse.de>
23080 PR tree-optimization/84417
23081 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
23082 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
23083 (non_rewritable_lvalue_p): Likewise, use poly-ints.
23085 2018-02-16 Martin Liska <mliska@suse.cz>
23088 * internal-fn.def (ASAN_CHECK): Set proper flags.
23089 (ASAN_MARK): Likewise.
23091 2018-02-16 Julia Koval <julia.koval@intel.com>
23093 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
23094 from PTA_CANNONLAKE.
23096 2018-02-16 Jakub Jelinek <jakub@redhat.com>
23099 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
23100 Use ++iter rather than iter++ for std::list iterators.
23101 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
23102 defer deleting them until all nodes in the forest are processed. Do
23103 free even leaf nodes. Change to_process into auto_vec.
23106 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
23107 * vec.h (vec_default_construct): Use memset instead of placement new
23108 if BROKEN_VALUE_INITIALIZATION is defined.
23109 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
23110 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
23113 PR rtl-optimization/83723
23114 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
23115 * lra.c (lra_substitute_pseudo): Likewise. If true, use
23116 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
23118 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
23120 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
23122 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
23124 PR rtl-optimization/81443
23125 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
23126 from inner REGs to paradoxical SUBREGs.
23128 2018-02-16 Richard Biener <rguenther@suse.de>
23130 PR tree-optimization/84399
23131 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
23132 For operands we can analyze at their definition make sure we can
23133 analyze them at each use as well.
23135 2018-02-16 Richard Biener <rguenther@suse.de>
23137 PR tree-optimization/84190
23138 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
23139 volatile accesses if the decl isn't volatile.
23141 2018-02-15 Jason Merrill <jason@redhat.com>
23143 PR c++/84314 - ICE with templates and fastcall attribute.
23144 * attribs.c (build_type_attribute_qual_variant): Don't clobber
23145 TYPE_CANONICAL on an existing type.
23147 2018-02-15 Jakub Jelinek <jakub@redhat.com>
23149 PR tree-optimization/84383
23150 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
23151 dstoff nor call operand_equal_p if dstbase is NULL.
23153 PR tree-optimization/84334
23154 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
23155 also a CONSTANT_CLASS_P, punt.
23157 2018-02-14 Jim Wilson <jimw@sifive.com>
23159 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
23160 first SMALL_OPERAND check. New local min_second_step. Move assert
23161 to where locals are set. Add TARGET_RVC support.
23162 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
23164 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
23166 * doc/invoke.texi: Correct -Wformat-overflow code sample.
23168 2018-02-14 Martin Sebor <msebor@redhat.com>
23170 PR tree-optimization/83698
23171 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
23172 arrays constrain the offset range to their bounds.
23173 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
23174 (builtin_access::overlap): Avoid setting the size of overlap if it's
23176 (maybe_diag_overlap): Also consider arrays when deciding what values
23177 of offsets to include in diagnostics.
23179 2018-02-14 Martin Sebor <msebor@redhat.com>
23182 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
23183 that correspond to the kind of a declaration.
23185 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
23188 * config/pa/pa.md: Load address of PIC label using the linkage table
23189 if the label is nonlocal.
23191 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23193 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
23194 warning message if user requests -maltivec=be.
23195 * doc/invoke.texi: Document deprecation of -maltivec=be.
23197 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
23200 * config/rs6000/rs6000-c.c: Update definitions for
23201 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
23202 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
23204 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23207 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
23208 add _get_ssp intrinsics. Remove argument from
23209 __builtin_ia32_rdssp[d|q].
23210 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
23211 * config/i386/i386-builtin.def: Remove argument from
23212 __builtin_ia32_rdssp[d|q].
23213 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
23214 ix86_expand_special_args_builtin for _rdssp[d|q].
23215 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
23216 Clear register before usage.
23217 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
23218 Add documentation for new _get_ssp and _inc_ssp intrinsics.
23220 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
23222 PR tree-optimization/84357
23223 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
23224 operand 1 of an ARRAY_REF too.
23226 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
23229 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
23230 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23232 (set_of_reg): New struct.
23233 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
23234 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
23235 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23237 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
23238 Split into bitclr, bitset, bitinvert patterns if appropriate.
23239 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
23240 use rx_fuse_in_memory_bitop.
23241 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
23242 to named insn, correct maximum insn length.
23244 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
23247 * machmode.def: Define a complex mode for PARTIAL_INT.
23248 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
23250 * doc/rtl.texi: Document CSPImode.
23251 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
23253 (msp430_hard_regno_nregs_with_padding): Likewise.
23255 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
23258 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
23260 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23262 PR rtl-optimization/84169
23263 * combine.c (try_combine): New variable split_i2i3. Set it to true if
23264 we generated a parallel as new i3 and we split that to new i2 and i3
23265 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
23266 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
23267 those to i2, not i1. Partially rewrite this scan code.
23269 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23272 * stor-layout.c (place_field): For variable length fields, adjust
23273 offset_align afterwards not just based on the field's alignment,
23274 but also on the size.
23276 PR middle-end/84309
23277 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
23278 of exps and logs in the use_exp2 case.
23280 2018-02-13 Jeff Law <law@redhat.com>
23282 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
23283 entry for "vector".
23285 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
23288 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
23292 * common.opt (gas-loc-support, gas-locview-support): New.
23293 (ginline-points, ginternal-reset-location-views): New.
23294 * doc/invoke.texi: Document them. Use @itemx where intended.
23295 (gvariable-location-views): Adjust.
23296 * target.def (reset_location_view): New.
23297 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
23298 (TARGET_RESET_LOCATION_VIEW): New.
23299 * doc/tm.texi: Rebuilt.
23300 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
23301 (dwarf2out_default_as_locview_support): New.
23302 (output_asm_line_debug_info): Use option variables.
23303 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
23304 (output_loc_list): Likewise.
23305 (add_high_low_attributes): Check option variables.
23306 Don't output entry view attribute in strict mode.
23307 (gen_inlined_subroutine_die): Check option variables.
23308 (dwarf2out_inline_entry): Likewise.
23309 (init_sections_and_labels): Likewise.
23310 (dwarf2out_early_finish): Likewise.
23311 (maybe_reset_location_view): New, from...
23312 (dwarf2out_var_location): ... here. Call it.
23313 * debug.h (dwarf2out_default_as_loc_support): Declare.
23314 (dwarf2out_default_as_locview_support): Declare.
23315 * hooks.c (hook_int_rtx_insn_0): New.
23316 * hooks.h (hook_int_rtx_insn_0): Declare.
23317 * toplev.c (process_options): Take -gas-loc-support and
23318 -gas-locview-support from dwarf2out. Enable
23319 -gvariable-location-views by default only with locview
23320 assembler support. Enable -ginternal-reset-location-views by
23321 default only if the target defines the corresponding hook.
23322 Enable -ginline-points by default if location views are
23323 enabled; force it disabled if statement frontiers are
23325 * tree-inline.c (expand_call_inline): Check option variables.
23326 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
23328 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
23330 PR tree-optimization/84321
23331 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
23332 handling. Also check whether the anti-range contains any values
23333 that satisfy the mask; switch to a VR_RANGE if not.
23335 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
23338 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
23340 2018-02-13 Martin Jambor <mjambor@suse.cz>
23343 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
23344 of call statements, also set location of a load to a temporary.
23346 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
23348 * config/rl78/rl78.c (add_vector_labels): New function.
23349 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
23350 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
23351 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
23352 which checks that no arguments are passed.
23353 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
23354 * doc/extend.texi: Documentation for the new attribute.
23356 2018-02-13 Andreas Schwab <schwab@suse.de>
23358 * config/riscv/linux.h (CPP_SPEC): Define.
23360 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23363 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
23364 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
23365 OPTION_MASK_ISA_AES as first argument to def_builtin_const
23366 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
23367 instead of OPTION_MASK_ISA_PCLMUL as first argument to
23368 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
23369 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
23370 temporarily for AES and PCLMUL builtins.
23372 PR tree-optimization/84339
23373 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
23374 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
23377 PR middle-end/84309
23378 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
23379 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
23380 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
23382 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
23384 * omp-simd-clone.h: New file.
23385 * omp-simd-clone.c: Include omp-simd-clone.h.
23386 (expand_simd_clones): No longer static.
23387 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
23388 cgraph.h and omp-simd-clone.h.
23389 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
23390 (vect_recog_widen_shift_pattern): Formatting fix.
23391 (vect_pattern_recog_1): Don't check optab for calls.
23394 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
23395 operands[2] into a REG before using gen_lowpart on it.
23397 2018-02-12 Jeff Law <law@redhat.com>
23400 * config/sh/sh.c (find_barrier): Consider a sibling call
23403 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
23404 successfully back substituting a reg.
23406 2018-02-12 Richard Biener <rguenther@suse.de>
23408 PR tree-optimization/84037
23409 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
23410 parameter, move visited init to caller.
23411 (vect_slp_analyze_operations): Separate cost from validity
23412 check, initialize visited once for all instances.
23413 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
23415 * tree-vect-stmts.c (vect_model_simple_cost): Make early
23417 (vect_model_promotion_demotion_cost): Likewise.
23418 (vectorizable_bswap): Guard cost modeling with !slp_node
23419 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
23421 (vectorizable_call): Likewise.
23422 (vectorizable_conversion): Likewise.
23423 (vectorizable_assignment): Likewise.
23424 (vectorizable_shift): Likewise.
23425 (vectorizable_operation): Likewise.
23426 (vectorizable_store): Likewise.
23427 (vectorizable_load): Likewise.
23428 (vectorizable_condition): Likewise.
23429 (vectorizable_comparison): Likewise.
23431 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
23434 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
23435 (ASAN_MARK): Fix fnspec to account for return value, change pointer
23436 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
23438 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23440 PR middle-end/83665
23441 * params.def (inline-min-speedup): Increase from 8 to 15.
23442 (max-inline-insns-auto): Decrease from 40 to 30.
23443 * ipa-split.c (consider_split): Add some buffer for function to
23444 be considered inlining candidate.
23445 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
23448 2018-02-12 Richard Biener <rguenther@suse.de>
23450 PR tree-optimization/84037
23451 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
23452 matched stmts if we cannot swap the non-matched ones.
23454 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23456 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
23457 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
23458 _mm_maskz_scalef_round_ss): New intrinsics.
23459 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
23460 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
23461 __builtin_ia32_scalefss_round): Remove.
23462 (__builtin_ia32_scalefsd_mask_round,
23463 __builtin_ia32_scalefss_mask_round): New intrinsics.
23464 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
23465 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
23466 ((match_operand:VF_128 2 "<round_nimm_predicate>"
23467 "<round_constraint>")): Changed to ...
23468 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
23469 "<round_scalar_constraint>")): ... this.
23470 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
23471 %0, %1, %2<round_op3>}"): Changed to ...
23472 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
23473 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
23474 %2<round_scalar_mask_op3>}"): ... this.
23475 * config/i386/subst.md (round_scalar_nimm_predicate): New.
23477 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23479 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
23480 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
23481 (_mm_maskz_sqrt_round_ss): New intrinsics.
23482 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
23483 (__builtin_ia32_sqrtsd_mask_round)
23484 (__builtin_ia32_sqrtss_mask_round): New builtins.
23485 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
23486 (__builtin_ia32_sqrtss_round): Remove.
23487 (__builtin_ia32_sqrtsd_mask_round)
23488 (__builtin_ia32_sqrtss_mask_round): New builtins.
23489 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
23490 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
23491 ((match_operand:VF_128 1 "vector_operand"
23492 "xBm,<round_constraint>")): Changed to ...
23493 ((match_operand:VF_128 1 "vector_operand"
23494 "xBm,<round_scalar_constraint>")): ... this.
23495 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
23496 %0, %2, %<iptr>1<round_op3>}): Changed to ...
23497 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
23498 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
23499 %<iptr>1<round_scalar_mask_op3>}): ... this.
23500 ((set_attr "prefix" "<round_prefix>")): Changed to ...
23501 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
23503 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
23506 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
23507 Cast vec_cmpeq result to correct type.
23508 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
23509 Cast vec_cmpgt result to correct type.
23511 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
23513 * final.c (final_scan_insn_1): Renamed from...
23514 (final_scan_insn): ... this. New wrapper, to recover
23515 seen from the outermost call in recursive ones.
23516 * config/sparc/sparc.c (output_return): Drop seen from call.
23517 (output_sibcall): Likewise.
23518 * config/visium/visium.c (output_branch): Likewise.
23520 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
23522 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
23525 2018-02-10 Alan Modra <amodra@gmail.com>
23528 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
23529 Specify LR as an input.
23531 2018-02-10 Jakub Jelinek <jakub@redhat.com>
23534 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
23535 remove_member_access_dummy_vars): New functions.
23536 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
23537 lower_omp_1, execute_lower_omp): Use them.
23539 PR rtl-optimization/84308
23540 * shrink-wrap.c (spread_components): Release todo vector.
23542 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
23544 PR rtl-optimization/57193
23545 * ira-color.c (struct allocno_color_data): Add member
23546 conflict_allocno_hard_prefs.
23547 (update_conflict_allocno_hard_prefs): New.
23548 (bucket_allocno_compare_func): Add a preference based on
23549 conflict_allocno_hard_prefs.
23550 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23551 (color_allocnos): Remove a dead code. Initiate
23552 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23554 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23557 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23558 constraint from =wa to wa. Avoid a subreg on the output operand,
23559 instead use a pseudo and subreg it in a move.
23560 (p9_xxbrd_<mode>): Changed to ...
23561 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23562 (p9_xxbrd_v2df): New expander.
23563 (p9_xxbrw_<mode>): Changed to ...
23564 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23565 (p9_xxbrw_v4sf): New expander.
23567 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23569 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23571 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23574 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23575 multiply in 32-bit mode.
23576 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23577 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23580 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23582 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23583 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23584 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23585 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23587 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23590 * dwarf2out.c (is_trivial_indirect_ref): New function.
23591 (dwarf2out_late_global_decl): Do not generate a location
23592 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23593 and that are not defined in the current unit.
23595 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23597 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23598 instead of a libcall for UNORDERED.
23600 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23603 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23604 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23606 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23609 * config/s390/s390.c (s390_set_current_function): Invoke
23610 s390_indirect_branch_settings also if fndecl didn't change.
23612 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23614 * config/rs6000/rs6000.md (blockage): Set length to zero.
23616 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23618 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23620 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23623 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23624 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23628 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23629 PARALLEL incoming that failed vt_get_decl_and_offset check.
23631 PR middle-end/84237
23632 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23633 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23635 (get_variable_section): For decls in named .bss* sections pass true as
23636 second argument to bss_initializer_p.
23638 2018-02-09 Marek Polacek <polacek@redhat.com>
23639 Jakub Jelinek <jakub@redhat.com>
23642 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23643 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23644 Sync some changes from cxx_fold_indirect_ref.
23646 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23648 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23650 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23651 (BLOCK_INLINE_ENTRY_LABEL): New.
23652 (dwarf2out_var_location): Disregard inline entry markers.
23653 (inline_entry_data): New struct.
23654 (inline_entry_data_hasher): New hashtable type.
23655 (inline_entry_data_hasher::hash): New.
23656 (inline_entry_data_hasher::equal): New.
23657 (inline_entry_data_table): New variable.
23658 (add_high_low_attributes): Add DW_AT_entry_pc and
23659 DW_AT_GNU_entry_view attributes if a pending entry is found
23660 in inline_entry_data_table. Add old entry_pc attribute only
23661 if debug nonbinding markers are disabled.
23662 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23663 markers are enabled.
23664 (block_within_block_p, dwarf2out_inline_entry): New.
23665 (dwarf2out_finish): Check that no entries remained in
23666 inline_entry_data_table.
23667 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23668 (final_scan_insn, notice_source_line): Likewise.
23669 (rest_of_clean_state): Skip inline entry markers.
23670 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23672 * gimple.c (gimple_build_debug_inline_entry): New.
23673 * gimple.h (enum gimple_debug_subcode): Add
23674 GIMPLE_DEBUG_INLINE_ENTRY.
23675 (gimple_build_debug_inline_entry): Declare.
23676 (gimple_debug_inline_entry_p): New.
23677 (gimple_debug_nonbind_marker_p): Adjust.
23678 * insn-notes.def (INLINE_ENTRY): New.
23679 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23680 inline entry marker notes.
23681 (print_insn): Likewise.
23682 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23683 (INSN_DEBUG_MARKER_KIND): Likewise.
23684 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23685 * tree-inline.c (expand_call_inline): Build and insert
23686 debug_inline_entry stmt.
23687 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23688 inline entry blocks early, if nonbind markers are enabled.
23689 (dump_scope_block): Dump fragment info.
23690 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23691 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23692 (gimple_build_debug_inline_entry): New.
23693 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23694 Enable/disable inline entry points too.
23695 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23696 (DEBUG_INSN): Describe inline entry markers.
23698 * common.opt (gvariable-location-views): New.
23699 (gvariable-location-views=incompat5): New.
23700 * config.in: Rebuilt.
23701 * configure: Rebuilt.
23702 * configure.ac: Test assembler for view support.
23703 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23704 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23705 * dwarf2out.c (var_loc_view): New typedef.
23706 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23707 (dwarf2out_locviews_in_attribute): New.
23708 (dwarf2out_locviews_in_loclist): New.
23709 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23710 (enum dw_line_info_opcode): Add LI_adv_address.
23711 (struct dw_line_info_table): Add view.
23712 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23713 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23714 (zero_view_p): New variable.
23715 (ZERO_VIEW_P): New macro.
23716 (output_asm_line_debug_info): New.
23717 (struct var_loc_node): Add view.
23718 (add_AT_view_list, AT_loc_list): New.
23719 (add_var_loc_to_decl): Add view param. Test it against last.
23720 (new_loc_list): Add view params. Record them.
23721 (AT_loc_list_ptr): Handle loc and view lists.
23722 (view_list_to_loc_list_val_node): New.
23723 (print_dw_val): Handle dw_val_class_view_list.
23724 (size_of_die): Likewise.
23725 (value_format): Likewise.
23726 (loc_list_has_views): New.
23727 (gen_llsym): Set vl_symbol too.
23728 (maybe_gen_llsym, skip_loc_list_entry): New.
23729 (dwarf2out_maybe_output_loclist_view_pair): New.
23730 (output_loc_list): Output view list or entries too.
23731 (output_view_list_offset): New.
23732 (output_die): Handle dw_val_class_view_list.
23733 (output_dwarf_version): New.
23734 (output_compilation_unit_header): Use it.
23735 (output_skeleton_debug_sections): Likewise.
23736 (output_rnglists, output_line_info): Likewise.
23737 (output_pubnames, output_aranges): Update version comments.
23738 (output_one_line_info_table): Output view numbers in asm comments.
23739 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23740 Call maybe_gen_llsym.
23741 (loc_list_from_tree_1): Adjust.
23742 (add_AT_location_description): Create view list attribute if
23743 needed, check it's absent otherwise.
23744 (convert_cfa_to_fb_loc_list): Adjust.
23745 (maybe_emit_file): Call output_asm_line_debug_info for test.
23746 (dwarf2out_var_location): Reset views as needed. Precompute
23747 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23748 attribute. Set view.
23749 (new_line_info_table): Reset next view.
23750 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23751 (dwarf2out_source_line): Likewise. Output view resets and labels to
23752 the assembler, or select appropriate line info opcodes.
23753 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23754 (optimize_string_length): Catch it. Adjust.
23755 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23756 dw_val_class_view_list, and remove it if no longer needed.
23757 (hash_loc_list): Hash view numbers.
23758 (loc_list_hasher::equal): Compare them.
23759 (optimize_location_lists): Check whether a view list symbol is
23760 needed, and whether the locview attribute is present, and
23761 whether they match. Remove the locview attribute if no longer
23763 (index_location_lists): Call skip_loc_list_entry for test.
23764 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23765 Use output_dwarf_version.
23766 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23767 (struct dw_val_node): Add val_view_list.
23768 * final.c (SEEN_NEXT_VIEW): New.
23769 (set_next_view_needed): New.
23770 (clear_next_view_needed): New.
23771 (maybe_output_next_view): New.
23772 (final_start_function): Rename to...
23773 (final_start_function_1): ... this. Take pointer to FIRST,
23774 add SEEN parameter. Emit param bindings in the initial view.
23775 (final_start_function): Reintroduce SEEN-less interface.
23776 (final): Rename to...
23777 (final_1): ... this. Take SEEN parameter. Output final pending
23778 next view at the end.
23779 (final): Reintroduce seen-less interface.
23780 (final_scan_insn): Output pending next view before switching
23781 sections or ending a block. Mark the next view as needed when
23782 outputting variable locations. Notify debug backend of section
23783 changes, and of location view changes.
23784 (rest_of_handle_final): Adjust.
23785 * toplev.c (process_options): Autodetect value for debug variable
23786 location views option. Warn on incompat5 without -gdwarf-5.
23787 * doc/invoke.texi (gvariable-location-views): New.
23788 (gvariable-location-views=incompat5): New.
23789 (gno-variable-location-views): New.
23791 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23793 PR tree-optimization/84136
23794 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23795 that the result of find_edge is non-NULL.
23797 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23800 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23801 storing integer register in SImode. Fix cost of 256 and 512
23802 byte aligned SSE register store.
23804 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23806 * config/i386/i386.c (ix86_multiplication_cost): Fix
23807 multiplication cost for TARGET_AVX512DQ.
23809 2018-02-08 Marek Polacek <polacek@redhat.com>
23811 PR tree-optimization/84238
23812 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23815 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23817 PR tree-optimization/84265
23818 * tree-vect-stmts.c (vectorizable_store): Don't treat
23819 VMAT_CONTIGUOUS accesses as grouped.
23820 (vectorizable_load): Likewise.
23822 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23824 PR tree-optimization/81635
23825 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23826 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23827 (test_round_for_mask): New functions.
23828 (wide_int_cc_tests): Call test_round_for_mask.
23829 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23830 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23831 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23832 range returned by get_range_info.
23834 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23837 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23838 * symtab.c: Include builtins.h
23839 (symtab_node::output_to_lto_symbol_table_p): Move here
23840 from lto-streamer-out.c:output_symbol_p.
23841 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23842 (output_symbol_p): Move all logic to symtab.c
23843 (produce_symtab): Update.
23845 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23847 * config/s390/s390-opts.h (enum indirect_branch): Define.
23848 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23849 (s390_indirect_branch_via_thunk)
23850 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23851 (enum s390_indirect_branch_type): Define.
23852 * config/s390/s390.c (struct s390_frame_layout, struct
23853 machine_function): Remove.
23854 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23855 (indirect_branch_table_label_no, indirect_branch_table_name):
23857 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23858 (enum s390_indirect_branch_option): Define.
23859 (s390_return_addr_from_memory): New function.
23860 (s390_handle_string_attribute): New function.
23861 (s390_attribute_table): Add new attribute handler.
23862 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23863 (s390_indirect_branch_via_thunk): New function.
23864 (s390_indirect_branch_via_inline_thunk): New function.
23865 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23866 sibling call optimization for non z10 compiles.
23867 (s390_emit_call): Force indirect branch target to be a single
23868 register. Add r1 clobber for non-z10 compiles.
23869 (s390_emit_epilogue): Emit return jump via return_use expander.
23870 (s390_reorg): Handle JUMP_INSNs as execute targets.
23871 (s390_option_override_internal): Perform validity checks for the
23872 new command line options.
23873 (s390_indirect_branch_attrvalue): New function.
23874 (s390_indirect_branch_settings): New function.
23875 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23876 (s390_output_indirect_thunk_function): New function.
23877 (s390_code_end): Implement target hook.
23878 (s390_case_values_threshold): Implement target hook.
23879 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23881 * config/s390/s390.h (struct s390_frame_layout)
23882 (struct machine_function): Move here from s390.c.
23883 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23884 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23885 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23886 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23887 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23888 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23889 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23890 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23891 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23892 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23893 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23894 (mnemonic attribute): Add values which aren't recognized
23896 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23897 pattern for branch conversion. Fix mnemonic attribute.
23898 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23899 indirect branch via thunk if requested.
23900 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23901 ("*indirect_jump"): Disable for branch conversion using out of
23903 ("indirect_jump_via_thunk<mode>_z10")
23904 ("indirect_jump_via_thunk<mode>")
23905 ("indirect_jump_via_inlinethunk<mode>_z10")
23906 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23907 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23908 ("casesi_jump_via_inlinethunk<mode>_z10")
23909 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23910 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23911 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23912 ("*indirect2_jump"): Disable for branch conversion.
23913 ("casesi_jump"): Turn into expander and expand patterns for branch
23915 ("return_use"): New expander.
23916 ("*return"): Emit return via thunk and rename it to ...
23917 ("*return<mode>"): ... this one.
23918 * config/s390/s390.opt: Add new options and and enum for the
23921 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23923 * lra-constraints.c (match_reload): Unconditionally use
23924 gen_lowpart_SUBREG, rather than selecting between that
23925 and equivalent gen_rtx_SUBREG code.
23927 2018-02-08 Richard Biener <rguenther@suse.de>
23929 PR tree-optimization/84233
23930 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23931 changed flag instead of boguously re-using phi_inserted.
23933 2018-02-08 Martin Jambor <mjambor@suse.cz>
23935 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23936 static local variables.
23938 2018-02-08 Richard Biener <rguenther@suse.de>
23940 PR tree-optimization/84278
23941 * tree-vect-stmts.c (vectorizable_store): When looking for
23942 smaller vector types to perform grouped strided loads/stores
23943 make sure the mode is supported by the target.
23944 (vectorizable_load): Likewise.
23946 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23948 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23949 Increase LDP/STP opportunities by adding adjacent callee-saves.
23951 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23953 PR rtl-optimization/84068
23954 PR rtl-optimization/83459
23955 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23957 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23959 PR tree-optimization/84224
23960 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23961 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23962 non-zero arguments.
23964 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23967 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23968 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23969 count, remove one USE.
23971 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23973 * doc/install.texi (Configuration): Document the
23974 --with-long-double-format={ibm,ieee} PowerPC configuration
23978 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23979 Convert from define_expand to be define_insn_and_split. Rework
23980 float/double/_Float128 conversions to QI/HI/SImode to work with
23981 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23982 conversions to QI/HImode types did a store and then a load to
23983 truncate the value. For conversions to VSX registers, don't split
23984 the insn, instead emit the code directly. Use the code iterator
23985 any_fix to combine signed and unsigned conversions.
23986 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23987 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23988 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23989 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23990 (fix_<mode>di2_hw): Likewise.
23991 (fixuns_<mode>di2_hw): Likewise.
23992 (fix_<mode>si2_hw): Likewise.
23993 (fixuns_<mode>si2_hw): Likewise.
23994 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23995 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23996 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23997 fix<uns>_trunc<SFDF:mode>si2_p8.
23998 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23999 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
24000 (fix<uns>_<mode>_mem): Likewise.
24001 (fctiw<u>z_<mode>_mem): Likewise.
24002 (fix<uns>_<mode>_mem): Likewise.
24003 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
24004 the register allocator from doing a direct move to the GPRs to do
24005 a store, and instead use the ISA 3.0 store byte/half-word from
24006 vector register instruction. For IEEE 128-bit floating point,
24007 also optimize stores of 32-bit ints.
24008 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
24010 2018-02-07 Alan Hayward <alan.hayward@arm.com>
24012 * genextract.c (push_pathstr_operand): New function to support
24014 (walk_rtx): Call push_pathstr_operand.
24015 (print_path): Support [a-zA-Z].
24017 2018-02-07 Richard Biener <rguenther@suse.de>
24019 PR tree-optimization/84037
24020 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
24021 (cse_and_gimplify_to_preheader): Declare.
24022 (vect_get_place_in_interleaving_chain): Likewise.
24023 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
24025 (_loop_vec_info::~_loop_vec_info): Delete it.
24026 (cse_and_gimplify_to_preheader): New function.
24027 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
24028 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
24029 (vectorizable_load): Likewise. For grouped stores always base
24030 the IV on the first element.
24031 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
24032 condition before gimplifying.
24034 2018-02-07 Jakub Jelinek <jakub@redhat.com>
24036 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
24037 *DIV_EXPR and *MOD_EXPR.
24039 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
24042 * config/i386/i386.c (ix86_option_override_internal): Mask out
24043 the CF_SET bit when checking -fcf-protection.
24045 2018-02-07 Tom de Vries <tom@codesourcery.com>
24048 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
24051 2018-02-07 Richard Biener <rguenther@suse.de>
24053 PR tree-optimization/84204
24054 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
24057 PR tree-optimization/84205
24058 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
24059 special-case isl_ast_op_zdiv_r.
24061 PR tree-optimization/84223
24062 * graphite-scop-detection.c (gather_bbs::before_dom_children):
24063 Only add conditions from within the region.
24064 (gather_bbs::after_dom_children): Adjust.
24066 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
24069 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
24070 * config/avr/avr.md: Only post-reload split REG-REG moves if
24071 either register is GENERAL_REG_P.
24073 2018-02-07 Jakub Jelinek <jakub@redhat.com>
24075 PR tree-optimization/84235
24076 * tree-ssa-scopedtables.c
24077 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
24078 if the subtraction is performed in floating point type where NaNs are
24079 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
24080 build 1. Formatting fix.
24082 2018-02-06 Jakub Jelinek <jakub@redhat.com>
24085 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
24086 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
24087 and skip it regardless of bb boundaries. Use CALL_P macro,
24088 don't test INSN_P (insn) together with CALL_P or JUMP_P check
24089 unnecessarily, formatting fix.
24091 2018-02-06 Michael Collison <michael.collison@arm.com>
24093 * config/arm/thumb2.md:
24094 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
24095 (*thumb_mov_notscc): Ditto.
24097 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
24100 * config/rs6000/rs6000.md (su code attribute): Use "u" for
24101 unsigned_fix, not "s".
24103 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24105 * configure.ac (gcc_fn_eh_frame_ro): New function.
24106 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
24107 correct .eh_frame permissions.
24108 * configure: Regenerate.
24110 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
24112 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
24113 irrelevant options.
24115 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24117 * config/rs6000/rs6000.c (rs6000_option_override_internal):
24118 Display warning message for -mno-speculate-indirect-jumps.
24120 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
24122 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
24124 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
24126 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
24128 PR tree-optimization/84225
24129 * tree-eh.c (find_trapping_overflow): Only call
24130 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
24132 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24135 * config/i386/i386.c: Reimplement the check of possible options
24136 -mibt/-mshstk conbination. Change error messages.
24137 * doc/invoke.texi: Fix a typo: remove extra '='.
24139 2018-02-06 Marek Polacek <polacek@redhat.com>
24141 PR tree-optimization/84228
24142 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
24144 2018-02-06 Tamar Christina <tamar.christina@arm.com>
24147 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
24148 emitted arch directives.
24149 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
24150 __ARM_FEATURE_COPROC before changing architectures.
24152 2018-02-06 Richard Biener <rguenther@suse.de>
24154 * config/i386/i386.c (print_reg): Fix typo.
24155 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
24157 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
24159 * configure: Regenerate.
24161 2018-02-05 Martin Sebor <msebor@redhat.com>
24163 PR tree-optimization/83369
24164 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
24167 2018-02-05 Martin Liska <mliska@suse.cz>
24169 * doc/invoke.texi: Cherry-pick upstream r323995.
24171 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
24173 * ira.c (ira_init_register_move_cost): Adjust comment.
24175 2018-02-05 Martin Liska <mliska@suse.cz>
24177 PR gcov-profile/84137
24178 * doc/gcov.texi: Fix typo in documentation.
24180 2018-02-05 Martin Liska <mliska@suse.cz>
24182 PR gcov-profile/83879
24183 * doc/gcov.texi: Document necessity of --dynamic-list-data when
24184 using dlopen functionality.
24186 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
24188 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
24189 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
24190 _mm_maskz_range_ss, _mm_mask_range_round_ss,
24191 _mm_maskz_range_round_ss): New intrinsics.
24192 (__builtin_ia32_rangesd128_round)
24193 (__builtin_ia32_rangess128_round): Remove.
24194 (__builtin_ia32_rangesd128_mask_round,
24195 __builtin_ia32_rangess128_mask_round): New builtins.
24196 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
24197 __builtin_ia32_rangess128_round): Remove.
24198 (__builtin_ia32_rangesd128_mask_round,
24199 __builtin_ia32_rangess128_mask_round): New builtins.
24200 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
24201 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
24202 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
24203 "<round_saeonly_constraint>")): Changed to ...
24204 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
24205 "<round_saeonly_scalar_constraint>")): ... this.
24206 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24207 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
24208 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
24209 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
24210 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
24212 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
24214 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
24216 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
24217 Remove all values except native, 8540 and 8548.
24219 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
24221 * config/i386/i386.c (ix86_output_function_return): Pass
24222 INVALID_REGNUM, instead of -1, as invalid register number to
24223 indirect_thunk_name and output_indirect_thunk.
24225 2018-02-02 Julia Koval <julia.koval@intel.com>
24227 * config.gcc: Add -march=icelake.
24228 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
24229 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
24230 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
24231 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
24232 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
24233 (processor_target_table): Add icelake.
24234 (ix86_option_override_internal): Handle new PTAs.
24235 (get_builtin_code_for_version): Handle icelake.
24236 (M_INTEL_COREI7_ICELAKE): New.
24237 (fold_builtin_cpu): Handle icelake.
24238 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
24239 * doc/invoke.texi: Add -march=icelake.
24241 2018-02-02 Julia Koval <julia.koval@intel.com>
24243 * config/i386/i386.c (ix86_option_override_internal): Change flags type
24244 to wide_int_bitmask.
24245 * wide-int-bitmask.h: New.
24247 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24250 * config/i386/i386.md: Replace Pmode with word_mode in
24251 builtin_setjmp_setup and builtin_longjmp to support x32.
24253 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
24257 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
24259 (rs6000_supported_cpu_names): New static variable.
24260 (linux_cpu_translation_table): Likewise.
24261 (elf_platform) <cpu>: Define new static variable and use it.
24262 Translate kernel AT_PLATFORM name to canonical name if needed.
24263 Error if platform name is unknown.
24265 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
24268 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
24270 2018-02-01 Jeff Law <law@redhat.com>
24273 * config/i386/i386.c (release_scratch_register_on_entry): Add new
24274 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
24275 the scratch if RELEASE_VIA_POP is false.
24276 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
24277 If we have to save a temporary register, decrement SIZE appropriately.
24278 Pass new arguments to release_scratch_register_on_entry.
24279 (ix86_adjust_stack_and_probe): Likewise.
24280 (ix86_emit_probe_stack_range): Pass new arguments to
24281 release_scratch_register_on_entry.
24283 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
24285 PR rtl-optimization/84157
24286 * combine.c (change_zero_ext): Use REG_P predicate in
24287 front of HARD_REGISTER_P predicate.
24289 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
24291 * config/avr/avr.c (avr_option_override): Move disabling of
24292 -fdelete-null-pointer-checks to...
24293 * common/config/avr/avr-common.c (avr_option_optimization_table):
24296 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24298 PR tree-optimization/81635
24299 * tree-data-ref.c (split_constant_offset_1): For types that
24300 wrap on overflow, try to use range info to prove that wrapping
24303 2018-02-01 Renlin Li <renlin.li@arm.com>
24306 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
24307 TAILCALL_ADDR_REGS.
24308 (aarch64_register_move_cost): Likewise.
24309 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
24310 TAILCALL_ADDR_REGS.
24311 (REG_CLASS_NAMES): Likewise.
24312 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
24313 TAILCALL_ADDR_REGS. Remove IP registers.
24314 * config/aarch64/aarch64.md (Ucs): Update register constraint.
24316 2018-02-01 Richard Biener <rguenther@suse.de>
24318 * domwalk.h (dom_walker::dom_walker): Add additional constructor
24319 for specifying RPO order and allow NULL for that.
24320 * domwalk.c (dom_walker::dom_walker): Likewise.
24321 (dom_walker::walk): Handle NULL RPO order.
24322 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
24324 (rewrite_update_dom_walker): Likewise.
24325 (mark_def_dom_walker): Likewise.
24327 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24329 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
24330 (aarch64_maybe_expand_sve_subreg_move): Declare.
24331 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
24332 * config/aarch64/predicates.md (aarch64_any_register_operand): New
24334 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
24335 that are semantically a reverse operation.
24336 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
24337 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
24338 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
24340 (aarch64_can_change_mode_class): For big-endian, forbid changes
24341 between two SVE modes if they have different element sizes.
24343 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24345 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
24346 the TImode handling for big-endian targets.
24348 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24350 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
24351 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
24353 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
24354 Remove BSWAP handing for big-endian targets and use the form of
24355 LD1RQ appropariate for the mode.
24357 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24359 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
24360 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
24361 duplicated element.
24363 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24366 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
24367 check for operands that need to go through aarch64_sve_reload_be.
24369 2018-02-01 Jakub Jelinek <jakub@redhat.com>
24371 PR tree-optimization/81661
24372 PR tree-optimization/84117
24373 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
24374 * tree-eh.c: Include gimplify.h.
24375 (find_trapping_overflow, replace_trapping_overflow,
24376 rewrite_to_non_trapping_overflow): New functions.
24377 * tree-vect-loop.c: Include tree-eh.h.
24378 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
24379 * tree-data-ref.c: Include tree-eh.h.
24380 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
24382 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
24384 PR rtl-optimization/84123
24385 * combine.c (change_zero_ext): Check if hard register satisfies
24386 can_change_dest_mode before calling gen_lowpart_SUBREG.
24388 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
24391 * ira.c (ira_init_register_move_cost): Remove assert.
24393 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24395 PR rtl-optimization/84071
24396 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
24397 * doc/tm.texi: Regenerate.
24399 2018-01-31 Richard Biener <rguenther@suse.de>
24401 PR tree-optimization/84132
24402 * tree-data-ref.c (analyze_miv_subscript): Properly
24403 check whether evolution_function_is_affine_multivariate_p
24404 before calling gcd_of_steps_may_divide_p.
24406 2018-01-31 Julia Koval <julia.koval@intel.com>
24409 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24410 * config/i386/i386.md (rdpid_rex64) New.
24411 (rdpid): Make 32bit only.
24413 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
24416 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
24417 an IDENTIFIER_NODE for FUNCTION_TYPE's.
24419 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24422 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
24424 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
24426 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24428 PR rtl-optimization/84071
24429 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
24430 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
24432 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24434 * config/arc/arc.c (arc_handle_aux_attribute): New function.
24435 (arc_attribute_table): Add 'aux' attribute.
24436 (arc_in_small_data_p): Consider aux like variables.
24437 (arc_is_aux_reg_p): New function.
24438 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
24439 (arc_get_aux_arg): New function.
24440 (prepare_move_operands): Handle aux-register access.
24441 (arc_handle_aux_attribute): New function.
24442 * doc/extend.texi (ARC Variable attributes): Add subsection.
24444 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24446 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
24447 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
24448 (arc_attribute_table): Add 'uncached' attribute.
24449 (arc_print_operand): Print '.di' flag for uncached memory
24451 (arc_in_small_data_p): Do not consider for small data the uncached
24453 (arc_is_uncached_mem_p): New function.
24454 * config/arc/predicates.md (compact_store_memory_operand): Check
24455 for uncached memory accesses.
24456 (nonvol_nonimm_operand): Likewise.
24457 * doc/extend.texi (ARC Type Attribute): New subsection.
24459 2018-01-31 Jakub Jelinek <jakub@redhat.com>
24462 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
24463 falign-loops=): Add Optimization flag.
24465 2018-01-30 Jeff Law <law@redhat.com>
24468 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
24469 INT_REGISTERS_SAVED. Check it prior to calling
24470 get_scratch_register_on_entry.
24471 (ix86_adjust_stack_and_probe): Similarly.
24472 (ix86_emit_probe_stack_range): Similarly.
24473 (ix86_expand_prologue): Corresponding changes.
24475 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24478 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
24479 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
24481 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
24484 * lra-constraints.c (curr_insn_transform): Process AND in the
24487 2018-01-30 Jakub Jelinek <jakub@redhat.com>
24489 PR rtl-optimization/83986
24490 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
24491 dependence against last_pending_memory_flush in addition to
24492 pending_jump_insns.
24494 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
24496 PR tree-optimization/81611
24497 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
24500 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24503 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
24506 2018-01-30 Richard Biener <rguenther@suse.de>
24507 Jakub Jelinek <jakub@redhat.com>
24509 PR tree-optimization/84111
24510 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
24511 inner loops added during recursion, as they don't have up-to-date
24514 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24517 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
24518 (can_inline_edge_by_limits_p): ... here.
24519 (can_early_inline_edge_p, check_callers,
24520 update_caller_keys, update_callee_keys, recursive_inlining,
24521 add_new_edges_to_heap, speculation_useful_p,
24522 inline_small_functions,
24523 inline_small_functions, flatten_function,
24524 inline_to_all_callers_1): Update.
24526 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24528 * profile-count.c (profile_count::combine_with_ipa_count): Handle
24531 2018-01-30 Richard Biener <rguenther@suse.de>
24533 PR tree-optimization/83008
24534 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
24535 invariant and constant vector uses in stmts when they need
24536 more than one stmt.
24538 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24541 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
24542 * configure: Regenerate.
24544 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24546 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
24548 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
24549 Use gen_rtx_REG rather than gen_lowpart.
24551 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24553 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24554 rather than 0 when creating partial subregs.
24556 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24558 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24561 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24564 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24565 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24566 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24567 flags. This restores the settings used before the 2017-07-24.
24568 Turning off pre increment/decrement/modify allows IVOPTS to
24569 optimize DF/SF loops where the index is an int.
24571 2018-01-29 Richard Biener <rguenther@suse.de>
24572 Kelvin Nilsen <kelvin@gcc.gnu.org>
24575 * tree-vect-stmts.c (vectorizable_call): Don't call
24576 targetm.vectorize_builtin_md_vectorized_function if callee is
24579 2018-01-22 Carl Love <cel@us.ibm.com>
24581 * doc/extend.tex: Fix typo in second arg in
24582 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24584 2018-01-29 Richard Biener <rguenther@suse.de>
24586 PR tree-optimization/84086
24587 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24588 (flush_ssaname_freelist): When SSA names were released reset
24589 the SCEV hash table.
24591 2018-01-29 Richard Biener <rguenther@suse.de>
24593 PR tree-optimization/84057
24594 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24595 removed paths when removing edges.
24597 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24599 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24600 -mfunction-return=@var{choice}.
24602 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24604 PR diagnostic/84034
24605 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24606 Handle CR like TAB.
24607 (layout::print_source_line): Likewise.
24608 (test_get_line_width_without_trailing_whitespace): Add test cases.
24610 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24612 PR middle-end/84040
24613 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24616 2018-01-26 Jim Wilson <jimw@sifive.com>
24618 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24620 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24623 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24625 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24626 and CMP + SUB-immediate -> SUBS.
24628 2018-01-26 Martin Sebor <msebor@redhat.com>
24630 PR tree-optimization/83896
24631 * tree-ssa-strlen.c (get_string_len): Rename...
24632 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24633 Avoid assuming length is constant.
24634 (handle_char_store): Use HOST_WIDE_INT for string length.
24636 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24639 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24640 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24642 2018-01-26 Richard Biener <rguenther@suse.de>
24644 PR rtl-optimization/84003
24645 * dse.c (record_store): Only record redundant stores when
24646 the earlier store aliases at least all accesses the later one does.
24648 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24650 PR rtl-optimization/83985
24651 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24652 REG_CFA_RESTORE insns.
24653 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24654 REG_CFA_RESTORE insns here.
24657 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24658 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24660 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24662 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24663 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24664 (arc_init): Likewise.
24665 (arc_override_options): Likewise.
24666 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24668 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24670 * config/arc/arc.h (TARGET_DBNZ): Define.
24671 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24672 properly set the tune attribute.
24673 (dbnz): Use TARGET_DBNZ guard.
24674 * config/arc/arc.opt (mtune): Add core3 option.
24676 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24678 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24679 recognize new pic like addresses.
24680 (arc_delegitimize_address): Clean up.
24682 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24684 * config/arc/arc-arches.def: Option mrf16 valid for all
24686 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24687 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24688 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24689 * config/arc/arc-tables.opt: Regenerate.
24690 * config/arc/arc.c (arc_conditional_register_usage): Handle
24691 reduced register file case.
24692 (arc_file_start): Set must have build attributes.
24693 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24694 mrf16 option value.
24695 * config/arc/arc.opt (mrf16): Add new option.
24696 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24697 * config/arc/genmultilib.awk: Handle new mrf16 option.
24698 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24699 * config/arc/t-multilib: Regenerate.
24700 * doc/invoke.texi (ARC Options): Document mrf16 option.
24702 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24704 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24705 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24706 (arc_attribute_table): Add 'secure_call' attribute.
24707 (arc_print_operand): Print secure call operand.
24708 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24710 (arc_is_secure_call_p): New function. * config/arc/arc.md
24711 (call_i): Add support for sjli instruction.
24712 (call_value_i): Likewise.
24713 * config/arc/constraints.md (Csc): New constraint.
24715 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24716 John Eric Martin <John.Martin@emmicro-us.com>
24718 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24719 * config/arc/arc.c (_arc_jli_section): New struct.
24720 (arc_jli_section): New type.
24721 (rc_jli_sections): New static variable.
24722 (arc_handle_jli_attribute): New function.
24723 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24724 (arc_file_end): New function.
24725 (TARGET_ASM_FILE_END): Define.
24726 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24727 (arc_add_jli_section): New function.
24728 (jli_call_scan): Likewise.
24729 (arc_reorg): Call jli_call_scan.
24730 (arc_output_addsi): Remove 'S' from printing asm operand.
24731 (arc_is_jli_call_p): New function.
24732 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24734 (movhi_insn): Likewise.
24735 (movsi_insn): Likewise.
24736 (movsi_set_cc_insn): Likewise.
24737 (loadqi_update): Likewise.
24738 (load_zeroextendqisi_update): Likewise.
24739 (load_signextendqisi_update): Likewise.
24740 (loadhi_update): Likewise.
24741 (load_zeroextendhisi_update): Likewise.
24742 (load_signextendhisi_update): Likewise.
24743 (loadsi_update): Likewise.
24744 (loadsf_update): Likewise.
24745 (movsicc_insn): Likewise.
24746 (bset_insn): Likewise.
24747 (bxor_insn): Likewise.
24748 (bclr_insn): Likewise.
24749 (bmsk_insn): Likewise.
24750 (bicsi3_insn): Likewise.
24751 (cmpsi_cc_c_insn): Likewise.
24752 (movsi_ne): Likewise.
24753 (movsi_cond_exec): Likewise.
24754 (clrsbsi2): Likewise.
24755 (norm_f): Likewise.
24765 (call_i): Remove 'S' asm letter, add jli instruction.
24766 (call_value_i): Likewise.
24767 * config/arc/arc.op (mjli-always): New option.
24768 * config/arc/constraints.md (Cji): New constraint.
24769 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24771 (subsf3_fpx): Likewise.
24772 (mulsf3_fpx): Likewise.
24773 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24775 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24776 function attrbutes.
24777 * doc/invoke.texi (ARC): Document mjli-always option.
24779 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24781 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24782 avoid addition with 0 and use incw and decw where possible.
24784 2018-01-26 Richard Biener <rguenther@suse.de>
24786 PR tree-optimization/81082
24787 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24788 association if it requires casting to unsigned.
24789 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24790 from fold_plusminus_mult_expr to catch important cases late when
24791 range info is available.
24793 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24795 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24796 * configure.ac (hidden_linkonce): New test.
24797 * configure: Regenerate.
24798 * config.in: Regenerate.
24800 2018-01-26 Julia Koval <julia.koval@intel.com>
24802 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24803 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24804 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24805 _mm_mask_bitshuffle_epi64_mask): Fix type.
24806 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24807 USI_FTYPE_V4DI_V4DI_USI): Remove.
24808 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24809 __builtin_ia32_vpshufbitqmb256_mask,
24810 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24811 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24812 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24814 2018-01-26 Alan Modra <amodra@gmail.com>
24817 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24818 UNSPEC_VBPERMQ. Sort other unspecs.
24820 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24822 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24824 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24826 PR middle-end/83055
24827 * predict.c (drop_profile): Do not push/pop cfun; update also
24829 (handle_missing_profiles): Fix logic looking for zero profiles.
24831 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24833 PR middle-end/83977
24834 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24835 on functions with #pragma omp declare simd or functions with simd
24837 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24838 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24839 Remove trailing \n from warning_at calls.
24841 2018-01-25 Tom de Vries <tom@codesourcery.com>
24844 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24845 for neutered workers.
24847 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24850 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24851 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24853 2018-01-24 Jeff Law <law@redhat.com>
24856 * i386.c (get_probe_interval): Move to earlier point.
24857 (ix86_compute_frame_layout): If -fstack-clash-protection and
24858 the frame is larger than the probe interval, then use pushes
24859 to save registers rather than reg->mem moves.
24860 (ix86_expand_prologue): Remove conditional for int_registers_saved
24863 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24866 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24867 min/max for never referenced object.
24869 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24871 PR middle-end/83977
24872 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24874 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24875 attributes from DECL_ATTRIBUTES (decl) without affecting
24876 DECL_ATTRIBUTES (current_function_decl).
24877 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24878 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24880 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24882 PR tree-optimization/83979
24883 * fold-const.c (fold_comparison): Use constant_boolean_node
24884 instead of boolean_{true,false}_node.
24886 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24888 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24891 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24893 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24894 Simplify the clause that sets the length attribute.
24895 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24896 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24897 clause that sets the length attribute.
24898 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24900 2018-01-24 Tom de Vries <tom@codesourcery.com>
24903 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24904 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24905 Add strict parameter.
24906 (prevent_branch_around_nothing): Insert dummy insn between branch to
24907 label and label with no ptx insn inbetween.
24908 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24910 2018-01-24 Tom de Vries <tom@codesourcery.com>
24913 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24914 for neutered threads in warp.
24915 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24917 2018-01-24 Richard Biener <rguenther@suse.de>
24919 PR tree-optimization/83176
24920 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24923 2018-01-24 Richard Biener <rguenther@suse.de>
24925 PR tree-optimization/82819
24926 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24927 code generating pluses that are no-ops in the target precision.
24929 2018-01-24 Richard Biener <rguenther@suse.de>
24931 PR middle-end/84000
24932 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24934 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24936 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24937 to merge probabilities.
24938 * predict.c (probably_never_executed): Also mark as cold functions
24939 with global 0 profile and guessed local profile.
24940 * profile-count.c (profile_probability::combine_with_count): New
24942 * profile-count.h (profile_probability::operator*,
24943 profile_probability::operator*=, profile_probability::operator/,
24944 profile_probability::operator/=): Reduce precision to adjusted
24945 and set value to guessed on contradictory divisions.
24946 (profile_probability::combine_with_freq): Remove.
24947 (profile_probability::combine_wiht_count): Declare.
24948 (profile_count::force_nonzero):: Set to adjusted.
24949 (profile_count::probability_in):: Set quality to adjusted.
24950 * tree-ssa-tail-merge.c (replace_block_by): Use
24951 combine_with_count.
24953 2018-01-23 Andrew Waterman <andrew@sifive.com>
24954 Jim Wilson <jimw@sifive.com>
24956 * config/riscv/riscv.c (riscv_stack_boundary): New.
24957 (riscv_option_override): Set riscv_stack_boundary. Handle
24958 riscv_preferred_stack_boundary_arg.
24959 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24960 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24961 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24962 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24963 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24964 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24966 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24969 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24970 of struct ix86_frame.
24971 (ix86_expand_epilogue): Likewise. Add a local variable for
24972 the reg_save_offset field in struct ix86_frame.
24974 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24976 PR tree-optimization/82604
24977 * tree-loop-distribution.c (enum partition_kind): New enum item
24978 PKIND_PARTIAL_MEMSET.
24979 (partition_builtin_p): Support above new enum item.
24980 (generate_code_for_partition): Ditto.
24981 (compute_access_range): Differentiate cases that equality can be
24982 proven at all loops, the innermost loops or no loops.
24983 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24984 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24985 (finalize_partitions, distribute_loop): Don't fuse partition of
24986 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24987 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24988 parloop is enabled.
24990 2018-01-23 Martin Liska <mliska@suse.cz>
24992 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24993 order to ignore the predictor.
24994 (PRED_POLYMORPHIC_CALL): Likewise.
24995 (PRED_RECURSIVE_CALL): Likewise.
24997 2018-01-23 Martin Liska <mliska@suse.cz>
24999 * tree-profile.c (tree_profiling): Print function header to
25000 aware reader which function we are working on.
25001 * value-prof.c (gimple_find_values_to_profile): Do not print
25002 not interesting value histograms.
25004 2018-01-23 Martin Liska <mliska@suse.cz>
25006 * profile-count.h (enum profile_quality): Add
25007 profile_uninitialized as the first value. Do not number values
25008 as they are zero based.
25009 (profile_count::verify): Update sanity check.
25010 (profile_probability::verify): Likewise.
25012 2018-01-23 Nathan Sidwell <nathan@acm.org>
25014 * doc/invoke.texi (ffor-scope): Deprecate.
25016 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25018 PR tree-optimization/83510
25019 * domwalk.c (set_all_edges_as_executable): New function.
25020 (dom_walker::dom_walker): Convert bool param
25021 "skip_unreachable_blocks" to enum reachability. Move setup of
25022 edge flags to set_all_edges_as_executable and only do it when
25023 reachability is REACHABLE_BLOCKS.
25024 * domwalk.h (enum dom_walker::reachability): New enum.
25025 (dom_walker::dom_walker): Convert bool param
25026 "skip_unreachable_blocks" to enum reachability.
25027 (set_all_edges_as_executable): New decl.
25028 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
25029 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
25031 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
25032 but converting true to REACHABLE_BLOCKS.
25033 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
25035 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
25036 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
25037 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
25039 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
25040 if check_all_array_refs will be called.
25042 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25044 * tree.c (selftest::test_location_wrappers): Add more test
25047 2018-01-23 David Malcolm <dmalcolm@redhat.com>
25049 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
25050 (selftest::test_bit_in_range): Likewise.
25052 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
25055 * doc/sourcebuild.texi (vect_float): Say that the selector
25056 only describes the situation when -funsafe-math-optimizations is on.
25057 (vect_float_strict): Document.
25059 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
25061 PR tree-optimization/83965
25062 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
25063 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
25064 instead of checking only for a reduction.
25065 (vect_recog_widen_sum_pattern): Likewise.
25067 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25069 * predict.c (probably_never_executed): Only use precise profile info.
25070 (compute_function_frequency): Skip after inlining hack since we now
25071 have quality checking.
25073 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
25075 * profile-count.h (profile_probability::very_unlikely,
25076 profile_probability::unlikely, profile_probability::even): Set
25077 precision to guessed.
25079 2018-01-23 Richard Biener <rguenther@suse.de>
25081 PR tree-optimization/83963
25082 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25083 Properly terminate dominator walk when crossing the exit edge not
25084 when visiting its source block.
25086 2018-01-23 Jakub Jelinek <jakub@redhat.com>
25089 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
25090 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
25092 2018-01-22 Jakub Jelinek <jakub@redhat.com>
25094 PR tree-optimization/83957
25095 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
25096 semicolon after for body surrounded by braces.
25098 PR tree-optimization/83081
25099 * profile-count.h (profile_probability::split): New method.
25100 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
25101 Use profile_probability::split.
25102 (do_compare_rtx_and_jump): Fix adjustment of probabilities
25103 when splitting a single conditional jump into 2.
25105 2018-01-22 David Malcolm <dmalcolm@redhat.com>
25107 PR tree-optimization/69452
25108 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
25111 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25113 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
25114 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
25115 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
25117 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25119 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
25121 * config/rl78/rl78.md (movdi): New define_expand.
25122 * config/rl78/rl78.c (rl78_split_movdi): New function.
25124 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
25127 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
25129 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
25130 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
25131 128-bit to produce an UNSPEC move to get the double word with the
25132 signbit and then a shift directly to do signbit.
25133 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
25134 implementation with a new version that just does either a direct
25135 move or a regular move. Move memory interface to separate insns.
25136 Move insns so they are next to the expander.
25137 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
25138 with signbit move. Split big and little endian case.
25139 (signbit<mode>2_dm_mem_le): Likewise.
25140 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
25141 (signbit<mode>2_dm2): Likewise.
25143 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25145 * config/rl78/rl78.md (anddi3): New define_expand.
25147 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25149 * config/rl78/rl78.md (umindi3): New define_expand.
25151 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25153 * config/rl78/rl78.md (smindi3): New define_expand.
25155 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25157 * config/rl78/rl78.md (smaxdi3): New define_expand.
25159 2018-01-22 Carl Love <cel@us.ibm.com>
25161 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
25162 LVX_V1TI): Add macro expansion.
25163 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
25164 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
25165 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
25166 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
25167 Change check to determine if the instruction is a byte reversing
25168 entry. Fix typo in comment.
25169 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
25170 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
25171 Add def_builtin calls for new builtins.
25172 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
25173 Add define_insn expansion.
25175 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25177 * config/rl78/rl78.md (umaxdi3): New define_expand.
25179 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25181 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
25182 for non-QImode registers.
25184 2018-01-22 Richard Biener <rguenther@suse.de>
25186 PR tree-optimization/83963
25187 * graphite-scop-detection.c (scop_detection::get_sese): Delay
25188 including the loop exit block.
25189 (scop_detection::merge_sese): Likewise.
25190 (scop_detection::add_scop): Do it here instead.
25192 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25194 * doc/sourcebuild.texi (arm_softfloat): Document.
25196 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
25199 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
25200 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
25201 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
25203 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25204 David Edelsohn <dje.gcc@gmail.com>
25207 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25208 Change "crset eq" to "crset 2".
25209 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25210 (*call_indirect_aix<mode>_nospec): Likewise.
25211 (*call_value_indirect_aix<mode>_nospec): Likewise.
25212 (*call_indirect_elfv2<mode>_nospec): Likewise.
25213 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
25214 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
25215 change assembly output from . to $.
25216 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25217 (indirect_jump<mode>_nospec): Change assembly output from . to $.
25218 (*tablejump<mode>_internal1_nospec): Likewise.
25220 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
25223 * config/sh/sh_optimize_sett_clrt.cc:
25224 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
25226 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
25228 PR tree-optimization/83940
25229 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
25230 offset_dt to vect_constant_def rather than vect_unknown_def_type.
25231 (vect_check_load_store_mask): Add a mask_dt_out parameter and
25232 use it to pass back the definition type.
25233 (vect_check_store_rhs): Likewise rhs_dt_out.
25234 (vect_build_gather_load_calls): Add a mask_dt argument and use
25235 it instead of a call to vect_is_simple_use.
25236 (vectorizable_store): Update calls to vect_check_load_store_mask
25237 and vect_check_store_rhs. Use the dt returned by the latter instead
25238 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
25239 instead of calls to vect_is_simple_use. Pass the scalar rather
25240 than the vector operand to vect_is_simple_use when handling
25241 second and subsequent copies of an rhs value.
25242 (vectorizable_load): Update calls to vect_check_load_store_mask
25243 and vect_build_gather_load_calls. Use the cached mask_dt and
25244 gs_info.offset_dt instead of calls to vect_is_simple_use.
25246 2018-01-20 Jakub Jelinek <jakub@redhat.com>
25248 PR middle-end/83945
25249 * tree-emutls.c: Include gimplify.h.
25250 (lower_emutls_2): New function.
25251 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
25252 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
25253 it before further processing.
25256 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
25257 UINTVAL (trueop1) instead of INTVAL (op1).
25259 2018-01-19 Jakub Jelinek <jakub@redhat.com>
25263 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
25264 INCOMING_FRAME_SP_OFFSET if not defined.
25265 (scan_trace): Add ENTRY argument. If true and
25266 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
25267 emit a note to adjust the CFA offset.
25268 (create_cfi_notes): Adjust scan_trace callers.
25269 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
25270 INCOMING_FRAME_SP_OFFSET in the CIE.
25271 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
25272 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
25274 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
25275 * doc/tm.texi: Regenerated.
25277 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25279 PR rtl-optimization/83147
25280 * lra-constraints.c (remove_inheritance_pseudos): Use
25281 lra_substitute_pseudo_within_insn.
25283 2018-01-19 Tom de Vries <tom@codesourcery.com>
25284 Cesar Philippidis <cesar@codesourcery.com>
25287 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
25289 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
25292 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
25293 spaces for function labels.
25295 2018-01-19 Martin Liska <mliska@suse.cz>
25297 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
25298 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
25299 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
25300 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
25301 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
25302 (PRED_CONST_RETURN): Change from 69 to 65.
25303 (PRED_NULL_RETURN): Change from 91 to 71.
25304 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
25305 (PRED_LOOP_GUARD): Change from 66 to 73.
25307 2018-01-19 Martin Liska <mliska@suse.cz>
25309 * predict.c (predict_insn_def): Add new assert.
25310 (struct branch_predictor): Change type to signed integer.
25311 (test_prediction_value_range): Amend test to cover
25312 PROB_UNINITIALIZED.
25313 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
25314 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
25315 (PRED_LOOP_ITERATIONS_MAX): Likewise.
25316 (PRED_LOOP_IV_COMPARE): Likewise.
25317 * predict.h (PROB_UNINITIALIZED): Define new constant.
25319 2018-01-19 Martin Liska <mliska@suse.cz>
25321 * predict.c (dump_prediction): Add new format for
25322 analyze_brprob.py script which is enabled with -details
25324 * profile-count.h (precise_p): New function.
25326 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25328 PR tree-optimization/83922
25329 * tree-vect-loop.c (vect_verify_full_masking): Return false if
25330 there are no statements that need masking.
25331 (vect_active_double_reduction_p): New function.
25332 (vect_analyze_loop_operations): Use it when handling phis that
25333 are not in the loop header.
25335 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25337 PR tree-optimization/83914
25338 * tree-vect-loop.c (vectorizable_induction): Don't convert
25339 init_expr or apply the peeling adjustment for inductions
25340 that are nested within the vectorized loop.
25342 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25344 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
25347 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25351 * function.h (gimplify_parameters): Add gimple_seq * argument.
25352 * function.c: Include gimple.h and options.h.
25353 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
25354 for the added local temporaries if needed.
25355 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
25356 if there are any parameter cleanups, wrap whole body into a
25357 try/finally with the cleanups.
25359 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25362 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25363 Use GET_MODE_CLASS for scalar floating point.
25365 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25369 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
25370 Fix call of call_cgraph_insertion_hooks.
25372 2018-01-18 Martin Sebor <msebor@redhat.com>
25374 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
25376 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25379 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
25382 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
25385 * common.opt: (-ffile-prefix-map): New option.
25386 * opts.c (common_handle_option): Defer it.
25387 * opts-global.c (handle_common_deferred_options): Handle it.
25388 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
25389 * file-prefix-map.h: New file.
25390 (remap_debug_filename, add_debug_prefix_map): ...here.
25391 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
25392 * final.c (debug_prefix_map, add_debug_prefix_map
25393 remap_debug_filename): Move to...
25394 * file-prefix-map.c: New file.
25395 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
25396 generalize, get rid of alloca(), use strrchr() instead of strchr().
25397 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
25398 Implement in terms of add_prefix_map().
25399 (remap_macro_filename, remap_debug_filename): Implement in term of
25401 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
25402 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
25403 * dbxout.c: Include file-prefix-map.h.
25404 * varasm.c: Likewise.
25405 * vmsdbgout.c: Likewise.
25406 * xcoffout.c: Likewise.
25407 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
25408 * doc/cppopts.texi (-fmacro-prefix-map): Document.
25409 * doc/invoke.texi (-ffile-prefix-map): Document.
25410 (-fdebug-prefix-map): Update description.
25412 2018-01-18 Martin Liska <mliska@suse.cz>
25414 * config/i386/i386.c (indirect_thunk_name): Document that also
25416 (output_indirect_thunk): Document why both instructions
25417 (pause and lfence) are generated.
25419 2018-01-18 Richard Biener <rguenther@suse.de>
25421 PR tree-optimization/83887
25422 * graphite-scop-detection.c
25423 (scop_detection::get_nearest_dom_with_single_entry): Remove.
25424 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
25425 (scop_detection::merge_sese): Re-implement with a flood-fill
25426 algorithm that properly finds a SESE region if it exists.
25428 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25431 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
25432 pointer_diff optimizations use view_convert instead of convert.
25434 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25436 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25437 Generate different code for -mno-speculate-indirect-jumps.
25438 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25439 (*call_indirect_aix<mode>): Disable for
25440 -mno-speculate-indirect-jumps.
25441 (*call_indirect_aix<mode>_nospec): New define_insn.
25442 (*call_value_indirect_aix<mode>): Disable for
25443 -mno-speculate-indirect-jumps.
25444 (*call_value_indirect_aix<mode>_nospec): New define_insn.
25445 (*sibcall_nonlocal_sysv<mode>): Generate different code for
25446 -mno-speculate-indirect-jumps.
25447 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25449 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
25451 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
25452 long double type, set the flags for noting the default long double
25453 type, even if we don't pass or return a long double type.
25455 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
25458 * ipa-inline.c (flatten_function): Do not overwrite final inlining
25461 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
25463 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
25464 support for merge[hl].
25465 (fold_mergehl_helper): New helper function.
25466 (tree-vector-builder.h): New #include for tree_vector_builder usage.
25467 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
25468 (altivec_vmrglw_direct): Add xxmrglw insn.
25470 2018-01-17 Andrew Waterman <andrew@sifive.com>
25472 * config/riscv/riscv.c (riscv_conditional_register_usage): If
25473 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
25475 2018-01-17 David Malcolm <dmalcolm@redhat.com>
25478 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
25479 call the lto_location_cache before reading the
25480 DECL_SOURCE_LOCATION of the types.
25482 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25483 Richard Sandiford <richard.sandiford@linaro.org>
25485 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
25486 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
25487 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
25488 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
25489 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
25491 * config/aarch64/constraints.md (aarch64_movti_operand):
25493 * config/aarch64/predicates.md (Uti): Add new constraint.
25495 2018-01-17 Carl Love <cel@us.ibm.com>
25497 * config/rs6000/vsx.md (define_expand xl_len_r,
25498 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
25499 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
25501 (define_expand, define_insn): Move the shift left from the
25502 define_insn to the define_expand for lxvl and stxvl instructions.
25503 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
25504 and XL_LEN_R definitions to PURE.
25506 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
25508 * config/i386/i386.c (indirect_thunk_name): Declare regno
25509 as unsigned int. Compare regno with INVALID_REGNUM.
25510 (output_indirect_thunk): Ditto.
25511 (output_indirect_thunk_function): Ditto.
25512 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
25513 in the call to output_indirect_thunk_function.
25515 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
25517 PR middle-end/83884
25518 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
25519 rather than the size of inner_type to determine the stack slot size
25520 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
25522 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
25525 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
25528 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25530 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
25531 endian Linux systems to optionally enable multilibs for selecting
25532 the long double type if the user configured an explicit type.
25533 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
25534 have no long double multilibs if not defined.
25535 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
25536 warn if the user used -mabi={ieee,ibm}longdouble and we built
25537 multilibs for long double.
25538 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
25539 appropriate multilib option.
25540 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
25542 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
25543 for building long double multilibs.
25544 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
25546 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
25548 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25551 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25553 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25556 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25559 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25562 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25564 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25565 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25567 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25569 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25570 different rtl trees depending on TARGET_64BIT.
25571 (rs6000_gen_lvx): Likewise.
25573 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25575 * config/visium/visium.md (nop): Tweak comment.
25576 (hazard_nop): Likewise.
25578 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25580 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25581 -mspeculate-indirect-jumps.
25582 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25583 for -mno-speculate-indirect-jumps.
25584 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25585 (*call_value_indirect_elfv2<mode>): Disable for
25586 -mno-speculate-indirect-jumps.
25587 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25588 (indirect_jump): Emit different RTL for
25589 -mno-speculate-indirect-jumps.
25590 (*indirect_jump<mode>): Disable for
25591 -mno-speculate-indirect-jumps.
25592 (*indirect_jump<mode>_nospec): New define_insn.
25593 (tablejump): Emit different RTL for
25594 -mno-speculate-indirect-jumps.
25595 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25596 (tablejumpsi_nospec): New define_expand.
25597 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25598 (tablejumpdi_nospec): New define_expand.
25599 (*tablejump<mode>_internal1): Disable for
25600 -mno-speculate-indirect-jumps.
25601 (*tablejump<mode>_internal1_nospec): New define_insn.
25602 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25605 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25607 * caller-save.c (insert_save): Drop unnecessary parameter. All
25610 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25611 Richard Biener <rguenth@suse.de>
25614 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25615 return early, inline manually is_gimple_sizepos. Make sure if we
25616 call gimplify_expr we don't end up with a gimple constant.
25617 * tree.c (variably_modified_type_p): Don't return true for
25618 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25619 * gimplify.h (is_gimple_sizepos): Remove.
25621 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25623 PR tree-optimization/83857
25624 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25625 vectorizable_live_operation for pure SLP statements.
25626 (vectorizable_live_operation): Handle PHIs.
25628 2018-01-16 Richard Biener <rguenther@suse.de>
25630 PR tree-optimization/83867
25631 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25632 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25634 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25637 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25638 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25639 If off is not INTEGER_CST, issue a may not be aligned warning
25640 rather than isn't aligned. Use isn%'t rather than isn't.
25641 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25643 <case MULT_EXPR>: Improve the case when bottom and one of the
25644 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25645 operand, in that case check if the other operand is multiple of
25646 bottom divided by the INTEGER_CST operand.
25648 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25651 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25652 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25653 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25654 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25655 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25656 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25657 (pa_function_arg_size): New function.
25659 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25661 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25662 in a separate statement.
25664 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25666 PR tree-optimization/83847
25667 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25668 group gathers and scatters.
25670 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25672 PR rtl-optimization/86620
25673 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25675 PR rtl-optimization/83213
25676 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25677 to last if both are JUMP_INSNs.
25679 PR tree-optimization/83843
25680 * gimple-ssa-store-merging.c
25681 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25682 store_immediate_info for bswap/nop orig_stores.
25684 2018-01-15 Andrew Waterman <andrew@sifive.com>
25686 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25688 <UDIV>: Increase cost if !TARGET_DIV.
25690 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25692 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25693 (define_attr "cr_logical_3op"): New.
25694 (cceq_ior_compare): Adjust.
25695 (cceq_ior_compare_complement): Adjust.
25696 (*cceq_rev_compare): Adjust.
25697 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25698 (is_cracked_insn): Adjust.
25699 (insn_must_be_first_in_group): Adjust.
25700 * config/rs6000/40x.md: Adjust.
25701 * config/rs6000/440.md: Adjust.
25702 * config/rs6000/476.md: Adjust.
25703 * config/rs6000/601.md: Adjust.
25704 * config/rs6000/603.md: Adjust.
25705 * config/rs6000/6xx.md: Adjust.
25706 * config/rs6000/7450.md: Adjust.
25707 * config/rs6000/7xx.md: Adjust.
25708 * config/rs6000/8540.md: Adjust.
25709 * config/rs6000/cell.md: Adjust.
25710 * config/rs6000/e300c2c3.md: Adjust.
25711 * config/rs6000/e500mc.md: Adjust.
25712 * config/rs6000/e500mc64.md: Adjust.
25713 * config/rs6000/e5500.md: Adjust.
25714 * config/rs6000/e6500.md: Adjust.
25715 * config/rs6000/mpc.md: Adjust.
25716 * config/rs6000/power4.md: Adjust.
25717 * config/rs6000/power5.md: Adjust.
25718 * config/rs6000/power6.md: Adjust.
25719 * config/rs6000/power7.md: Adjust.
25720 * config/rs6000/power8.md: Adjust.
25721 * config/rs6000/power9.md: Adjust.
25722 * config/rs6000/rs64.md: Adjust.
25723 * config/rs6000/titan.md: Adjust.
25725 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25727 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25728 ix86_indirect_branch_register logic.
25730 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25732 * config/i386/constraints.md (Bs): Update
25733 ix86_indirect_branch_register check. Don't check
25734 ix86_indirect_branch_register with GOT_memory_operand.
25736 * config/i386/predicates.md (GOT_memory_operand): Don't check
25737 ix86_indirect_branch_register here.
25738 (GOT32_symbol_operand): Likewise.
25740 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25742 * config/i386/predicates.md (constant_call_address_operand):
25743 Rewrite ix86_indirect_branch_register logic.
25744 (sibcall_insn_operand): Likewise.
25746 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25748 * config/i386/constraints.md (Bs): Replace
25749 ix86_indirect_branch_thunk_register with
25750 ix86_indirect_branch_register.
25752 * config/i386/i386.md (indirect_jump): Likewise.
25753 (tablejump): Likewise.
25754 (*sibcall_memory): Likewise.
25755 (*sibcall_value_memory): Likewise.
25756 Peepholes of indirect call and jump via memory: Likewise.
25757 * config/i386/i386.opt: Likewise.
25758 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25759 (GOT_memory_operand): Likewise.
25760 (call_insn_operand): Likewise.
25761 (sibcall_insn_operand): Likewise.
25762 (GOT32_symbol_operand): Likewise.
25764 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25766 PR middle-end/83837
25767 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25768 type rather than type addr's type points to.
25769 (expand_omp_atomic_mutex): Likewise.
25770 (expand_omp_atomic): Likewise.
25772 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25775 * config/i386/i386.c (output_indirect_thunk_function): Use
25776 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25777 for __x86_return_thunk.
25779 2018-01-15 Richard Biener <rguenther@suse.de>
25781 PR middle-end/83850
25782 * expmed.c (extract_bit_field_1): Fix typo.
25784 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25787 * config/arm/iterators.md (VF): New mode iterator.
25788 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25789 Remove integer-related logic from pattern.
25790 (neon_vabd<mode>_3): Likewise.
25792 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25794 PR middle-end/82694
25795 * common.opt (fstrict-overflow): No longer an alias.
25796 (fwrapv-pointer): New option.
25797 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25798 also for pointer types based on flag_wrapv_pointer.
25799 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25800 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25801 opts->x_flag_wrapv got set.
25802 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25803 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25804 POINTER_TYPE_OVERFLOW_UNDEFINED.
25805 * match.pd: Likewise in address comparison pattern.
25806 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25808 2018-01-15 Richard Biener <rguenther@suse.de>
25811 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25812 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25813 Reset type names to their identifier if their TYPE_DECL doesn't
25814 have linkage (and thus is used for ODR and devirt).
25815 (save_debug_info_for_decl): Remove.
25816 (save_debug_info_for_type): Likewise.
25817 (add_tree_to_fld_list): Adjust.
25818 * tree-pretty-print.c (dump_generic_node): Make dumping of
25819 type names more robust.
25821 2018-01-15 Richard Biener <rguenther@suse.de>
25823 * BASE-VER: Bump to 8.0.1.
25825 2018-01-14 Martin Sebor <msebor@redhat.com>
25828 * builtins.c (check_access): Avoid warning when the no-warning bit
25831 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25833 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25834 * ira-color (allocno_hard_regs_compare): Likewise.
25836 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25839 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25840 Use .pushsection/.popsection.
25842 2018-01-14 Martin Sebor <msebor@redhat.com>
25845 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25847 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25849 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25850 entry from extra_headers.
25851 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25852 extra_headers, make the list bitwise identical to the i?86-*-* one.
25854 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25856 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25857 -mcmodel=large with -mindirect-branch=thunk,
25858 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25859 -mfunction-return=thunk-extern.
25860 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25861 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25862 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25864 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25866 * config/i386/i386.c (print_reg): Print the name of the full
25867 integer register without '%'.
25868 (ix86_print_operand): Handle 'V'.
25869 * doc/extend.texi: Document 'V' modifier.
25871 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25873 * config/i386/constraints.md (Bs): Disallow memory operand for
25874 -mindirect-branch-register.
25876 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25877 (GOT_memory_operand): Likewise.
25878 (call_insn_operand): Likewise.
25879 (sibcall_insn_operand): Likewise.
25880 (GOT32_symbol_operand): Likewise.
25881 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25882 for -mindirect-branch-register.
25883 (tablejump): Likewise.
25884 (*sibcall_memory): Likewise.
25885 (*sibcall_value_memory): Likewise.
25886 Disallow peepholes of indirect call and jump via memory for
25887 -mindirect-branch-register.
25888 (*call_pop): Replace m with Bw.
25889 (*call_value_pop): Likewise.
25890 (*sibcall_pop_memory): Replace m with Bs.
25891 * config/i386/i386.opt (mindirect-branch-register): New option.
25892 * doc/invoke.texi: Document -mindirect-branch-register option.
25894 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25896 * config/i386/i386-protos.h (ix86_output_function_return): New.
25897 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25898 set function_return_type.
25899 (indirect_thunk_name): Add ret_p to indicate thunk for function
25901 (output_indirect_thunk_function): Pass false to
25902 indirect_thunk_name.
25903 (ix86_output_indirect_branch_via_reg): Likewise.
25904 (ix86_output_indirect_branch_via_push): Likewise.
25905 (output_indirect_thunk_function): Create alias for function
25906 return thunk if regno < 0.
25907 (ix86_output_function_return): New function.
25908 (ix86_handle_fndecl_attribute): Handle function_return.
25909 (ix86_attribute_table): Add function_return.
25910 * config/i386/i386.h (machine_function): Add
25911 function_return_type.
25912 * config/i386/i386.md (simple_return_internal): Use
25913 ix86_output_function_return.
25914 (simple_return_internal_long): Likewise.
25915 * config/i386/i386.opt (mfunction-return=): New option.
25916 (indirect_branch): Mention -mfunction-return=.
25917 * doc/extend.texi: Document function_return function attribute.
25918 * doc/invoke.texi: Document -mfunction-return= option.
25920 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25922 * config/i386/i386-opts.h (indirect_branch): New.
25923 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25924 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25925 with local indirect jump when converting indirect call and jump.
25926 (ix86_set_indirect_branch_type): New.
25927 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25928 (indirectlabelno): New.
25929 (indirect_thunk_needed): Likewise.
25930 (indirect_thunk_bnd_needed): Likewise.
25931 (indirect_thunks_used): Likewise.
25932 (indirect_thunks_bnd_used): Likewise.
25933 (INDIRECT_LABEL): Likewise.
25934 (indirect_thunk_name): Likewise.
25935 (output_indirect_thunk): Likewise.
25936 (output_indirect_thunk_function): Likewise.
25937 (ix86_output_indirect_branch_via_reg): Likewise.
25938 (ix86_output_indirect_branch_via_push): Likewise.
25939 (ix86_output_indirect_branch): Likewise.
25940 (ix86_output_indirect_jmp): Likewise.
25941 (ix86_code_end): Call output_indirect_thunk_function if needed.
25942 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25944 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25945 (ix86_attribute_table): Add indirect_branch.
25946 * config/i386/i386.h (machine_function): Add indirect_branch_type
25947 and has_local_indirect_jump.
25948 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25950 (tablejump): Likewise.
25951 (*indirect_jump): Use ix86_output_indirect_jmp.
25952 (*tablejump_1): Likewise.
25953 (simple_return_indirect_internal): Likewise.
25954 * config/i386/i386.opt (mindirect-branch=): New option.
25955 (indirect_branch): New.
25958 (thunk-inline): Likewise.
25959 (thunk-extern): Likewise.
25960 * doc/extend.texi: Document indirect_branch function attribute.
25961 * doc/invoke.texi: Document -mindirect-branch= option.
25963 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25966 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25968 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25970 * ipa-inline.c (want_inline_small_function_p): Return false if
25971 inlining has already failed with CIF_FINAL_ERROR.
25972 (update_caller_keys): Call want_inline_small_function_p before
25974 (update_callee_keys): Likewise.
25976 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25978 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25980 (rs6000_quadword_masked_address_p): Likewise.
25981 (quad_aligned_load_p): Likewise.
25982 (quad_aligned_store_p): Likewise.
25983 (const_load_sequence_p): Add comment to describe the outer-most loop.
25984 (mimic_memory_attributes_and_flags): New function.
25985 (rs6000_gen_stvx): Likewise.
25986 (replace_swapped_aligned_store): Likewise.
25987 (rs6000_gen_lvx): Likewise.
25988 (replace_swapped_aligned_load): Likewise.
25989 (replace_swapped_load_constant): Capitalize argument name in
25990 comment describing this function.
25991 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25992 and stores that access quad-word aligned addresses and replace
25993 with stvx or lvx instructions when appropriate.
25994 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25995 New function prototype.
25996 (rs6000_quadword_masked_address_p): Likewise.
25997 (rs6000_gen_lvx): Likewise.
25998 (rs6000_gen_stvx): Likewise.
25999 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
26000 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
26001 when memory address is aligned.
26002 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
26003 this split to select lvx instruction when memory address is aligned.
26004 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
26005 instruction when memory address is aligned.
26006 (*vsx_le_perm_load_v16qi): Likewise.
26007 (four unnamed splitters): Modify to select the stvx instruction
26008 when memory is aligned.
26010 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
26012 * predict.c (determine_unlikely_bbs): Handle correctly BBs
26013 which appears in the queue multiple times.
26015 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26016 Alan Hayward <alan.hayward@arm.com>
26017 David Sherwood <david.sherwood@arm.com>
26019 * tree-vectorizer.h (vec_lower_bound): New structure.
26020 (_loop_vec_info): Add check_nonzero and lower_bounds.
26021 (LOOP_VINFO_CHECK_NONZERO): New macro.
26022 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
26023 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
26024 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
26025 fields. Make seg_len the distance travelled, not including the
26027 (dr_direction_indicator): Declare.
26028 (dr_zero_step_indicator): Likewise.
26029 (dr_known_forward_stride_p): Likewise.
26030 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
26032 (runtime_alias_check_p): Allow runtime alias checks with
26034 (operator ==): Compare access_size and align.
26035 (prune_runtime_alias_test_list): Rework for new distinction between
26036 the access_size and seg_len.
26037 (create_intersect_range_checks_index): Likewise. Cope with polynomial
26039 (get_segment_min_max): New function.
26040 (create_intersect_range_checks): Use it.
26041 (dr_step_indicator): New function.
26042 (dr_direction_indicator): Likewise.
26043 (dr_zero_step_indicator): Likewise.
26044 (dr_known_forward_stride_p): Likewise.
26045 * tree-loop-distribution.c (data_ref_segment_size): Return
26046 DR_STEP * (niters - 1).
26047 (compute_alias_check_pairs): Update call to the dr_with_seg_len
26049 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
26050 (vect_preserves_scalar_order_p): New function, split out from...
26051 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
26052 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
26053 (vect_vfa_access_size): New function.
26054 (vect_vfa_align): Likewise.
26055 (vect_compile_time_alias): Take access_size_a and access_b arguments.
26056 (dump_lower_bound): New function.
26057 (vect_check_lower_bound): Likewise.
26058 (vect_small_gap_p): Likewise.
26059 (vectorizable_with_step_bound_p): Likewise.
26060 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
26061 depencies if the vectorization factor is 1. Convert the checks
26062 for nonzero steps into checks on the bounds of DR_STEP. Try using
26063 a bunds check for variable steps if the minimum required step is
26064 relatively small. Update calls to the dr_with_seg_len
26065 constructor and to vect_compile_time_alias.
26066 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
26068 (vect_loop_versioning): Call it.
26069 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
26071 (vect_estimate_min_profitable_iters): Account for any bounds checks.
26073 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26074 Alan Hayward <alan.hayward@arm.com>
26075 David Sherwood <david.sherwood@arm.com>
26077 * doc/sourcebuild.texi (vect_scatter_store): Document.
26078 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
26080 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
26082 * genopinit.c (main): Add supports_vec_scatter_store and
26083 supports_vec_scatter_store_cached to target_optabs.
26084 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
26085 IFN_MASK_SCATTER_STORE.
26086 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
26088 * internal-fn.h (internal_store_fn_p): Declare.
26089 (internal_fn_stored_value_index): Likewise.
26090 * internal-fn.c (scatter_store_direct): New macro.
26091 (expand_scatter_store_optab_fn): New function.
26092 (direct_scatter_store_optab_supported_p): New macro.
26093 (internal_store_fn_p): New function.
26094 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
26095 IFN_MASK_SCATTER_STORE.
26096 (internal_fn_mask_index): Likewise.
26097 (internal_fn_stored_value_index): New function.
26098 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
26099 for scatter stores.
26100 * optabs-query.h (supports_vec_scatter_store_p): Declare.
26101 * optabs-query.c (supports_vec_scatter_store_p): New function.
26102 * tree-vectorizer.h (vect_get_store_rhs): Declare.
26103 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
26104 true for scatter stores.
26105 (vect_gather_scatter_fn_p): Handle scatter stores too.
26106 (vect_check_gather_scatter): Consider using scatter stores if
26107 supports_vec_scatter_store_p.
26108 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
26109 scatter stores too.
26110 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26111 internal_fn_stored_value_index.
26112 (check_load_store_masking): Handle scatter stores too.
26113 (vect_get_store_rhs): Make public.
26114 (vectorizable_call): Use internal_store_fn_p.
26115 (vectorizable_store): Handle scatter store internal functions.
26116 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
26117 when deciding whether the end of the group has been reached.
26118 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
26119 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
26120 (mask_scatter_store<mode>): New insns.
26122 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26123 Alan Hayward <alan.hayward@arm.com>
26124 David Sherwood <david.sherwood@arm.com>
26126 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
26127 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
26128 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
26130 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
26131 Use vect_truncate_gather_scatter_offset if we can't treat the
26132 operation as a normal gather load or scatter store.
26133 (get_group_load_store_type): Take the gather_scatter_info
26134 as argument. Try using a gather load or scatter store for
26135 single-element groups.
26136 (get_load_store_type): Update calls to get_group_load_store_type
26137 and vect_use_strided_gather_scatters_p.
26139 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26140 Alan Hayward <alan.hayward@arm.com>
26141 David Sherwood <david.sherwood@arm.com>
26143 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
26144 optional tree argument.
26145 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
26147 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
26148 but continue to use the current value as a fallback.
26149 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
26150 to compare the updates.
26151 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
26152 (get_load_store_type): Use it when handling a strided access.
26153 (vect_get_strided_load_store_ops): New function.
26154 (vect_get_data_ptr_increment): Likewise.
26155 (vectorizable_load): Handle strided gather loads. Always pass
26156 a step to vect_create_data_ref_ptr and bump_vector_ptr.
26158 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26159 Alan Hayward <alan.hayward@arm.com>
26160 David Sherwood <david.sherwood@arm.com>
26162 * doc/md.texi (gather_load@var{m}): Document.
26163 (mask_gather_load@var{m}): Likewise.
26164 * genopinit.c (main): Add supports_vec_gather_load and
26165 supports_vec_gather_load_cached to target_optabs.
26166 * optabs-tree.c (init_tree_optimization_optabs): Use
26167 ggc_cleared_alloc to allocate target_optabs.
26168 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
26169 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
26171 * internal-fn.h (internal_load_fn_p): Declare.
26172 (internal_gather_scatter_fn_p): Likewise.
26173 (internal_fn_mask_index): Likewise.
26174 (internal_gather_scatter_fn_supported_p): Likewise.
26175 * internal-fn.c (gather_load_direct): New macro.
26176 (expand_gather_load_optab_fn): New function.
26177 (direct_gather_load_optab_supported_p): New macro.
26178 (direct_internal_fn_optab): New function.
26179 (internal_load_fn_p): Likewise.
26180 (internal_gather_scatter_fn_p): Likewise.
26181 (internal_fn_mask_index): Likewise.
26182 (internal_gather_scatter_fn_supported_p): Likewise.
26183 * optabs-query.c (supports_at_least_one_mode_p): New function.
26184 (supports_vec_gather_load_p): Likewise.
26185 * optabs-query.h (supports_vec_gather_load_p): Declare.
26186 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
26187 and memory_type field.
26188 (NUM_PATTERNS): Bump to 15.
26189 * tree-vect-data-refs.c: Include internal-fn.h.
26190 (vect_gather_scatter_fn_p): New function.
26191 (vect_describe_gather_scatter_call): Likewise.
26192 (vect_check_gather_scatter): Try using internal functions for
26193 gather loads. Recognize existing calls to a gather load function.
26194 (vect_analyze_data_refs): Consider using gather loads if
26195 supports_vec_gather_load_p.
26196 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
26197 (vect_get_gather_scatter_offset_type): Likewise.
26198 (vect_convert_mask_for_vectype): Likewise.
26199 (vect_add_conversion_to_patterm): Likewise.
26200 (vect_try_gather_scatter_pattern): Likewise.
26201 (vect_recog_gather_scatter_pattern): New pattern recognizer.
26202 (vect_vect_recog_func_ptrs): Add it.
26203 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26204 internal_fn_mask_index and internal_gather_scatter_fn_p.
26205 (check_load_store_masking): Take the gather_scatter_info as an
26206 argument and handle gather loads.
26207 (vect_get_gather_scatter_ops): New function.
26208 (vectorizable_call): Check internal_load_fn_p.
26209 (vectorizable_load): Likewise. Handle gather load internal
26211 (vectorizable_store): Update call to check_load_store_masking.
26212 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
26213 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
26214 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
26215 (aarch64_gather_scale_operand_d): New predicates.
26216 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
26217 (mask_gather_load<mode>): New insns.
26219 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26220 Alan Hayward <alan.hayward@arm.com>
26221 David Sherwood <david.sherwood@arm.com>
26223 * optabs.def (fold_left_plus_optab): New optab.
26224 * doc/md.texi (fold_left_plus_@var{m}): Document.
26225 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
26226 * internal-fn.c (fold_left_direct): Define.
26227 (expand_fold_left_optab_fn): Likewise.
26228 (direct_fold_left_optab_supported_p): Likewise.
26229 * fold-const-call.c (fold_const_fold_left): New function.
26230 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
26231 * tree-parloops.c (valid_reduction_p): New function.
26232 (gather_scalar_reductions): Use it.
26233 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
26234 (vect_finish_replace_stmt): Declare.
26235 * tree-vect-loop.c (fold_left_reduction_fn): New function.
26236 (needs_fold_left_reduction_p): New function, split out from...
26237 (vect_is_simple_reduction): ...here. Accept reductions that
26238 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
26239 (vect_force_simple_reduction): Also store the reduction type in
26240 the assignment's STMT_VINFO_REDUC_TYPE.
26241 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
26242 (merge_with_identity): New function.
26243 (vect_expand_fold_left): Likewise.
26244 (vectorize_fold_left_reduction): Likewise.
26245 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
26246 scalar phi in place for it. Check for target support and reject
26247 cases that would reassociate the operation. Defer the transform
26248 phase to vectorize_fold_left_reduction.
26249 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
26250 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
26251 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
26253 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26255 * tree-if-conv.c (predicate_mem_writes): Remove redundant
26256 call to ifc_temp_var.
26258 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26259 Alan Hayward <alan.hayward@arm.com>
26260 David Sherwood <david.sherwood@arm.com>
26262 * target.def (legitimize_address_displacement): Take the original
26263 offset as a poly_int.
26264 * targhooks.h (default_legitimize_address_displacement): Update
26266 * targhooks.c (default_legitimize_address_displacement): Likewise.
26267 * doc/tm.texi: Regenerate.
26268 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
26269 as an argument, moving assert of ad->disp == ad->disp_term to...
26270 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
26271 Try calling targetm.legitimize_address_displacement before expanding
26272 the address rather than afterwards, and adjust for the new interface.
26273 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
26274 Match the new hook interface. Handle SVE addresses.
26275 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
26276 new hook interface.
26278 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26280 * Makefile.in (OBJS): Add early-remat.o.
26281 * target.def (select_early_remat_modes): New hook.
26282 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
26283 * doc/tm.texi: Regenerate.
26284 * targhooks.h (default_select_early_remat_modes): Declare.
26285 * targhooks.c (default_select_early_remat_modes): New function.
26286 * timevar.def (TV_EARLY_REMAT): New timevar.
26287 * passes.def (pass_early_remat): New pass.
26288 * tree-pass.h (make_pass_early_remat): Declare.
26289 * early-remat.c: New file.
26290 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
26292 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
26294 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26295 Alan Hayward <alan.hayward@arm.com>
26296 David Sherwood <david.sherwood@arm.com>
26298 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
26299 vfm1 with a bound_epilog parameter.
26300 (vect_do_peeling): Update calls accordingly, and move the prologue
26301 call earlier in the function. Treat the base bound_epilog as 0 for
26302 fully-masked loops and retain vf - 1 for other loops. Add 1 to
26303 this base when peeling for gaps.
26304 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
26305 with fully-masked loops.
26306 (vect_estimate_min_profitable_iters): Handle the single peeled
26307 iteration in that case.
26309 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26310 Alan Hayward <alan.hayward@arm.com>
26311 David Sherwood <david.sherwood@arm.com>
26313 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
26314 single-element interleaving even if the size is not a power of 2.
26315 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
26316 accesses for single-element interleaving if the group size is
26319 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26320 Alan Hayward <alan.hayward@arm.com>
26321 David Sherwood <david.sherwood@arm.com>
26323 * doc/md.texi (fold_extract_last_@var{m}): Document.
26324 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
26325 * optabs.def (fold_extract_last_optab): New optab.
26326 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
26327 * internal-fn.c (fold_extract_direct): New macro.
26328 (expand_fold_extract_optab_fn): Likewise.
26329 (direct_fold_extract_optab_supported_p): Likewise.
26330 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
26331 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26332 EXTRACT_LAST_REDUCTION.
26333 (get_initial_def_for_reduction): Do not create an initial vector
26334 for EXTRACT_LAST_REDUCTION reductions.
26335 (vectorizable_reduction): Leave the scalar phi in place for
26336 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
26337 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
26338 epilogue code for EXTRACT_LAST_REDUCTION and defer the
26339 transform phase to vectorizable_condition.
26340 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
26342 (vect_finish_stmt_generation): ...here.
26343 (vect_finish_replace_stmt): New function.
26344 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
26345 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
26347 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
26349 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26350 Alan Hayward <alan.hayward@arm.com>
26351 David Sherwood <david.sherwood@arm.com>
26353 * doc/md.texi (extract_last_@var{m}): Document.
26354 * optabs.def (extract_last_optab): New optab.
26355 * internal-fn.def (EXTRACT_LAST): New internal function.
26356 * internal-fn.c (cond_unary_direct): New macro.
26357 (expand_cond_unary_optab_fn): Likewise.
26358 (direct_cond_unary_optab_supported_p): Likewise.
26359 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
26360 loops using EXTRACT_LAST.
26361 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
26362 (extract_last_<mode>): ...this optab.
26363 (vec_extract<mode><Vel>): Update accordingly.
26365 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26366 Alan Hayward <alan.hayward@arm.com>
26367 David Sherwood <david.sherwood@arm.com>
26369 * target.def (empty_mask_is_expensive): New hook.
26370 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
26371 * doc/tm.texi: Regenerate.
26372 * targhooks.h (default_empty_mask_is_expensive): Declare.
26373 * targhooks.c (default_empty_mask_is_expensive): New function.
26374 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
26375 if the target says that empty masks are expensive.
26376 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
26378 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
26380 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26381 Alan Hayward <alan.hayward@arm.com>
26382 David Sherwood <david.sherwood@arm.com>
26384 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
26385 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
26386 (vect_use_loop_mask_for_alignment_p): New function.
26387 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
26388 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
26389 niters_skip argument. Make sure that the first niters_skip elements
26390 of the first iteration are inactive.
26391 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
26392 Update call to vect_set_loop_masks_directly.
26393 (get_misalign_in_elems): New function, split out from...
26394 (vect_gen_prolog_loop_niters): ...here.
26395 (vect_update_init_of_dr): Take a code argument that specifies whether
26396 the adjustment should be added or subtracted.
26397 (vect_update_init_of_drs): Likewise.
26398 (vect_prepare_for_masked_peels): New function.
26399 (vect_do_peeling): Skip prologue peeling if we're using a mask
26400 instead. Update call to vect_update_inits_of_drs.
26401 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26403 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
26404 alignment. Do not include the number of peeled iterations in
26405 the minimum threshold in that case.
26406 (vectorizable_induction): Adjust the start value down by
26407 LOOP_VINFO_MASK_SKIP_NITERS iterations.
26408 (vect_transform_loop): Call vect_prepare_for_masked_peels.
26409 Take the number of skipped iterations into account when calculating
26411 * tree-vect-stmts.c (vect_gen_while_not): New function.
26413 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26414 Alan Hayward <alan.hayward@arm.com>
26415 David Sherwood <david.sherwood@arm.com>
26417 * doc/sourcebuild.texi (vect_fully_masked): Document.
26418 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
26419 default value to 0.
26420 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
26422 (vect_analyze_loop_2): ...here. Don't check the vectorization
26423 factor against the number of loop iterations if the loop is
26426 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26427 Alan Hayward <alan.hayward@arm.com>
26428 David Sherwood <david.sherwood@arm.com>
26430 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
26431 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
26432 (dump_groups): Update accordingly.
26433 (iv_use::mem_type): New member variable.
26434 (address_p): New function.
26435 (record_use): Add a mem_type argument and initialize the new
26437 (record_group_use): Add a mem_type argument. Use address_p.
26438 Remove obsolete null checks of base_object. Update call to record_use.
26439 (find_interesting_uses_op): Update call to record_group_use.
26440 (find_interesting_uses_cond): Likewise.
26441 (find_interesting_uses_address): Likewise.
26442 (get_mem_type_for_internal_fn): New function.
26443 (find_address_like_use): Likewise.
26444 (find_interesting_uses_stmt): Try find_address_like_use before
26445 calling find_interesting_uses_op.
26446 (addr_offset_valid_p): Use the iv mem_type field as the type
26447 of the addressed memory.
26448 (add_autoinc_candidates): Likewise.
26449 (get_address_cost): Likewise.
26450 (split_small_address_groups_p): Use address_p.
26451 (split_address_groups): Likewise.
26452 (add_iv_candidate_for_use): Likewise.
26453 (autoinc_possible_for_pair): Likewise.
26454 (rewrite_groups): Likewise.
26455 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
26456 (determine_group_iv_cost): Update after split of USE_ADDRESS.
26457 (get_alias_ptr_type_for_ptr_address): New function.
26458 (rewrite_use_address): Rewrite address uses in calls that were
26459 identified by find_address_like_use.
26461 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26462 Alan Hayward <alan.hayward@arm.com>
26463 David Sherwood <david.sherwood@arm.com>
26465 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
26467 * gimple-expr.h (is_gimple_addressable: Likewise.
26468 * gimple-expr.c (is_gimple_address): Likewise.
26469 * internal-fn.c (expand_call_mem_ref): New function.
26470 (expand_mask_load_optab_fn): Use it.
26471 (expand_mask_store_optab_fn): Likewise.
26473 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26474 Alan Hayward <alan.hayward@arm.com>
26475 David Sherwood <david.sherwood@arm.com>
26477 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
26478 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
26479 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
26480 (cond_umax@var{mode}): Document.
26481 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
26482 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
26483 (cond_umin_optab, cond_umax_optab): New optabs.
26484 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
26485 (COND_IOR, COND_XOR): New internal functions.
26486 * internal-fn.h (get_conditional_internal_fn): Declare.
26487 * internal-fn.c (cond_binary_direct): New macro.
26488 (expand_cond_binary_optab_fn): Likewise.
26489 (direct_cond_binary_optab_supported_p): Likewise.
26490 (get_conditional_internal_fn): New function.
26491 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
26492 Cope with reduction statements that are vectorized as calls rather
26494 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
26495 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
26496 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
26497 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
26498 (UNSPEC_COND_EOR): New unspecs.
26499 (optab): Add mappings for them.
26500 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
26501 (sve_int_op, sve_fp_op): New int attributes.
26503 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26504 Alan Hayward <alan.hayward@arm.com>
26505 David Sherwood <david.sherwood@arm.com>
26507 * optabs.def (while_ult_optab): New optab.
26508 * doc/md.texi (while_ult@var{m}@var{n}): Document.
26509 * internal-fn.def (WHILE_ULT): New internal function.
26510 * internal-fn.h (direct_internal_fn_supported_p): New override
26511 that takes two types as argument.
26512 * internal-fn.c (while_direct): New macro.
26513 (expand_while_optab_fn): New function.
26514 (convert_optab_supported_p): Likewise.
26515 (direct_while_optab_supported_p): New macro.
26516 * wide-int.h (wi::udiv_ceil): New function.
26517 * tree-vectorizer.h (rgroup_masks): New structure.
26518 (vec_loop_masks): New typedef.
26519 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
26520 and fully_masked_p.
26521 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
26522 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
26523 (vect_max_vf): New function.
26524 (slpeel_make_loop_iterate_ntimes): Delete.
26525 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
26526 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
26527 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
26528 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
26529 internal-fn.h, stor-layout.h and optabs-query.h.
26530 (vect_set_loop_mask): New function.
26531 (add_preheader_seq): Likewise.
26532 (add_header_seq): Likewise.
26533 (interleave_supported_p): Likewise.
26534 (vect_maybe_permute_loop_masks): Likewise.
26535 (vect_set_loop_masks_directly): Likewise.
26536 (vect_set_loop_condition_masked): Likewise.
26537 (vect_set_loop_condition_unmasked): New function, split out from
26538 slpeel_make_loop_iterate_ntimes.
26539 (slpeel_make_loop_iterate_ntimes): Rename to..
26540 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
26541 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
26542 (vect_do_peeling): Update call accordingly.
26543 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
26545 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26546 mask_compare_type, can_fully_mask_p and fully_masked_p.
26547 (release_vec_loop_masks): New function.
26548 (_loop_vec_info): Use it to free the loop masks.
26549 (can_produce_all_loop_masks_p): New function.
26550 (vect_get_max_nscalars_per_iter): Likewise.
26551 (vect_verify_full_masking): Likewise.
26552 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26553 retries, and free the mask rgroups before retrying. Check loop-wide
26554 reasons for disallowing fully-masked loops. Make the final decision
26555 about whether use a fully-masked loop or not.
26556 (vect_estimate_min_profitable_iters): Do not assume that peeling
26557 for the number of iterations will be needed for fully-masked loops.
26558 (vectorizable_reduction): Disable fully-masked loops.
26559 (vectorizable_live_operation): Likewise.
26560 (vect_halve_mask_nunits): New function.
26561 (vect_double_mask_nunits): Likewise.
26562 (vect_record_loop_mask): Likewise.
26563 (vect_get_loop_mask): Likewise.
26564 (vect_transform_loop): Handle the case in which the final loop
26565 iteration might handle a partial vector. Call vect_set_loop_condition
26566 instead of slpeel_make_loop_iterate_ntimes.
26567 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26568 (check_load_store_masking): New function.
26569 (prepare_load_store_mask): Likewise.
26570 (vectorizable_store): Handle fully-masked loops.
26571 (vectorizable_load): Likewise.
26572 (supportable_widening_operation): Use vect_halve_mask_nunits for
26574 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26575 (vect_gen_while): New function.
26576 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26577 (aarch64_uqdec<mode>): New insn.
26579 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26580 Alan Hayward <alan.hayward@arm.com>
26581 David Sherwood <david.sherwood@arm.com>
26583 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26584 (reduc_xor_scal_optab): New optabs.
26585 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26586 (reduc_xor_scal_@var{m}): Document.
26587 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26588 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26589 internal functions.
26590 * fold-const-call.c (fold_const_call): Handle them.
26591 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26592 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26593 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26594 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26595 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26596 (UNSPEC_XORV): New unspecs.
26597 (optab): Add entries for them.
26598 (BITWISEV): New int iterator.
26599 (bit_reduc_op): New int attributes.
26601 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26602 Alan Hayward <alan.hayward@arm.com>
26603 David Sherwood <david.sherwood@arm.com>
26605 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26606 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26607 * optabs.def (vec_shl_insert_optab): New optab.
26608 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26609 (duplicate_and_interleave): Likewise.
26610 * tree-vect-loop.c: Include internal-fn.h.
26611 (neutral_op_for_slp_reduction): New function, split out from
26612 get_initial_defs_for_reduction.
26613 (get_initial_def_for_reduction): Handle option 2 for variable-length
26614 vectors by loading the neutral value into a vector and then shifting
26615 the initial value into element 0.
26616 (get_initial_defs_for_reduction): Replace the code argument with
26617 the neutral value calculated by neutral_op_for_slp_reduction.
26618 Use gimple_build_vector for constant-length vectors.
26619 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26620 but the first group_size elements have a neutral value.
26621 Use duplicate_and_interleave otherwise.
26622 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26623 Update call to get_initial_defs_for_reduction. Handle SLP
26624 reductions for variable-length vectors by creating one vector
26625 result for each scalar result, with the elements associated
26626 with other scalar results stubbed out with the neutral value.
26627 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26628 Require IFN_VEC_SHL_INSERT for double reductions on
26629 variable-length vectors, or SLP reductions that have
26630 a neutral value. Require can_duplicate_and_interleave_p
26631 support for variable-length unchained SLP reductions if there
26632 is no neutral value, such as for MIN/MAX reductions. Also require
26633 the number of vector elements to be a multiple of the number of
26634 SLP statements when doing variable-length unchained SLP reductions.
26635 Update call to vect_create_epilog_for_reduction.
26636 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26637 and remove initial values.
26638 (duplicate_and_interleave): Make public.
26639 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26640 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26642 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26643 Alan Hayward <alan.hayward@arm.com>
26644 David Sherwood <david.sherwood@arm.com>
26646 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26647 (can_duplicate_and_interleave_p): New function.
26648 (vect_get_and_check_slp_defs): Take the vector of statements
26649 rather than just the current one. Remove excess parentheses.
26650 Restriction rejectinon of vect_constant_def and vect_external_def
26651 for variable-length vectors to boolean types, or types for which
26652 can_duplicate_and_interleave_p is false.
26653 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26654 (duplicate_and_interleave): New function.
26655 (vect_get_constant_vectors): Use gimple_build_vector for
26656 constant-length vectors and suitable variable-length constant
26657 vectors. Use duplicate_and_interleave for other variable-length
26658 vectors. Don't defer the update when inserting new statements.
26660 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26661 Alan Hayward <alan.hayward@arm.com>
26662 David Sherwood <david.sherwood@arm.com>
26664 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26665 min_profitable_iters doesn't go negative.
26667 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26668 Alan Hayward <alan.hayward@arm.com>
26669 David Sherwood <david.sherwood@arm.com>
26671 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26672 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26673 * optabs.def (vec_mask_load_lanes_optab): New optab.
26674 (vec_mask_store_lanes_optab): Likewise.
26675 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26676 (MASK_STORE_LANES): Likewise.
26677 * internal-fn.c (mask_load_lanes_direct): New macro.
26678 (mask_store_lanes_direct): Likewise.
26679 (expand_mask_load_optab_fn): Handle masked operations.
26680 (expand_mask_load_lanes_optab_fn): New macro.
26681 (expand_mask_store_optab_fn): Handle masked operations.
26682 (expand_mask_store_lanes_optab_fn): New macro.
26683 (direct_mask_load_lanes_optab_supported_p): Likewise.
26684 (direct_mask_store_lanes_optab_supported_p): Likewise.
26685 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26687 (vect_load_lanes_supported): Likewise.
26688 * tree-vect-data-refs.c (strip_conversion): New function.
26689 (can_group_stmts_p): Likewise.
26690 (vect_analyze_data_ref_accesses): Use it instead of checking
26691 for a pair of assignments.
26692 (vect_store_lanes_supported): Take a masked_p parameter.
26693 (vect_load_lanes_supported): Likewise.
26694 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26695 vect_store_lanes_supported and vect_load_lanes_supported.
26696 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26697 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26698 parameter. Don't allow gaps for masked accesses.
26699 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26700 and vect_load_lanes_supported.
26701 (get_load_store_type): Take a masked_p parameter and update
26702 call to get_group_load_store_type.
26703 (vectorizable_store): Update call to get_load_store_type.
26704 Handle IFN_MASK_STORE_LANES.
26705 (vectorizable_load): Update call to get_load_store_type.
26706 Handle IFN_MASK_LOAD_LANES.
26708 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26709 Alan Hayward <alan.hayward@arm.com>
26710 David Sherwood <david.sherwood@arm.com>
26712 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26714 * config/aarch64/aarch64-protos.h
26715 (aarch64_sve_struct_memory_operand_p): Declare.
26716 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26717 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26718 (VPRED, vpred): Handle SVE structure modes.
26719 * config/aarch64/constraints.md (Utx): New constraint.
26720 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26721 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26722 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26723 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26724 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26725 structure modes. Split into pieces after RA.
26726 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26727 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26729 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26730 SVE structure modes.
26731 (aarch64_classify_address): Likewise.
26732 (sizetochar): Move earlier in file.
26733 (aarch64_print_operand): Handle SVE register lists.
26734 (aarch64_array_mode): New function.
26735 (aarch64_sve_struct_memory_operand_p): Likewise.
26736 (TARGET_ARRAY_MODE): Redefine.
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 * target.def (array_mode): New target hook.
26743 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26744 * doc/tm.texi: Regenerate.
26745 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26746 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26747 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26748 targetm.array_mode.
26749 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26752 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26753 Alan Hayward <alan.hayward@arm.com>
26754 David Sherwood <david.sherwood@arm.com>
26756 * fold-const.c (fold_binary_loc): Check the argument types
26757 rather than the result type when testing for a vector operation.
26759 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26761 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26762 * doc/tm.texi: Regenerate.
26764 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26765 Alan Hayward <alan.hayward@arm.com>
26766 David Sherwood <david.sherwood@arm.com>
26768 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26769 (sve): Document new AArch64 extension.
26770 * doc/md.texi (w): Extend the description of the AArch64
26771 constraint to include SVE vectors.
26772 (Upl, Upa): Document new AArch64 predicate constraints.
26773 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26775 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26776 (msve-vector-bits=): New option.
26777 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26778 SVE when these are disabled.
26779 (sve): New extension.
26780 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26781 modes. Adjust their number of units based on aarch64_sve_vg.
26782 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26783 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26784 aarch64_addr_query_type.
26785 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26786 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26787 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26788 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26789 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26790 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26791 (aarch64_simd_imm_zero_p): Delete.
26792 (aarch64_check_zero_based_sve_index_immediate): Declare.
26793 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26794 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26795 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26796 (aarch64_sve_float_mul_immediate_p): Likewise.
26797 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26798 rather than an rtx.
26799 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26800 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26801 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26802 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26803 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26804 (aarch64_regmode_natural_size): Likewise.
26805 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26806 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26808 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26809 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26810 for VG and the SVE predicate registers.
26811 (V_ALIASES): Add a "z"-prefixed alias.
26812 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26813 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26814 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26815 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26816 (REG_CLASS_NAMES): Add entries for them.
26817 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26818 and the predicate registers.
26819 (aarch64_sve_vg): Declare.
26820 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26821 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26822 (REGMODE_NATURAL_SIZE): Define.
26823 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26825 * config/aarch64/aarch64.c: Include cfgrtl.h.
26826 (simd_immediate_info): Add a constructor for series vectors,
26827 and an associated step field.
26828 (aarch64_sve_vg): New variable.
26829 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26830 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26831 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26832 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26833 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26834 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26835 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26836 (aarch64_get_mask_mode): New functions.
26837 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26838 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26839 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26840 predicate modes and predicate registers. Explicitly restrict
26841 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26842 to store a vector mode if it is recognized by
26843 aarch64_classify_vector_mode.
26844 (aarch64_regmode_natural_size): New function.
26845 (aarch64_hard_regno_caller_save_mode): Return the original mode
26847 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26848 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26849 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26850 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26852 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26853 does not overlap dest if the function is frame-related. Handle
26855 (aarch64_split_add_offset): New function.
26856 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26857 them aarch64_add_offset.
26858 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26859 and update call to aarch64_sub_sp.
26860 (aarch64_add_cfa_expression): New function.
26861 (aarch64_expand_prologue): Pass extra temporary registers to the
26862 functions above. Handle the case in which we need to emit new
26863 DW_CFA_expressions for registers that were originally saved
26864 relative to the stack pointer, but now have to be expressed
26865 relative to the frame pointer.
26866 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26868 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26869 IP0 and IP1 values for SVE frames.
26870 (aarch64_expand_vec_series): New function.
26871 (aarch64_expand_sve_widened_duplicate): Likewise.
26872 (aarch64_expand_sve_const_vector): Likewise.
26873 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26874 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26875 into the register, rather than emitting a SET directly.
26876 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26877 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26878 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26879 (offset_9bit_signed_scaled_p): New functions.
26880 (aarch64_replicate_bitmask_imm): New function.
26881 (aarch64_bitmask_imm): Use it.
26882 (aarch64_cannot_force_const_mem): Reject expressions involving
26883 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26884 (aarch64_classify_index): Handle SVE indices, by requiring
26885 a plain register index with a scale that matches the element size.
26886 (aarch64_classify_address): Handle SVE addresses. Assert that
26887 the mode of the address is VOIDmode or an integer mode.
26888 Update call to aarch64_classify_symbol.
26889 (aarch64_classify_symbolic_expression): Update call to
26890 aarch64_classify_symbol.
26891 (aarch64_const_vec_all_in_range_p): New function.
26892 (aarch64_print_vector_float_operand): Likewise.
26893 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26894 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26895 and the FP immediates 1.0 and 0.5.
26896 (aarch64_print_address_internal): Handle SVE addresses.
26897 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26898 (aarch64_regno_regclass): Handle predicate registers.
26899 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26901 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26902 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26903 (aarch64_convert_sve_vector_bits): New function.
26904 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26905 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26906 rather than an rtx.
26907 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26908 Handle SVE vector and predicate modes. Accept VL-based constants
26909 that need only one temporary register, and VL offsets that require
26910 no temporary registers.
26911 (aarch64_conditional_register_usage): Mark the predicate registers
26912 as fixed if SVE isn't available.
26913 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26914 Return true for SVE vector and predicate modes.
26915 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26916 rather than an unsigned int. Handle SVE modes.
26917 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26919 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26921 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26922 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26923 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26924 (aarch64_sve_float_mul_immediate_p): New functions.
26925 (aarch64_sve_valid_immediate): New function.
26926 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26927 Explicitly reject structure modes. Check for INDEX constants.
26928 Handle PTRUE and PFALSE constants.
26929 (aarch64_check_zero_based_sve_index_immediate): New function.
26930 (aarch64_simd_imm_zero_p): Delete.
26931 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26932 vector modes. Accept constants in the range of CNT[BHWD].
26933 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26934 ask for an Advanced SIMD mode.
26935 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26936 (aarch64_simd_vector_alignment): Handle SVE predicates.
26937 (aarch64_vectorize_preferred_vector_alignment): New function.
26938 (aarch64_simd_vector_alignment_reachable): Use it instead of
26940 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26941 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26943 (MAX_VECT_LEN): Delete.
26944 (expand_vec_perm_d): Add a vec_flags field.
26945 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26946 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26947 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26949 (aarch64_evpc_rev): Rename to...
26950 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26951 (aarch64_evpc_rev_global): New function.
26952 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26953 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26955 (aarch64_evpc_sve_tbl): New function.
26956 (aarch64_expand_vec_perm_const_1): Update after rename of
26957 aarch64_evpc_rev. Handle SVE permutes too, trying
26958 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26959 than aarch64_evpc_tbl.
26960 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26961 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26962 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26963 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26964 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26965 (aarch64_expand_sve_vcond): New functions.
26966 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26967 of aarch64_vector_mode_p.
26968 (aarch64_dwarf_poly_indeterminate_value): New function.
26969 (aarch64_compute_pressure_classes): Likewise.
26970 (aarch64_can_change_mode_class): Likewise.
26971 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26972 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26973 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26974 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26975 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26976 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26977 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26978 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26980 (Dn, Dl, Dr): Accept const as well as const_vector.
26981 (Dz): Likewise. Compare against CONST0_RTX.
26982 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26983 of "vector" where appropriate.
26984 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26985 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26986 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26987 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26988 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26989 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26990 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26991 (v_int_equiv): Extend to SVE modes.
26992 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26994 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26995 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26996 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26997 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26998 (SVE_COND_FP_CMP): New int iterators.
26999 (perm_hilo): Handle the new unpack unspecs.
27000 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
27002 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
27003 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
27004 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
27005 (aarch64_equality_operator, aarch64_constant_vector_operand)
27006 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
27007 (aarch64_sve_nonimmediate_operand): Likewise.
27008 (aarch64_sve_general_operand): Likewise.
27009 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
27010 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
27011 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
27012 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
27013 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
27014 (aarch64_sve_float_arith_immediate): Likewise.
27015 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
27016 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
27017 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
27018 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
27019 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
27020 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
27021 (aarch64_sve_float_arith_operand): Likewise.
27022 (aarch64_sve_float_arith_with_sub_operand): Likewise.
27023 (aarch64_sve_float_mul_operand): Likewise.
27024 (aarch64_sve_vec_perm_operand): Likewise.
27025 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
27026 (aarch64_mov_operand): Accept const_poly_int and const_vector.
27027 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
27028 as well as const_vector.
27029 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
27030 in file. Use CONST0_RTX and CONSTM1_RTX.
27031 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
27032 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
27033 Use aarch64_simd_imm_zero.
27034 * config/aarch64/aarch64-sve.md: New file.
27035 * config/aarch64/aarch64.md: Include it.
27036 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
27037 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
27038 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
27039 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
27040 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
27041 (sve): New attribute.
27042 (enabled): Disable instructions with the sve attribute unless
27044 (movqi, movhi): Pass CONST_POLY_INT operaneds through
27045 aarch64_expand_mov_immediate.
27046 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
27047 CNT[BHSD] immediates.
27048 (movti): Split CONST_POLY_INT moves into two halves.
27049 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
27050 Split additions that need a temporary here if the destination
27051 is the stack pointer.
27052 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
27053 (*add<mode>3_poly_1): New instruction.
27054 (set_clobber_cc): New expander.
27056 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27058 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
27059 parameter and use it instead of GET_MODE_SIZE (innermode). Use
27060 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
27061 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
27062 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
27063 Change innermode from fixed_mode_size to machine_mode.
27064 (simplify_subreg): Update call accordingly. Handle a constant-sized
27065 subreg of a variable-length CONST_VECTOR.
27067 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27068 Alan Hayward <alan.hayward@arm.com>
27069 David Sherwood <david.sherwood@arm.com>
27071 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
27072 (add_offset_to_base): New function, split out from...
27073 (create_mem_ref): ...here. When handling a scale other than 1,
27074 check first whether the address is valid without the offset.
27075 Add it into the base if so, leaving the index and scale as-is.
27077 2018-01-12 Jakub Jelinek <jakub@redhat.com>
27080 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
27081 fold_for_warn before checking if arg2 is INTEGER_CST.
27083 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
27085 * config/rs6000/predicates.md (load_multiple_operation): Delete.
27086 (store_multiple_operation): Delete.
27087 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
27088 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
27089 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
27090 guarded by TARGET_STRING.
27091 (rs6000_output_load_multiple): Delete.
27092 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27093 OPTION_MASK_STRING / TARGET_STRING handling.
27094 (print_operand) <'N', 'O'>: Add comment that these are unused now.
27095 (const rs6000_opt_masks) <"string">: Change mask to 0.
27096 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
27097 (MASK_STRING): Delete.
27098 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
27100 (load_multiple): Delete.
27107 (store_multiple): Delete.
27114 (movmemsi_8reg): Delete.
27115 (corresponding unnamed define_insn): Delete.
27116 (movmemsi_6reg): Delete.
27117 (corresponding unnamed define_insn): Delete.
27118 (movmemsi_4reg): Delete.
27119 (corresponding unnamed define_insn): Delete.
27120 (movmemsi_2reg): Delete.
27121 (corresponding unnamed define_insn): Delete.
27122 (movmemsi_1reg): Delete.
27123 (corresponding unnamed define_insn): Delete.
27124 * config/rs6000/rs6000.opt (mno-string): New.
27125 (mstring): Replace by deprecation warning stub.
27126 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
27128 2018-01-12 Jakub Jelinek <jakub@redhat.com>
27130 * regrename.c (regrename_do_replace): If replacing the same
27131 reg multiple times, try to reuse last created gen_raw_REG.
27134 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
27135 main to workaround a bug in GDB.
27137 2018-01-12 Tom de Vries <tom@codesourcery.com>
27140 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
27142 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
27144 PR rtl-optimization/80481
27145 * ira-color.c (get_cap_member): New function.
27146 (allocnos_conflict_by_live_ranges_p): Use it.
27147 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
27148 (setup_slot_coalesced_allocno_live_ranges): Ditto.
27150 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
27153 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
27154 (*saddl_se_1): Ditto.
27155 (*ssubsi_1): Ditto.
27156 (*ssubl_se_1): Ditto.
27158 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27160 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
27161 rather than wi::to_widest for DR_INITs.
27162 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
27163 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
27164 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
27166 (vect_analyze_group_access_1): Note that here.
27168 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27170 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
27171 polynomial type sizes.
27173 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27175 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
27176 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
27177 (gimple_add_tmp_var): Likewise.
27179 2018-01-12 Martin Liska <mliska@suse.cz>
27181 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
27182 (gimple_alloc_sizes): Likewise.
27183 (dump_gimple_statistics): Use PRIu64 in printf format.
27184 * gimple.h: Change uint64_t to int.
27186 2018-01-12 Martin Liska <mliska@suse.cz>
27188 * tree-core.h: Use uint64_t instead of int.
27189 * tree.c (tree_node_counts): Likewise.
27190 (tree_node_sizes): Likewise.
27191 (dump_tree_statistics): Use PRIu64 in printf format.
27193 2018-01-12 Martin Liska <mliska@suse.cz>
27195 * Makefile.in: As qsort_chk is implemented in vec.c, add
27196 vec.o to linkage of gencfn-macros.
27197 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
27198 passing the info to record_node_allocation_statistics.
27199 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
27201 * ggc-common.c (struct ggc_usage): Add operator== and use
27202 it in operator< and compare function.
27203 * mem-stats.h (struct mem_usage): Likewise.
27204 * vec.c (struct vec_usage): Remove operator< and compare
27205 function. Can be simply inherited.
27207 2018-01-12 Martin Jambor <mjambor@suse.cz>
27210 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
27211 * tree-ssa-math-opts.c: Include domwalk.h.
27212 (convert_mult_to_fma_1): New function.
27213 (fma_transformation_info): New type.
27214 (fma_deferring_state): Likewise.
27215 (cancel_fma_deferring): New function.
27216 (result_of_phi): Likewise.
27217 (last_fma_candidate_feeds_initial_phi): Likewise.
27218 (convert_mult_to_fma): Added deferring logic, split actual
27219 transformation to convert_mult_to_fma_1.
27220 (math_opts_dom_walker): New type.
27221 (math_opts_dom_walker::after_dom_children): New method, body moved
27222 here from pass_optimize_widening_mul::execute, added deferring logic
27224 (pass_optimize_widening_mul::execute): Moved most of code to
27225 math_opts_dom_walker::after_dom_children.
27226 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
27227 * config/i386/i386.c (ix86_option_override_internal): Added
27228 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
27230 2018-01-12 Richard Biener <rguenther@suse.de>
27233 * dwarf2out.c (gen_variable_die): Do not reset old_die for
27234 inline instance vars.
27236 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
27239 * config/rx/rx.c (rx_is_restricted_memory_address):
27240 Handle SUBREG case.
27242 2018-01-12 Richard Biener <rguenther@suse.de>
27244 PR tree-optimization/80846
27245 * target.def (split_reduction): New target hook.
27246 * targhooks.c (default_split_reduction): New function.
27247 * targhooks.h (default_split_reduction): Declare.
27248 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
27249 target requests first reduce vectors by combining low and high
27251 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
27252 (get_vectype_for_scalar_type_and_size): Export.
27253 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
27254 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
27255 * doc/tm.texi: Regenerate.
27256 * config/i386/i386.c (ix86_split_reduction): Implement
27257 TARGET_VECTORIZE_SPLIT_REDUCTION.
27259 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27262 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
27263 in PIC mode except for TARGET_VXWORKS_RTP.
27264 * config/sparc/sparc.c: Include cfgrtl.h.
27265 (TARGET_INIT_PIC_REG): Define.
27266 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
27267 (sparc_pic_register_p): New predicate.
27268 (sparc_legitimate_address_p): Use it.
27269 (sparc_legitimize_pic_address): Likewise.
27270 (sparc_delegitimize_address): Likewise.
27271 (sparc_mode_dependent_address_p): Likewise.
27272 (gen_load_pcrel_sym): Remove 4th parameter.
27273 (load_got_register): Adjust call to above. Remove obsolete stuff.
27274 (sparc_expand_prologue): Do not call load_got_register here.
27275 (sparc_flat_expand_prologue): Likewise.
27276 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
27277 (sparc_use_pseudo_pic_reg): New function.
27278 (sparc_init_pic_reg): Likewise.
27279 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
27280 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
27282 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
27284 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27285 Add item for branch_cost.
27287 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27289 PR rtl-optimization/83565
27290 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
27291 not extend the result to a larger mode for rotate operations.
27292 (num_sign_bit_copies1): Likewise.
27294 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27297 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
27299 Use values-Xc.o for -pedantic.
27300 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
27302 2018-01-12 Martin Liska <mliska@suse.cz>
27305 * ipa-devirt.c (final_warning_record::grow_type_warnings):
27307 (possible_polymorphic_call_targets): Use it.
27308 (ipa_devirt): Likewise.
27310 2018-01-12 Martin Liska <mliska@suse.cz>
27312 * profile-count.h (enum profile_quality): Use 0 as invalid
27313 enum value of profile_quality.
27315 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
27317 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
27318 -mext-string options.
27320 2018-01-12 Richard Biener <rguenther@suse.de>
27322 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
27323 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
27324 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
27326 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
27328 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27330 * configure.ac (--with-long-double-format): Add support for the
27331 configuration option to change the default long double format on
27333 * config.gcc (powerpc*-linux*-*): Likewise.
27334 * configure: Regenerate.
27335 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
27336 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
27337 used without modification.
27339 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27341 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
27342 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
27343 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
27344 MISC_BUILTIN_SPEC_BARRIER.
27345 (rs6000_init_builtins): Likewise.
27346 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
27348 (speculation_barrier): New define_insn.
27349 * doc/extend.texi: Document __builtin_speculation_barrier.
27351 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27354 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
27355 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
27356 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
27358 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
27359 integral modes instead of "ss" and "sd".
27360 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
27361 vectors with 32-bit and 64-bit elements.
27362 (vecdupssescalarmodesuffix): New mode attribute.
27363 (vec_dup<mode>): Use it.
27365 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
27368 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
27369 frame if argument is passed on stack.
27371 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27374 * ree.c (combine_reaching_defs): Optimize also
27375 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
27376 reg2=any_extend(exp); reg1=reg2;, formatting fix.
27378 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27380 PR middle-end/83189
27381 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
27383 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27385 PR middle-end/83718
27386 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
27387 after they are computed.
27389 2018-01-11 Bin Cheng <bin.cheng@arm.com>
27391 PR tree-optimization/83695
27392 * gimple-loop-linterchange.cc
27393 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
27394 reset cached scev information after interchange.
27395 (pass_linterchange::execute): Remove call to scev_reset_htab.
27397 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27399 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
27400 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
27401 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
27402 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
27403 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
27404 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
27405 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
27406 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
27407 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
27408 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
27409 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
27410 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
27411 (V_lane_reg): Likewise.
27412 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
27414 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
27415 (vfmal_lane_low<mode>_intrinsic,
27416 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
27417 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
27418 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
27419 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
27420 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
27421 vfmsl_lane_high<mode>_intrinsic): New define_insns.
27423 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27425 * config/arm/arm-cpus.in (fp16fml): New feature.
27426 (ALL_SIMD): Add fp16fml.
27427 (armv8.2-a): Add fp16fml as an option.
27428 (armv8.3-a): Likewise.
27429 (armv8.4-a): Add fp16fml as part of fp16.
27430 * config/arm/arm.h (TARGET_FP16FML): Define.
27431 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
27433 * config/arm/arm-modes.def (V2HF): Define.
27434 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
27435 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
27436 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
27437 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
27438 vfmsl_low, vfmsl_high): New set of builtins.
27439 * config/arm/iterators.md (PLUSMINUS): New code iterator.
27440 (vfml_op): New code attribute.
27441 (VFMLHALVES): New int iterator.
27442 (VFML, VFMLSEL): New mode attributes.
27443 (V_reg): Define mapping for V2HF.
27444 (V_hi, V_lo): New mode attributes.
27445 (VF_constraint): Likewise.
27446 (vfml_half, vfml_half_selector): New int attributes.
27447 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
27449 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
27450 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
27452 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
27453 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
27454 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
27455 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
27457 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
27458 Document new effective target and option set.
27460 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27462 * config/arm/arm-cpus.in (armv8_4): New feature.
27463 (ARMv8_4a): New fgroup.
27464 (armv8.4-a): New arch.
27465 * config/arm/arm-tables.opt: Regenerate.
27466 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
27467 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
27468 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
27469 Add matching rules for -march=armv8.4-a and extensions.
27470 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
27472 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
27475 * config/rx/rx.md (BW): New mode attribute.
27476 (sync_lock_test_and_setsi): Add mode suffix to insn output.
27478 2018-01-11 Richard Biener <rguenther@suse.de>
27480 PR tree-optimization/83435
27481 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
27482 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
27483 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
27485 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27486 Alan Hayward <alan.hayward@arm.com>
27487 David Sherwood <david.sherwood@arm.com>
27489 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
27491 (aarch64_classify_address): Initialize it. Track polynomial offsets.
27492 (aarch64_print_address_internal): Use it to check for a zero offset.
27494 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27495 Alan Hayward <alan.hayward@arm.com>
27496 David Sherwood <david.sherwood@arm.com>
27498 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
27499 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
27500 Return a poly_int64 rather than a HOST_WIDE_INT.
27501 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
27502 rather than a HOST_WIDE_INT.
27503 * config/aarch64/aarch64.h (aarch64_frame): Protect with
27504 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
27505 hard_fp_offset, frame_size, initial_adjust, callee_offset and
27506 final_offset from HOST_WIDE_INT to poly_int64.
27507 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
27508 to_constant when getting the number of units in an Advanced SIMD
27510 (aarch64_builtin_vectorized_function): Check for a constant number
27512 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
27514 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
27515 attribute instead of GET_MODE_NUNITS.
27516 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
27517 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
27518 GET_MODE_SIZE for fixed-size registers.
27519 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
27520 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
27521 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
27522 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
27523 (aarch64_print_operand, aarch64_print_address_internal)
27524 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
27525 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
27526 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
27527 Handle polynomial GET_MODE_SIZE.
27528 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
27529 wider than SImode without modification.
27530 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
27531 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
27532 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
27533 passing and returning SVE modes.
27534 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
27535 rather than GEN_INT.
27536 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
27537 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
27538 (aarch64_allocate_and_probe_stack_space): Likewise.
27539 (aarch64_layout_frame): Cope with polynomial offsets.
27540 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
27541 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
27542 polynomial offsets.
27543 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
27544 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
27545 poly_int64 rather than a HOST_WIDE_INT.
27546 (aarch64_get_separate_components, aarch64_process_components)
27547 (aarch64_expand_prologue, aarch64_expand_epilogue)
27548 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
27549 (aarch64_anchor_offset): New function, split out from...
27550 (aarch64_legitimize_address): ...here.
27551 (aarch64_builtin_vectorization_cost): Handle polynomial
27552 TYPE_VECTOR_SUBPARTS.
27553 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27555 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27556 number of elements from the PARALLEL rather than the mode.
27557 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27558 rather than GET_MODE_BITSIZE.
27559 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27560 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27561 (aarch64_expand_vec_perm_const_1): Handle polynomial
27562 d->perm.length () and d->perm elements.
27563 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27564 Apply to_constant to d->perm elements.
27565 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27566 polynomial CONST_VECTOR_NUNITS.
27567 (aarch64_move_pointer): Take amount as a poly_int64 rather
27569 (aarch64_progress_pointer): Avoid temporary variable.
27570 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27571 the mode attribute instead of GET_MODE.
27573 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27574 Alan Hayward <alan.hayward@arm.com>
27575 David Sherwood <david.sherwood@arm.com>
27577 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27578 x exists before using it.
27579 (aarch64_add_constant_internal): Rename to...
27580 (aarch64_add_offset_1): ...this. Replace regnum with separate
27581 src and dest rtxes. Handle the case in which they're different,
27582 including when the offset is zero. Replace scratchreg with an rtx.
27583 Use 2 additions if there is no spare register into which we can
27584 move a 16-bit constant.
27585 (aarch64_add_constant): Delete.
27586 (aarch64_add_offset): Replace reg with separate src and dest
27587 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27588 Use aarch64_add_offset_1.
27589 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27590 an rtx rather than an int. Take the delta as a poly_int64
27591 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27592 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27593 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27594 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27595 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27596 and aarch64_add_sp.
27597 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27598 aarch64_add_constant.
27600 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27602 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27603 Use scalar_float_mode.
27605 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27607 * config/aarch64/aarch64-simd.md
27608 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27609 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27610 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27611 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27612 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27613 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27614 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27615 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27616 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27617 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27619 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27622 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27623 targ_options->x_arm_arch_string is non NULL.
27625 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27627 * config/aarch64/aarch64.h
27628 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27630 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27633 * expmed.c (emit_store_flag_force): Swap if const op0
27634 and change VOIDmode to mode of op0.
27636 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27638 PR rtl-optimization/83761
27639 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27640 than bytes to mode_for_size.
27642 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27644 PR middle-end/83189
27645 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27646 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27649 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27651 PR middle-end/83575
27652 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27653 when in layout mode.
27654 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27655 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27658 2018-01-10 Michael Collison <michael.collison@arm.com>
27660 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27661 * config/aarch64/aarch64-option-extension.def: Add
27662 AARCH64_OPT_EXTENSION of 'fp16fml'.
27663 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27664 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27665 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27666 * config/aarch64/constraints.md (Ui7): New constraint.
27667 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27668 (VFMLA_SEL_W): Ditto.
27671 (VFMLA16_LOW): New int iterator.
27672 (VFMLA16_HIGH): Ditto.
27673 (UNSPEC_FMLAL): New unspec.
27674 (UNSPEC_FMLSL): Ditto.
27675 (UNSPEC_FMLAL2): Ditto.
27676 (UNSPEC_FMLSL2): Ditto.
27677 (f16mac): New code attribute.
27678 * config/aarch64/aarch64-simd-builtins.def
27679 (aarch64_fmlal_lowv2sf): Ditto.
27680 (aarch64_fmlsl_lowv2sf): Ditto.
27681 (aarch64_fmlalq_lowv4sf): Ditto.
27682 (aarch64_fmlslq_lowv4sf): Ditto.
27683 (aarch64_fmlal_highv2sf): Ditto.
27684 (aarch64_fmlsl_highv2sf): Ditto.
27685 (aarch64_fmlalq_highv4sf): Ditto.
27686 (aarch64_fmlslq_highv4sf): Ditto.
27687 (aarch64_fmlal_lane_lowv2sf): Ditto.
27688 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27689 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27690 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27691 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27692 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27693 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27694 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27695 (aarch64_fmlal_lane_highv2sf): Ditto.
27696 (aarch64_fmlsl_lane_highv2sf): Ditto.
27697 (aarch64_fmlal_laneq_highv2sf): Ditto.
27698 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27699 (aarch64_fmlalq_lane_highv4sf): Ditto.
27700 (aarch64_fmlsl_lane_highv4sf): Ditto.
27701 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27702 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27703 * config/aarch64/aarch64-simd.md:
27704 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27705 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27706 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27707 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27708 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27709 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27710 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27711 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27712 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27713 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27714 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27715 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27716 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27717 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27718 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27719 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27720 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27721 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27722 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27723 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27724 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27725 (vfmlsl_low_u32): Ditto.
27726 (vfmlalq_low_u32): Ditto.
27727 (vfmlslq_low_u32): Ditto.
27728 (vfmlal_high_u32): Ditto.
27729 (vfmlsl_high_u32): Ditto.
27730 (vfmlalq_high_u32): Ditto.
27731 (vfmlslq_high_u32): Ditto.
27732 (vfmlal_lane_low_u32): Ditto.
27733 (vfmlsl_lane_low_u32): Ditto.
27734 (vfmlal_laneq_low_u32): Ditto.
27735 (vfmlsl_laneq_low_u32): Ditto.
27736 (vfmlalq_lane_low_u32): Ditto.
27737 (vfmlslq_lane_low_u32): Ditto.
27738 (vfmlalq_laneq_low_u32): Ditto.
27739 (vfmlslq_laneq_low_u32): Ditto.
27740 (vfmlal_lane_high_u32): Ditto.
27741 (vfmlsl_lane_high_u32): Ditto.
27742 (vfmlal_laneq_high_u32): Ditto.
27743 (vfmlsl_laneq_high_u32): Ditto.
27744 (vfmlalq_lane_high_u32): Ditto.
27745 (vfmlslq_lane_high_u32): Ditto.
27746 (vfmlalq_laneq_high_u32): Ditto.
27747 (vfmlslq_laneq_high_u32): Ditto.
27748 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27749 (AARCH64_FL_FOR_ARCH8_4): New.
27750 (AARCH64_ISA_F16FML): New ISA flag.
27751 (TARGET_F16FML): New feature flag for fp16fml.
27752 (doc/invoke.texi): Document new fp16fml option.
27754 2018-01-10 Michael Collison <michael.collison@arm.com>
27756 * config/aarch64/aarch64-builtins.c:
27757 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27758 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27759 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27760 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27761 (AARCH64_ISA_SHA3): New ISA flag.
27762 (TARGET_SHA3): New feature flag for sha3.
27763 * config/aarch64/iterators.md (sha512_op): New int attribute.
27764 (CRYPTO_SHA512): New int iterator.
27765 (UNSPEC_SHA512H): New unspec.
27766 (UNSPEC_SHA512H2): Ditto.
27767 (UNSPEC_SHA512SU0): Ditto.
27768 (UNSPEC_SHA512SU1): Ditto.
27769 * config/aarch64/aarch64-simd-builtins.def
27770 (aarch64_crypto_sha512hqv2di): New builtin.
27771 (aarch64_crypto_sha512h2qv2di): Ditto.
27772 (aarch64_crypto_sha512su0qv2di): Ditto.
27773 (aarch64_crypto_sha512su1qv2di): Ditto.
27774 (aarch64_eor3qv8hi): Ditto.
27775 (aarch64_rax1qv2di): Ditto.
27776 (aarch64_xarqv2di): Ditto.
27777 (aarch64_bcaxqv8hi): Ditto.
27778 * config/aarch64/aarch64-simd.md:
27779 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27780 (aarch64_crypto_sha512su0qv2di): Ditto.
27781 (aarch64_crypto_sha512su1qv2di): Ditto.
27782 (aarch64_eor3qv8hi): Ditto.
27783 (aarch64_rax1qv2di): Ditto.
27784 (aarch64_xarqv2di): Ditto.
27785 (aarch64_bcaxqv8hi): Ditto.
27786 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27787 (vsha512h2q_u64): Ditto.
27788 (vsha512su0q_u64): Ditto.
27789 (vsha512su1q_u64): Ditto.
27790 (veor3q_u16): Ditto.
27791 (vrax1q_u64): Ditto.
27792 (vxarq_u64): Ditto.
27793 (vbcaxq_u16): Ditto.
27794 * config/arm/types.md (crypto_sha512): New type attribute.
27795 (crypto_sha3): Ditto.
27796 (doc/invoke.texi): Document new sha3 option.
27798 2018-01-10 Michael Collison <michael.collison@arm.com>
27800 * config/aarch64/aarch64-builtins.c:
27801 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27802 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27803 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27804 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27805 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27806 (AARCH64_ISA_SM4): New ISA flag.
27807 (TARGET_SM4): New feature flag for sm4.
27808 * config/aarch64/aarch64-simd-builtins.def
27809 (aarch64_sm3ss1qv4si): Ditto.
27810 (aarch64_sm3tt1aq4si): Ditto.
27811 (aarch64_sm3tt1bq4si): Ditto.
27812 (aarch64_sm3tt2aq4si): Ditto.
27813 (aarch64_sm3tt2bq4si): Ditto.
27814 (aarch64_sm3partw1qv4si): Ditto.
27815 (aarch64_sm3partw2qv4si): Ditto.
27816 (aarch64_sm4eqv4si): Ditto.
27817 (aarch64_sm4ekeyqv4si): Ditto.
27818 * config/aarch64/aarch64-simd.md:
27819 (aarch64_sm3ss1qv4si): Ditto.
27820 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27821 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27822 (aarch64_sm4eqv4si): Ditto.
27823 (aarch64_sm4ekeyqv4si): Ditto.
27824 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27825 (sm3part_op): Ditto.
27826 (CRYPTO_SM3TT): Ditto.
27827 (CRYPTO_SM3PART): Ditto.
27828 (UNSPEC_SM3SS1): New unspec.
27829 (UNSPEC_SM3TT1A): Ditto.
27830 (UNSPEC_SM3TT1B): Ditto.
27831 (UNSPEC_SM3TT2A): Ditto.
27832 (UNSPEC_SM3TT2B): Ditto.
27833 (UNSPEC_SM3PARTW1): Ditto.
27834 (UNSPEC_SM3PARTW2): Ditto.
27835 (UNSPEC_SM4E): Ditto.
27836 (UNSPEC_SM4EKEY): Ditto.
27837 * config/aarch64/constraints.md (Ui2): New constraint.
27838 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27839 * config/arm/types.md (crypto_sm3): New type attribute.
27840 (crypto_sm4): Ditto.
27841 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27842 (vsm3tt1aq_u32): Ditto.
27843 (vsm3tt1bq_u32): Ditto.
27844 (vsm3tt2aq_u32): Ditto.
27845 (vsm3tt2bq_u32): Ditto.
27846 (vsm3partw1q_u32): Ditto.
27847 (vsm3partw2q_u32): Ditto.
27848 (vsm4eq_u32): Ditto.
27849 (vsm4ekeyq_u32): Ditto.
27850 (doc/invoke.texi): Document new sm4 option.
27852 2018-01-10 Michael Collison <michael.collison@arm.com>
27854 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27855 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27856 (AARCH64_FL_FOR_ARCH8_4): New.
27857 (AARCH64_FL_V8_4): New flag.
27858 (doc/invoke.texi): Document new armv8.4-a option.
27860 2018-01-10 Michael Collison <michael.collison@arm.com>
27862 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27863 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27864 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27865 * config/aarch64/aarch64-option-extension.def: Add
27866 AARCH64_OPT_EXTENSION of 'sha2'.
27867 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27868 (crypto): Disable sha2 and aes if crypto disabled.
27869 (crypto): Enable aes and sha2 if enabled.
27870 (simd): Disable sha2 and aes if simd disabled.
27871 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27873 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27874 (TARGET_SHA2): New feature flag for sha2.
27875 (TARGET_AES): New feature flag for aes.
27876 * config/aarch64/aarch64-simd.md:
27877 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27878 conditional on TARGET_AES.
27879 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27880 (aarch64_crypto_sha1hsi): Make pattern conditional
27882 (aarch64_crypto_sha1hv4si): Ditto.
27883 (aarch64_be_crypto_sha1hv4si): Ditto.
27884 (aarch64_crypto_sha1su1v4si): Ditto.
27885 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27886 (aarch64_crypto_sha1su0v4si): Ditto.
27887 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27888 (aarch64_crypto_sha256su0v4si): Ditto.
27889 (aarch64_crypto_sha256su1v4si): Ditto.
27890 (doc/invoke.texi): Document new aes and sha2 options.
27892 2018-01-10 Martin Sebor <msebor@redhat.com>
27894 PR tree-optimization/83781
27895 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27898 2018-01-11 Martin Sebor <msebor@gmail.com>
27899 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27901 PR tree-optimization/83501
27902 PR tree-optimization/81703
27904 * tree-ssa-strlen.c (get_string_cst): Rename...
27905 (get_string_len): ...to this. Handle global constants.
27906 (handle_char_store): Adjust.
27908 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27909 Jim Wilson <jimw@sifive.com>
27911 * config/riscv/riscv-protos.h (riscv_output_return): New.
27912 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27913 (riscv_attribute_table, riscv_output_return),
27914 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27915 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27916 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27918 (riscv_expand_prologue): Add early return for naked function.
27919 (riscv_expand_epilogue): Likewise.
27920 (riscv_function_ok_for_sibcall): Return false for naked function.
27921 (riscv_set_current_function): New.
27922 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27923 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27924 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27925 * doc/extend.texi (RISC-V Function Attributes): New.
27927 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27929 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27930 check for 128-bit long double before checking TCmode.
27931 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27932 128-bit long doubles before checking TFmode or TCmode.
27933 (FLOAT128_IBM_P): Likewise.
27935 2018-01-10 Martin Sebor <msebor@redhat.com>
27937 PR tree-optimization/83671
27938 * builtins.c (c_strlen): Unconditionally return zero for the empty
27940 Use -Warray-bounds for warnings.
27941 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27942 for non-constant array indices with COMPONENT_REF, arrays of
27943 arrays, and pointers to arrays.
27944 (gimple_fold_builtin_strlen): Determine and set length range for
27945 non-constant character arrays.
27947 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27949 PR middle-end/81897
27950 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27953 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27955 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27957 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27960 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27961 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27962 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27963 indexed_or_indirect_operand predicate.
27964 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27965 (*vsx_le_perm_load_v8hi): Likewise.
27966 (*vsx_le_perm_load_v16qi): Likewise.
27967 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27968 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27969 (*vsx_le_perm_store_v8hi): Likewise.
27970 (*vsx_le_perm_store_v16qi): Likewise.
27971 (eight unnamed splitters): Likewise.
27973 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27975 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27976 * config/rs6000/emmintrin.h: Likewise.
27977 * config/rs6000/mmintrin.h: Likewise.
27978 * config/rs6000/xmmintrin.h: Likewise.
27980 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27983 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27985 * tree.c (tree_nop_conversion): Return true for location wrapper
27987 (maybe_wrap_with_location): New function.
27988 (selftest::check_strip_nops): New function.
27989 (selftest::test_location_wrappers): New function.
27990 (selftest::tree_c_tests): Call it.
27991 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27992 (maybe_wrap_with_location): New decl.
27993 (EXPR_LOCATION_WRAPPER_P): New macro.
27994 (location_wrapper_p): New inline function.
27995 (tree_strip_any_location_wrapper): New inline function.
27997 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
28000 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
28001 stack_realign_offset for the largest alignment of stack slot
28003 (ix86_find_max_used_stack_alignment): New function.
28004 (ix86_finalize_stack_frame_flags): Use it. Set
28005 max_used_stack_alignment if we don't realign stack.
28006 * config/i386/i386.h (machine_function): Add
28007 max_used_stack_alignment.
28009 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
28011 * config/arm/arm.opt (-mbranch-cost): New option.
28012 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
28015 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
28018 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
28019 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
28021 2018-01-10 Richard Biener <rguenther@suse.de>
28024 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
28025 early out so it also covers the case where we have a non-NULL
28028 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
28030 PR tree-optimization/83753
28031 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
28032 for non-strided grouped accesses if the number of elements is 1.
28034 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
28037 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
28038 * i386.h (TARGET_USE_GATHER): Define.
28039 * x86-tune.def (X86_TUNE_USE_GATHER): New.
28041 2018-01-10 Martin Liska <mliska@suse.cz>
28044 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
28045 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
28047 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
28048 CLEANUP_NO_PARTITIONING is not set.
28050 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
28052 * doc/rtl.texi: Remove documentation of (const ...) wrappers
28053 for vectors, as a partial revert of r254296.
28054 * rtl.h (const_vec_p): Delete.
28055 (const_vec_duplicate_p): Don't test for vector CONSTs.
28056 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
28057 * expmed.c (make_tree): Likewise.
28060 * common.md (E, F): Use CONSTANT_P instead of checking for
28062 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
28063 checking for CONST_VECTOR.
28065 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28067 PR middle-end/83575
28068 * predict.c (force_edge_cold): Handle in more sane way edges
28069 with no prediction.
28071 2018-01-09 Carl Love <cel@us.ibm.com>
28073 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
28075 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
28076 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
28077 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
28078 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
28079 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
28080 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
28081 * config/rs6000/rs6000-protos.h: Add extern defition for
28082 rs6000_generate_float2_double_code.
28083 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
28085 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
28086 (float2_v2df): Add define_expand.
28088 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
28091 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
28092 op_mode in the force_to_mode call.
28094 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28096 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
28097 instead of checking each element individually.
28098 (aarch64_evpc_uzp): Likewise.
28099 (aarch64_evpc_zip): Likewise.
28100 (aarch64_evpc_ext): Likewise.
28101 (aarch64_evpc_rev): Likewise.
28102 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
28103 instead of checking each element individually. Return true without
28105 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
28106 whether all selected elements come from the same input, instead of
28107 checking each element individually. Remove calls to gen_rtx_REG,
28108 start_sequence and end_sequence and instead assert that no rtl is
28111 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28113 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
28114 order of HIGH and CONST checks.
28116 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
28118 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
28119 if the destination isn't an SSA_NAME.
28121 2018-01-09 Richard Biener <rguenther@suse.de>
28123 PR tree-optimization/83668
28124 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
28126 (canonicalize_loop_form): ... here, renamed from ...
28127 (canonicalize_loop_closed_ssa_form): ... this and amended to
28128 swap successor edges for loop exit blocks to make us use
28129 the RPO order we need for initial schedule generation.
28131 2018-01-09 Joseph Myers <joseph@codesourcery.com>
28133 PR tree-optimization/64811
28134 * match.pd: When optimizing comparisons with Inf, avoid
28135 introducing or losing exceptions from comparisons with NaN.
28137 2018-01-09 Martin Liska <mliska@suse.cz>
28140 * asan.c (shadow_mem_size): Add gcc_assert.
28142 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
28144 Don't save registers in main().
28147 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
28148 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
28149 * config/avr/avr.c (avr_set_current_function): Don't error if
28150 naked, OS_task or OS_main are specified at the same time.
28151 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
28153 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
28155 * common/config/avr/avr-common.c (avr_option_optimization_table):
28156 Switch on -mmain-is-OS_task for optimizing compilations.
28158 2018-01-09 Richard Biener <rguenther@suse.de>
28160 PR tree-optimization/83572
28161 * graphite.c: Include cfganal.h.
28162 (graphite_transform_loops): Connect infinite loops to exit
28163 and remove fake edges at the end.
28165 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28167 * ipa-inline.c (edge_badness): Revert accidental checkin.
28169 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28172 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
28173 symbols; not inline clones.
28175 2018-01-09 Jakub Jelinek <jakub@redhat.com>
28178 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
28179 hard registers. Formatting fixes.
28181 PR preprocessor/83722
28182 * gcc.c (try_generate_repro): Pass
28183 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
28184 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
28187 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
28188 Kito Cheng <kito.cheng@gmail.com>
28190 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
28191 (riscv_leaf_function_p): Delete.
28192 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
28194 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28196 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
28198 (do_ifelse): New function.
28199 (do_isel): New function.
28200 (do_sub3): New function.
28201 (do_add3): New function.
28202 (do_load_mask_compare): New function.
28203 (do_overlap_load_compare): New function.
28204 (expand_compare_loop): New function.
28205 (expand_block_compare): Call expand_compare_loop() when appropriate.
28206 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
28207 option description.
28208 (-mblock-compare-inline-loop-limit): New option.
28210 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28213 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
28214 Reverse order of second and third operands in first alternative.
28215 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
28216 of first and second elements in UNSPEC_VPERMR vector.
28217 (altivec_expand_vec_perm_le): Likewise.
28219 2018-01-08 Jeff Law <law@redhat.com>
28221 PR rtl-optimizatin/81308
28222 * tree-switch-conversion.c (cfg_altered): New file scoped static.
28223 (process_switch): If group_case_labels makes a change, then set
28225 (pass_convert_switch::execute): If a switch is converted, then
28226 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
28228 PR rtl-optimization/81308
28229 * recog.c (split_all_insns): Conditionally cleanup the CFG after
28232 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
28234 PR target/83663 - Revert r255946
28235 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
28236 generation for cases where splatting a value is not useful.
28237 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
28238 across a vec_duplicate and a paradoxical subreg forming a vector
28239 mode to a vec_concat.
28241 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28243 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
28244 -march=armv8.3-a variants.
28245 * config/arm/t-multilib: Likewise.
28246 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
28248 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28250 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
28252 (cceq_ior_compare_complement): Give it a name so I can use it, and
28253 change boolean_or_operator predicate to boolean_operator so it can
28254 be used to generate a crand.
28255 (eqne): New code iterator.
28256 (bd/bd_neg): New code_attrs.
28257 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
28258 a single define_insn.
28259 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
28260 decrement (bdnzt/bdnzf/bdzt/bdzf).
28261 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
28262 with the new names of the branch decrement patterns, and added the
28263 names of the branch decrement conditional patterns.
28265 2018-01-08 Richard Biener <rguenther@suse.de>
28267 PR tree-optimization/83563
28268 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
28271 2018-01-08 Richard Biener <rguenther@suse.de>
28273 PR middle-end/83713
28274 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
28276 2018-01-08 Richard Biener <rguenther@suse.de>
28278 PR tree-optimization/83685
28279 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
28280 references to abnormals.
28282 2018-01-08 Richard Biener <rguenther@suse.de>
28285 * dwarf2out.c (output_indirect_strings): Handle empty
28286 skeleton_debug_str_hash.
28287 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
28289 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28291 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
28292 (emit_store_direct): Likewise.
28293 (arc_trampoline_adjust_address): Likewise.
28294 (arc_asm_trampoline_template): New function.
28295 (arc_initialize_trampoline): Use asm_trampoline_template.
28296 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
28297 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
28298 * config/arc/arc.md (flush_icache): Delete pattern.
28300 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28302 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
28303 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
28306 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28309 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
28310 by not USED_FOR_TARGET.
28311 (make_pass_resolve_sw_modes): Likewise.
28313 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28315 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
28318 2018-01-08 Richard Biener <rguenther@suse.de>
28320 PR middle-end/83580
28321 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
28323 2018-01-08 Richard Biener <rguenther@suse.de>
28325 PR middle-end/83517
28326 * match.pd ((t * 2) / 2) -> t): Add missing :c.
28328 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
28330 PR middle-end/81897
28331 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
28332 basic blocks with a small number of successors.
28333 (convert_control_dep_chain_into_preds): Improve handling of
28335 (dump_predicates): Split apart into...
28336 (dump_pred_chain): ...here...
28337 (dump_pred_info): ...and here.
28338 (can_one_predicate_be_invalidated_p): Add debugging printfs.
28339 (can_chain_union_be_invalidated_p): Improve check for invalidation
28341 (uninit_uses_cannot_happen): Avoid unnecessary if
28342 convert_control_dep_chain_into_preds yielded nothing.
28344 2018-01-06 Martin Sebor <msebor@redhat.com>
28346 PR tree-optimization/83640
28347 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
28348 subtracting negative offset from size.
28349 (builtin_access::overlap): Adjust offset bounds of the access to fall
28350 within the size of the object if possible.
28352 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
28354 PR rtl-optimization/83699
28355 * expmed.c (extract_bit_field_1): Restrict the vector usage of
28356 extract_bit_field_as_subreg to cases in which the extracted
28357 value is also a vector.
28359 * lra-constraints.c (process_alt_operands): Test for the equivalence
28360 substitutions when detecting a possible reload cycle.
28362 2018-01-06 Jakub Jelinek <jakub@redhat.com>
28365 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
28366 by default if flag_selective_schedling{,2}. Formatting fixes.
28368 PR rtl-optimization/83682
28369 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
28370 if it has non-VECTOR_MODE element mode.
28371 (vec_duplicate_p): Likewise.
28373 PR middle-end/83694
28374 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
28375 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
28377 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28380 * config/i386/i386-builtin.def
28381 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
28382 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
28383 Require also OPTION_MASK_ISA_AVX512F in addition to
28384 OPTION_MASK_ISA_GFNI.
28385 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
28386 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
28387 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
28388 to OPTION_MASK_ISA_GFNI.
28389 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
28390 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
28391 OPTION_MASK_ISA_AVX512BW.
28392 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
28393 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
28394 addition to OPTION_MASK_ISA_GFNI.
28395 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
28396 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
28397 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
28398 to OPTION_MASK_ISA_GFNI.
28399 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
28400 a requirement for all ISAs rather than any of them with a few
28402 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
28404 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
28405 bitmasks to be enabled with 3 exceptions, instead of requiring any
28406 enabled ISA with lots of exceptions.
28407 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
28408 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
28409 Change avx512bw in isa attribute to avx512f.
28410 * config/i386/sgxintrin.h: Add license boilerplate.
28411 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
28412 to __AVX512F__ and __AVX512VL to __AVX512VL__.
28413 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
28414 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
28416 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
28417 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
28418 temporarily sse2 rather than sse if not enabled already.
28421 * config/i386/sse.md (VI248_VLBW): Rename to ...
28422 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
28423 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
28424 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
28425 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
28426 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
28427 mode iterator instead of VI248_VLBW.
28429 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
28431 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
28432 (record_modified): Skip clobbers; add debug output.
28433 (param_change_prob): Use sreal frequencies.
28435 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28437 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
28438 punt for user-aligned variables.
28440 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28442 * tree-chrec.c (chrec_contains_symbols): Return true for
28445 2018-01-05 Sudakshina Das <sudi.das@arm.com>
28448 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
28449 of (x|y) == x for BICS pattern.
28451 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28453 PR tree-optimization/83605
28454 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
28455 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
28458 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
28460 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
28461 * config/epiphany/rtems.h: New file.
28463 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28464 Uros Bizjak <ubizjak@gmail.com>
28467 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
28468 QIreg_operand instead of register_operand predicate.
28469 * config/i386/i386.c (ix86_rop_should_change_byte_p,
28470 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
28471 comments instead of -fmitigate[-_]rop.
28473 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28476 * cgraphunit.c (symbol_table::compile): Switch to text_section
28477 before calling assembly_start debug hook.
28478 * run-rtl-passes.c (run_rtl_passes): Likewise.
28481 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28483 * tree-vrp.c (extract_range_from_binary_expr_1): Check
28484 range_int_cst_p rather than !symbolic_range_p before calling
28485 extract_range_from_multiplicative_op_1.
28487 2018-01-04 Jeff Law <law@redhat.com>
28489 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
28490 redundant test in assertion.
28492 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28494 * doc/rtl.texi: Document machine_mode wrapper classes.
28496 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28498 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
28499 using tree_to_uhwi.
28501 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28503 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
28504 the VEC_PERM_EXPR fold to fail.
28506 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28509 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
28510 to switched_sections.
28512 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28515 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
28516 test for d.testing.
28518 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
28521 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
28522 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
28524 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28527 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
28528 is BLKmode and bitpos not zero or mode change is needed.
28530 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28533 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
28536 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
28539 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
28540 instead of MULT rtx. Update all corresponding splitters.
28541 (*saddl_se): Ditto.
28542 (*ssub<modesuffix>): Ditto.
28543 (*ssubl_se): Ditto.
28544 (*cmp_sadd_di): Update split patterns.
28545 (*cmp_sadd_si): Ditto.
28546 (*cmp_sadd_sidi): Ditto.
28547 (*cmp_ssub_di): Ditto.
28548 (*cmp_ssub_si): Ditto.
28549 (*cmp_ssub_sidi): Ditto.
28550 * config/alpha/predicates.md (const23_operand): New predicate.
28551 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28552 Look for ASHIFT, not MULT inner operand.
28553 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28555 2018-01-04 Martin Liska <mliska@suse.cz>
28557 PR gcov-profile/83669
28558 * gcov.c (output_intermediate_file): Add version to intermediate
28560 * doc/gcov.texi: Document new field 'version' in intermediate
28561 file format. Fix location of '-k' option of gcov command.
28563 2018-01-04 Martin Liska <mliska@suse.cz>
28566 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28568 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28570 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28572 2018-01-03 Martin Sebor <msebor@redhat.com>
28574 PR tree-optimization/83655
28575 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28576 checking calls with invalid arguments.
28578 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28580 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28581 (vectorizable_mask_load_store): Delete.
28582 (vectorizable_call): Return false for masked loads and stores.
28583 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28584 instead of gimple_assign_rhs1.
28585 (vectorizable_load): Handle IFN_MASK_LOAD.
28586 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28588 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28590 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28592 (vectorizable_mask_load_store): ...here.
28593 (vectorizable_load): ...and here.
28595 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28597 * tree-vect-stmts.c (vect_build_all_ones_mask)
28598 (vect_build_zero_merge_argument): New functions, split out from...
28599 (vectorizable_load): ...here.
28601 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28603 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28605 (vectorizable_mask_load_store): ...here.
28606 (vectorizable_store): ...and here.
28608 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28610 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28612 (vectorizable_mask_load_store): ...here.
28614 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28616 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28617 (vect_model_store_cost): Take a vec_load_store_type instead of a
28619 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28620 (vect_model_store_cost): Take a vec_load_store_type instead of a
28622 (vectorizable_mask_load_store): Update accordingly.
28623 (vectorizable_store): Likewise.
28624 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28626 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28628 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28629 IFN_MASK_LOAD calls here rather than...
28630 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28632 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28633 Alan Hayward <alan.hayward@arm.com>
28634 David Sherwood <david.sherwood@arm.com>
28636 * expmed.c (extract_bit_field_1): For vector extracts,
28637 fall back to extract_bit_field_as_subreg if vec_extract
28640 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28641 Alan Hayward <alan.hayward@arm.com>
28642 David Sherwood <david.sherwood@arm.com>
28644 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28645 they are variable or constant sized.
28646 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28647 slots for constant-sized data.
28649 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28650 Alan Hayward <alan.hayward@arm.com>
28651 David Sherwood <david.sherwood@arm.com>
28653 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28654 handling COND_EXPRs with boolean comparisons, try to find a better
28655 basis for the mask type than the boolean itself.
28657 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28659 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28660 is calculated and how it can be overridden.
28661 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28662 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28664 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28667 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28668 Alan Hayward <alan.hayward@arm.com>
28669 David Sherwood <david.sherwood@arm.com>
28671 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28672 Remove the mode argument.
28673 (aarch64_simd_valid_immediate): Remove the mode and inverse
28675 * config/aarch64/iterators.md (bitsize): New iterator.
28676 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28677 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28678 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28679 aarch64_simd_valid_immediate.
28680 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28681 (aarch64_reg_or_bic_imm): Likewise.
28682 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28683 with an insn_type enum and msl with a modifier_type enum.
28684 Replace element_width with a scalar_mode. Change the shift
28685 to unsigned int. Add constructors for scalar_float_mode and
28686 scalar_int_mode elements.
28687 (aarch64_vect_float_const_representable_p): Delete.
28688 (aarch64_can_const_movi_rtx_p)
28689 (aarch64_simd_scalar_immediate_valid_for_move)
28690 (aarch64_simd_make_constant): Update call to
28691 aarch64_simd_valid_immediate.
28692 (aarch64_advsimd_valid_immediate_hs): New function.
28693 (aarch64_advsimd_valid_immediate): Likewise.
28694 (aarch64_simd_valid_immediate): Remove mode and inverse
28695 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28696 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28697 and aarch64_float_const_representable_p on the result.
28698 (aarch64_output_simd_mov_immediate): Remove mode argument.
28699 Update call to aarch64_simd_valid_immediate and use of
28700 simd_immediate_info.
28701 (aarch64_output_scalar_simd_mov_immediate): Update call
28704 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28705 Alan Hayward <alan.hayward@arm.com>
28706 David Sherwood <david.sherwood@arm.com>
28708 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28709 (mode_nunits): Likewise CONST_MODE_NUNITS.
28710 * machmode.def (ADJUST_NUNITS): Document.
28711 * genmodes.c (mode_data::need_nunits_adj): New field.
28712 (blank_mode): Update accordingly.
28713 (adj_nunits): New variable.
28714 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28716 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28717 listed in adj_nunits.
28718 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28719 listed in adj_nunits. Don't emit case statements for such modes.
28720 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28721 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28722 nothing if adj_nunits is nonnull.
28723 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28724 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28725 (emit_mode_fbit): Update use of print_maybe_const_decl.
28726 (emit_move_size): Likewise. Treat the array as non-const
28728 (emit_mode_adjustments): Handle adj_nunits.
28730 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28732 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28733 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28734 (VECTOR_MODES): Use it.
28735 (make_vector_modes): Take the prefix as an argument.
28737 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28738 Alan Hayward <alan.hayward@arm.com>
28739 David Sherwood <david.sherwood@arm.com>
28741 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28742 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28743 for MODE_VECTOR_BOOL.
28744 * machmode.def (VECTOR_BOOL_MODE): Document.
28745 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28746 (make_vector_bool_mode): New function.
28747 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28749 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28750 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28752 * stor-layout.c (int_mode_for_mode): Likewise.
28753 * tree.c (build_vector_type_for_mode): Likewise.
28754 * varasm.c (output_constant_pool_2): Likewise.
28755 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28756 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28757 for MODE_VECTOR_BOOL.
28758 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28759 of mode class checks.
28760 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28761 instead of a list of mode class checks.
28762 (expand_vector_scalar_condition): Likewise.
28763 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28765 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28766 Alan Hayward <alan.hayward@arm.com>
28767 David Sherwood <david.sherwood@arm.com>
28769 * machmode.h (mode_size): Change from unsigned short to
28771 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28772 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28773 or if measurement_type is not polynomial.
28774 (fixed_size_mode::includes_p): Check for constant-sized modes.
28775 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28776 return a poly_uint16 rather than an unsigned short.
28777 (emit_mode_size): Change the type of mode_size from unsigned short
28778 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28779 (emit_mode_adjustments): Cope with polynomial vector sizes.
28780 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28782 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28784 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28785 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28786 * caller-save.c (setup_save_areas): Likewise.
28787 (replace_reg_with_saved_mem): Likewise.
28788 * calls.c (emit_library_call_value_1): Likewise.
28789 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28790 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28791 (gen_lowpart_for_combine): Likewise.
28792 * convert.c (convert_to_integer_1): Likewise.
28793 * cse.c (equiv_constant, cse_insn): Likewise.
28794 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28795 (cselib_subst_to_values): Likewise.
28796 * dce.c (word_dce_process_block): Likewise.
28797 * df-problems.c (df_word_lr_mark_ref): Likewise.
28798 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28799 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28800 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28801 (rtl_for_decl_location): Likewise.
28802 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28803 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28804 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28805 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28806 (expand_expr_real_1): Likewise.
28807 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28808 (pad_below): Likewise.
28809 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28810 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28811 * ira.c (get_subreg_tracking_sizes): Likewise.
28812 * ira-build.c (ira_create_allocno_objects): Likewise.
28813 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28814 (ira_sort_regnos_for_alter_reg): Likewise.
28815 * ira-costs.c (record_operand_costs): Likewise.
28816 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28817 (resolve_simple_move): Likewise.
28818 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28819 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28820 (lra_constraints): Likewise.
28821 (CONST_POOL_OK_P): Reject variable-sized modes.
28822 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28823 (add_pseudo_to_slot, lra_spill): Likewise.
28824 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28825 * optabs-query.c (get_best_extraction_insn): Likewise.
28826 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28827 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28828 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28829 * recog.c (offsettable_address_addr_space_p): Likewise.
28830 * regcprop.c (maybe_mode_change): Likewise.
28831 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28832 * regrename.c (build_def_use): Likewise.
28833 * regstat.c (dump_reg_info): Likewise.
28834 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28835 (find_reloads, find_reloads_subreg_address): Likewise.
28836 * reload1.c (eliminate_regs_1): Likewise.
28837 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28838 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28839 (simplify_binary_operation_1, simplify_subreg): Likewise.
28840 * targhooks.c (default_function_arg_padding): Likewise.
28841 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28842 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28843 (verify_gimple_assign_ternary): Likewise.
28844 * tree-inline.c (estimate_move_cost): Likewise.
28845 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28846 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28847 (get_address_cost_ainc): Likewise.
28848 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28849 (vect_supportable_dr_alignment): Likewise.
28850 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28851 (vectorizable_reduction): Likewise.
28852 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28853 (vectorizable_operation, vectorizable_load): Likewise.
28854 * tree.c (build_same_sized_truth_vector_type): Likewise.
28855 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28856 * var-tracking.c (emit_note_insn_var_location): Likewise.
28857 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28858 (ADDR_VEC_ALIGN): Likewise.
28860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28861 Alan Hayward <alan.hayward@arm.com>
28862 David Sherwood <david.sherwood@arm.com>
28864 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28866 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28867 or if measurement_type is polynomial.
28868 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28869 * combine.c (make_extraction): Likewise.
28870 * dse.c (find_shift_sequence): Likewise.
28871 * dwarf2out.c (mem_loc_descriptor): Likewise.
28872 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28873 (extract_bit_field, extract_low_bits): Likewise.
28874 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28875 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28876 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28877 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28878 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28879 * reload.c (find_reloads): Likewise.
28880 * reload1.c (alter_reg): Likewise.
28881 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28882 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28883 * tree-if-conv.c (predicate_mem_writes): Likewise.
28884 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28885 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28886 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28887 * valtrack.c (dead_debug_insert_temp): Likewise.
28888 * varasm.c (mergeable_constant_section): Likewise.
28889 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28891 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28892 Alan Hayward <alan.hayward@arm.com>
28893 David Sherwood <david.sherwood@arm.com>
28895 * expr.c (expand_assignment): Cope with polynomial mode sizes
28896 when assigning to a CONCAT.
28898 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28899 Alan Hayward <alan.hayward@arm.com>
28900 David Sherwood <david.sherwood@arm.com>
28902 * machmode.h (mode_precision): Change from unsigned short to
28904 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28906 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28907 or if measurement_type is not polynomial.
28908 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28909 in which the mode is already known to be a scalar_int_mode.
28910 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28911 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28913 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28914 for GET_MODE_PRECISION.
28915 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28916 for GET_MODE_PRECISION.
28917 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28919 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28920 (expand_field_assignment, make_extraction): Likewise.
28921 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28922 (get_last_value): Likewise.
28923 * convert.c (convert_to_integer_1): Likewise.
28924 * cse.c (cse_insn): Likewise.
28925 * expr.c (expand_expr_real_1): Likewise.
28926 * lra-constraints.c (simplify_operand_subreg): Likewise.
28927 * optabs-query.c (can_atomic_load_p): Likewise.
28928 * optabs.c (expand_atomic_load): Likewise.
28929 (expand_atomic_store): Likewise.
28930 * ree.c (combine_reaching_defs): Likewise.
28931 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28932 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28933 * tree.h (type_has_mode_precision_p): Likewise.
28934 * ubsan.c (instrument_si_overflow): Likewise.
28936 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28937 Alan Hayward <alan.hayward@arm.com>
28938 David Sherwood <david.sherwood@arm.com>
28940 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28941 polynomial numbers of units.
28942 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28943 (valid_vector_subparts_p): New function.
28944 (build_vector_type): Remove temporary shim and take the number
28945 of units as a poly_uint64 rather than an int.
28946 (build_opaque_vector_type): Take the number of units as a
28947 poly_uint64 rather than an int.
28948 * tree.c (build_vector_from_ctor): Handle polynomial
28949 TYPE_VECTOR_SUBPARTS.
28950 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28951 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28952 (build_vector_from_val): If the number of units is variable,
28953 use build_vec_duplicate_cst for constant operands and
28954 VEC_DUPLICATE_EXPR otherwise.
28955 (make_vector_type): Remove temporary is_constant ().
28956 (build_vector_type, build_opaque_vector_type): Take the number of
28957 units as a poly_uint64 rather than an int.
28958 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28960 * cfgexpand.c (expand_debug_expr): Likewise.
28961 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28962 (store_constructor, expand_expr_real_1): Likewise.
28963 (const_scalar_mask_from_tree): Likewise.
28964 * fold-const-call.c (fold_const_reduction): Likewise.
28965 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28966 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28967 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28968 (fold_relational_const): Likewise.
28969 (native_interpret_vector): Likewise. Change the size from an
28970 int to an unsigned int.
28971 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28972 TYPE_VECTOR_SUBPARTS.
28973 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28974 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28975 duplicating a non-constant operand into a variable-length vector.
28976 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28977 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28978 * ipa-icf.c (sem_variable::equals): Likewise.
28979 * match.pd: Likewise.
28980 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28981 * print-tree.c (print_node): Likewise.
28982 * stor-layout.c (layout_type): Likewise.
28983 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28984 * tree-cfg.c (verify_gimple_comparison): Likewise.
28985 (verify_gimple_assign_binary): Likewise.
28986 (verify_gimple_assign_ternary): Likewise.
28987 (verify_gimple_assign_single): Likewise.
28988 * tree-pretty-print.c (dump_generic_node): Likewise.
28989 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28990 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28991 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28992 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28993 (vect_shift_permute_load_chain): Likewise.
28994 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28995 (expand_vector_condition, optimize_vector_constructor): Likewise.
28996 (lower_vec_perm, get_compute_type): Likewise.
28997 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28998 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28999 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
29000 (vect_recog_mask_conversion_pattern): Likewise.
29001 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
29002 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
29003 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29004 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
29005 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
29006 (vectorizable_shift, vectorizable_operation, vectorizable_store)
29007 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
29008 (supportable_widening_operation): Likewise.
29009 (supportable_narrowing_operation): Likewise.
29010 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
29012 * varasm.c (output_constant): Likewise.
29014 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29015 Alan Hayward <alan.hayward@arm.com>
29016 David Sherwood <david.sherwood@arm.com>
29018 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
29019 so that both the length == 3 and length != 3 cases set up their
29020 own permute vectors. Add comments explaining why we know the
29021 number of elements is constant.
29022 (vect_permute_load_chain): Likewise.
29024 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29025 Alan Hayward <alan.hayward@arm.com>
29026 David Sherwood <david.sherwood@arm.com>
29028 * machmode.h (mode_nunits): Change from unsigned char to
29030 (ONLY_FIXED_SIZE_MODES): New macro.
29031 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
29032 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
29033 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
29035 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
29036 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
29037 or if measurement_type is not polynomial.
29038 * genmodes.c (ZERO_COEFFS): New macro.
29039 (emit_mode_nunits_inline): Make mode_nunits_inline return a
29041 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
29042 Use ZERO_COEFFS when emitting initializers.
29043 * data-streamer.h (bp_pack_poly_value): New function.
29044 (bp_unpack_poly_value): Likewise.
29045 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
29046 for GET_MODE_NUNITS.
29047 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
29048 for GET_MODE_NUNITS.
29049 * tree.c (make_vector_type): Remove temporary shim and make
29050 the real function take the number of units as a poly_uint64
29051 rather than an int.
29052 (build_vector_type_for_mode): Handle polynomial nunits.
29053 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
29054 * emit-rtl.c (const_vec_series_p_1): Likewise.
29055 (gen_rtx_CONST_VECTOR): Likewise.
29056 * fold-const.c (test_vec_duplicate_folding): Likewise.
29057 * genrecog.c (validate_pattern): Likewise.
29058 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
29059 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
29060 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
29061 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
29062 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
29063 * rtlanal.c (subreg_get_info): Likewise.
29064 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29065 (vect_grouped_load_supported): Likewise.
29066 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
29067 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
29068 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
29069 (simplify_const_unary_operation, simplify_binary_operation_1)
29070 (simplify_const_binary_operation, simplify_ternary_operation)
29071 (test_vector_ops_duplicate, test_vector_ops): Likewise.
29072 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
29073 instead of CONST_VECTOR_NUNITS.
29074 * varasm.c (output_constant_pool_2): Likewise.
29075 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
29076 explicit-encoded elements in the XVEC for variable-length vectors.
29078 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29080 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
29082 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29083 Alan Hayward <alan.hayward@arm.com>
29084 David Sherwood <david.sherwood@arm.com>
29086 * coretypes.h (fixed_size_mode): Declare.
29087 (fixed_size_mode_pod): New typedef.
29088 * builtins.h (target_builtins::x_apply_args_mode)
29089 (target_builtins::x_apply_result_mode): Change type to
29090 fixed_size_mode_pod.
29091 * builtins.c (apply_args_size, apply_result_size, result_vector)
29092 (expand_builtin_apply_args_1, expand_builtin_apply)
29093 (expand_builtin_return): Update accordingly.
29095 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29097 * cse.c (hash_rtx_cb): Hash only the encoded elements.
29098 * cselib.c (cselib_hash_rtx): Likewise.
29099 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
29100 CONST_VECTOR encoding.
29102 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29103 Jeff Law <law@redhat.com>
29106 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
29107 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
29108 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
29109 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
29112 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
29113 explicitly probe *sp in a noreturn function if there were any callee
29114 register saves or frame pointer is needed.
29116 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29119 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
29120 BLKmode for ternary, binary or unary expressions.
29123 * var-tracking.c (delete_vta_debug_insn): New inline function.
29124 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
29125 insns from get_insns () to NULL instead of each bb separately.
29126 Use delete_vta_debug_insn. No longer static.
29127 (vt_debug_insns_local, variable_tracking_main_1): Adjust
29128 delete_vta_debug_insns callers.
29129 * rtl.h (delete_vta_debug_insns): Declare.
29130 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
29131 instead of variable_tracking_main.
29133 2018-01-03 Martin Sebor <msebor@redhat.com>
29135 PR tree-optimization/83603
29136 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
29137 arguments past the endof the argument list in functions declared
29138 without a prototype.
29139 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
29140 Avoid checking when arguments are null.
29142 2018-01-03 Martin Sebor <msebor@redhat.com>
29145 * doc/extend.texi (attribute const): Fix a typo.
29146 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
29147 issuing -Wsuggest-attribute for void functions.
29149 2018-01-03 Martin Sebor <msebor@redhat.com>
29151 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
29152 offset_int::from instead of wide_int::to_shwi.
29153 (maybe_diag_overlap): Remove assertion.
29154 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
29155 * gimple-ssa-sprintf.c (format_directive): Same.
29156 (parse_directive): Same.
29157 (sprintf_dom_walker::compute_format_length): Same.
29158 (try_substitute_return_value): Same.
29160 2018-01-03 Jeff Law <law@redhat.com>
29162 PR middle-end/83654
29163 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
29164 non-constant residual for zero at runtime and avoid probing in
29165 that case. Reorganize code for trailing problem to mirror handling
29168 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29170 PR tree-optimization/83501
29171 * tree-ssa-strlen.c (get_string_cst): New.
29172 (handle_char_store): Call get_string_cst.
29174 2018-01-03 Martin Liska <mliska@suse.cz>
29176 PR tree-optimization/83593
29177 * tree-ssa-strlen.c: Include tree-cfg.h.
29178 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
29179 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
29180 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
29182 (strlen_dom_walker::before_dom_children): Call
29183 gimple_purge_dead_eh_edges. Dump tranformation with details
29185 (strlen_dom_walker::before_dom_children): Update call by adding
29186 new argument cleanup_eh.
29187 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
29189 2018-01-03 Martin Liska <mliska@suse.cz>
29192 * cif-code.def (VARIADIC_THUNK): New enum value.
29193 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
29196 2018-01-03 Jan Beulich <jbeulich@suse.com>
29198 * sse.md (mov<mode>_internal): Tighten condition for when to use
29199 vmovdqu<ssescalarsize> for TI and OI modes.
29201 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29203 Update copyright years.
29205 2018-01-03 Martin Liska <mliska@suse.cz>
29208 * ipa-visibility.c (function_and_variable_visibility): Skip
29209 functions with noipa attribure.
29211 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29213 * gcc.c (process_command): Update copyright notice dates.
29214 * gcov-dump.c (print_version): Ditto.
29215 * gcov.c (print_version): Ditto.
29216 * gcov-tool.c (print_version): Ditto.
29217 * gengtype.c (create_file): Ditto.
29218 * doc/cpp.texi: Bump @copying's copyright year.
29219 * doc/cppinternals.texi: Ditto.
29220 * doc/gcc.texi: Ditto.
29221 * doc/gccint.texi: Ditto.
29222 * doc/gcov.texi: Ditto.
29223 * doc/install.texi: Ditto.
29224 * doc/invoke.texi: Ditto.
29226 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29228 * vector-builder.h (vector_builder::m_full_nelts): Change from
29229 unsigned int to poly_uint64.
29230 (vector_builder::full_nelts): Update prototype accordingly.
29231 (vector_builder::new_vector): Likewise.
29232 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
29233 (vector_builder::operator ==): Likewise.
29234 (vector_builder::finalize): Likewise.
29235 * int-vector-builder.h (int_vector_builder::int_vector_builder):
29236 Take the number of elements as a poly_uint64 rather than an
29238 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
29239 from unsigned int to poly_uint64.
29240 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
29241 (vec_perm_indices::new_vector): Likewise.
29242 (vec_perm_indices::length): Likewise.
29243 (vec_perm_indices::nelts_per_input): Likewise.
29244 (vec_perm_indices::input_nelts): Likewise.
29245 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
29246 number of elements per input as a poly_uint64 rather than an
29247 unsigned int. Use the original encoding for variable-length
29248 vectors, rather than clamping each individual element.
29249 For the second and subsequent elements in each pattern,
29250 clamp the step and base before clamping their sum.
29251 (vec_perm_indices::series_p): Handle polynomial element counts.
29252 (vec_perm_indices::all_in_range_p): Likewise.
29253 (vec_perm_indices_to_tree): Likewise.
29254 (vec_perm_indices_to_rtx): Likewise.
29255 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
29256 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
29257 (tree_vector_builder::new_binary_operation): Handle polynomial
29258 element counts. Return false if we need to know the number
29259 of elements at compile time.
29260 * fold-const.c (fold_vec_perm): Punt if the number of elements
29261 isn't known at compile time.
29263 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29265 * vec-perm-indices.h (vec_perm_builder): Change element type
29266 from HOST_WIDE_INT to poly_int64.
29267 (vec_perm_indices::element_type): Update accordingly.
29268 (vec_perm_indices::clamp): Handle polynomial element_types.
29269 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29270 (vec_perm_indices::all_in_range_p): Likewise.
29271 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
29273 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
29274 polynomial vec_perm_indices element types.
29275 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
29276 * fold-const.c (fold_vec_perm): Likewise.
29277 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
29278 * tree-vect-generic.c (lower_vec_perm): Likewise.
29279 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29280 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
29281 element type to HOST_WIDE_INT.
29283 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29284 Alan Hayward <alan.hayward@arm.com>
29285 David Sherwood <david.sherwood@arm.com>
29287 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
29288 rather than an int. Use plus_constant.
29289 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
29290 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
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 * calls.c (emit_call_1, expand_call): Change struct_value_size from
29297 a HOST_WIDE_INT to a poly_int64.
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 * calls.c (load_register_parameters): Cope with polynomial
29304 mode sizes. Require a constant size for BLKmode parameters
29305 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
29306 forces a parameter to be padded at the lsb end in order to
29307 fill a complete number of words, require the parameter size
29308 to be ordered wrt UNITS_PER_WORD.
29310 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29311 Alan Hayward <alan.hayward@arm.com>
29312 David Sherwood <david.sherwood@arm.com>
29314 * reload1.c (spill_stack_slot_width): Change element type
29315 from unsigned int to poly_uint64_pod.
29316 (alter_reg): Treat mode sizes as polynomial.
29318 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29319 Alan Hayward <alan.hayward@arm.com>
29320 David Sherwood <david.sherwood@arm.com>
29322 * reload.c (complex_word_subreg_p): New function.
29323 (reload_inner_reg_of_subreg, push_reload): Use it.
29325 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29326 Alan Hayward <alan.hayward@arm.com>
29327 David Sherwood <david.sherwood@arm.com>
29329 * lra-constraints.c (process_alt_operands): Reject matched
29330 operands whose sizes aren't ordered.
29331 (match_reload): Refer to this check here.
29333 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29334 Alan Hayward <alan.hayward@arm.com>
29335 David Sherwood <david.sherwood@arm.com>
29337 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
29338 that the mode size is in the set {1, 2, 4, 8, 16}.
29340 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29341 Alan Hayward <alan.hayward@arm.com>
29342 David Sherwood <david.sherwood@arm.com>
29344 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
29345 Use plus_constant instead of gen_rtx_PLUS.
29347 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29348 Alan Hayward <alan.hayward@arm.com>
29349 David Sherwood <david.sherwood@arm.com>
29351 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
29352 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
29353 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
29354 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
29355 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
29356 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
29357 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
29358 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
29359 * config/i386/i386.c (ix86_push_rounding): ...this new function.
29360 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
29362 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
29363 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
29364 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
29365 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
29366 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
29367 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
29368 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
29369 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
29370 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
29371 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
29373 * expr.c (emit_move_resolve_push): Treat the input and result
29374 of PUSH_ROUNDING as a poly_int64.
29375 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
29376 (emit_push_insn): Likewise.
29377 * lra-eliminations.c (mark_not_eliminable): Likewise.
29378 * recog.c (push_operand): Likewise.
29379 * reload1.c (elimination_effects): Likewise.
29380 * rtlanal.c (nonzero_bits1): Likewise.
29381 * calls.c (store_one_arg): Likewise. Require the padding to be
29382 known at compile time.
29384 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29385 Alan Hayward <alan.hayward@arm.com>
29386 David Sherwood <david.sherwood@arm.com>
29388 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
29389 Use plus_constant instead of gen_rtx_PLUS.
29391 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29392 Alan Hayward <alan.hayward@arm.com>
29393 David Sherwood <david.sherwood@arm.com>
29395 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
29396 rather than an int.
29398 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29399 Alan Hayward <alan.hayward@arm.com>
29400 David Sherwood <david.sherwood@arm.com>
29402 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
29403 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
29404 via stack temporaries. Treat the mode size as polynomial too.
29406 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29407 Alan Hayward <alan.hayward@arm.com>
29408 David Sherwood <david.sherwood@arm.com>
29410 * expr.c (expand_expr_real_2): When handling conversions involving
29411 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
29412 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
29413 as a poly_uint64 too.
29415 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29416 Alan Hayward <alan.hayward@arm.com>
29417 David Sherwood <david.sherwood@arm.com>
29419 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
29421 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29422 Alan Hayward <alan.hayward@arm.com>
29423 David Sherwood <david.sherwood@arm.com>
29425 * combine.c (can_change_dest_mode): Handle polynomial
29426 REGMODE_NATURAL_SIZE.
29427 * expmed.c (store_bit_field_1): Likewise.
29428 * expr.c (store_constructor): Likewise.
29429 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
29430 and polynomial REGMODE_NATURAL_SIZE.
29431 (gen_lowpart_common): Likewise.
29432 * reginfo.c (record_subregs_of_mode): Likewise.
29433 * rtlanal.c (read_modify_subreg_p): Likewise.
29435 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29436 Alan Hayward <alan.hayward@arm.com>
29437 David Sherwood <david.sherwood@arm.com>
29439 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
29440 numbers of elements.
29442 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29443 Alan Hayward <alan.hayward@arm.com>
29444 David Sherwood <david.sherwood@arm.com>
29446 * match.pd: Cope with polynomial numbers of vector elements.
29448 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29449 Alan Hayward <alan.hayward@arm.com>
29450 David Sherwood <david.sherwood@arm.com>
29452 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
29453 in a POINTER_PLUS_EXPR.
29455 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29456 Alan Hayward <alan.hayward@arm.com>
29457 David Sherwood <david.sherwood@arm.com>
29459 * omp-simd-clone.c (simd_clone_subparts): New function.
29460 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
29461 (ipa_simd_modify_function_body): Likewise.
29463 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29464 Alan Hayward <alan.hayward@arm.com>
29465 David Sherwood <david.sherwood@arm.com>
29467 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
29468 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
29469 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
29470 (expand_vector_condition, vector_element): Likewise.
29471 (subparts_gt): New function.
29472 (get_compute_type): Use subparts_gt.
29473 (count_type_subparts): Delete.
29474 (expand_vector_operations_1): Use subparts_gt instead of
29475 count_type_subparts.
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-data-refs.c (vect_no_alias_p): Replace with...
29482 (vect_compile_time_alias): ...this new function. Do the calculation
29483 on poly_ints rather than trees.
29484 (vect_prune_runtime_alias_test_list): Update call accordingly.
29486 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29487 Alan Hayward <alan.hayward@arm.com>
29488 David Sherwood <david.sherwood@arm.com>
29490 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
29492 (vect_schedule_slp_instance): Likewise.
29494 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29495 Alan Hayward <alan.hayward@arm.com>
29496 David Sherwood <david.sherwood@arm.com>
29498 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
29499 constant and extern definitions for variable-length vectors.
29500 (vect_get_constant_vectors): Note that the number of units
29501 is known to be constant.
29503 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29504 Alan Hayward <alan.hayward@arm.com>
29505 David Sherwood <david.sherwood@arm.com>
29507 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
29508 of units as polynomial. Choose between WIDE and NARROW based
29511 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29512 Alan Hayward <alan.hayward@arm.com>
29513 David Sherwood <david.sherwood@arm.com>
29515 * tree-vect-stmts.c (simd_clone_subparts): New function.
29516 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
29518 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29519 Alan Hayward <alan.hayward@arm.com>
29520 David Sherwood <david.sherwood@arm.com>
29522 * tree-vect-stmts.c (vectorizable_call): Treat the number of
29523 vectors as polynomial. Use build_index_vector for
29524 IFN_GOMP_SIMD_LANE.
29526 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29527 Alan Hayward <alan.hayward@arm.com>
29528 David Sherwood <david.sherwood@arm.com>
29530 * tree-vect-stmts.c (get_load_store_type): Treat the number of
29531 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
29532 for variable-length vectors.
29533 (vectorizable_mask_load_store): Treat the number of units as
29534 polynomial, asserting that it is constant if the condition has
29535 already been enforced.
29536 (vectorizable_store, vectorizable_load): Likewise.
29538 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29539 Alan Hayward <alan.hayward@arm.com>
29540 David Sherwood <david.sherwood@arm.com>
29542 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
29543 of units as polynomial. Punt if we can't tell at compile time
29544 which vector contains the final result.
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-loop.c (vectorizable_induction): Treat the number
29551 of units as polynomial. Punt on SLP inductions. Use an integer
29552 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29553 cast of such a series for variable-length floating-point
29556 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29557 Alan Hayward <alan.hayward@arm.com>
29558 David Sherwood <david.sherwood@arm.com>
29560 * tree.h (build_index_vector): Declare.
29561 * tree.c (build_index_vector): New function.
29562 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29563 of units as polynomial, forcibly converting it to a constant if
29564 vectorizable_reduction has already enforced the condition.
29565 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29566 to create a {1,2,3,...} vector.
29567 (vectorizable_reduction): Treat the number of units as polynomial.
29568 Choose vectype_in based on the largest scalar element size rather
29569 than the smallest number of units. Enforce the restrictions
29572 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29573 Alan Hayward <alan.hayward@arm.com>
29574 David Sherwood <david.sherwood@arm.com>
29576 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29577 number of units as polynomial.
29579 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29580 Alan Hayward <alan.hayward@arm.com>
29581 David Sherwood <david.sherwood@arm.com>
29583 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29584 * target.def (autovectorize_vector_sizes): Return the vector sizes
29585 by pointer, using vector_sizes rather than a bitmask.
29586 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29587 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29588 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29590 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29591 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29592 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29593 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29594 * omp-general.c (omp_max_vf): Likewise.
29595 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29596 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29597 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29598 * tree-vect-slp.c (vect_slp_bb): Likewise.
29599 * doc/tm.texi: Regenerate.
29600 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29602 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29603 the vector size as a poly_uint64 rather than an unsigned int.
29604 (current_vector_size): Change from an unsigned int to a poly_uint64.
29605 (get_vectype_for_scalar_type): Update accordingly.
29606 * tree.h (build_truth_vector_type): Take the size and number of
29607 units as a poly_uint64 rather than an unsigned int.
29608 (build_vector_type): Add a temporary overload that takes
29609 the number of units as a poly_uint64 rather than an unsigned int.
29610 * tree.c (make_vector_type): Likewise.
29611 (build_truth_vector_type): Take the number of units as a poly_uint64
29612 rather than an unsigned int.
29614 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29615 Alan Hayward <alan.hayward@arm.com>
29616 David Sherwood <david.sherwood@arm.com>
29618 * target.def (get_mask_mode): Take the number of units and length
29619 as poly_uint64s rather than unsigned ints.
29620 * targhooks.h (default_get_mask_mode): Update accordingly.
29621 * targhooks.c (default_get_mask_mode): Likewise.
29622 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29623 * doc/tm.texi: Regenerate.
29625 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29626 Alan Hayward <alan.hayward@arm.com>
29627 David Sherwood <david.sherwood@arm.com>
29629 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29630 * omp-general.c (omp_max_vf): Likewise.
29631 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29632 (expand_omp_simd): Handle polynomial safelen.
29633 * omp-low.c (omplow_simd_context): Add a default constructor.
29634 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29635 (lower_rec_simd_input_clauses): Update accordingly.
29636 (lower_rec_input_clauses): Likewise.
29638 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29639 Alan Hayward <alan.hayward@arm.com>
29640 David Sherwood <david.sherwood@arm.com>
29642 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29643 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29644 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29645 (vect_analyze_slp_cost): Likewise.
29646 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29647 (vect_model_load_cost): Likewise.
29649 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29650 Alan Hayward <alan.hayward@arm.com>
29651 David Sherwood <david.sherwood@arm.com>
29653 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29654 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29655 from an unsigned int * to a poly_uint64_pod *.
29656 (calculate_unrolling_factor): New function.
29657 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29659 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29660 Alan Hayward <alan.hayward@arm.com>
29661 David Sherwood <david.sherwood@arm.com>
29663 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29664 from an unsigned int to a poly_uint64.
29665 (_loop_vec_info::slp_unrolling_factor): Likewise.
29666 (_loop_vec_info::vectorization_factor): Change from an int
29668 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29669 (vect_get_num_vectors): New function.
29670 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29671 (vect_get_num_copies): Use vect_get_num_vectors.
29672 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29673 to an unsigned int *.
29674 (vect_analyze_data_refs): Change min_vf from an int * to a
29676 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29677 than an unsigned HOST_WIDE_INT.
29678 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29679 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29680 to an unsigned int *.
29681 (vect_analyze_data_ref_dependences): Likewise.
29682 (vect_compute_data_ref_alignment): Handle polynomial vf.
29683 (vect_enhance_data_refs_alignment): Likewise.
29684 (vect_prune_runtime_alias_test_list): Likewise.
29685 (vect_shift_permute_load_chain): Likewise.
29686 (vect_supportable_dr_alignment): Likewise.
29687 (dependence_distance_ge_vf): Take the vectorization factor as a
29688 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29689 (vect_analyze_data_refs): Change min_vf from an int * to a
29691 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29692 vfm1 as a poly_uint64 rather than an int. Make the same change
29693 for the returned bound_scalar.
29694 (vect_gen_vector_loop_niters): Handle polynomial vf.
29695 (vect_do_peeling): Likewise. Update call to
29696 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29697 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29699 * tree-vect-loop.c (vect_determine_vectorization_factor)
29700 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29701 (vect_get_known_peeling_cost): Likewise.
29702 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29703 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29704 (vect_transform_loop): Likewise. Use the lowest possible VF when
29705 updating the upper bounds of the loop.
29706 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29707 rather than an int.
29708 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29709 polynomial unroll factors.
29710 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29711 (vect_make_slp_decision): Likewise.
29712 (vect_supported_load_permutation_p): Likewise, and polynomial
29714 (vect_analyze_slp_cost): Handle polynomial vf.
29715 (vect_slp_analyze_node_operations): Likewise.
29716 (vect_slp_analyze_bb_1): Likewise.
29717 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29718 than an unsigned HOST_WIDE_INT.
29719 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29720 (vectorizable_load): Handle polynomial vf.
29721 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29723 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29725 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29726 Alan Hayward <alan.hayward@arm.com>
29727 David Sherwood <david.sherwood@arm.com>
29729 * match.pd: Handle bit operations involving three constants
29730 and try to fold one pair.
29732 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29734 * tree-vect-loop-manip.c: Include gimple-fold.h.
29735 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29736 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29737 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29738 Add a path that uses a step of VF instead of 1, but disable it
29740 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29741 and niters_no_overflow parameters. Update calls to
29742 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29743 Create a new SSA name if the latter choses to use a ste other
29744 than zero, and return it via niters_vector_mult_vf_var.
29745 * tree-vect-loop.c (vect_transform_loop): Update calls to
29746 vect_do_peeling, vect_gen_vector_loop_niters and
29747 slpeel_make_loop_iterate_ntimes.
29748 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29749 (vect_gen_vector_loop_niters): Update declarations after above changes.
29751 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29753 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29754 128-bit round to integer instructions.
29755 (ceil<mode>2): Likewise.
29756 (btrunc<mode>2): Likewise.
29757 (round<mode>2): Likewise.
29759 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29761 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29762 unaligned VSX load/store on P8/P9.
29763 (expand_block_clear): Allow the use of unaligned VSX
29764 load/store on P8/P9.
29766 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29768 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29770 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29771 swap associated with both a load and a store.
29773 2018-01-02 Andrew Waterman <andrew@sifive.com>
29775 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29776 * config/riscv/riscv.md (clear_cache): Use it.
29778 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29780 * web.c: Remove out-of-date comment.
29782 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29784 * expr.c (fixup_args_size_notes): Check that any existing
29785 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29786 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29787 (emit_single_push_insn): ...here.
29789 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29791 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29792 (const_vector_encoded_nelts): New function.
29793 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29794 (const_vector_int_elt, const_vector_elt): Declare.
29795 * emit-rtl.c (const_vector_int_elt_1): New function.
29796 (const_vector_elt): Likewise.
29797 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29798 of CONST_VECTOR_ELT.
29800 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29802 * expr.c: Include rtx-vector-builder.h.
29803 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29804 directly on the tree encoding.
29805 (const_vector_from_tree): Likewise.
29806 * optabs.c: Include rtx-vector-builder.h.
29807 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29808 sequence of "u" values.
29809 * vec-perm-indices.c: Include rtx-vector-builder.h.
29810 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29811 directly on the vec_perm_indices encoding.
29813 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29815 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29816 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29817 * rtx-vector-builder.h: New file.
29818 * rtx-vector-builder.c: Likewise.
29819 * rtl.h (rtx_def::u2): Add a const_vector field.
29820 (CONST_VECTOR_NPATTERNS): New macro.
29821 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29822 (CONST_VECTOR_DUPLICATE_P): Likewise.
29823 (CONST_VECTOR_STEPPED_P): Likewise.
29824 (CONST_VECTOR_ENCODED_ELT): Likewise.
29825 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29826 (unwrap_const_vec_duplicate): Likewise.
29827 (const_vec_series_p): Check for a non-duplicated vector encoding.
29828 Say that the function only returns true for integer vectors.
29829 * emit-rtl.c: Include rtx-vector-builder.h.
29830 (gen_const_vec_duplicate_1): Delete.
29831 (gen_const_vector): Call gen_const_vec_duplicate instead of
29832 gen_const_vec_duplicate_1.
29833 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29834 (gen_const_vec_duplicate): Use rtx_vector_builder.
29835 (gen_const_vec_series): Likewise.
29836 (gen_rtx_CONST_VECTOR): Likewise.
29837 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29838 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29839 Build a new vector rather than modifying a CONST_VECTOR in-place.
29840 (handle_special_swappables): Update call accordingly.
29841 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29842 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29843 Build a new vector rather than modifying a CONST_VECTOR in-place.
29844 (handle_special_swappables): Update call accordingly.
29846 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29848 * simplify-rtx.c (simplify_const_binary_operation): Use
29849 CONST_VECTOR_ELT instead of XVECEXP.
29851 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29853 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29854 the selector elements to be different from the data elements
29855 if the selector is a VECTOR_CST.
29856 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29857 ssizetype for the selector.
29859 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29861 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29862 before testing each element individually.
29863 * tree-vect-generic.c (lower_vec_perm): Likewise.
29865 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29867 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29868 * selftest-run-tests.c (selftest::run_tests): Call it.
29869 * vector-builder.h (vector_builder::operator ==): New function.
29870 (vector_builder::operator !=): Likewise.
29871 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29872 (vec_perm_indices::all_from_input_p): New function.
29873 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29874 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29875 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29876 instead of reading the VECTOR_CST directly. Detect whether both
29877 vector inputs are the same before constructing the vec_perm_indices,
29878 and update the number of inputs argument accordingly. Use the
29879 utility functions added above. Only construct sel2 if we need to.
29881 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29883 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29884 the broadcast of the low byte.
29885 (expand_mult_highpart): Use an explicit encoding for the permutes.
29886 * optabs-query.c (can_mult_highpart_p): Likewise.
29887 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29888 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29889 (vectorizable_bswap): Likewise.
29890 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29891 explicit encoding for the power-of-2 permutes.
29892 (vect_permute_store_chain): Likewise.
29893 (vect_grouped_load_supported): Likewise.
29894 (vect_permute_load_chain): Likewise.
29896 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29898 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29899 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29900 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29901 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29902 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29903 (vect_gen_perm_mask_any): Likewise.
29905 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29907 * int-vector-builder.h: New file.
29908 * vec-perm-indices.h: Include int-vector-builder.h.
29909 (vec_perm_indices): Redefine as an int_vector_builder.
29910 (auto_vec_perm_indices): Delete.
29911 (vec_perm_builder): Redefine as a stand-alone class.
29912 (vec_perm_indices::vec_perm_indices): New function.
29913 (vec_perm_indices::clamp): Likewise.
29914 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29915 (vec_perm_indices::new_vector): New function.
29916 (vec_perm_indices::new_expanded_vector): Update for new
29917 vec_perm_indices class.
29918 (vec_perm_indices::rotate_inputs): New function.
29919 (vec_perm_indices::all_in_range_p): Operate directly on the
29920 encoded form, without computing elided elements.
29921 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29922 encoding. Update for new vec_perm_indices class.
29923 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29924 the given vec_perm_builder.
29925 (expand_vec_perm_var): Update vec_perm_builder constructor.
29926 (expand_mult_highpart): Use vec_perm_builder instead of
29927 auto_vec_perm_indices.
29928 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29929 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29930 or double series encoding as appropriate.
29931 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29932 vec_perm_indices instead of auto_vec_perm_indices.
29933 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29934 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29935 (vect_permute_store_chain): Likewise.
29936 (vect_grouped_load_supported): Likewise.
29937 (vect_permute_load_chain): Likewise.
29938 (vect_shift_permute_load_chain): Likewise.
29939 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29940 (vect_transform_slp_perm_load): Likewise.
29941 (vect_schedule_slp_instance): Likewise.
29942 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29943 (vectorizable_mask_load_store): Likewise.
29944 (vectorizable_bswap): Likewise.
29945 (vectorizable_store): Likewise.
29946 (vectorizable_load): Likewise.
29947 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29948 vec_perm_indices instead of auto_vec_perm_indices. Use
29949 tree_to_vec_perm_builder to read the vector from a tree.
29950 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29951 vec_perm_builder instead of a vec_perm_indices.
29952 (have_whole_vector_shift): Use vec_perm_builder and
29953 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29954 truncation to calc_vec_perm_mask_for_shift.
29955 (vect_create_epilog_for_reduction): Likewise.
29956 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29957 from auto_vec_perm_indices to vec_perm_indices.
29958 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29959 instead of changing individual elements.
29960 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29961 the vector in d.perm.
29962 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29963 from auto_vec_perm_indices to vec_perm_indices.
29964 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29965 instead of changing individual elements.
29966 (arm_vectorize_vec_perm_const): Use new_vector to install
29967 the vector in d.perm.
29968 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29969 Update vec_perm_builder constructor.
29970 (rs6000_expand_interleave): Likewise.
29971 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29972 (rs6000_expand_interleave): Likewise.
29974 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29976 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29977 to qimode could truncate the indices.
29978 * optabs.c (expand_vec_perm_var): Likewise.
29980 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29982 * Makefile.in (OBJS): Add vec-perm-indices.o.
29983 * vec-perm-indices.h: New file.
29984 * vec-perm-indices.c: Likewise.
29985 * target.h (vec_perm_indices): Replace with a forward class
29987 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29988 * optabs.h: Include vec-perm-indices.h.
29989 (expand_vec_perm): Delete.
29990 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29991 (expand_vec_perm_const): Declare.
29992 * target.def (vec_perm_const_ok): Replace with...
29993 (vec_perm_const): ...this new hook.
29994 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29995 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29996 * doc/tm.texi: Regenerate.
29997 * optabs.def (vec_perm_const): Delete.
29998 * doc/md.texi (vec_perm_const): Likewise.
29999 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
30000 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
30001 expand_vec_perm for constant permutation vectors. Assert that
30002 the mode of variable permutation vectors is the integer equivalent
30003 of the mode that is being permuted.
30004 * optabs-query.h (selector_fits_mode_p): Declare.
30005 * optabs-query.c: Include vec-perm-indices.h.
30006 (selector_fits_mode_p): New function.
30007 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
30008 is defined, instead of checking whether the vec_perm_const_optab
30009 exists. Use targetm.vectorize.vec_perm_const instead of
30010 targetm.vectorize.vec_perm_const_ok. Check whether the indices
30011 fit in the vector mode before using a variable permute.
30012 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
30013 vec_perm_indices instead of an rtx.
30014 (expand_vec_perm): Replace with...
30015 (expand_vec_perm_const): ...this new function. Take the selector
30016 as a vec_perm_indices rather than an rtx. Also take the mode of
30017 the selector. Update call to shift_amt_for_vec_perm_mask.
30018 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
30019 Use vec_perm_indices::new_expanded_vector to expand the original
30020 selector into bytes. Check whether the indices fit in the vector
30021 mode before using a variable permute.
30022 (expand_vec_perm_var): Make global.
30023 (expand_mult_highpart): Use expand_vec_perm_const.
30024 * fold-const.c: Includes vec-perm-indices.h.
30025 * tree-ssa-forwprop.c: Likewise.
30026 * tree-vect-data-refs.c: Likewise.
30027 * tree-vect-generic.c: Likewise.
30028 * tree-vect-loop.c: Likewise.
30029 * tree-vect-slp.c: Likewise.
30030 * tree-vect-stmts.c: Likewise.
30031 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
30033 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
30034 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
30035 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
30036 (aarch64_vectorize_vec_perm_const): ...this new function.
30037 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30038 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30039 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
30040 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
30041 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30042 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30043 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
30045 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
30046 check for NEON modes.
30047 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
30048 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
30049 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
30050 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
30052 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
30053 the old VEC_PERM_CONST conditions.
30054 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
30055 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
30056 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
30057 (ia64_vectorize_vec_perm_const_ok): Merge into...
30058 (ia64_vectorize_vec_perm_const): ...this new function.
30059 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
30060 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
30061 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
30062 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
30063 * config/mips/mips.c (mips_expand_vec_perm_const)
30064 (mips_vectorize_vec_perm_const_ok): Merge into...
30065 (mips_vectorize_vec_perm_const): ...this new function.
30066 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
30067 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
30068 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
30069 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
30070 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
30071 (rs6000_expand_vec_perm_const): Delete.
30072 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
30074 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30075 (altivec_expand_vec_perm_const_le): Take each operand individually.
30076 Operate on constant selectors rather than rtxes.
30077 (altivec_expand_vec_perm_const): Likewise. Update call to
30078 altivec_expand_vec_perm_const_le.
30079 (rs6000_expand_vec_perm_const): Delete.
30080 (rs6000_vectorize_vec_perm_const_ok): Delete.
30081 (rs6000_vectorize_vec_perm_const): New function.
30082 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
30083 an element count and rtx array.
30084 (rs6000_expand_extract_even): Update call accordingly.
30085 (rs6000_expand_interleave): Likewise.
30086 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
30087 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
30088 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
30089 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
30090 (rs6000_expand_vec_perm_const): Delete.
30091 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
30092 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30093 (altivec_expand_vec_perm_const_le): Take each operand individually.
30094 Operate on constant selectors rather than rtxes.
30095 (altivec_expand_vec_perm_const): Likewise. Update call to
30096 altivec_expand_vec_perm_const_le.
30097 (rs6000_expand_vec_perm_const): Delete.
30098 (rs6000_vectorize_vec_perm_const_ok): Delete.
30099 (rs6000_vectorize_vec_perm_const): New function. Remove stray
30100 reference to the SPE evmerge intructions.
30101 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
30102 an element count and rtx array.
30103 (rs6000_expand_extract_even): Update call accordingly.
30104 (rs6000_expand_interleave): Likewise.
30105 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
30106 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
30108 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
30110 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30112 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
30113 vector mode and that that mode matches the mode of the data
30115 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
30116 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
30117 directly using expand_vec_perm_1 when forcing selectors into
30119 (expand_vec_perm_var): New function, split out from expand_vec_perm.
30121 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30123 * optabs-query.h (can_vec_perm_p): Delete.
30124 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
30125 * optabs-query.c (can_vec_perm_p): Split into...
30126 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
30127 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
30128 particular selector is valid.
30129 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30130 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30131 (vect_grouped_load_supported): Likewise.
30132 (vect_shift_permute_load_chain): Likewise.
30133 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
30134 (vect_transform_slp_perm_load): Likewise.
30135 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30136 (vectorizable_bswap): Likewise.
30137 (vect_gen_perm_mask_checked): Likewise.
30138 * fold-const.c (fold_ternary_loc): Likewise. Don't take
30139 implementations of variable permutation vectors into account
30140 when deciding which selector to use.
30141 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
30142 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
30143 with a false third argument.
30144 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
30145 to test whether the constant selector is valid and can_vec_perm_var_p
30146 to test whether a variable selector is valid.
30148 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30150 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
30151 * optabs-query.c (can_vec_perm_p): Likewise.
30152 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
30153 instead of vec_perm_indices.
30154 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
30155 (vect_gen_perm_mask_checked): Likewise,
30156 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
30157 (vect_gen_perm_mask_checked): Likewise,
30159 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30161 * optabs-query.h (qimode_for_vec_perm): Declare.
30162 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
30163 (qimode_for_vec_perm): ...this new function.
30164 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
30166 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30168 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
30169 does not have a conditional at the top.
30171 2018-01-02 Richard Biener <rguenther@suse.de>
30173 * ipa-inline.c (big_speedup_p): Fix expression.
30175 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30178 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
30181 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30185 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
30186 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
30187 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
30188 cond_taken_branch_cost 3->4.
30190 2018-01-01 Jakub Jelinek <jakub@redhat.com>
30192 PR tree-optimization/83581
30193 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
30194 TODO_cleanup_cfg if any changes have been made.
30196 PR middle-end/83608
30197 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
30198 convert_modes if target mode has the right side, but different mode
30201 PR middle-end/83609
30202 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
30203 last argument when extracting from CONCAT. If either from_real or
30204 from_imag is NULL, use expansion through memory. If result is not
30205 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
30206 the parts directly to inner mode, if even that fails, use expansion
30209 PR middle-end/83623
30210 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
30211 check for bswap in mode rather than HImode and use that in expand_unop
30214 Copyright (C) 2018 Free Software Foundation, Inc.
30216 Copying and distribution of this file, with or without modification,
30217 are permitted in any medium without royalty provided the copyright
30218 notice and this notice are preserved.