1 2018-11-13 David Malcolm <dmalcolm@redhat.com>
3 * builtins.c: Replace "source_location" with "location_t".
4 * diagnostic-show-locus.c: Likewise.
5 * diagnostic.c: Likewise.
6 * dumpfile.c: Likewise.
7 * gcc-rich-location.h: Likewise.
8 * genmatch.c: Likewise.
10 * gimplify.c: Likewise.
12 * input.h: Likewise. Eliminate the typedef.
13 * omp-expand.c: Likewise.
14 * selftest.h: Likewise.
15 * substring-locations.h (get_source_location_for_substring):
17 (get_location_within_string): ...this.
18 * tree-cfg.c: Replace "source_location" with "location_t".
19 * tree-cfgcleanup.c: Likewise.
20 * tree-diagnostic.c: Likewise.
21 * tree-into-ssa.c: Likewise.
22 * tree-outof-ssa.c: Likewise.
23 * tree-parloops.c: Likewise.
24 * tree-phinodes.c: Likewise.
25 * tree-phinodes.h: Likewise.
26 * tree-ssa-loop-ivopts.c: Likewise.
27 * tree-ssa-loop-manip.c: Likewise.
28 * tree-ssa-phiopt.c: Likewise.
29 * tree-ssa-phiprop.c: Likewise.
30 * tree-ssa-threadupdate.c: Likewise.
31 * tree-ssa.c: Likewise.
32 * tree-ssa.h: Likewise.
33 * tree-vect-loop-manip.c: Likewise.
35 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
37 * doc/extend.texi: Fix typo in the weakref description.
39 2018-11-13 Richard Biener <rguenther@suse.de>
41 PR tree-optimization/86991
42 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
43 group building until we have successfully detected the SLP
45 (vect_is_simple_reduction): Remove fixup code here.
47 2018-11-13 David Malcolm <dmalcolm@redhat.com>
49 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
50 (dump_gimple_stmt): Use it.
51 (dump_gimple_stmt_loc): Likewise.
52 (dump_gimple_expr): Likewise.
53 (dump_gimple_expr_loc): Likewise.
54 (dump_generic_expr): Likewise.
55 (dump_generic_expr_loc): Likewise.
56 (dump_printf): Likewise.
57 (dump_printf_loc): Likewise.
61 (dump_symtab_node): Likewise.
62 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
63 Guard dump call with dump_enabled_p.
64 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
65 * graphite-optimize-isl.c (optimize_isl): Likewise.
66 * graphite.c (graphite_transform_loops): Likewise.
67 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
68 * tree-parloops.c (parallelize_loops): Likewise.
69 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
70 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
71 (vect_prune_runtime_alias_test_list): Likewise.
72 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
73 (vect_estimate_min_profitable_iters): Likewise.
74 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
75 (vect_build_slp_tree_2): Likewise.
76 (vect_supported_load_permutation_p): Likewise.
77 (vect_slp_analyze_operations): Likewise.
78 (vect_slp_analyze_bb_1): Likewise.
79 (vect_slp_bb): Likewise.
80 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
81 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
82 (pass_slp_vectorize::execute): Likewise.
83 (increase_alignment): Likewise.
85 2018-11-13 David Malcolm <dmalcolm@redhat.com>
88 * ipa-inline.c (report_inline_failed_reason): Guard calls to
89 cl_target_option_print_diff and cl_optimization_print_diff with
92 2018-11-13 David Malcolm <dmalcolm@redhat.com>
94 * doc/invoke.texi (-fsave-optimization-record): Note that the
96 * optinfo-emit-json.cc: Include <zlib.h>.
97 (optrecord_json_writer::write): Compress the output.
99 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
101 * tree-vrp.c (value_range_base::dump): Dump type.
102 Do not use INF nomenclature for 1-bit types.
103 (dump_value_range): Group all variants to common dumping code.
104 (debug): New overloaded functions for value_ranges.
105 (value_range_base::dump): Remove no argument version.
106 (value_range::dump): Same.
108 2018-11-13 Richard Biener <rguenther@suse.de>
110 PR tree-optimization/87931
111 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
112 nested cycles we support to latch computations vectorizable_reduction
115 2018-11-13 Martin Liska <mliska@suse.cz>
117 PR tree-optimization/87885
118 * cfghooks.c (account_profile_record): Rename
120 (profile_record_check_consistency): ... this.
121 Calculate missing num_mismatched_freq_in.
122 (profile_record_account_profile): New function
123 that calculates time and size of a function.
124 * cfghooks.h (struct profile_record): Remove
126 (struct cfg_hooks): Remove after_pass flag.
127 (account_profile_record): Rename to ...
128 (profile_record_check_consistency): ... this.
129 (profile_record_account_profile): New.
130 * cfgrtl.c (rtl_account_profile_record): Remove
132 * passes.c (check_profile_consistency): Do only
134 (account_profile): Calculate size and time of
136 (pass_manager::dump_profile_report): Reformat
138 (execute_one_ipa_transform_pass): Call
139 consistency check before clean upand call account_profile
140 after a clean up is done.
141 (execute_one_pass): Call check_profile_consistency and
142 account_profile instead of using after_pass flag..
143 * tree-cfg.c (gimple_account_profile_record): Likewise.
145 2018-11-13 Martin Liska <mliska@suse.cz>
148 * config/i386/i386.c (ix86_option_override_internal): Error
149 about usage -mabi=ms and -fsanitize={,kernel-}address.
151 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
153 * config/aarch64/aarch64.c
154 (aarch64_vectorize_preferred_vector_alignment): Change return type to
156 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
157 alignment being a poly int.
158 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
159 return type to poly_uint64.
160 * target.def (default_preferred_vector_alignment): Likewise.
161 * targhooks.c (default_preferred_vector_alignment): Likewise.
162 * targhooks.h (default_preferred_vector_alignment): Likewise.
163 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
164 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
166 (vect_update_misalignment_for_peel): Likewise.
167 (vect_enhance_data_refs_alignment): Likewise.
168 (vect_find_same_alignment_drs): Likewise.
169 (vect_duplicate_ssa_name_ptr_info): Likewise.
170 (vect_setup_realignment): Likewise.
171 (vect_can_force_dr_alignment_p): Change alignment parameter type to
173 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
174 mask with a compile time variable vector alignment.
175 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
177 (vect_do_peeling): Exit early if vector alignment is not constant.
178 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
180 (vectorizable_store): Likewise.
181 (vectorizable_load): Likweise.
182 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
184 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
186 (vect_can_force_dr_alignment_p): Change alignment parameter type to
189 2018-11-13 Richard Biener <rguenther@suse.de>
191 PR tree-optimization/87962
192 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
193 detect outer reduction for disqualifying in-loop uses.
195 2018-11-13 Richard Biener <rguenther@suse.de>
197 PR tree-optimization/87967
198 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
199 for constants for the scalar loop.
201 2018-11-13 Alan Modra <amodra@gmail.com>
203 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
204 integer constants when -mcmodel=medium.
206 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
208 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
209 * config/arc/elf.h (PROFILE_HOOK): Define.
210 * config/arc/linux.h (PROFILE_HOOK): Likewise.
212 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
214 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
215 jump table data in the text section.
217 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
219 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
220 to fit the eh_return pattern.
221 * config/arc/arc.md (eh_return): Define.
222 (VUNSPEC_ARC_EH_RETURN): Likewise.
223 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
225 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
228 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
230 * common/config/arc/arc-common.c (arc_option_optimization_table):
231 Millicode optimization is default on for size optimizations.
232 * config/arc/arc-protos.h (arc_check_multi): New function.
233 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
234 (ENTER_LEAVE_START_REG): Define.
235 (ENTER_LEAVE_END_REG): Likewise.
236 (arc_override_options): Disable millicode when long calls option
238 (arc_frame_info): Change it from int to bool.
239 (arc_compute_frame_size): Clean up.
240 (arc_save_restore): Remove.
241 (frame_save_reg): New function.
242 (frame_restore_reg): Likewise.
243 (arc_enter_leave_p): Likewise.
244 (arc_save_callee_saves): Likewise.
245 (arc_restore_callee_saves): Likewise.
246 (arc_save_callee_enter): Likewise.
247 (arc_restore_callee_leave): Likewise.
248 (arc_save_callee_milli): Likewise.
249 (arc_restore_callee_milli): Likewise.
250 (arc_expand_prologue): Reimplement to emit enter/leave
252 (arc_expand_epilogue): Likewise.
253 (arc_check_multi): New function.
254 * config/arc/arc.md (push_multi_fp): New pattern.
255 (push_multi_fp_blink): Likewise.
256 (pop_multi_fp): Likewise.
257 (pop_multi_fp_blink): Likewise.
258 (pop_multi_fp_ret): Likewise.
259 (pop_multi_fp_blink_ret): Likewise.
260 * config/arc/arc.opt (mmillicode): Update option.
261 (mcode-density-frame): New option.
262 * config/arc/predicates.md (push_multi_operand): New predicate.
263 (pop_multi_operand): Likewise.
264 * doc/invoke.texi (ARC): Update ARC options information.
267 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
269 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
270 * config/arc/arc.c (operands_ok_ldd_std): New function.
271 (mem_ok_for_ldd_std): Likewise.
272 (gen_operands_ldd_std): Likewise.
273 * config/arc/arc.md: Add peephole2 rules for std/ldd.
275 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
277 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
278 test on stack_usage_file.
279 (lang_dependent_init): Do not open the .su file if generating LTO.
281 2018-11-13 Jakub Jelinek <jakub@redhat.com>
283 PR rtl-optimization/87918
284 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
285 simplify_gen_relational rather than simplify_gen_binary.
287 2018-11-13 Richard Biener <rguenther@suse.de>
289 * tree-ssanames.h (set_range_info): Use value_range_base.
290 (get_range_info): Likewise.
291 * tree-ssanames.c (set_range_info): Likewise.
292 (get_range_info): Likewise.
293 * tree-vrp.c (value_range_base::union_helper): Split
294 out common parts of value_range[_base]::union_.
295 (value_range_base::union_): Update.
296 (value_range::union_): Likewise.
297 (determine_value_range_1): Use value_range_base.
298 (determine_value_range): Likewise.
299 * tree-vrp.h (value_range_base::union_helper): Move ...
300 (value_range::union_helper): ... from here.
302 2018-11-13 Alan Modra <amodra@gmail.com>
304 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
306 (rs6000_secondary_reload_gpr): Don't call find_replacement.
308 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
311 * doc/invoke.texi (Optimize Options): Clarify that the
312 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
313 apply to data cache size.
315 2018-11-13 Alan Modra <amodra@gmail.com>
317 * config/rs6000/predicates.md (logical_const_operand),
318 (logical_operand): Correct comment.
319 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
321 2018-11-13 Alan Modra <amodra@gmail.com>
323 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
325 2018-11-13 Alan Modra <amodra@gmail.com>
327 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
329 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
330 covered by alternative.
331 (movcc_internal1): Ignore h for register preference.
332 (mov<mode>_hardfloat64): Likewise.
333 (mov<mode>_softfloat): Ignore c, l, h for register preference.
335 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
337 PR preprocessor/47823
338 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
341 * doc/extend.texi (Pragmas): Note additional pragmas documented
344 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
348 * tree-scalar-evolution.c (expression_expensive_p): Make BUILTIN POPCOUNT
349 as expensive when backend does not define it.
351 2018-11-12 Fredrik Noring <noring@nocrew.org>
353 * config.gcc: Update with-llsc defaults for MIPS r5900.
355 2018-11-12 Martin Liska <mliska@suse.cz>
357 PR gcov-profile/87442
358 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
360 * doc/invoke.texi: Document them.
361 * tree-profile.c (parse_profile_filter): New.
362 (parse_profile_file_filtering): Likewise.
363 (release_profile_file_filtering): Likewise.
364 (include_source_file_for_profile): Likewise.
365 (tree_profiling): Filter source files based on the
368 2018-11-12 Martin Liska <mliska@suse.cz>
371 * doc/extend.texi: Add missing values for __builtin_cpu_is and
372 __builtin_cpu_supports for x86 target.
374 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
377 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
379 (Arithmetic): Likewise for "lo_sum".
381 2018-11-12 Renlin Li <renlin.li@arm.com>
384 * dse.c (get_stored_val): Add check for compile-time constantness
387 2018-11-12 Sudakshina Das <sudi.das@arm.com>
389 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
390 (ARMv8_5a): New fgroup.
391 (armv8.5-a): New arch.
392 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
394 * config/arm/arm-tables.opt: Regenerate.
395 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
396 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
397 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
398 Add matching rules for -march=armv8.5-a and extensions.
399 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
400 Add sb and predres to all armv8-a except armv8.5-a.
402 2018-11-12 Richard Biener <rguenther@suse.de>
404 * tree-vrp.h (value_range[_base]::set): Make public. Provide
405 overload for single value.
406 (value_range[_base]::set_nonnull): New.
407 (value_range[_base]::set_null): Likewise.
408 (value_range): Document bitmap copying behavior, mark
409 copy constructor and assignment operator deleted.
410 (value_range::move): New.
411 (value_range::set_and_canonicalize): Default bitmap to zero.
412 (set_value_range_to_nonnull): Remove.
413 (set_value_range_to_null): Likewise.
414 (set_value_range): Likewise.
415 (set_value_range_to_value): Likewise.
416 (extract_range_from_unary_expr): Work on value_range_base.
417 (extract_range_from_binary_expr_1): Likewise. Rename to...
418 (extract_range_from_binary_expr): ... this.
419 * tree-vrp.c (value_range::update): Clear equiv bitmap
421 (value_range::move): New, move equiv bitmap.
422 (value_range_base::set_undefined): Avoid assignment.
423 (value_range::set_undefined): Likewise.
424 (value_range_base::set_varying): Likewise.
425 (value_range::set_varying): Likewise.
426 (set_value_range): Remove.
427 (value_range_base::set): New overload for value.
428 (value_range::set): Likewise.
429 (set_value_range_to_nonnull): Remove.
430 (value_range_base::set_nonnull): New.
431 (value_range::set_nonnull): Likewise.
432 (set_value_range_to_null): Remove.
433 (value_range_base::set_null): New.
434 (value_range::set_null): Likewise.
435 (range_is_null): Work on value_range_base.
436 (range_is_nonnull): Likewise.
437 (ranges_from_anti_range): Likewise.
438 (extract_range_into_wide_ints): Likewise.
439 (extract_range_from_multiplicative_op): Likewise.
440 (extract_range_from_binary_expr): Likewise. Update for API changes.
441 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
443 (value_range::intersect_helper): Avoid copy and assignment.
444 (value_range::union_helper): Likewise.
445 (determine_value_range_1): Adjust.
446 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
447 Avoid assignment by using move.
448 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
449 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
451 * tree-ssanames.c (get_range_info): Likewise.
452 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
453 * vr-values.c (vr_values::get_value_range): Adjust.
454 (vr_values::update_value_range): Likewise.
455 (symbolic_range_based_on_p): Work on value_range_base.
456 (vr_values::extract_range_from_binary_expr): Use value_range_base.
457 (vr_values::extract_range_from_unary_expr): Likewise.
458 (vr_values::extract_range_from_cond_expr): Avoid assignment.
459 (vr_values::extract_range_from_comparison): Adjust.
460 (vr_values::check_for_binary_op_overflow): Use value_range_base.
461 (vr_values::extract_range_basic): Adjust.
462 (vr_values::adjust_range_with_scev): Likewise.
463 (vr_values::vrp_visit_assignment_or_call): Likewise.
464 (vr_values::get_vr_for_comparison): Change API to avoid
465 assignment and copy construction.
466 (vr_values::compare_name_with_value): Adjust accordingly.
467 (vr_values::compare_names): Likewise.
468 (vr_values::extract_range_from_phi_node): Avoid assignment and
469 bogus in-place modify of equiv bitmap.
470 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
471 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
472 for extract_range_from_unary_expr API change.
473 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
475 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
477 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
479 2018-11-12 Richard Biener <rguenther@suse.de>
481 * tree-vrp.h (value_range_base::symbolic_p,
482 value_range_base::constant_p, value_range_base::zero_p,
483 value_range_base::singleton_p): Move from value_range.
484 (value_range::dump): Add.
485 * gimple-ssa-evrp-analyze.c
486 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
487 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
488 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
490 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
491 (value_range::constant_p): Likewise.
492 (value_range::singleton_p): Likewise.
493 (value_range_base::dump): Add.
494 (set_value_range_to_undefined): Remove.
495 (set_value_range_to_varying): Likewise.
496 (range_int_cst_p): Take value_range_base argument.
497 (range_int_cst_singleton_p): Likewise.
498 (value_range_constant_singleton): Likewise.
499 (vrp_set_zero_nonzero_bits): Likewise.
500 (extract_range_from_multiplicative_op): Use set_varying.
501 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
502 (extract_range_from_unary_expr): Likewise.
503 (dump_value_range_base): Change to overload of dump_value_range.
504 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
505 (vrp_prop::visit_stmt): Likewise.
506 (value_range::intersect_helper): Likewise.
507 (value_range::union_helper): Likewise.
508 (determine_value_range_1): Likewise.
510 2018-11-12 Richard Biener <rguenther@suse.de>
512 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
513 (set_value_range_to_null): Likewise.
514 * vr-values.c (vr_values::extract_range_from_comparison):
515 Clear equiv for constant singleton ranges.
517 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
519 * config/i386/sse.md: Combine VFIXUPIMM* patterns
520 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
521 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
522 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
523 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
524 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
525 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
527 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
530 * doc/extend.texi (Common Type Attributes): For the align type
531 attribute, copy language about decreasing alignment from the
532 corresponding variable attribute.
534 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
536 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
539 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
542 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
543 -frtti and -fno-rtti code.
545 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
548 * doc/extend.texi (Other Builtins): Document probability associated
549 with __builtin_expect.
551 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
554 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
555 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
556 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
557 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
559 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
561 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
563 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
565 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
567 2018-11-11 Richard Biener <rguenther@suse.de>
569 * tree-vrp.h (class value_range_base): New base class for
570 value_range containing all but the m_equiv member.
571 (dump_value_range_base): Add.
572 (range_includes_zero_p): Work on value_range_base.
573 * tree-vrp.c (value_range_base::set): Split out base handling
575 (value_range::set): this.
576 (value_range::set_equiv): New.
577 (value_range_base::value_range_base): New constructors.
578 (value_range_base::check): Split out base handling from...
579 (value_range::check): this.
580 (value_range::equal_p): Refactor in terms of
581 ignore_equivs_equal_p which is now member of the base.
582 (value_range_base::set_undefined): New.
583 (value_range_base::set_varying): Likewise.
584 (value_range_base::dump):Split out base handling from...
585 (value_range::dump): this.
586 (value_range_base::set_and_canonicalize): Split out base handling
588 (value_range::set_and_canonicalize): this.
589 (value_range_base::union_): New.
590 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
592 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
593 instead of value_range everywhere.
594 (ipcp_vr_lattice::print): Use dump_value_range_base.
595 (ipcp_vr_lattice::meet_with): Adjust.
596 (ipcp_vr_lattice::meet_with_1): Likewise.
597 (ipa_vr_operation_and_type_effects): Likewise.
598 (propagate_vr_across_jump_function): Likewise.
599 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
600 (ipa_get_value_range): Likewise.
601 (ipa_set_jfunc_vr): Likewise.
602 (ipa_compute_jump_functions_for_edge): Likewise.
604 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
607 * doc/invoke.texi (Optimize Options): Add @opindex entries
608 for the positive forms of -fno-xxx and -mno-xxx options
609 that were lacking them.
611 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
613 * combine.c (make_more_copies): Only make an intermediate copy if the
614 dest of a move is a pseudo.
616 2018-11-09 Maya Rashish <coypu@sdf.org>
619 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
620 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
622 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
626 * doc/invoke.texi (Optimize Options): Clarify default behavior
627 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
629 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
631 PR tree-optimization/87940
632 * expr.c (string_constant): Don't strip NOPS in subexpressions.
633 Fold PLUS_EXPR correctly.
635 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
638 * config/s390/s390.c (s390_safe_relative_long_p): New function.
639 (annotate_constant_pool_refs): Skip insns which support
641 (annotate_constant_pool_refs_1): New helper function.
642 (find_constant_pool_ref): Skip insns which support relative
644 (find_constant_pool_ref_1): New helper function.
645 (replace_constant_pool_ref): Skip insns which support
647 (replace_constant_pool_ref_1): New helper function.
648 (s390_mainpool_start): Adapt to the new signature.
649 (s390_mainpool_finish): Likewise.
650 (s390_chunkify_start): Likewise.
651 (s390_chunkify_finish): Likewise.
652 (pass_s390_early_mach::execute): Likewise.
653 (s390_prologue_plus_offset): Likewise.
654 (s390_emit_prologue): Likewise.
655 (s390_emit_epilogue): Likewise.
657 2018-11-09 Jakub Jelinek <jakub@redhat.com>
659 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
660 but unsupported lastprivate with conditional modifier.
662 2018-11-09 Jeff Law <law@redhat.com>
664 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
665 unused argument better. Add gcc_unreachable to silence warning.
667 2018-11-09 Martin Sebor <msebor@redhat.com>
670 * attribs.c (has_attribute): New helper function.
671 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
672 * attribs.h (decls_mismatched_attributes): Declare.
673 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
674 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
675 * common.opt (-Wattribute-alias): Take an argument.
676 (-Wno-attribute-alias): New option.
677 * doc/extend.texi (Common Function Attributes): Document copy.
678 (Common Variable Attributes): Same.
679 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
680 (-Wattribute-alias): Document new option argument.
682 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
684 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
685 matching criteria. Remove unused array initializer.
687 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
688 Jinsong Ji <jji@us.ibm.com>
690 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
691 constraints by introducing a new temporary.
692 (_mm_cvtss_si64): Likewise.
694 2018-11-09 Martin Liska <mliska@suse.cz>
696 * common.opt: Add -fipa-stack-alignment flag.
697 * doc/invoke.texi: Document it.
698 * final.c (rest_of_clean_state): Guard stack
701 2018-11-09 Martin Liska <mliska@suse.cz>
703 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
705 (ipa_discover_variable_flags): ... this.
706 * common.opt: Come up with new flag -fipa-reference-addressable.
707 * doc/invoke.texi: Document it.
708 * ipa-reference.c (propagate): Call the renamed fn.
709 * ipa-visibility.c (whole_program_function_and_variable_visibility):
711 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
713 (ipa_discover_variable_flags): ... this. Discover
714 non-addressable variables only with the newly added flag.
715 * opts.c: Enable the newly added flag with -O1 and higher
718 2018-11-09 David Malcolm <dmalcolm@redhat.com>
720 * json.cc (selftest::test_writing_literals): Fix comment.
722 2018-11-09 Martin Liska <mliska@suse.cz>
724 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
725 string to a stack buffer.
726 (aarch64_parse_cpu): Likewise.
727 (aarch64_parse_tune): Likewise.
729 2018-11-09 Richard Biener <rguenther@suse.de>
731 PR tree-optimization/87953
732 * tree-vect-loop.c (vectorizable_reduction): For analysis
733 always pass ops[0] to vectorizable_condition.
735 2018-11-09 Stafford Horne <shorne@gmail.com>
736 Richard Henderson <rth@twiddle.net>
737 Joel Sherrill <joel@rtems.org>
739 * common/config/or1k/or1k-common.c: New file.
740 * config/or1k/*: New.
741 * config.gcc (or1k*-*-*): New.
742 * configure.ac (or1k*-*-*): New test for openrisc tls.
743 * configure: Regenerated.
744 * doc/install.texi: Document OpenRISC triplets.
745 * doc/invoke.texi: Document OpenRISC arguments.
746 * doc/md.texi: Document OpenRISC.
748 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
750 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
751 (arm7tdmi-s): Delete CPU.
752 (arm710t): Add aliases for arm720t and arm740t.
753 (arm720t, arm740t): Delete CPUs.
754 (arm920t): Add aliases for arm920, arm922t and arm940t.
755 (arm920, arm922t, arm940t): Delete CPUs.
756 (arm10tdmi): Add alias for arm1020t.
757 (arm1020t): Delete CPU.
758 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
759 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
760 (arm10e): Add aliases for arm1020e and arm1022e.
761 (arm1020e, arm1022e): Delete CPU.
762 * config/arm/arm.md (generic_sched): Remove entries that are now
764 (generic_vfp): Likewise.
765 * config/arm/arm1020e.md: Simplify tuning selection based on alias
767 * config/arm/arm-tune.md: Regenerated.
768 * config/arm/arm-tables.opt: Regenerated.
770 2018-11-09 Richard Biener <rguenther@suse.de>
772 PR tree-optimization/87621
773 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
774 op with only phi inputs.
775 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
776 (ch_base::copy_headers): Run CSE on copied loop headers.
777 (pass_ch_vect::process_loop_p): Simplify.
779 2018-11-09 Alexandre Oliva <oliva@adacore.com>
781 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
782 for non-w64 x86_64 biarch.
784 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
786 PR rtl-optimization/86438
787 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
788 of in_b for the compare if in_b is SET_DEST.
791 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
794 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
796 * tree-vrp.c (value_range::check): Do not access internals
798 (value_range::singleton_p): Same.
799 (value_range::type): Same.
800 (vrp_finalize): Use value_range API.
802 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
804 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
806 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
808 * vr-values.c (vr_values::get_value_range): Use value_range API
809 instead of piecing together ranges.
810 (vr_values::update_value_range): Same.
812 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
814 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
815 of performing ad-hoc calculations.
816 * tree-ssanames.c (set_range_info): New overloaded function
817 accepting value_range &.
818 (get_range_info): Same.
819 * tree-ssanames.h (set_range_info_raw): Remove.
820 (set_range_info): New prototype.
821 (get_range_info): Same.
822 * tree-vrp.h (value_range::null_p): Rename to zero_p.
823 * tree-vrp.c (value_range::null_p): Same.
825 2018-11-09 Jan Hubicka <jh@suse.cz>
827 * tree.c (fld_type_variant_equal_p): Test user align flag.
828 (flt_type_variant): Copy user align flag.
829 (fld_incomplete_type_of): Clear it.
831 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
833 * config/arm/neon.md (div<mode>3): New pattern.
835 2018-11-08 Andi Kleen <ak@linux.intel.com>
837 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
838 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
839 (ix86_handle_option): Handle OPT_mptwrite.
840 * config/i386/cpuid.h (bit_PTWRITE): Add.
841 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
842 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
843 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
844 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
845 (ix86_option_override_internal): Handle PTA_PTWRITE.
846 (ix86_valid_target_attribute_inner_p): Define ptwrite.
847 (def_builtin2): Force UINT64 to be 64bit only.
848 * config/i386/i386.h (TARGET_PTWRITE): Add.
849 (TARGET_PTWRITE_P): Add.
851 * config/i386/i386.md: Define ptwrite.
852 * config/i386/i386.opt: Add -mptwrite.
853 * config/i386/immintrin.h (_ptwrite64): Add.
855 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
856 * doc/invoke.texi: Document -mptwrite.
858 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
860 PR rtl-optimization/87600
861 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
862 * lra-constraints.c (process_alt_operands): Skip illegal hard
863 register usage. Prefer reloading non hard register operands.
865 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
868 * doc/invoke.texi (Optimize Options): Clarify default behavior
869 for -fno-sched-interblock and -fno-sched-spec.
871 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
873 * collect2.c (linker_select): Add USE_LLD_LD.
874 (ld_suffixes): Add ld.lld.
875 (main): Handle -fuse-ld=lld.
876 * common.opt (-fuse-ld=lld): New option.
877 * doc/invoke.texi (-fuse-ld=lld): Document.
878 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
880 2018-11-08 Paul Koning <ni1d@arrl.net>
882 * config/pdp11/constraints.md: Add "Z" series constraints for use
883 with pre-dec and post-inc addressing.
884 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
885 (pdp11_expand_operands): Add int argument (word count).
886 (pdp11_sp_frame_offset): Delete.
887 (pdp11_cmp_length): New function.
888 (pushpop_regeq): New function.
889 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
891 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
893 (pdp11_initial_elimination_offset): Ditto.
894 (pdp11_expand_operands): Add word count argument. Bugfixes.
895 (output_move_multiple): Change how pointer adjustment is done.
896 (pdp11_gen_int_label): Correct format.
897 (output_ascii): Ditto.
898 (pdp11_asm_output_var): Add code for DEC assembler case.
899 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
901 (legitimate_const_double_p): Ditto.
902 (pdp11_register_move_cost): Adjust for new register classes.
903 (pdp11_regno_reg_class): Ditto.
904 (expand_block_move): Delete.
905 (pushpop_regeq): New function.
906 (pdp11_legitimate_address_p): Bugfix in check for constant
908 (pdp11_sp_frame_offset): Delete.
909 (pdp11_reg_save_size): New helper function for new frame layout.
910 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
911 (pdp11_expand_shift): Bugfix in check for constant shift count.
912 (pdp11_shift_length): Ditto.
913 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
914 (pdp11_cmp_length): New function.
915 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
916 some compile options.
917 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
918 (CALL_USED_REGISTERS): Ditto.
919 (ELIMINABLE_REGS): Ditto.
920 (REGISTER_NAMES): Ditto.
921 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
923 (REG_CLASS_NAMES): Ditto.
924 (REG_CLASS_CONTENTS): Ditto. Also remove
925 HARD_FRAME_POINTER_REGNUM.
926 (CPU_REG_CLASS): New macro.
927 (CLASS_MAX_NREGS): Adjust for new register classes.
928 (FUNCTION_PROFILER): Make no-op.
929 (may_call_alloca): Remove unused declaration.
930 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
931 (ASM_OUTPUT_SKIP): Fix format.
932 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
933 (HARD_FRAME_POINTER_REGNUM): Remove.
935 (*rts): Rename. Remove epilogue related checks.
936 (cmpsi, cmpdi): New insn.
937 (cbranch<mode>4): Change to apply to SI and DI modes as well.
938 (mov<mode>): Change constraints to enforce that push/pop
939 destination cannot use the same register as source.
940 (*mov<mode><cc_cc>): Ditto.
941 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
942 at assembly output rather than as RTL expander.
943 (zero_extendqihi2): Bugfix in check for same registers.
944 (adddi3_nocc): Bugfix in check for constant operand.
945 (addsi3_nocc): Ditto.
946 (subdi3_nocc): Ditto.
947 (subsi3_nocc): Ditto.
948 (negdi2_nocc): Copy input to pdp11_expand_operands.
949 (negsi2_nocc): Ditto.
950 (bswap2_nocc): Ditto.
951 * config/pdp11/pdp11.opt (mlra): Fix documentation.
952 * config/pdp11/t-pdp11: Use -Os.
954 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
956 * config/arm/parsecpu.awk (/alias/): New parsing rule.
957 (/begin cpu/): Check that the cpu name hasn't been previously defined.
958 (gen_comm_data): Print out CPU alias tables.
959 (check_cpu): Match aliases when checking the CPU name.
960 * config/arm/arm-protos.h (cpu_alias): New structure.
961 (cpu_option): Add entry for aliases.
962 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
963 strongarm1100 and strongarm1110.
964 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
965 (config/arm/arm-generic.md): Remove redundant references to
966 strongarm110, strongarm1100 and strongarm1110.
967 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
968 Scan aliases for additional hints.
969 (arm_parse_cpu_option_name): Also match a cpu name against the list
971 * config/arm/arm-tables.opt: Regenerated.
972 * config/arm/arm-tune.md: Regenerated.
974 2018-11-08 Jakub Jelinek <jakub@redhat.com>
976 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
977 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
978 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
979 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
980 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
981 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
982 * gengtype.c (open_base_files): Add omp-general.h.
983 * gimple.c (gimple_build_omp_critical):
984 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
985 gimple_omp_taskgroup_set_clauses.
986 (gimple_build_omp_atomic_load): Add mo argument, call
987 gimple_omp_atomic_set_memory_order.
988 (gimple_build_omp_atomic_store): Likewise.
989 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
990 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
992 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
993 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
994 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
995 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
996 different value for GF_OMP_ATOMIC_NEED_VALUE.
997 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
999 (struct gimple_statement_omp_single_layout): And remove here.
1000 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
1001 than gimple_statement_omp_single_layout.
1002 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
1004 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
1005 (gimple_omp_subcode): Formatting fix.
1006 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
1007 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
1008 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
1009 gimple_omp_teams_host, gimple_omp_teams_set_host,
1010 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
1011 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
1012 gimple_omp_taskgroup_set_clauses): New inline functions.
1013 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
1014 (gimple_build_omp_atomic_store): Likewise.
1015 (gimple_omp_atomic_seq_cst_p): Remove.
1016 (gimple_omp_atomic_memory_order): New function.
1017 (gimple_omp_atomic_set_seq_cst): Remove.
1018 (gimple_omp_atomic_set_memory_order): New function.
1019 (gimple_build_omp_taskgroup): Add clauses argument.
1020 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
1021 (dump_gimple_omp_task): Print taskwait with depend clauses.
1022 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
1023 dump_omp_atomic_memory_order.
1024 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
1025 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
1026 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
1027 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
1028 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
1029 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
1030 ORT_UNTIED_TASKLOOP enumerators.
1031 (enum gimplify_defaultmap_kind): New.
1032 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
1033 target_map_pointers_as_0len_arrays members, add defaultmap.
1034 (new_omp_context): Initialize defaultmap member.
1035 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1036 (maybe_fold_stmt): Don't fold even in host teams regions.
1037 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
1038 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
1039 ctx->omp_firstprivatize_variable.
1040 (omp_add_variable): Don't add private/firstprivate for VLAs in
1042 (omp_default_clause): Print "taskloop" rather than "task" if
1044 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1045 Handle new defaultmap clause kinds.
1046 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
1047 iterator to be lastprivate or private. Fix up diagnostics if linear
1048 is used on collapse>1 simd iterator.
1049 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1050 (gimplify_omp_depend): New function.
1051 (gimplify_scan_omp_clauses): Add shared clause on parallel for
1052 combined parallel master taskloop{, simd} if taskloop has
1053 firstprivate, lastprivate or reduction clause. Handle
1054 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
1055 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
1056 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
1057 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
1058 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
1060 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
1061 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
1062 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
1063 OMP_CLAUSE_{TASK,IN}_REDUCTION.
1064 (gimplify_omp_task): Handle taskwait with depend clauses.
1065 (gimplify_omp_for): Add shared clause on parallel for combined
1066 parallel master taskloop{, simd} if taskloop has firstprivate,
1067 lastprivate or reduction clause. Use ORT_TASKLOOP or
1068 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
1069 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
1070 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
1071 __for_end and __for_range temporaries on OMP_PARALLEL for
1072 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
1073 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
1074 sandwiched in between two taskloops.
1075 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
1076 instead of ctx->omp_firstprivatize_variable.
1077 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
1078 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
1079 host teams, use gimplify_and_return_first etc. for body like
1080 for target or target data constructs, and at the end call
1081 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
1082 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
1083 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
1084 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
1085 gimple_omp_atomic_set_seq_cst calls.
1086 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
1087 case, handle taskgroup clauses.
1088 * lto-streamer-out.c (hash_tree): Handle
1089 OMP_CLAUSE_{TASK,IN}_REDUCTION.
1090 * Makefile.in (GTFILES): Add omp-general.h.
1091 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
1092 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
1093 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
1094 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
1095 BUILT_IN_GOMP_LOOP_DOACROSS_START,
1096 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
1097 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
1098 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
1099 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
1100 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
1101 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
1102 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
1103 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
1104 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
1105 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
1106 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
1107 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
1108 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
1109 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
1110 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
1111 * omp-expand.c (workshare_safe_to_combine_p): Return false for
1112 non-worksharing loops.
1113 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
1114 (determine_parallel_type): Don't combine parallel with worksharing
1115 which has _reductemp_ clause.
1116 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
1117 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
1118 if there is nonmonotonic modifier or if there is no modifier and no
1119 ordered clause. For dynamic and guided schedule without monotonic
1120 and nonmonotonic modifier, default to nonmonotonic.
1121 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
1122 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
1124 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
1125 there are any reduction clauses.
1126 (expand_taskwait_call): New function.
1127 (expand_teams_call): New function.
1128 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
1129 expand_teams_call for it. Formatting fix. Handle taskwait with
1131 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
1132 of worksharing loops with task reductions.
1133 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
1134 expansion of worksharing loops with task reductions.
1135 (expand_omp_sections): Handle expansion of sections with task
1137 (expand_omp_synch): For host teams call expand_omp_taskreg.
1138 (omp_memory_order_to_memmodel): New function.
1139 (expand_omp_atomic_load, expand_omp_atomic_store,
1140 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
1141 instead of gimple_omp_atomic_seq_cst_p.
1142 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
1143 depend clauses as a standalone directive.
1144 * omp-general.c (enum omp_requires): New variable.
1145 (omp_extract_for_data): Initialize have_reductemp member. Allow
1146 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
1147 GT_EXPR loops depending on incr sign. Formatting fixes.
1148 * omp-general.h (struct omp_for_data): Add have_reductemp member.
1149 (enum omp_requires): New enum.
1150 (omp_requires_mask): Declare.
1151 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
1153 * omp-low.c (struct omp_context): Add task_reductions and
1154 task_reduction_map fields.
1155 (is_host_teams_ctx): New function.
1156 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
1157 (use_pointer_for_field): Use is_global_var instead of
1158 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
1160 (build_outer_var_ref): Ignore taskgroup outer contexts.
1161 (delete_omp_context): Release task_reductions and task_reduction_map.
1162 (scan_sharing_clauses): Don't add any fields for reduction clause on
1163 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
1164 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
1165 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
1166 Handle OMP_CLAUSE_NONTEMPORAL.
1167 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
1169 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
1170 clauses with task modifier.
1171 (scan_omp_task): Handle taskwait with depend clauses.
1172 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
1173 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
1174 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
1175 (scan_omp_for): Fix comment formatting.
1176 (scan_omp_teams): Handle host teams constructs.
1177 (check_omp_nesting_restrictions): Allow teams with no outer
1178 OpenMP context. Adjust diagnostics for teams strictly nested into
1179 some explicit OpenMP construct other than target. Allow OpenMP atomics
1180 inside of simd regions.
1181 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
1182 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
1183 taskreg_nesting_level while scanning host teams construct.
1184 (task_reduction_read): New function.
1185 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
1186 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
1187 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
1188 second argument create_tmp_var if it is NULL. Don't ignore shared
1189 clauses in is_host_teams_ctx contexts. Handle
1190 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
1192 (lower_reduction_clauses): Ignore reduction clauses with task
1193 modifier. Remove second argument create_tmp_var if it is NULL.
1194 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
1195 (lower_send_clauses): Ignore reduction clauses with task modifier.
1196 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
1197 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
1198 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
1199 rely that it is the last stmt in body so far. Ignore outer taskgroup
1201 (omp_task_reductions_find_first, omp_task_reduction_iterate,
1202 lower_omp_task_reductions): New functions.
1203 (lower_omp_sections): Handle reduction clauses with taskgroup
1204 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
1205 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
1206 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
1208 (lower_omp_taskgroup): Handle taskgroup reductions.
1209 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
1210 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
1211 (lower_depend_clauses): If there are any
1212 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
1213 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
1214 seen, assume lowering is done already and return early. Set kind
1215 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
1216 (lower_omp_taskreg): Handle reduction clauses with task modifier on
1217 parallel construct. Handle reduction clause on taskloop construct.
1218 Handle taskwait with depend clauses.
1219 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
1220 for host teams constructs.
1221 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
1222 nontemporal and _reductemp_ clause entries.
1223 (omp_clause_code_name): Likewise.
1224 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
1225 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
1226 * tree-core.h (enum omp_clause_code): Add
1227 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
1228 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
1229 (struct tree_base): Add omp_atomic_memory_order field into union.
1230 Remove OMP_ATOMIC_SEQ_CST comment.
1231 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
1232 and OMP_CLAUSE_DEPEND_DEPOBJ.
1233 (struct tree_omp_clause): Add subcode.defaultmap_kind.
1234 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
1235 OpenMP constructs with body and clauses operands.
1236 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
1237 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
1238 (OMP_TASKGROUP_CLAUSES): Define.
1239 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
1240 OMP_CLAUSE__LOOPTEMP_.
1241 (OMP_ATOMIC_SEQ_CST): Remove.
1242 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
1243 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
1244 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
1245 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
1246 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
1247 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
1248 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
1249 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
1250 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
1251 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
1253 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
1254 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
1255 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
1256 (convert_local_omp_clauses): Likewise. Remove useless test.
1257 * tree-parloops.c (create_call_for_reduction_1): Pass
1258 OMP_MEMORY_ORDER_RELAXED as new argument to
1259 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
1260 * tree-pretty-print.c (dump_omp_iterators): New function.
1261 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
1262 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
1263 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
1264 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
1265 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
1266 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
1267 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
1268 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
1269 (dump_omp_atomic_memory_order): New function.
1270 (dump_generic_node): Use it. Print taskgroup clauses. Print
1271 taskwait with depend clauses.
1272 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
1273 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
1274 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
1275 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
1276 write_ts_omp_clause_tree_pointers): Likewise.
1278 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1281 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
1282 "-missed", "-note", and "-all" sub-options.
1283 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
1285 (can_early_inline_edge_p): Likewise.
1286 (want_early_inline_function_p): Likewise.
1287 (want_inline_self_recursive_call_p): Likewise.
1288 (recursive_inlining): Likewise.
1289 (inline_small_functions): Likewise.
1290 (flatten_function): Likewise.
1291 (ipa_inline): Likewise.
1292 (inline_always_inline_functions): Likewise.
1293 (early_inline_small_functions): Likewise.
1294 (early_inliner): Likewise.
1295 * tree-inline.c (expand_call_inline): Likewise.
1297 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1299 * pretty-print.c (pp_format): Handle %f.
1300 (selftest::test_pp_format): Add test of %f.
1301 * pretty-print.h (pp_double): New macro.
1303 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1305 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
1306 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
1307 (dump_pretty_printer::decode_format): Implement "%C" for
1309 (selftest::test_capture_of_dump_calls): Rename "where" to
1310 "stmt_loc". Convert test_decl to a function decl and set its
1311 location. Add a symbol_table_test RAII instance and a
1312 cgraph_node, using it to test "%C" and dump_symtab_node.
1314 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
1317 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
1319 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1321 * cgraph.c: Include "selftest.h".
1322 (saved_symtab): New variable.
1323 (selftest::symbol_table_test::symbol_table_test): New ctor.
1324 (selftest::symbol_table_test::~symbol_table_test): New dtor.
1325 (selftest::test_symbol_table_test): New test.
1326 (selftest::cgraph_c_tests): New.
1327 * cgraph.h (saved_symtab): New decl.
1328 (selftest::symbol_table_test): New class.
1329 * selftest-run-tests.c (selftest::run_tests): Call
1330 selftest::cgraph_c_tests.
1331 * selftest.h (selftest::cgraph_c_tests): New decl.
1333 2018-11-08 Richard Biener <rguenther@suse.de>
1335 * tree-data-ref.h (lambda_int): New typedef.
1336 (lambda_vector_gcd): Adjust.
1337 (lambda_vector_new): Likewise.
1338 (lambda_matrix_new): Likewise.
1339 * tree-data-ref.c (print_lambda_vector): Adjust.
1341 2018-11-08 Richard Biener <rguenther@suse.de>
1343 PR tree-optimization/87929
1344 * tree-complex.c (expand_complex_comparison): Clean EH.
1346 2018-11-08 Martin Liska <mliska@suse.cz>
1348 * doc/extend.texi: Reword.
1349 * predict.c (expr_expected_value_1): Likewise.
1351 2018-11-08 Richard Biener <rguenther@suse.de>
1353 PR tree-optimization/87913
1354 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
1355 of extreme values to ordered comparisons.
1357 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1360 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
1362 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1364 * doc/invoke.texi: Remove leading dash from @opindex entries
1365 throughout the file.
1367 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1370 * doc/invoke.texi (Option Summary): Add -e and --entry.
1371 (Link Options): Likewise.
1373 2018-11-07 Nathan Sidwell <nathan@acm.org>
1376 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
1377 --disable-checking bootstrap.
1379 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
1381 * configure: Regenerated.
1383 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1386 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
1387 to the mode of the widest field iff the widest field has mode class
1388 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
1391 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
1393 * common.opt: Add -Wattribute-warning.
1394 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
1395 * expr.c (expand_expr_real_1): Add new attribute to warning_at
1396 call to allow user configure behavior of "warning" attribute.
1398 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1400 * target.def: Put @: after every vs., e.g., and i.e. where it is
1401 followed by whitespace.
1402 * doc/extend.texi: Ditto.
1403 * doc/fragments.texi: Ditto.
1404 * doc/gimple.texi: Ditto.
1405 * doc/implement-c.texi: Ditto.
1406 * doc/install.texi: Ditto.
1407 * doc/invoke.texi: Ditto.
1408 * doc/md.texi: Ditto.
1409 * doc/plugins.texi: Ditto.
1410 * doc/rtl.texi: Ditto.
1411 * doc/sourcebuild.texi: Ditto.
1412 * doc/tm.texi.in: Ditto.
1413 * doc/ux.texi: Ditto.
1414 * doc/tm.texi: Regenerate.
1416 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1418 * config/arm/arm-cpus.in (ares): New entry.
1419 * config/arm/arm-tables.opt: Regenerate.
1420 * config/arm/arm-tune.md: Likewise.
1421 * doc/invoke.texi (ARM Options): Document ares.
1423 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1425 * config/aarch64/aarch64-cores.def (ares): Define.
1426 * config/aarch64/aarch64-tune.md: Regenerate.
1427 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
1429 2018-11-07 Jan Hubicka <jh@suse.cz>
1431 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
1432 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
1433 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
1436 2018-11-07 Richard Biener <rguenther@suse.de>
1438 PR tree-optimization/87914
1439 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
1441 (vectorizable_reduction): Handle shifts and rotates by dispatching
1442 to vectorizable_shift.
1443 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
1444 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
1446 (vectorizable_shift): Export and handle being called as
1448 (vect_analyze_stmt): Call vectorizable_shift after
1449 vectorizable_reduction.
1450 * tree-vectorizer.h (vectorizable_shift): Declare.
1452 2018-11-07 Jan Hubicka <jh@suse.cz>
1454 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
1455 than const decls in TREE_VALUE of enum.
1456 (dump_type_inheritance_graph): Improve duplicate dumping.
1457 (free_enum_values): New.
1458 (build_type_inheritance_graph): Use it.
1459 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
1460 which are not main variants or not ODR types.
1461 (verify_type_variant): Expect variants to have no TYPE_VALUES.
1463 2018-11-07 Richard Biener <rguenther@suse.de>
1465 * ipa-inline.c (want_inline_small_function_p): Compute
1466 big_speedup_p lazily and last.
1468 2018-11-07 Jan Hubicka <jh@suse.cz>
1470 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
1471 building incomplete variant of complete type.
1472 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
1473 variant of complete type.
1475 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1477 * config/mips/mips.c: Fix typo in documentation of
1478 mips_loongson_ext2_prefetch_cookie.
1479 (mips_option_override): fix brain twister logical.
1480 * config/mips/mips.h: Fix typo in documentation of
1481 ISA_HAS_CTZ_CTO and define pattern.
1482 * config/mips/mips.md (prefetch): Hoist EXT2 above
1484 (prefetch_indexed): Hoist EXT2 above the EXT block.
1486 2018-11-07 Jan Hubicka <jh@suse.cz>
1488 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
1489 return and parameter types of function and method types.
1490 (free_lang_data_in_cgraph): Update.
1492 2018-11-07 Martin Liska <mliska@suse.cz>
1494 PR rtl-optimization/87868
1495 * postreload-gcse.c (eliminate_partially_redundant_load): Set
1496 threshold to max_count if we would overflow.
1497 * profile-count.h: Make max_count a public constant.
1499 2018-11-07 Martin Liska <mliska@suse.cz>
1501 * mem-stats.h: Fix GNU coding style.
1503 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1505 * config/mips/gs264e.md: New.
1506 * config/mips/mips-cpus.def: Define gs264e.
1507 * config/mips/mips-tables.opt: Regenerate.
1508 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
1510 (mips_issue_rate): Add support for gs264e.
1511 (mips_multipass_dfa_lookahead): Likewise.
1512 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
1513 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
1514 (MIPS_ASE_MSA_SPEC): New.
1515 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
1516 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
1517 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
1518 * config/mips/mips.md: Include gs264e.md.
1519 (processor): Add gs264e.
1520 * config/mips/mips.opt (MSA): Use Mask instead of Var.
1521 * doc/invoke.texi: Add gs264e to supported architectures.
1523 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1525 * config/mips/gs464e.md: New.
1526 * config/mips/mips-cpus.def: Define gs464e.
1527 * config/mips/mips-tables.opt: Regenerate.
1528 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
1530 (mips_issue_rate): Add support for gs464e.
1531 (mips_multipass_dfa_lookahead): Likewise.
1532 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
1533 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
1534 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
1535 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
1536 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
1537 * config/mips/mips.md: Include gs464e.md.
1538 (processor): Add gs464e.
1539 * doc/invoke.texi: Add gs464e to supported architectures.
1541 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1543 * config/mips/loongson3a.md: Rename to ...
1544 * config/mips/gs464.md: ... here.
1545 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
1546 as an alias of gs464 processor.
1547 * config/mips/mips-tables.opt: Regenerate.
1548 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
1549 instead of PROCESSOR_LOONGSON_3A.
1550 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
1552 (mips_option_override): Enable MMI and EXT for gs464.
1553 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
1554 Rename TUNE_LOONGSON_3A to TUNE_GS464.
1555 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
1556 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
1557 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
1559 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
1560 (processor): Add gs464;
1561 * doc/invoke.texi: Add gs464 to supported architectures.
1563 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1565 * config/mips/mips-protos.h
1566 (mips_loongson_ext2_prefetch_cookie): New prototype.
1567 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
1568 (mips_option_override): Enable TARGET_LOONGSON_EXT when
1569 TARGET_LOONGSON_EXT2 is true.
1570 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
1571 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
1572 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
1573 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
1574 TARGET_LOONGSON_EXT2.
1575 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
1576 (define_insn "ctz<mode>2"): New insn pattern.
1577 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
1578 (define_insn "prefetch_indexed_<mode>"): Include
1579 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
1580 * config/mips/mips.opt (-mloongson-ext2): Add option.
1581 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
1583 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1585 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
1586 __mips_loongson_ext.
1587 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
1588 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
1590 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
1591 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
1592 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
1593 instead of TARGET_LOONGSON_3A.
1594 * config/mips/mips.opt (-mloongson-ext): Add option.
1595 * gcc/doc/invoke.texi (-mloongson-ext): Document.
1597 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1599 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
1600 * config/mips/loongson.md: Move to ...
1601 * config/mips/loongson-mmi.md: here; Adjustment.
1602 * config/mips/loongson.h: Move to ...
1603 State as deprecated. Include loongson-mmiintrin.h for back
1604 compatibility and warning.
1605 * config/mips/loongson-mmiintrin.h: ... here.
1606 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
1607 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
1608 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
1609 (mips_option_override): Make sure MMI use hard float;
1610 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
1611 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
1612 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
1613 TARGET_LOONGSON_VECTORS.
1614 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
1615 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
1616 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
1617 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
1619 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
1620 TARGET_LOONGSON_VECTORS.
1621 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
1622 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
1623 (Loongson MMI patterns): Include loongson-mmi.md instead of
1625 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
1626 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
1628 2018-11-07 Richard Biener <rguenther@suse.de>
1631 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
1632 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
1634 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
1636 PR rtl-optimization/87874
1637 * lra.c (lra_substitute_pseudo): Do not create a subreg for
1640 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
1642 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
1643 if not in indexed or indirect form.
1644 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
1645 (bswap<mode>2_store): Ditto.
1647 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
1649 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
1652 2018-11-06 Richard Biener <rguenther@suse.de>
1654 PR tree-optimization/86850
1655 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
1656 instead of src.m_vec.
1658 2018-11-06 Jan Hubicka <jh@suse.cz>
1660 * tree.c (fld_simplified_type_name): Break out form ...
1661 (free_lang_data_in_type): ... here.
1662 (fld_type_variant_equal_p): Use it.
1664 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1666 * config/default-d.c: Include memmodel.h.
1668 * config/sol2-d.c: New file.
1669 * config/t-sol2 (sol2-d.o): New rule.
1670 * config.gcc <*-*-solaris2*>: Set d_target_objs,
1671 target_has_targetdm.
1673 2018-11-06 Jan Hubicka <jh@suse.cz>
1675 * tree.c (fld_type_variant): Also copy alignment; be sure that
1676 new variant is equal.
1678 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
1681 * config/s390/s390.md: Add relative_long attribute.
1683 2018-11-06 Jan Hubicka <jh@suse.cz>
1685 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
1686 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
1687 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
1688 stream TYPE_NEEDS_CONSTRUCTING.
1689 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
1690 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
1692 2018-11-06 Richard Biener <rguenther@suse.de>
1694 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
1696 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
1698 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1700 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
1701 REG_CLASS_CONTENTS[GEN_REGS].
1702 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
1704 2018-11-06 Jan Hubicka <jh@suse.cz>
1706 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
1708 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1711 * config/sol2.h (ASAN_CC1_SPEC): Define.
1712 (LD_WHOLE_ARCHIVE_OPTION): Define.
1713 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
1714 (ASAN_REJECT_SPEC): Provide default.
1715 (LIBASAN_EARLY_SPEC): Define.
1716 (LIBTSAN_EARLY_SPEC): Define.
1717 (LIBLSAN_EARLY_SPEC): Define.
1718 * config/i386/sol2.h (CC1_SPEC): Redefine.
1719 (ASAN_REJECT_SPEC): Define.
1721 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
1722 (TARGET_ASAN_SHADOW_OFFSET): Define.
1723 (sparc_asan_shadow_offset): New function.
1724 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
1725 (ASAN_REJECT_SPEC): Define.
1727 2018-11-06 Jan Hubicka <jh@suse.cz>
1729 * tree.c (fld_type_variant): Copy canonical type.
1730 (fld_incomplete_type_of): Check that canonical types looks sane;
1731 copy canonical type.
1732 (verify_type): Accept when incomplete type has complete canonical type.
1734 2018-11-06 Jan Hubicka <jh@suse.cz>
1736 * tree.c (free_lang_data): Reset overwite_assembler_name,
1737 print_xnode, print_decl, print_type and print_identifier of
1740 2018-11-06 Richard Biener <rguenther@suse.de>
1742 PR tree-optimization/87889
1743 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
1744 Do nothing if old and new arg are the same
1746 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
1749 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
1750 attributes for operands 3 and 4.
1752 2018-11-06 Richard Biener <rguenther@suse.de>
1755 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
1756 matching bitfield insertion.
1758 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
1760 * auto-inc-dec.c: Include valtrack.h. Improve comments.
1761 (reg_next_debug_use): New.
1762 (attempt_change): Propagate adjusted expression into affected
1764 (merge_in_block): Track uses in debug insns.
1765 (pass_inc_dec::execute): Allocate and release
1768 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
1770 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
1771 (_mm512_fixupimm_round_pd): Update parameters and builtin.
1772 (_mm512_maskz_fixupimm_round_pd): Ditto.
1773 (_mm512_fixupimm_round_ps): Ditto.
1774 (_mm512_maskz_fixupimm_round_ps): Ditto.
1775 (_mm_fixupimm_round_sd): Ditto.
1776 (_mm_maskz_fixupimm_round_sd): Ditto.
1777 (_mm_fixupimm_round_ss): Ditto.
1778 (_mm_maskz_fixupimm_round_ss): Ditto.
1779 (_mm512_fixupimm_pd): Ditto.
1780 (_mm512_maskz_fixupimm_pd): Ditto.
1781 (_mm512_fixupimm_ps): Ditto.
1782 (_mm512_maskz_fixupimm_ps): Ditto.
1783 (_mm_fixupimm_sd): Ditto.
1784 (_mm_maskz_fixupimm_sd): Ditto.
1785 (_mm_fixupimm_ss): Ditto.
1786 (_mm_maskz_fixupimm_ss): Ditto.
1787 (_mm512_mask_fixupimm_round_pd): Update builtin.
1788 (_mm512_mask_fixupimm_round_ps): Ditto.
1789 (_mm_mask_fixupimm_round_sd): Ditto.
1790 (_mm_mask_fixupimm_round_ss): Ditto.
1791 (_mm512_mask_fixupimm_pd): Ditto.
1792 (_mm512_mask_fixupimm_ps): Ditto.
1793 (_mm_mask_fixupimm_sd): Ditto.
1794 (_mm_mask_fixupimm_ss): Ditto.
1795 * config/i386/avx512vlintrin.h:
1796 (_mm256_fixupimm_pd): Update parameters and builtin.
1797 (_mm256_maskz_fixupimm_pd): Ditto.
1798 (_mm256_fixupimm_ps): Ditto.
1799 (_mm256_maskz_fixupimm_ps): Ditto.
1800 (_mm_fixupimm_pd): Ditto.
1801 (_mm_maskz_fixupimm_pd): Ditto.
1802 (_mm_fixupimm_ps): Ditto.
1803 (_mm_maskz_fixupimm_ps): Ditto.
1804 (_mm256_mask_fixupimm_pd): Update builtin.
1805 (_mm256_mask_fixupimm_ps): Ditto.
1806 (_mm_mask_fixupimm_pd): Ditto.
1807 (_mm_mask_fixupimm_ps): Ditto.
1808 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
1809 * config/i386/i386-builtin.def: Update builtin definitions.
1810 * config/i386/i386.c: Handle new builtin types and remove useless ones.
1811 * config/i386/sse.md: Update VFIXUPIMM* patterns.
1812 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1813 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1814 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
1815 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1816 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1817 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
1818 * config/i386/subst.md:
1819 (round_saeonly_sd_mask_operand4): Add new subst_attr.
1820 (round_saeonly_sd_mask_op4): Ditto.
1821 (round_saeonly_expand_operand5): Ditto.
1822 (round_saeonly_expand): Update.
1824 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
1826 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
1828 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
1830 PR rtl-optimization/87871
1831 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
1833 2018-11-05 Paul Koning <ni1d@arrl.net>
1835 * doc/sourcebuild.texi (target attributes): Document new "inf"
1836 effective target keyword.
1838 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
1840 * config/s390/s390.c (s390_register_move_cost): Increase costs for
1841 moves involving the CC reg.
1843 2018-11-05 Richard Biener <rguenther@suse.de>
1845 PR tree-optimization/87873
1846 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
1848 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
1849 * tree-vect-loop.c (vect_transform_loop): When splitting the
1850 loop exit also create forwarder PHIs for constants.
1851 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
1852 Handle constant to_arg, add extra checking we match up the correct
1855 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
1857 * config/s390/s390.md: QImode and HImode for load on condition.
1859 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
1861 * config/s390/predicates.md: Fix typo.
1862 * config/s390/s390.md: Allow immediates for load on condition.
1864 2018-11-05 Martin Liska <mliska@suse.cz>
1866 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
1867 * bitmap.h (struct bitmap_usage): Likewise.
1868 * ggc-common.c (SCALE): Remove.
1870 (struct ggc_usage): Use SIZE_AMOUNT. And update
1872 * ggc-page.c (SCALE): Remove.
1873 (STAT_LABEL): Remove.
1874 (ggc_print_statistics): Use SIZE_AMOUNT.
1875 * gimple.h (SCALE): Remove.
1877 * input.c (ONE_K): Remove.
1880 (STAT_LABEL): Likewise.
1881 (FORMAT_AMOUNT): Likewise.
1882 (dump_line_table_statistics): Use SIZE_AMOUNT.
1883 * mem-stats.h (struct mem_usage): Likewise.
1884 * rtl.c (dump_rtx_statistics): Likewise.
1885 (rtx_alloc_counts): Change type to size_t.
1886 (rtx_alloc_sizes): Likewise.
1887 (rtx_count_cmp): New.
1888 (dump_rtx_statistics): Sort first based on counts.
1889 * tree.c (tree_nodes_cmp): New.
1890 (tree_codes_cmp): New.
1891 (dump_tree_statistics): Sort first based on counts.
1892 * system.h (ONE_K): New.
1894 (SIZE_SCALE): Likewise.
1895 (SIZE_LABEL): Likewise.
1896 (SIZE_AMOUNT): Likewise.
1897 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
1898 * tree-dfa.c (dump_dfa_stats): Likewise.
1899 * tree-phinodes.c (phinodes_print_statistics): Likewise.
1900 * tree-ssanames.c (ssanames_print_statistics): Likewise.
1901 * tree.c (dump_tree_statistics): Likewise.
1902 * vec.c (struct vec_usage): Likewise.
1903 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
1904 trigger a -Werror=format-overflow with
1905 --enable-gather-detailed-stats.
1907 2018-11-05 Martin Liska <mliska@suse.cz>
1909 * mem-stats.h (mem_alloc_description::release_instance_overhead):
1911 * vec.c (struct vec_usage): Register m_element_size.
1912 (vec_prefix::register_overhead): New arguments: elements and
1914 (vec_prefix::release_overhead): Subtract elements.
1915 * vec.h (struct vec_prefix): Change signature.
1916 (va_heap::reserve): Pass proper arguments.
1917 (va_heap::release): Likewise.
1919 2018-11-05 Martin Liska <mliska@suse.cz>
1921 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
1925 2018-11-05 Richard Biener <rguenther@suse.de>
1927 * tree-scalar-evolution.h (final_value_replacement_loop): Update
1929 * tree-scalar-evolution.c (final_value_replacement_loop): Return
1930 whether anything was done.
1931 (scev_const_prop): Remove constant propagation part, fold
1933 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
1934 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
1937 2018-11-05 Jakub Jelinek <jakub@redhat.com>
1939 PR tree-optimization/87859
1940 * gimple-ssa-store-merging.c (struct merged_store_group): Add
1941 only_constants and first_nonmergeable_order members.
1942 (merged_store_group::merged_store_group): Initialize them.
1943 (merged_store_group::do_merge): Clear only_constants member if
1944 adding something other than INTEGER_CST store.
1945 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
1946 stores with order >= first_nonmergeable_order. Use
1947 merged_store->only_constants instead of always recomputing it.
1948 Set merged_store->first_nonmergeable_order if we've skipped any
1949 stores. Attempt to merge overlapping INTEGER_CST stores that
1950 we would otherwise skip.
1953 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
1955 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
1958 * config/i386/emmintrin.h (__v16qs): New to cope with option
1960 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
1961 (_mm_cmplt_epi8): Likewise.
1962 (_mm_cmpgt_epi8): Likewise.
1964 2018-11-05 Richard Biener <rguenther@suse.de>
1966 PR rtl-optimization/87852
1967 * fwprop.c (use_killed_between): Only consider single-defs of the
1968 use whose definition statement dominates the use.
1970 2018-11-05 Martin Liska <mliska@suse.cz>
1973 * doc/invoke.texi: Remove options that are
1974 not disabled with -Os.
1976 2018-11-05 Martin Liska <mliska@suse.cz>
1979 * doc/extend.texi: Update constrain about the last argument
1980 of __builtin_expect_with_probability.
1982 2018-11-05 Martin Liska <mliska@suse.cz>
1985 * predict.c (expr_expected_value_1): Verify
1986 that last argument is a real constants and emit
1989 2018-11-05 Martin Liska <mliska@suse.cz>
1991 PR gcov-profile/77698
1992 * ipa-profile.c (ipa_profile): Adjust hotness threshold
1995 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1997 PR tree-optimization/86572
1998 * builtins.c (c_strlen): Handle negative offsets in a safe way.
2000 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2002 PR tree-optimization/87672
2003 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
2004 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
2006 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
2009 * cfgexpand.c (pass_expand::execute): Move the call to
2010 finish_eh_generation in front of the call to expand_stack_alignment.
2012 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2014 * common/config/i386/i386-common.c (processor_alias_table): Add
2016 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
2017 (case ${target}): Add znver2.
2018 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
2019 -march=native recognize znver2 processors.
2020 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
2021 * config/i386/i386.c (m_znver2): New definition.
2022 (m_ZNVER): New definition.
2023 (m_AMD_MULTIPLE): Includes m_znver2.
2024 (processor_cost_table): Add znver2 entry.
2025 (processor_target_table): Add znver2 entry.
2026 (get_builtin_code_for_version): Set priority for
2028 (processor_model): Add M_AMDFAM17H_ZNVER2.
2029 (arch_names_table): Ditto.
2030 (ix86_reassociation_width): Include znver2.
2031 * config/i386/i386.h (TARGET_znver2): New definition.
2032 (struct ix86_size_cost): Add TARGET_ZNVER2.
2033 (enum processor_type): Add PROCESSOR_ZNVER2.
2034 * config/i386/i386.md (define_attr "cpu"): Add znver2.
2035 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
2036 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
2037 (ix86_adjust_cost): Add znver2.
2038 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
2039 * gcc/doc/extend.texi: Add details about znver2.
2040 * gcc/doc/invoke.texi: Add details about znver2.
2042 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
2046 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
2049 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2051 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
2052 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
2054 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
2056 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
2057 shrink-wrapping when -mtrack-speculation.
2059 2018-11-02 Richard Biener <rguenther@suse.de>
2061 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
2062 (pop_cost_one_pair): Do not free pair.
2063 (pop_best_coalesce): Likewise.
2064 (create_coalesce_list): Initialize obstack.
2065 (delete_coalesce_list): Free obstack.
2066 (find_coalesce_pair): Obstack-allocate coalesce pairs.
2067 (add_cost_one_coalesce): Likewise.
2068 (struct live_track): Remove bitmap pointer indirections.
2069 (new_live_track): Adjust.
2070 (delete_live_track): Likewise.
2071 (live_track_remove_partition): Likewise.
2072 (live_track_add_partition): Likewise.
2073 (live_track_live_p): Likewise.
2074 (live_track_process_def): Likewise.
2075 (live_track_clear_base_vars): Likewise.
2077 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2079 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
2080 * configure: Regenerate.
2081 * config.in: Regenerate.
2082 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
2083 HAVE_AS_REGISTER_PSEUDO_OP guard.
2084 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
2085 !HAVE_AS_REGISTER_PSEUDO_OP support.
2086 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
2089 2018-11-02 Richard Biener <rguenther@suse.de>
2091 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
2094 2018-11-02 Richard Biener <rguenther@suse.de>
2096 PR tree-optimization/87776
2097 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
2098 executable when iterating but running into rpo-vn-max-loop-depth
2099 and not eliding the iteration.
2101 2018-11-30 Jan Hubicka <jh@suse.cz>
2103 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
2106 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
2108 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
2110 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
2111 Combine with rs6000_address_for_fpconvert.
2112 (rs6000_address_for_fpconvert): Combine with
2113 rs6000_force_indexed_or_indirect_mem.
2114 (rs6000_expand_vector_init): Change function call from
2115 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
2116 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
2117 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
2118 (floatsi<mode>2_lfiwax_mem): Ditto.
2119 (floatunssi<mode>2_lfiwzx): Ditto.
2120 (floatunssi<mode>2_lfiwzx_mem): Ditto.
2121 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
2122 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
2123 (fix_trunc<mode>si2_stfiwx): Ditto.
2124 (fixuns_trunc<mode>si2_stfiwx): Ditto.
2125 (float_<mode>si2_hw): Ditto.
2126 (floatuns_<mode>si2_hw): Ditto.
2127 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
2128 (vsx_splat_<mode>): Ditto.
2130 2018-11-01 Joseph Myers <joseph@codesourcery.com>
2132 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
2133 at end of assembler input text.
2134 * configure: Regenerate.
2136 2018-11-01 Jakub Jelinek <jakub@redhat.com>
2138 PR tree-optimization/87826
2139 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
2140 negative or larger or equal to type's precision.
2142 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
2144 * opts.c (default_options_table): Do not enable
2145 OPT_fdelayed_branch at -Og.
2146 * doc/invoke.texi (-fdelayed-branch): Document it.
2148 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
2150 * optabs-libfuncs.c (build_libfunc_function_visibility):
2151 New, split out from...
2152 (build_libfunc_function): ... here.
2153 (init_one_libfunc_visibility): New, split out from ...
2154 (init_one_libfunc): ... here.
2156 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
2157 scratch register need not be early-clobber. Document the reason
2158 why we cannot use ST<OP>.
2160 2018-10-31 Joseph Myers <joseph@codesourcery.com>
2163 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
2164 line for second argument of AC_DEFINE_UNQUOTED.
2165 * doc/install.texi (Tools/packages necessary for modifying GCC):
2166 Update to autoconf 2.69 and automake 1.15.1.
2167 * aclocal.m4, config.in, configure: Regenerate.
2169 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
2171 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
2174 2018-10-31 Martin Liska <mliska@suse.cz>
2177 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
2178 Add new argument invalid_extension.
2179 (aarch64_get_all_extension_candidates): New function.
2180 (aarch64_rewrite_selected_cpu): Add NULL to function call.
2181 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
2183 (aarch64_get_all_extension_candidates): New function.
2184 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
2185 argument invalid_extension.
2186 (aarch64_parse_cpu): Likewise.
2187 (aarch64_print_hint_for_extensions): New function.
2188 (aarch64_validate_mcpu): Provide hint about invalid extension.
2189 (aarch64_validate_march): Likewise.
2190 (aarch64_handle_attr_arch): Pass new argument.
2191 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
2192 (aarch64_handle_attr_isa_flags): Likewise.
2194 2018-10-31 Richard Biener <rguenther@suse.de>
2198 * tree-outof-ssa.c (insert_backedge_copies): Restrict
2199 copy generation to useful cases. Place the copy before
2200 the definition of the backedge value when possible.
2202 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2204 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
2205 * config/arc/arc.c (arc_active_insn): New function.
2206 (check_store_cacheline_hazard): Likewise.
2207 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
2208 (arc_override_options): Disable delay slot scheduler for older
2210 (arc_store_addr_hazard_p): New implementation, old one renamed to
2212 (arc_store_addr_hazard_internal_p): Renamed.
2213 (arc_reorg): Don't combine into brcc instructions which are part
2214 of hardware hazard solution.
2215 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
2216 (tune_arc700): Likewise.
2217 * config/arc/arc.opt (arc7xx): New tune value.
2218 * config/arc/arc700.md: Improve A7 scheduler.
2220 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2222 * config/arc/arc.c (arc_override_options): Remove
2223 TARGET_COMPACT_CASESI.
2224 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
2225 (CASE_VECTOR_MODE): Likewise.
2226 (CASE_VECTOR_PC_RELATIVE): Likewise.
2227 (CASE_VECTOR_SHORTEN_MODE): Likewise.
2228 (CASE_VECTOR_SHORTEN_MODE1): Delete.
2229 (ADDR_VEC_ALIGN): Update.
2230 (ASM_OUTPUT_CASE_LABEL): Undefine.
2231 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
2232 (TARGET_BI_BIH): Define.
2233 (DEFAULT_BRANCH_INDEX): Likewise.
2234 * config/arc/arc.md (casesi): Rework to accept BI/BIH
2235 instructions, remove compact_casesi use case.
2236 (casesi_compact_jump): Remove.
2237 (casesi_dispatch): New pattern.
2238 * config/arc/arc.opt: Add mbranch-index option. Deprecate
2239 compact_casesi option.
2240 * doc/invoke.texi: Document mbranch-index option.
2242 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2244 * config/arc/arc.c (arc_get_tp): Remove function.
2245 (arc_emit_call_tls_get_addr): Likewise.
2246 (arc_call_tls_get_addr): New function.
2247 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
2248 * config/arc/arc.md (tls_load_tp_soft): Remove.
2249 (tls_gd_get_addr): Likewise.
2251 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2253 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
2254 (mulsi3_600_lib): Remove pattern.
2255 (umulsi3_highpart_600_lib_le): Likewise.
2256 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
2257 (umulsidi3): Remove call to umulsidi3_600_lib.
2258 (umulsidi3_600_lib): Remove pattern.
2259 (peephole2): Remove peephole using the above deprecated patterns.
2261 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2264 * config/arm/arm.c (arm_option_check_internal): Disable the combined
2265 use of -mslow-flash-data and -mword-relocations.
2266 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
2267 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
2269 * doc/invoke.texi (-mword-relocations): Mention conflict with
2271 (-mslow-flash-data): Reciprocally.
2273 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
2275 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
2276 16-byte modes held in GP registers to use an even regno.
2278 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
2279 (aarch64_atomic_ldop_supported_p): Remove.
2280 (aarch64_gen_atomic_ldop): Remove.
2281 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
2282 Fully expand LSE operations here.
2283 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
2284 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
2285 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
2286 and use ATOMIC_LDOP instead; use register_operand for the input;
2287 drop the split and emit insns directly.
2288 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
2289 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
2290 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
2292 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
2293 (aarch64_gen_atomic_ldop): Don't call it.
2294 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
2295 Use aarch64_reg_or_zero.
2296 (aarch64_atomic_exchange<ALLI>): Likewise.
2297 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
2298 operand 0; use aarch64_reg_or_zero for input; merge ...
2299 (@aarch64_atomic_swp<ALLI>): ... this and remove.
2301 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
2302 (aarch64_split_compare_and_swap): Use it.
2303 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
2304 test oldval against the proper predicate.
2305 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
2306 Use nonmemory_operand for expected.
2307 (cas_short_expected_pred): New.
2308 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
2309 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
2310 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
2311 (aarch64_plushi_operand): New.
2313 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
2314 Force oldval into the rval register for TARGET_LSE; emit the compare
2315 during initial expansion so that it may be deleted if unused.
2316 (aarch64_gen_atomic_cas): Remove.
2317 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
2318 Change =&r to +r for operand 0; use match_dup for operand 2;
2319 remove is_weak and mod_f operands as unused. Drop the split
2321 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
2322 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
2323 (@aarch64_atomic_cas<GPI>): Similarly.
2325 2018-10-31 Richard Biener <rguenther@suse.de>
2327 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
2330 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
2332 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
2333 * config/aarch64/aarch64.md: Include saphira.md
2334 * config/aarch64/saphira.md: New file for pipeline description.
2336 2018-10-30 Martin Sebor <msebor@redhat.com>
2339 * gimple-ssa-sprintf.c (format_directive): Use %G to include
2341 (sprintf_dom_walker::compute_format_length):
2342 Avoid setting POSUNDER4K here.
2343 (get_destination_size): Handle null argument values.
2344 (get_user_idx_format): New function.
2345 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
2346 functions, including user-defined with attribute format printf.
2347 Use %G to include inlining context.
2348 Set POSUNDER4K here.
2350 2018-10-30 Jan Hubicka <jh@suse.cz>
2352 * params.def (lto-partitions): Bump from 32 to 128.
2354 2018-10-30 Jan Hubicka <jh@suse.cz>
2357 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
2359 (free_lang_data_in_type): Forward declare.
2360 (fld_type_variant_equal_p): New function.
2361 (fld_type_variant): New function
2362 (fld_incomplete_types): New hash.
2363 (fld_incomplete_type_of): New function
2364 (fld_simplfied-type): New function.
2365 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
2366 (free_lang_data): Allocate and free fld_incomplete_type; update call
2367 of free_lang_data_in_decl.
2369 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
2371 * gcov.c (output_lines): Remove duplicate line.
2373 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
2375 * config/rs6000/rs6000.md (bswapdi2): Force address into register
2376 if not in indexed or indirect form.
2377 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
2378 (bswapdi2_store): Ditto.
2379 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
2381 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
2382 Prototype for helper function.
2384 2018-10-30 Martin Sebor <msebor@redhat.com>
2386 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
2387 (target, pragma GCC optimize, pragma GCC target): Ditto.
2389 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
2391 * doc/extend.texi: Fix prototype and description of
2392 __builtin_expect_with_probability.
2394 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2396 * cgraph.h (clone_function_name_1): Replaced by new
2397 clone_function_name_numbered that takes name as string; for
2398 privatize_symbol_name_1 use only.
2399 (clone_function_name): Renamed to
2400 clone_function_name_numbered to be explicit about numbering.
2401 (clone_function_name): New two-argument function that does
2402 not number its output.
2403 (clone_function_name): New three-argument function that
2404 takes a number to append to its output.
2405 * cgraphclones.c (duplicate_thunk_for_node):
2406 (clone_function_name_1): Renamed.
2407 (clone_function_name_numbered): Two new functions.
2408 (clone_function_name): Improved documentation.
2409 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
2410 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
2411 * final.c (final_scan_insn_1): Use the new clone_function_name
2413 * multiple_target.c (create_dispatcher_calls): Ditto.
2414 (create_target_clone): Ditto.
2415 * omp-expand.c (grid_expand_target_grid_body): Ditto.
2416 * omp-low.c (create_omp_child_function_name): Ditto.
2417 * omp-simd-clone.c (simd_clone_create): Ditto.
2418 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
2419 new clone_function_name without numbering.
2421 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
2423 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
2424 Assert that the allocation size is not zero.
2426 2018-10-30 Richard Biener <rguenther@suse.de>
2428 PR tree-optimization/87800
2429 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
2430 non-induction or reduction PHIs.
2432 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
2434 * config/aarch64/falkor-tag-collision-avoidance.c
2435 (execute_tag_collision_avoidance): Call df_note_add_problem.
2437 2018-10-30 Martin Liska <mliska@suse.cz>
2439 * doc/extend.texi: Fix typo in documentation
2440 of __builtin_expect_with_probability.
2442 2018-10-29 David Malcolm <dmalcolm@redhat.com>
2445 * input.c (get_substring_ranges_for_loc): Detect if
2446 linemap_resolve_location gives us a NULL map, and reject
2449 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
2451 * config.gcc (xstormy16-*-elf): Set tm_d_file.
2453 2018-10-29 David Malcolm <dmalcolm@redhat.com>
2454 Martin Sebor <msebor@redhat.com>
2455 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2457 * doc/ux.texi (Quoting): New subsection, adapted from material at
2458 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
2459 MartinSebor and ManuelLopezIbanez.
2460 (Fix-it hints): Note that fix-it hints shouldn't be marked for
2463 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
2466 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
2469 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
2471 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
2473 2018-10-29 Paul Koning <ni1d@arrl.net>
2475 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
2477 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
2479 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
2480 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
2481 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
2482 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
2483 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
2484 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
2485 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
2486 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
2487 Change 'vector' to '__vector'.
2488 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
2489 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
2490 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
2491 _mm_avg_pu16): Likewise. And, whitespace corrections.
2493 2018-10-29 Richard Biener <rguenther@suse.de>
2495 PR tree-optimization/87785
2496 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
2499 2018-10-29 Olivier Hainque <hainque@adacore.com>
2501 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
2503 2018-10-29 Olivier Hainque <hainque@adacore.com>
2505 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
2506 ports configuration macro, defaults to "gnu".
2507 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
2508 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
2510 2018-10-29 Olivier Hainque <hainque@adacore.com>
2512 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
2514 2018-10-29 Richard Biener <rguenther@suse.de>
2516 PR tree-optimization/87790
2517 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
2518 (vect_make_slp_decision): Adjust.
2519 (vect_slp_analyze_bb_1): Likewise.
2520 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
2523 2018-10-29 Richard Biener <rguenther@suse.de>
2525 PR tree-optimization/87785
2526 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
2528 (vect_build_slp_tree): Likewise.
2529 (vect_gather_slp_loads): New function.
2530 (vect_analyze_slp_instance): Gather loads separately from the
2533 2018-10-29 Martin Liska <mliska@suse.cz>
2535 * Makefile.in: Make dependency to json.o.
2536 * doc/gcov.texi: Document new JSON format, remove
2537 old intermediate format documentation.
2538 * gcov.c (struct function_info): Come up with m_name and
2540 (function_info::function_info): Initialize it.
2541 (function_info::~function_info): Release it.
2542 (main): Rename flag_intermediate_format to flag_json_format.
2543 (print_usage): Describe --json-format.
2544 (process_args): Set flag_json_format.
2545 (output_intermediate_line): Remove.
2546 (output_intermediate_json_line): Likewise.
2547 (get_gcov_intermediate_filename): Return new extension
2549 (output_intermediate_file): Implement JSON emission.
2550 (output_json_intermediate_file): Implement JSON emission.
2551 (generate_results): Use ::get_name for function name.
2552 Handle JSON output file.
2553 (read_graph_file): Use ::get_name instead of cplus_demangle.
2554 (read_count_file): Likewise.
2555 (solve_flow_graph): Likewise.
2556 (add_line_counts): Likewise.
2557 (accumulate_line_counts): Use new flag_json_format.
2558 (output_function_details): Use ::get_name instead of cplus_demangle.
2559 (output_lines): Likewise.
2560 * json.cc (test_writing_literals): Add new tests.
2561 * json.h (class literal): Add new boolean constructor.
2563 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
2565 PR rtl-optimization/87701
2566 PR rtl-optimization/87780
2567 * combine.c (make_more_copies): Rewrite.
2569 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
2571 * doc/generic.texi (ABSU_EXPR): Document.
2572 * match.pd (absu(x)*absu(x) -> x*x): Handle.
2573 (absu(absu(X)) -> absu(X)): Likewise.
2574 (absu(-X) -> absu(X)): Likewise.
2575 (absu(X) where X is nonnegative -> X): Likewise.
2577 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
2579 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
2580 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
2581 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
2582 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
2583 (s-tm-texi): Also check timestamp on d-target.def.
2584 (generated_files): Add TM_D_H and d-target-hooks-def.h.
2585 (build/genhooks.o): Also depend on D_TARGET_DEF.
2586 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
2588 * config/aarch64/aarch64-d.c: New file.
2589 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
2591 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
2593 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
2594 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
2595 * config/arm/arm-d.c: New file.
2596 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
2597 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
2598 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2599 * config/arm/t-arm (arm-d.o): New rule.
2600 * config/default-d.c: New file.
2601 * config/glibc-d.c: New file.
2602 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2603 * config/i386/i386-d.c: New file.
2604 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
2605 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
2606 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2607 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
2608 * config/i386/t-i386 (i386-d.o): New rule.
2609 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2610 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2611 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
2612 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2613 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2614 * config/mips/mips-d.c: New file.
2615 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
2616 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
2617 * config/mips/t-mips (mips-d.o): New rule.
2618 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2619 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2620 * config/powerpcspe/powerpcspe-d.c: New file.
2621 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
2623 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
2624 Support GNU D by using 0 as the language type.
2625 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
2626 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
2627 * config/riscv/riscv-d.c: New file.
2628 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
2630 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
2631 * config/riscv/t-riscv (riscv-d.o): New rule.
2632 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2633 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2634 * config/rs6000/rs6000-d.c: New file.
2635 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
2637 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
2638 Support GNU D by using 0 as the language type.
2639 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
2640 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
2641 * config/s390/s390-d.c: New file.
2642 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
2643 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
2644 * config/s390/t-s390 (s390-d.o): New rule.
2645 * config/sparc/sparc-d.c: New file.
2646 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
2648 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
2649 * config/sparc/t-sparc (sparc-d.o): New rule.
2650 * config/t-glibc (glibc-d.o): New rule.
2651 * configure: Regenerated.
2652 * configure.ac (tm_d_file): New variable.
2653 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
2654 * doc/contrib.texi (Contributors): Add self for the D frontend.
2655 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
2656 * doc/install.texi (Configuration): Mention libphobos as an option for
2657 --enable-shared. Mention d as an option for --enable-languages.
2658 (Testing): Mention check-d as a target.
2659 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
2660 name suffixes. Mention d as a -x option.
2661 * doc/sourcebuild.texi (Top Level): Mention libphobos.
2662 * doc/standards.texi (Standards): Add section on D language.
2663 * doc/tm.texi: Regenerated.
2664 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
2665 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
2666 * dwarf2out.c (is_dlang): New function.
2667 (gen_compile_unit_die): Use DW_LANG_D for D.
2668 (declare_in_namespace): Return module die for D, instead of adding
2669 extra declarations into the namespace.
2670 (gen_namespace_die): Generate DW_TAG_module for D.
2671 (gen_decl_die): Handle CONST_DECLSs for D.
2672 (dwarf2out_decl): Likewise.
2673 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
2674 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
2675 kinds of aggregates.
2676 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
2677 * genhooks.c: Include d/d-target.def.
2679 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
2682 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
2683 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
2684 (STACK_DYNAMIC_OFFSET): Likewise.
2686 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
2689 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
2692 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
2694 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
2695 return type and other typos.
2697 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
2699 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
2700 a shorter sequence with fewer branches.
2701 (emit_final_str_compare_gpr): Ditto.
2703 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
2705 * config/rs6000/tmmintrin.h: New file.
2706 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
2708 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
2710 * config/rs6000/mmintrin.h: Enable 32bit compilation.
2711 * config/rs6000/xmmintrin.h: Likewise.
2713 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
2715 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
2717 2018-10-26 Richard Biener <rguenther@suse.de>
2719 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
2721 (vect_mark_slp_stmts_relevant): Likewise.
2722 (vect_detect_hybrid_slp_stmts): Likewise.
2723 (vect_bb_slp_scalar_cost): Likewise.
2724 (vect_remove_slp_scalar_calls): Likewise.
2726 2018-10-26 Jan Hubicka <jh@suse.cz>
2728 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
2729 (warn_types_mismatch): Fix walk of DECL_NAME.
2730 (odr_types_equivalent_p): Fix overactive assert.
2732 2018-10-26 Richard Biener <rguenther@suse.de>
2734 PR tree-optimization/87105
2735 * tree-vectorizer.h (_slp_tree::refcnt): New member.
2736 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
2738 (vect_create_new_slp_node): Initialize refcnt to one.
2740 (scalar_stmts_set_t, bst_fail): Remove.
2741 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
2742 (vect_build_slp_tree): Add bst_map argument and lookup
2743 already created SLP nodes.
2744 (vect_print_slp_tree): Handle a SLP graph, print SLP node
2746 (vect_slp_rearrange_stmts): Handle a SLP graph.
2747 (vect_analyze_slp_instance): Adjust and free SLP nodes from
2748 the CSE map. Fix indenting.
2749 (vect_schedule_slp_instance): Add short-cut.
2751 2018-10-26 Martin Liska <mliska@suse.cz>
2754 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
2755 addr_expr and not with pointers.
2757 2018-10-26 Jan Hubicka <jh@suse.cz>
2759 * tree.c (free_lang_data_in_type): Only check main variants.
2760 * ipa-devirt.c (warn_odr): Make static.
2761 (types_same_for_odr): Drop strict variant.
2762 (types_odr_comparable): Likewise.
2763 (odr_or_derived_type_p): Look for main variants.
2764 (odr_name_hasher::equal): Cleanup comment.
2765 (odr_subtypes_equivalent): Add warn and warned arguments; check main
2767 (type_variants_equivalent_p): break out from ...
2768 (odr_types_equivalent): ... here; go for main variants where needed.
2769 (warn_odr): ... here; turn static.
2770 (warn_types_mismatch): Compare mangled names of main variants.
2771 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
2772 (type_with_linkage_p): Sanity check that we look at main variant.
2773 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
2774 * tree.h (types_same_for_odr): Drop strict argument.
2776 2018-10-26 Richard Biener <rguenther@suse.de>
2778 PR tree-optimization/87746
2779 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
2780 Simplify and fix WRT strided store groups with size not
2781 equal to step in element count.
2782 (vect_analyze_group_access_1): Dump the whole group.
2784 2018-10-25 Carl Love <cel@us.ibm.com>
2786 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
2787 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
2788 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
2789 precicion entry for each overloaded builtin.
2790 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
2791 VSCEDPUO): Rename overloaded name.
2792 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
2793 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
2794 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
2795 define_expand for xscmpexqp instruction.
2796 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
2798 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
2799 Jinsong Ji <jji@us.ibm.com>
2801 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
2802 function with vec_sl.
2803 (_mm_slli_epi32): Likewise.
2804 (_mm_slli_epi64): Likewise.
2805 (_mm_srai_epi16): Replace deprecated function with vec_sra.
2806 (_mm_srai_epi32): Likewise.
2807 (_mm_srli_epi16): Replace deprecated function with vec_sr.
2808 (_mm_srli_epi32): Likewise.
2809 (_mm_srli_epi64): Likewise.
2810 (_mm_sll_epi16): Replace deprecated function with vec_sl.
2811 (_mm_sll_epi32): Likewise.
2812 (_mm_sll_epi64): Likewise.
2813 (_mm_sra_epi16): Replace deprecated function with vec_sra.
2814 (_mm_sra_epi32): Likewise.
2815 (_mm_srl_epi16): Replace deprecated function with vec_sr.
2816 (_mm_srl_epi32): Likewise.
2817 (_mm_srl_epi64): Likewise.
2819 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
2820 Jinsong Ji <jji@us.ibm.com>
2822 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
2823 comparison operators with vec_cmp* for compatibility due to
2824 unfortunate history; clean up formatting and use types more
2826 (_mm_sll_epi32): Likewise.
2827 (_mm_sll_epi64): Likewise.
2828 (_mm_srl_epi16): Likewise.
2829 (_mm_srl_epi32): Likewise.
2830 (_mm_srl_epi64): Likewise.
2832 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
2833 Jinsong Ji <jji@us.ibm.com>
2835 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
2836 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
2837 __vector __bool int. Use vec_cmpgt in preference to deprecated
2838 function vec_vcmpgtfp.
2839 (_mm_max_ps): Likewise.
2841 2018-10-25 Jeff Law <law@redhat.com>
2843 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
2846 2018-10-25 Martin Sebor <msebor@redhat.com>
2848 * doc/extend.texi (aligned): Expand attribute description.
2849 (Alignment): Rename section. Discuss function arguments.
2851 2018-10-25 Jan Hubicka <jh@suse.cz>
2853 * ipa-devirt.c (main_odr_variant): Remove.
2854 (hash_odr_name, types_same_for_odr, types_odr_comparable,
2855 odr_name_hasher::equal, odr_subtypes_equivalent_p):
2856 Drop use of main_odr_variant.
2857 (add_type_duplicate): Silence confused warnings on integer types.
2858 (get_odr_type): Always look for main variant.
2859 (register_odr_type): Simplify.
2861 2018-10-25 Richard Biener <rguenther@suse.de>
2863 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
2864 Initialize ng to silence error with release checking bootstrap.
2866 2018-10-25 Richard Biener <rguenther@suse.de>
2868 * tree-if-conv.c: Include tree-ssa-sccvn.h.
2869 (tree_if_conversion): Run CSE on the if-converted loop body.
2871 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
2873 * config/s390/constraints.md (ZL): New constraint.
2874 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
2876 * config/s390/s390.md (movdi_larl): Remove.
2877 (movdi_64): Add the LARL alternative.
2879 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
2882 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
2883 (RTX_CODE_HWINT_P): New macro.
2884 (rtx_code_size): Use RTX_CODE_HWINT_P ().
2886 2018-10-25 Jan Hubicka <jh@suse.cz>
2888 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
2891 2018-10-25 Richard Biener <rguenther@suse.de>
2893 PR tree-optimization/87665
2894 PR tree-optimization/87745
2895 * tree-vectorizer.h (get_earlier_stmt): Remove.
2896 (get_later_stmt): Pick up UID from the original non-pattern stmt.
2898 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
2900 * options.texi (Deprecated): Move list to Var section.
2902 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
2903 Jinsong Ji <jji@us.ibm.com>
2905 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
2906 __vector long to __vector long long.
2907 (_mm_cvtpd_ps): Likewise.
2908 (_mm_cvttpd_epi32): Likewise.
2909 (_mm_cvtpi32_pd): Likewise.
2910 (_mm_unpackhi_epi64): Likewise.
2911 (_mm_unpacklo_epi64): Likewise.
2913 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
2915 PR rtl-optimization/87720
2916 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
2918 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
2920 * gimple-ssa-isolate-paths.c
2921 (find_implicit_erroneous_behavior): Do not change code if the
2922 pass is running for warnings only.
2923 (find_explicit_erroneous_behavior): Likewise.
2925 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
2927 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
2928 Define as rs6000_mangle_decl_assembler_name.
2929 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
2930 long double to IEEE long double, switch the names of the long
2931 double built-in functions to be <func>f128 instead of <func>l.
2933 2018-10-24 Martin Sebor <msebor@redhat.com>
2935 * doc/extend.texi (nonnull): List no-argument form. Reference
2936 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
2938 2018-10-24 Richard Biener <rguenther@suse.de>
2940 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
2942 2018-10-24 Martin Liska <mliska@suse.cz>
2944 PR tree-optimization/84436
2945 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
2947 (switch_conversion::contains_linear_function_p): New.
2948 (switch_conversion::build_one_array): Support linear
2949 transformation on input.
2950 * tree-switch-conversion.h (struct switch_conversion): Add
2951 contains_linear_function_p declaration.
2953 2018-10-24 Richard Biener <rguenther@suse.de>
2955 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
2956 if its argument is CONSTANT_CLASS_P.
2958 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
2960 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
2961 it is wrong for forward declarations.
2963 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
2965 * config/s390/s390.c (s390_check_qrst_address): Add the missing
2966 SYMBOL_REF_P () check.
2968 2018-10-24 Richard Biener <rguenther@suse.de>
2970 PR tree-optimization/87105
2971 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
2972 dump classification.
2973 (vect_analyze_data_ref_accesses): Handle duplicate loads and
2974 stores by splitting the affected group after the fact.
2975 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
2976 fail the SLP build because of size constraints.
2978 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2980 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
2981 * configure: Regenerate.
2982 * config.in: Regenerate.
2983 * varasm.c (mergeable_string_section): Use readonly_data_section
2984 if linker doesn't support SHF_MERGE with alignment > 8.
2985 (mergeable_constant_section): Likewise.
2987 2018-10-24 Richard Biener <rguenther@suse.de>
2989 PR tree-optimization/84013
2990 * tree-ssa-structalias.c (struct msdi_data): New struct for
2991 marshalling data to walk_stmt_load_store_ops.
2992 (maybe_set_dependence_info): Refactor as callback for
2993 walk_stmt_load_store_ops.
2994 (compute_dependence_clique): Set restrict info on all stmt kinds.
2996 2018-10-24 Martin Liska <mliska@suse.cz>
2998 * cgraph.c (cgraph_node::dump):
2999 Remove reduntant dumps and make tp_first_run dump more compact.
3001 2018-10-24 Richard Biener <rguenther@suse.de>
3003 PR tree-optimization/87665
3004 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
3007 2018-10-12 Jeff Law <law@redhat.com>
3009 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
3012 2018-10-23 Richard Biener <rguenther@suse.de>
3014 * tree-vrp.c (add_assert_info): Guard dump_printf with
3016 * gimple-ssa-evrp-analyze.c
3017 (evrp_range_analyzer::record_ranges_from_incoming_edge):
3018 Use value_range::ignore_equivs_equal_p.
3020 2018-10-23 Richard Biener <rguenther@suse.de>
3022 PR tree-optimization/87105
3023 PR tree-optimization/87608
3024 * passes.def (pass_all_early_optimizations): Add early phi-opt
3026 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
3027 addition to debug stmts.
3028 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
3029 and abs replacement early.
3030 * tree-cfg.c (gimple_empty_block_p): Likewise.
3032 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
3035 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
3036 specified to configure.
3037 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
3039 2018-10-23 Richard Biener <rguenther@suse.de>
3041 PR tree-optimization/87700
3042 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
3044 2018-10-23 Jakub Jelinek <jakub@redhat.com>
3047 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
3048 second argument from __mmask16 to __mmask8.
3049 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
3050 _mm_mask_packs_epi32): Likewise.
3051 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
3053 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
3055 2018-10-23 Richard Biener <rguenther@suse.de>
3057 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
3059 2018-10-23 Richard Biener <rguenther@suse.de>
3061 PR tree-optimization/86144
3062 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
3063 over simd attribute.
3065 2018-10-23 Richard Biener <rguenther@suse.de>
3067 PR tree-optimization/87693
3068 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
3069 the case we do not find the taken edge.
3071 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
3072 Jinsong Ji <jji@us.ibm.com>
3074 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
3075 (_mm_store_pd): Use unaligned vector type for pointer cast.
3076 (_mm_maskmoveu_si128): Likewise.
3077 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
3078 (_mm_store_ps): Use unaligned vector type for pointer cast.
3080 2018-10-22 Paul Koning <ni1d@arrl.net>
3082 * symtab.c (symtab_node::increase_alignment): Correct max
3085 2018-10-22 Yury Gribov <tetra2005@gmail.com>
3087 PR tree-optimization/87633
3088 * match.pd: Do not generate unordered integer comparisons.
3090 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
3092 PR rtl-optimization/87600
3093 * combine.c: Add include of expr.h.
3094 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
3095 register to a pseudo.
3096 (make_more_copies): New function, add a copy to a new pseudo after
3097 the moves from hard registers into pseudos.
3098 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
3099 later. Call make_more_copies.
3101 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
3103 * lra-constraints.c (process_alt_operands): New local array,
3104 matching_early_clobber. Check matching_early_clobber before
3105 decrementing reject, and set matching_early_clobber after.
3107 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
3110 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
3111 call output_addr_const and hope for the best.
3113 2018-10-22 Richard Biener <rguenther@suse.de>
3115 * gimple-ssa-evrp-analyze.c
3116 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
3117 smarter about what ranges to use.
3118 * tree-vrp.c (add_assert_info): Dump here.
3119 (register_edge_assert_for_2): Instead of here at multiple but
3122 * gcc.dg/tree-ssa/evrp12.c: New testcase.
3123 * gcc.dg/predict-6.c: Adjust.
3124 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
3125 * gcc.dg/tree-ssa/vrp02.c: Likewise.
3126 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
3128 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
3129 Richard Biener <rguenther@suse.de>
3131 * bitmap.h: Update data structure documentation, including a
3132 description of bitmap views as either linked-lists or splay trees.
3133 (struct bitmap_element_def): Update comments for splay tree bitmaps.
3134 (struct bitmap_head_def): Likewise.
3135 (bitmap_list_view, bitmap_tree_view): New prototypes.
3136 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
3138 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
3139 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
3140 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
3141 released bitmap element here.
3142 (bitmap_element_free): Remove.
3143 (bitmap_elt_clear_from): Work on splay tree bitmaps.
3144 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
3145 this function similar ones such that linked-list bitmap implementation
3146 functions are grouped.
3147 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
3148 and moved for grouping.
3149 (bitmap_list_insert_element_after): Renamed from
3150 bitmap_elt_insert_after, and moved for grouping.
3151 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
3152 (bitmap_tree_link_left, bitmap_tree_link_right,
3153 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
3154 bitmap_tree_link_element, bitmap_tree_unlink_element,
3155 bitmap_tree_find_element): New functions for splay-tree bitmap
3157 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
3158 Renamed and moved, see above entries.
3159 (bitmap_tree_listify_from): New function to convert part of a splay
3160 tree bitmap to a linked-list bitmap.
3161 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
3162 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
3163 (bitmap_find_bit): Remove.
3164 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
3165 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
3166 Handle splay tree bitmaps.
3167 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
3168 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
3169 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
3170 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
3171 bitmap_intersect_compl_p, bitmap_ior_and_compl,
3172 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
3173 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
3174 corresponding changes to use linked-list specific bitmap_element
3175 manipulation functions as applicable for efficiency.
3176 (bitmap_tree_to_vec): New function.
3177 (debug_bitmap_elt_file): New function split out from ...
3178 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
3179 (bitmap_print): Likewise.
3181 PR tree-optimization/63155
3182 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
3184 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
3187 2018-10-22 Martin Liska <mliska@suse.cz>
3189 PR tree-optimization/87686
3191 2018-08-29 Martin Liska <mliska@suse.cz>
3193 * tree-switch-conversion.c (switch_conversion::expand):
3194 Strenghten assumption about gswitch statements.
3196 2018-10-22 Martin Liska <mliska@suse.cz>
3198 * ipa-icf.c (sem_item::compare_attributes): Remove.
3199 (sem_item::compare_referenced_symbol_properties): Use
3200 attribute_list_equal instead.
3201 (sem_function::equals_wpa): Likewise.
3202 * ipa-icf.h: Remove compare_attributes.
3204 2018-10-22 Richard Biener <rguenther@suse.de>
3207 * mem-stats.h (mem_usage::operator==): Fix pasto.
3209 2018-10-22 Richard Biener <rguenther@suse.de>
3211 PR tree-optimization/87640
3212 * tree-vrp.c (set_value_range_with_overflow): Decompose
3214 (extract_range_from_binary_expr_1): Adjust.
3216 2018-10-22 Martin Jambor <mjambor@suse.cz>
3218 * tree-eh.h (stmt_could_throw_p): Add function parameter.
3219 (stmt_can_throw_external): Likewise.
3220 (stmt_can_throw_internal): Likewise.
3221 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
3222 (lower_eh_constructs_2): Likewise.
3223 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
3224 (stmt_can_throw_external): Likewise.
3225 (stmt_can_throw_internal): Likewise.
3226 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
3227 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
3228 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
3229 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
3230 (pass_lower_eh_dispatch::execute): Pass cfun to
3231 stmt_can_throw_external.
3232 (cleanup_empty_eh): Likewise.
3233 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
3234 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
3235 stmt_can_throw_external instead of pushing it to cfun.
3236 (symbol_table::create_edge): Likewise.
3237 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
3238 stmt_can_throw_internal.
3239 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
3240 to stmt_could_throw_p.
3241 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
3242 stmt_can_throw_internal.
3243 (pass_store_merging::execute): Likewise.
3244 * gimple-ssa-strength-reduction.c
3245 (find_candidates_dom_walker::before_dom_children): Pass cfun to
3247 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
3248 stmt_can_throw_internal.
3249 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
3250 to stmt_can_throw_external.
3251 (check_stmt): Pass cfun to stmt_could_throw_p.
3252 (check_stmt): Pass cfun to stmt_can_throw_external.
3253 (pass_nothrow::execute): Likewise.
3254 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
3255 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
3256 stmt_can_throw_internal.
3257 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
3258 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
3259 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
3260 * tree-complex.c (expand_complex_libcall): Pass cfun to
3261 stmt_could_throw_p and to stmt_can_throw_internal.
3262 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
3263 * tree-inline.c (copy_edges_for_bb): Likewise.
3264 (maybe_move_debug_stmts_to_successors): Likewise.
3265 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
3267 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
3268 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
3269 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
3270 stmt_can_throw_internal.
3271 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
3272 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
3274 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
3275 stmt_can_throw_internal.
3276 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
3277 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
3279 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
3280 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
3281 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3282 (convert_mult_to_fma_1): Likewise.
3283 (convert_to_divmod): Likewise.
3284 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
3285 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
3286 * tree-ssa-propagate.c
3287 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
3288 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
3289 (maybe_optimize_range_tests): Likewise.
3290 (linearize_expr_tree): Likewise.
3291 (reassociate_bb): Likewise.
3292 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
3293 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
3294 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
3295 (handle_char_store): Likewise.
3296 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
3297 stmt_can_throw_internal.
3298 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
3300 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
3301 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
3302 (vectorizable_simd_clone_call): Likewise.
3303 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
3304 (gimple_stringop_fixed_value): Likewise.
3306 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
3308 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
3309 literal pool references.
3310 (s390_check_qrst_address): Adapt to the new behavior of
3311 s390_loadrelative_operand_p ().
3313 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3316 * config/i386/sse.md (*andnot<mode>3_bcst): New.
3318 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3321 * config/i386/sse.md (*<code><mode>3_bcst): New.
3323 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3326 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
3327 V4DI, V16SI and V8DI.
3328 (*sub<mode>3<mask_name>_bcst): New.
3329 (*add<mode>3<mask_name>_bcst): Likewise.
3331 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
3332 Jinsong Ji <jji@us.ibm.com>
3334 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
3335 __m64 with __vector unsigned long long for compatibility.
3336 (_mm_movemask_epi8): Likewise.
3337 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
3338 (_mm_cvttps_pi32): Likewise.
3339 (_mm_cvtpi32_ps): Likewise.
3340 (_mm_cvtps_pi16): Likewise.
3341 (_mm_loadh_pi): Likewise.
3342 (_mm_storeh_pi): Likewise.
3343 (_mm_movehl_ps): Likewise.
3344 (_mm_movelh_ps): Likewise.
3345 (_mm_loadl_pi): Likewise.
3346 (_mm_storel_pi): Likewise.
3347 (_mm_movemask_ps): Likewise.
3348 (_mm_shuffle_pi16): Likewise.
3350 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3353 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
3354 __builtin_ia32_vfnmsubpd512_mask.
3355 (_mm512_mask_fnmsub_round_pd): Likewise.
3356 (_mm512_fnmsub_pd): Likewise.
3357 (_mm512_mask_fnmsub_pd): Likewise.
3358 (_mm512_maskz_fnmsub_round_pd): Use
3359 __builtin_ia32_vfnmsubpd512_maskz.
3360 (_mm512_maskz_fnmsub_pd): Likewise.
3361 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
3362 (_mm512_mask_fnmsub_round_ps): Likewise.
3363 (_mm512_fnmsub_ps): Likewise.
3364 (_mm512_mask_fnmsub_ps): Likewise.
3365 (_mm512_maskz_fnmsub_round_ps): Use
3366 __builtin_ia32_vfnmsubps512_maskz.
3367 (_mm512_maskz_fnmsub_ps): Likewise.
3368 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
3369 __builtin_ia32_vfnmsubpd256_mask.
3370 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
3371 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
3372 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
3373 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
3374 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
3375 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
3376 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
3377 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
3378 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
3379 __builtin_ia32_vfnmsubpd.
3380 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
3381 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
3382 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
3383 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
3384 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
3385 * config/i386/i386-builtin.def: Add
3386 __builtin_ia32_vfnmsubpd256_mask,
3387 __builtin_ia32_vfnmsubpd256_maskz,
3388 __builtin_ia32_vfnmsubpd128_mask,
3389 __builtin_ia32_vfnmsubpd128_maskz,
3390 __builtin_ia32_vfnmsubps256_mask,
3391 __builtin_ia32_vfnmsubps256_maskz,
3392 __builtin_ia32_vfnmsubps128_mask,
3393 __builtin_ia32_vfnmsubps128_maskz,
3394 __builtin_ia32_vfnmsubpd512_mask,
3395 __builtin_ia32_vfnmsubpd512_maskz,
3396 __builtin_ia32_vfnmsubps512_mask,
3397 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
3398 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
3399 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
3400 __builtin_ia32_vfnmsubpd256.
3401 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
3402 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
3403 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
3405 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
3407 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
3409 (fmai_vmfnmsub_<mode><round_name>): Likewise.
3411 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3414 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
3415 __builtin_ia32_vfnmaddpd512_mask.
3416 (_mm512_mask_fnmadd_round_pd): Likewise.
3417 (_mm512_fnmadd_pd): Likewise.
3418 (_mm512_mask_fnmadd_pd): Likewise.
3419 (_mm512_maskz_fnmadd_round_pd): Use
3420 __builtin_ia32_vfnmaddpd512_maskz.
3421 (_mm512_maskz_fnmadd_pd): Likewise.
3422 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
3423 (_mm512_mask_fnmadd_round_ps): Likewise.
3424 (_mm512_fnmadd_ps): Likewise.
3425 (_mm512_mask_fnmadd_ps): Likewise.
3426 (_mm512_maskz_fnmadd_round_ps): Use
3427 __builtin_ia32_vfnmaddps512_maskz.
3428 (_mm512_maskz_fnmadd_ps): Likewise.
3429 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
3430 __builtin_ia32_vfnmaddpd256_mask.
3431 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
3432 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
3433 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
3434 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
3435 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
3436 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
3437 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
3438 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
3439 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
3440 __builtin_ia32_vfnmaddpd.
3441 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
3442 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
3443 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
3444 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
3445 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
3446 * config/i386/i386-builtin.def: Add
3447 __builtin_ia32_vfnmaddpd256_mask,
3448 __builtin_ia32_vfnmaddpd256_maskz,
3449 __builtin_ia32_vfnmaddpd128_mask,
3450 __builtin_ia32_vfnmaddpd128_maskz,
3451 __builtin_ia32_vfnmaddps256_mask,
3452 __builtin_ia32_vfnmaddps256_maskz,
3453 __builtin_ia32_vfnmaddps128_mask,
3454 __builtin_ia32_vfnmaddps128_maskz,
3455 __builtin_ia32_vfnmaddpd512_mask,
3456 __builtin_ia32_vfnmaddpd512_maskz,
3457 __builtin_ia32_vfnmaddps512_mask,
3458 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
3459 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
3460 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
3461 __builtin_ia32_vfnmaddpd256.
3462 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
3463 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
3464 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
3466 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
3468 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
3470 (fmai_vmfnmadd_<mode><round_name>): Likewise.
3472 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3475 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
3476 __builtin_ia32_vfmsubpd512_mask.
3477 (_mm512_mask_fmsub_round_pd): Likewise.
3478 (_mm512_fmsub_pd): Likewise.
3479 (_mm512_mask_fmsub_pd): Likewise.
3480 (_mm512_maskz_fmsub_round_pd): Use
3481 __builtin_ia32_vfmsubpd512_maskz.
3482 (_mm512_maskz_fmsub_pd): Likewise.
3483 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
3484 (_mm512_mask_fmsub_round_ps): Likewise.
3485 (_mm512_fmsub_ps): Likewise.
3486 (_mm512_mask_fmsub_ps): Likewise.
3487 (_mm512_maskz_fmsub_round_ps): Use
3488 __builtin_ia32_vfmsubps512_maskz.
3489 (_mm512_maskz_fmsub_ps): Likewise.
3490 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
3491 __builtin_ia32_vfmsubpd256_mask.
3492 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
3493 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
3494 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
3495 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
3496 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
3497 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
3498 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
3499 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
3500 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
3501 __builtin_ia32_vfmsubpd.
3502 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
3503 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
3504 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
3505 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
3506 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
3507 * config/i386/i386-builtin.def: Add
3508 __builtin_ia32_vfmsubpd256_mask,
3509 __builtin_ia32_vfmsubpd256_maskz,
3510 __builtin_ia32_vfmsubpd128_mask,
3511 __builtin_ia32_vfmsubpd128_maskz,
3512 __builtin_ia32_vfmsubps256_mask,
3513 __builtin_ia32_vfmsubps256_maskz,
3514 __builtin_ia32_vfmsubps128_mask,
3515 __builtin_ia32_vfmsubps128_maskz,
3516 __builtin_ia32_vfmsubpd512_mask,
3517 __builtin_ia32_vfmsubpd512_maskz,
3518 __builtin_ia32_vfmsubps512_mask,
3519 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
3520 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
3521 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
3522 __builtin_ia32_vfmsubpd256.
3523 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
3524 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
3525 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
3527 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
3529 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
3531 (fmai_vmfmsub_<mode><round_name>): Likewise.
3533 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3535 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
3536 Remove plus. Renamed to ...
3537 (*sub<mode>3<mask_name>_bcst): This.
3538 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
3539 (*add<mode>3<mask_name>_bcst): This.
3541 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3544 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
3546 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3549 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
3550 (_mm_or_epi32): Likewise.
3551 (_mm256_xor_epi32): Likewise.
3552 (_mm_xor_epi32): Likewise.
3553 (_mm256_or_epi64): Likewise.
3554 (_mm_or_epi64): Likewise.
3555 (_mm256_xor_epi64): Likewise.
3556 (_mm_xor_epi64): Likewise.
3558 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
3561 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
3563 2018-10-20 Jakub Jelinek <jakub@redhat.com>
3566 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
3568 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
3570 * doc/ux.texi: Move @section directly after @node.
3572 2018-10-19 Jakub Jelinek <jakub@redhat.com>
3576 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
3577 depend closely nested inside of loop with ordered clause with
3580 2018-10-19 David Malcolm <dmalcolm@redhat.com>
3582 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
3583 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
3584 * doc/ux.texi: New file.
3586 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3588 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
3589 be the first CR field allocated.
3591 2018-10-19 Richard Biener <rguenther@suse.de>
3594 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
3595 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
3597 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
3600 * config/i386/sse.md
3601 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
3602 (*add<mode>3<mask_name>_bcst_2): Likewise.
3604 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
3606 * config/i386/sse.md
3607 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
3608 Replace nonimmediate_operand with register_operand.
3609 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
3611 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
3614 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
3616 PR rtl-optimization/87596
3617 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
3618 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
3619 for instructions in FROM..TO range.
3621 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
3623 * cfgexpand.c (expand_one_var): Use specific wording in error message
3624 for non-local frame variables.
3625 * stor-layout.c (layout_decl): Do not issue a warning for them.
3627 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
3629 * haifa-sched.c (priority): Add force_recompute parameter.
3630 (apply_replacement): Call priority () with force_recompute = true.
3631 (restore_pattern): Likewise.
3633 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3635 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
3636 HOST_BITS_PER_WIDE_INT.
3637 (test_vector_ops_duplicate): Likewise.
3639 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3642 * config/i386/sse.md (VF_AVX512): New.
3643 (avx512bcst): Likewise.
3644 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
3646 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
3648 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
3651 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
3653 * doc/invoke.texi (-dumpversion): Improve grammar.
3654 (-dumpfullversion): Make more consistent with -dumpversion.
3656 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
3658 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
3659 Set JUMP_LABEL to the jump insn.
3660 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
3661 Predict emitted jump and add label to jump insn.
3663 2018-10-18 David Malcolm <dmalcolm@redhat.com>
3665 PR tree-optimization/87562
3666 * input.c (get_substring_ranges_for_loc): Use
3667 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
3668 getting the linemap for the endpoint. Verify that it's either
3669 in the same linemap as the start point's spelling location, or
3670 at least in the same file.
3672 2018-10-18 Richard Biener <rguenther@suse.de>
3674 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
3675 feed width-specific load/store costs through ix86_vec_cost.
3676 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
3677 (k8_cost): Likewise.
3678 (bdver_cost): Likewise.
3679 (znver1_cost): Likewise.
3680 (btver1_cost): Likewise.
3681 (btver2_cost): Likewise.
3683 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3685 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
3686 to simplify subreg of vec_merge.
3688 2018-10-18 Richard Biener <rguenther@suse.de>
3690 * config/i386/i386.c: Fix costing of vector FMA.
3692 2018-10-18 Richard Biener <rguenther@suse.de>
3694 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
3696 (ix86_builtin_vectorization_cost): For vec_construct properly
3697 cost insertion into SSE regs.
3698 (...): Adjust calls to ix86_vec_cost.
3700 2018-10-18 Richard Biener <rguenther@suse.de>
3704 2018-02-07 Richard Biener <rguenther@suse.de>
3706 PR tree-optimization/84204
3707 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
3710 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3713 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
3715 (test_vector_ops_duplicate): Add test for a scalar subreg of a
3716 VEC_MERGE of a VEC_DUPLICATE.
3718 2018-10-17 Joseph Myers <joseph@codesourcery.com>
3720 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
3721 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
3722 * doc/standards.texi (C Language): Document C2X.
3723 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
3724 (rl78_option_override): Handle "GNU C2X" language name.
3726 2018-10-17 Joseph Myers <joseph@codesourcery.com>
3728 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
3729 Document C17 as published in 2018.
3731 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
3734 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
3735 bail out if both sides do not have the same storage order.
3737 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
3739 * bitmap.c (bitmap_head::dump): New.
3740 * bitmap.h (bitmap_head): Add dump().
3741 * gimple-ssa-evrp-analyze.c
3742 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
3743 (evrp_range_analyzer::set_ssa_range_info): Same.
3744 (evrp_range_analyzer::record_ranges_from_phis): Same.
3745 (evrp_range_analyzer::record_ranges_from_stmt): Same.
3746 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
3747 * gimple-ssa-sprintf.c (get_int_range): Same.
3748 (format_integer): Same.
3749 (sprintf_dom_walker::handle_gimple_call): Same.
3750 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
3751 (ipcp_vr_lattice::top_p): Same.
3752 (ipcp_vr_lattice::bottom_p): Same.
3753 (ipcp_vr_lattice::set_to_bottom): Same.
3754 (ipa_vr_operation_and_type_effects): Same.
3755 (propagate_vr_across_jump_function): Same.
3756 (ipcp_store_vr_results): Same.
3757 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
3758 (ipa_print_node_jump_functions_for_edge): Same.
3759 (ipa_get_value_range): Same.
3760 (ipa_compute_jump_functions_for_edge): Same.
3761 (ipa_write_jump_function): Same.
3762 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
3763 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3765 * vr-values.c (set_value_range_to_nonnegative): Same.
3766 (set_value_range_to_truthvalue): Same.
3767 (vr_values::get_value_range): Same.
3768 (vr_values::set_defs_to_varying): Same.
3769 (vr_values::update_value_range): Same.
3770 (symbolic_range_based_on_p): Same.
3771 (vr_values::op_with_boolean_value_range_p): Same.
3772 (vr_values::extract_range_for_var_from_comparison_expr): Same.
3773 (vr_values::extract_range_from_ssa_name): Same.
3774 (vr_values::extract_range_from_binary_expr): Same.
3775 (vr_values::extract_range_from_unary_expr): Same.
3776 (vr_values::extract_range_from_cond_expr): Same.
3777 (vr_values::extract_range_from_comparison): Same.
3778 (vr_values::check_for_binary_op_overflow): Same.
3779 (vr_values::extract_range_basic): Same.
3780 (vr_values::extract_range_from_assignment): Same.
3781 (compare_ranges): Same.
3782 (compare_range_with_value): Same.
3783 (vr_values::adjust_range_with_scev): Same.
3784 (vrp_valueize): Same.
3785 (vrp_valueize_1): Same.
3786 (vr_values::get_vr_for_comparison): Same.
3787 (vr_values::compare_name_with_value): Same.
3788 (vr_values::compare_names): Same.
3789 (vr_values::vrp_evaluate_conditional): Same.
3790 (find_case_label_ranges): Same.
3791 (vr_values::vrp_visit_switch_stmt): Same.
3792 (vr_values::extract_range_from_phi_node): Same.
3793 (vr_values::simplify_div_or_mod_using_ranges): Same.
3794 (vr_values::simplify_bit_ops_using_ranges): Same.
3795 (test_for_singularity): Same.
3796 (range_fits_type_p): Same.
3797 (vr_values::simplify_cond_using_ranges_1): Same.
3798 (vr_values::simplify_switch_using_ranges): Same.
3799 (vr_values::simplify_float_conversion_using_ranges): Same.
3800 (vr_values::two_valued_val_range_p): Same.
3801 (vr_values::add_equivalence): Move to value_range::equiv_add.
3802 * vr-values.h (vr_values::add_equivalence): Remove.
3803 (VR_INITIALIZER): Remove.
3804 * tree-vrp.c (value_range::set): New.
3805 (value_range::equiv_add): New.
3806 (value_range::value_range): New.
3807 (value_range::deep_copy): New.
3808 (value_range::check): New.
3809 (value_range::equal_p): New.
3810 (value_range::ignore_equivs_equal_p): New.
3811 (value_range::operator==): New.
3812 (value_range::operator!=): New.
3813 (value_range::symbolic_p): New.
3814 (value_range::numeric_p): New.
3815 (value_range::set_undefined): New.
3816 (value_range::set_varying): New.
3817 (value_range::may_contain_p): New.
3818 (value_range::equiv_clear): New.
3819 (value_range::singleton_p): New.
3820 (value_range::intersect): New.
3821 (value_range::dump): New.
3822 (value_range::set_and_canonicalize): New.
3823 (set_value_range): Adjust for value_range API.
3824 (set_value_range_to_undefined): Same.
3825 (set_value_range_to_varying): Same.
3826 (set_and_canonicalize_value_range): Same.
3827 (set_value_range_to_nonnull): Same.
3828 (set_value_range_to_null): Same.
3829 (range_is_null): Same.
3830 (range_is_nonnull): Same.
3831 (range_int_cst_p): Same.
3832 (range_int_cst_singleton_p): Same.
3833 (symbolic_range_p): Same.
3834 (range_includes_zero_p): Same.
3835 (value_range_constant_singleton): Same.
3836 (vrp_set_zero_nonzero_bits): Same.
3837 (ranges_from_anti_range): Same.
3838 (extract_range_into_wide_ints): Same.
3839 (extract_range_from_multiplicative_op): Same.
3840 (set_value_range_with_overflow): Same.
3841 (extract_range_from_binary_expr_1): Same.
3842 (extract_range_from_unary_expr): Same.
3843 (dump_value_range): Same.
3844 (debug_value_range): Same.
3845 (vrp_prop::check_array_ref): Same.
3846 (vrp_prop::check_mem_ref): Same.
3847 (vrp_prop::vrp_initialize): Same.
3848 (vrp_prop::visit_stmt): Same.
3849 (intersect_ranges): Same.
3850 (vrp_prop::visit_phi): Same.
3851 (vrp_prop::vrp_finalize): Same.
3852 (determine_value_range_1): Same.
3853 (determine_value_range): Same.
3854 (vrp_intersect_ranges_1): Rename to...
3855 (vrp_intersect_1): this.
3856 (vrp_intersect_ranges): Rename to...
3857 (value_range::intersect_helper): ...this.
3858 (vrp_meet_1): Rename to...
3859 (value_range::union_helper): ...this.
3860 (vrp_meet): Rename to...
3861 (value_range::union_): ...this.
3862 (copy_value_range): Remove.
3863 * tree-vrp.h (struct value_range): Rewrite into a proper class.
3864 (value_range::vrtype): New.
3865 (value_range::type): New.
3866 (value_range::equiv): New.
3867 (value_range::min): New.
3868 (value_range::max): New.
3869 (value_range::varying_p): New.
3870 (value_range::undefined_p): New.
3871 (value_range::null_p): New.
3872 (value_range::equiv_add): New.
3873 (copy_value_range): Remove.
3875 2018-10-17 David Malcolm <dmalcolm@redhat.com>
3877 * Makefile.in (SELFTEST_TARGETS): New.
3878 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
3879 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
3880 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
3882 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
3883 (selftest-c++-gdb, selftest-c++-valgrind): Move to
3885 * configure: Regenerate.
3886 * configure.ac (selftest_languages): New.
3888 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
3890 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
3891 overflow wraps argument.
3892 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
3893 wide_int_range_multiplicative_op.
3894 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
3895 overflow wraps argument.
3896 (wide_int_range_multiplicative_op): Same.
3897 (wide_int_range_lshift): Same.
3898 (wide_int_range_div): Same.
3899 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
3900 (wide_int_range_lshift): Same.
3901 (wide_int_range_div): Same.
3903 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
3905 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
3906 use sign as argument.
3907 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
3908 wide_int_range_shift_undefined_p.
3910 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3912 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
3914 (@despeculate_copy<ALLI_TI:mode>): ... This.
3915 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
3918 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3920 * config.gcc: Obsolete *-*-solaris2.10*.
3921 * doc/install.texi (Specific, *-*-solaris2*): Document it.
3923 2018-10-12 Jeff Law <law@redhat.com>
3925 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
3926 reg + sym +- const_int addressing modes.
3928 2018-10-15 David Malcolm <dmalcolm@redhat.com>
3930 * common.opt (fdiagnostics-minimum-margin-width=): New option.
3931 * diagnostic-show-locus.c (layout::layout): Apply the minimum
3933 (layout::start_annotation_line): Only print up to 3 of the
3934 margin character, to avoid touching the left-hand side.
3935 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
3936 minimum margin width, as set by test_diagnostic_context's ctor.
3937 (selftest::test_fixit_insert_containing_newline): Likewise.
3938 (selftest::test_fixit_insert_containing_newline_2): Likewise.
3939 (selftest::test_line_numbers_multiline_range): Clear
3940 dc.min_margin_width.
3941 * diagnostic.c (diagnostic_initialize): Initialize
3943 * diagnostic.h (struct diagnostic_context): Add field
3945 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
3946 * opts.c (common_handle_option): Handle
3947 OPT_fdiagnostics_minimum_margin_width_.
3948 * selftest-diagnostic.c
3949 (selftest::test_diagnostic_context::test_diagnostic_context):
3950 Initialize min_margin_width to 6.
3951 * toplev.c (general_init): Initialize global_dc->min_margin_width.
3953 2018-10-15 David Malcolm <dmalcolm@redhat.com>
3955 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
3956 Fix usage of "error_at_rich_loc" in the comment.
3958 2018-10-15 Renlin Li <renlin.li@arm.com>
3961 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
3962 if-conversioned loop when it contains ifn with types not
3963 supported by backend.
3964 * internal-fn.c (expand_direct_optab_fn): Add an assert.
3965 (direct_internal_fn_supported_p): New helper function.
3966 * internal-fn.h (direct_internal_fn_supported_p): Declare.
3968 2018-10-15 Jakub Jelinek <jakub@redhat.com>
3971 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
3972 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
3973 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
3975 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
3977 PR tree-optimization/87022
3978 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
3979 bits in dist vector rather than the first one.
3981 2018-10-15 Richard Biener <rguenther@suse.de>
3984 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
3985 (visit_loadstore): When a used restrict tag escaped verify that
3986 the points-to solution of "other" pointers do not include
3988 (compute_dependence_clique): If a used restrict tag escaped
3989 communicated that down to visit_loadstore.
3991 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
3993 * config/s390/s390.c (s390_expand_vec_init): Force vector element
3994 into reg if it isn't a general operand.
3996 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
3999 * config/i386/sse.md (*vec_dupv2di): Add register source to
4002 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4005 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4006 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
4007 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4008 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4010 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
4012 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
4013 (notice_args_size): Set it in the current trace if no insn that can
4014 throw internally has been seen yet.
4015 (connect_traces): When connecting args_size between traces, allow the
4016 incoming values not to match if there is an insn setting it before the
4017 first insn that can throw internally; in that case, force the creation
4018 of a CFI note on this latter insn.
4020 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
4022 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
4023 the base class, because using the injected-class-name was not clearly
4024 specified until DR 176.
4026 2018-10-12 Paul Koning <ni1d@arrl.net>
4028 * config/pdp11/pdp11.md (doloop_end): New expander.
4029 (doloop_end_insn): renamed from "doloop_end".
4030 (addqi3): New pattern.
4031 (subqi3): New pattern.
4032 * config/pdp11/predicates.md (incdec_operand): New predicate.
4034 2018-10-12 Yury Gribov <tetra2005@gmail.com>
4037 * real.c (format_helper::can_represent_integral_type_p): New function
4038 * real.h (format_helper::can_represent_integral_type_p): Ditto.
4039 * match.pd: New pattern.
4041 2018-10-12 Alexandre Oliva <oliva@adacore.com>
4043 * configure.ac: Introduce --enable-large-address-aware
4044 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
4045 * doc/install.texi: Document it.
4046 * configure, config.in: Rebuilt.
4047 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
4048 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
4049 (LINK_SPEC): Insert it.
4050 * config/i386/mingw-w64.h: Likewise.
4052 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
4054 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
4056 PR rtl-optimization/87600
4057 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
4059 2018-10-12 Paul Koning <ni1d@arrl.net>
4061 * doc/md.texi (doloop_end): Document that the pattern code may
4062 need to check operand mode.
4064 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
4066 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
4067 to zero-extend between int and floating-point registers.
4068 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
4069 ldp into floating-point registers. Add type and arch attributes.
4070 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
4071 Use f_loads for type attribute.
4073 2018-10-11 Martin Sebor <msebor@redhat.com>
4075 * doc/extend.texi (attribute packed): Correct typos.
4077 2018-10-11 Martin Sebor <msebor@redhat.com>
4079 * doc/extend.texi (attribute flatten): Mention interaction with
4082 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
4085 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
4086 Set new_decl virtual flag to zero.
4088 2018-10-11 Martin Sebor <msebor@redhat.com>
4091 * doc/extend.texi (attribute format_arg): Discuss using multiple
4092 attributes on a single function.
4094 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
4096 PR tree-optimization/86829
4097 * match.pd (sin (atan (x))): New simplification rules.
4098 (cos (atan (x))): Likewise.
4099 * real.c (build_sinatan_real): New function.
4100 * real.h (build_sinatan_real): Prototype.
4102 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4104 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
4106 (fold_mergeeo_helper): New helper function.
4107 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
4108 intrinsics. Correct some whitespace indentation issues.
4110 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
4113 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
4114 Use HOST_WIDE_INT_1U for shift.
4116 2018-10-11 Doug Rupp <rupp@adacore.com>
4117 Olivier Hainque <hainque@adacore.com>
4119 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
4120 Pass --relax to the linker for RTPs.
4121 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
4123 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
4124 Jan Hubicka <jh@suse.cz>
4125 Martin Jambor <mjambor@suse.cz>
4127 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
4128 the same elements are repeated rather than printing all of them.
4129 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
4130 "repeated" elements.
4131 * read-rtl-function.c (test_loading_repeat): New function.
4132 (read_rtl_function_c_tests): Call test_loading_repeat.
4133 * rtl-tests.c (test_dumping_repeat): New function.
4134 (rtl_tests_c_tests): Call test_dumping_repeat.
4136 2018-10-11 Richard Biener <rguenther@suse.de>
4138 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
4139 bdver?_cost): Unify to ...
4140 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
4141 * config/i386/i386.c (processor_cost_table): Adjust.
4143 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
4146 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
4147 the thunk when expanding to GIMPLE.
4149 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
4151 * varasm.c (mergeable_string_section): Don't try to move zero-length
4152 strings to the merge section.
4154 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
4157 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
4159 2018-10-10 Jakub Jelinek <jakub@redhat.com>
4162 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
4163 to special_args set.
4165 2018-10-10 Richard Biener <rguenther@suse.de>
4167 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
4168 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
4169 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
4170 and recursing and pattern terminating the recursion on SSE
4171 vector width using ix86_expand_reduc.
4172 (reduc_sminmax_scal_<mode>): Split into part reducing to half
4173 width and recursing and SSE2 vector variant doing the final
4174 reduction with ix86_expand_reduc.
4175 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
4176 with terminating the recursion at AVX level, splitting that
4179 2018-10-09 David Malcolm <dmalcolm@redhat.com>
4181 * genmatch.c (error_cb): Rename to...
4182 (diagnostic_cb): ...this, converting int params to enums.
4183 (fatal_at): Update for renaming.
4184 (warning_at): Likewise.
4186 * input.c (selftest::ebcdic_execution_charset::apply):
4187 Update for renaming of...
4188 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
4190 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
4191 converting level and reason to enums.
4192 (class selftest::lexer_error_sink): Rename to...
4193 (class selftest::lexer_test_options): ...this, renaming field
4194 "m_errors" to "m_diagnostics".
4195 (selftest::lexer_test_options::apply): Update for renaming of...
4196 (selftest::lexer_test_options::on_error): ...this, renaming to...
4197 (selftest::lexer_test_options::on_diagnostic): ...this
4198 converting level and reason to enums.
4199 (selftest::test_lexer_string_locations_raw_string_unterminated):
4200 Update for renamings.
4201 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
4204 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
4206 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
4207 * config/rs6000/pmmintrin.h: New file.
4209 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
4211 PR tree-optimization/86659
4212 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
4214 2018-10-09 Richard Biener <rguenther@suse.de>
4216 PR tree-optimization/63155
4217 * tree-ssa-structalias.c: Include tree-ssa.h.
4218 (get_constraint_for_ssa_var): For undefs return nothing_id.
4219 (find_func_aliases): Cleanup PHI handling.
4221 2018-10-09 Richard Biener <rguenther@suse.de>
4223 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
4226 2018-10-09 Martin Liska <mliska@suse.cz>
4228 * asan.c (asan_emit_stack_protection): If a stack variable
4229 is located in a same file as current function, then emit
4230 line info into variable definition string.
4232 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4234 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
4237 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4239 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
4242 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4244 PR tree-optimization/86659
4245 * gimple-match.h (struct gimple_match_op): Add reverse field.
4246 (gimple_match_op::set_op): New overloaded method.
4247 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
4248 the REF_REVERSE_STORAGE_ORDER flag on the value.
4249 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
4250 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
4252 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
4255 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
4256 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
4258 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
4261 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
4262 Defined with __builtin_ia32_vfmaddsubpd512_mask.
4264 2018-10-08 Richard Biener <rguenther@suse.de>
4266 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
4267 cost the same as AVX128 ones.
4269 2018-10-08 Paul Koning <ni1d@arrl.net>
4271 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
4272 (expand_block_move): New function.
4273 * config/pdp11/pdp11.c (output_block_move): Remove.
4274 (expand_block_move): New function.
4275 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
4276 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
4277 (*movmemhi1): Remove.
4279 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
4281 * config/s390/2827.md: Increase latencies for some FP instructions.
4283 2018-10-08 Richard Biener <rguenther@suse.de>
4285 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
4287 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
4288 * tree-vectorizer.h (dump_stmt_cost): Adjust.
4289 (add_stmt_cost): Dump return value of the hook.
4291 2018-10-08 Richard Biener <rguenther@suse.de>
4293 PR tree-optimization/63155
4294 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
4295 (ssa_propagation_engine::ssa_propagate): Remove redundant
4296 bitmap bit clearing.
4298 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
4300 PR rtl-optimization/86939
4301 PR rtl-optimization/87479
4302 * ira.h (non_conflicting_reg_copy_p): New prototype.
4303 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
4304 (make_hard_regno_dead): Don't add conflicts for register
4305 ignore_reg_for_conflicts.
4306 (make_object_dead): Likewise.
4307 (non_conflicting_reg_copy_p): New function.
4308 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
4309 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
4310 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
4311 (make_hard_regno_dead): Don't add conflicts for register
4312 ignore_reg_for_conflicts. Remove special conflict handling of
4313 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
4314 check_pic_pseudo_p and update callers.
4315 (mark_pseudo_dead): Don't add conflicts for register
4316 ignore_reg_for_conflicts.
4317 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
4319 2018-10-05 Andrew Waterman <andrew@sifive.com>
4320 Jim Wilson <jimw@sifive.com>
4322 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
4323 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
4324 new pattern using HONOR_SNANS that emits one extra instruction.
4326 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4328 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
4329 patterns): Merge SI and DI patterns to a GPR pattern.
4330 (unnamed define_insn and define_split for record form of that): Merge
4331 to a single define_insn_and_split pattern.
4333 2018-10-05 David Malcolm <dmalcolm@redhat.com>
4336 * input.c (expand_location_to_spelling_point): Add param "aspect"
4337 and use rather than hardcoding LOCATION_ASPECT_CARET.
4338 (get_substring_ranges_for_loc): Handle the case of a single token
4339 within a macro expansion.
4340 * input.h (expand_location_to_spelling_point): Add "aspect" param,
4341 defaulting to LOCATION_ASPECT_CARET.
4343 2018-10-05 Paul Koning <ni1d@arrl.net>
4345 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
4346 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
4347 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
4348 (pdp11_guard_type): New function.
4350 2018-10-05 Paul Koning <ni1d@arrl.net>
4352 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
4353 * config/pdp11/pdp11.opt (mfloat32): Remove.
4355 * doc/invoke.texi (pdp11 -mfloat32): Remove:
4356 (pdp11 -mfloat64): Remove.
4358 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4360 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
4361 (*cmp<mode>_cc_i387): Ditto.
4362 (*cmpu<mode>_cc_i387): Ditto.
4363 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4364 * config/i386/i386.c (ix86_expand_fp_compare): Remove
4366 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
4367 Emit x86_sahf_1 pattern.
4368 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
4369 (ix86_expand_carry_flag_compare): Ditto.
4371 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4373 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
4374 to reg_or_0_operand. Add "C" constraint.
4375 (*cmpxf_cc_i387): Ditto.
4376 (*cmp<mode>_i387): Change operand 2 predicate
4377 to nonimm_or_0_operand. Add "C" constraint.
4378 (*cmp<mode>_cc_i387): Ditto.
4379 (*cmp<mode>_0_i387): Remove insn pattern.
4380 (*cmp<mode>_0_cc_i387): Ditto.
4382 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4384 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
4385 * config/i386/predicates.md (nonimm_or_0_operand): Rename
4386 from vector_move_operand. Update all uses.
4388 2018-10-05 Martin Sebor <msebor@redhat.com>
4390 PR tree-optimization/87490
4391 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
4394 2018-10-05 Richard Biener <rguenther@suse.de>
4396 PR tree-optimization/63155
4397 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
4398 vertical space in dumpfiles.
4399 * tree-ssa-propagate.h
4400 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
4401 * tree-ssa-propagate.c (cfg_blocks_back): New global.
4402 (ssa_edge_worklist_back): Likewise.
4403 (curr_order): Likewise.
4404 (cfg_blocks_get): Remove abstraction.
4405 (cfg_blocks_add): Likewise.
4406 (cfg_blocks_empty_p): Likewise.
4407 (add_ssa_edge): Add to current or next worklist based on
4409 (add_control_edge): Likewise.
4410 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
4412 (ssa_propagation_engine::ssa_propagate): ... here. Unify
4413 iteration from CFG and SSA edge worklist so we process
4414 everything in RPO order, prioritizing forward progress
4416 (ssa_prop_init): Allocate new worklists, do not dump
4418 (ssa_prop_fini): Free new worklists.
4420 2018-10-05 Richard Biener <rguenther@suse.de>
4422 * tree-core.h (tree_block::abstract_flag): Remove.
4423 (tree_block::block_num): Make full 32bits.
4424 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
4425 * tree.h (BLOCK_ABSTRACT): Remove.
4426 * dwarf2out.c (gen_lexical_block_die): Remove dead code
4427 resulting from BLOCK_ABSTRACT being always false.
4428 (gen_inlined_subroutine_die): Likewise.
4429 (gen_block_die): Likewise.
4430 * tree.c (block_ultimate_origin): Likewise.
4431 * tree-pretty-print.c (dump_block_node): Remove code dealing
4432 with BLOCK_ABSTRACT.
4433 * tree-ssa-live.c (dump_scope_block): Likewise.
4434 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
4435 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
4437 2018-10-05 Richard Biener <rguenther@suse.de>
4439 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
4440 is asked for initialize mode to the component mode of the
4443 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
4446 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
4447 assembler for -mavx.
4448 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
4450 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4453 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
4455 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
4456 to be DImode. When using mffscrn, force the operand to a register.
4458 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
4460 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
4461 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
4462 X87MODEF mode iterator.
4463 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
4464 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
4465 X87MODEF mode iterator.
4467 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
4469 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
4470 -Wno-prio-ctor-dtor.
4472 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4474 * Makefile.in (OBJS): Add opt-problem.o.
4475 * dump-context.h: Include "selftest.h.
4476 (selftest::temp_dump_context): New forward decl.
4477 (class dump_context): Make friend of class
4478 selftest::temp_dump_context.
4479 (dump_context::dump_loc_immediate): New decl.
4480 (class dump_pretty_printer): Move here from dumpfile.c.
4481 (class temp_dump_context): Move to namespace selftest.
4482 (temp_dump_context::temp_dump_context): Add param
4483 "forcibly_enable_dumping".
4484 (selftest::verify_dumped_text):
4485 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
4486 (selftest::verify_item):
4487 (ASSERT_IS_TEXT): Move here from dumpfile.c.
4488 (ASSERT_IS_TREE): Likewise.
4489 (ASSERT_IS_GIMPLE): Likewise.
4490 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
4492 (dump_context::dump_loc_immediate): ...this new function.
4493 (class dump_pretty_printer): Move to dump-context.h.
4494 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
4495 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
4496 (temp_dump_context::temp_dump_context): Move to "selftest"
4497 namespace. Add param "forcibly_enable_dumping", and use it to
4498 conditionalize the use of m_pp;
4499 (selftest::verify_dumped_text): Make non-static.
4500 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
4501 (selftest::verify_item): Make non-static.
4502 (ASSERT_IS_TEXT): Move to dump-context.h.
4503 (ASSERT_IS_TREE): Likewise.
4504 (ASSERT_IS_GIMPLE): Likewise.
4505 (selftest::test_capture_of_dump_calls): Pass "true" for new
4506 param of temp_dump_context.
4507 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
4508 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
4510 * opt-problem.cc: New file.
4511 * opt-problem.h: New file.
4512 * optinfo-emit-json.cc
4513 (selftest::test_building_json_from_dump_calls): Pass "true" for
4514 new param of temp_dump_context.
4515 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
4516 (optinfo::emit_for_opt_problem): New function.
4517 (optinfo::emit): Clarity which emit_item is used.
4518 * optinfo.h (optinfo::get_dump_location): New accessor.
4519 (optinfo::emit_for_opt_problem): New decl.
4520 (optinfo::emit): Make const.
4521 * selftest-run-tests.c (selftest::run_tests): Call
4522 selftest::opt_problem_cc_tests.
4523 * selftest.h (selftest::opt_problem_cc_tests): New decl.
4524 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
4525 bool to opt_result, converting fprintf messages to
4526 opt_result::failure_at calls. Add "stmt" param for use by the
4528 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
4529 (runtime_alias_check_p): Convert return type from bool to
4530 opt_result, converting dump_printf calls to
4531 opt_result::failure_at, using the statement DDR_A for their
4533 (find_data_references_in_stmt): Convert return type from bool to
4534 opt_result, converting "return false" to opt_result::failure_at
4536 * tree-data-ref.h: Include "opt-problem.h".
4537 (dr_analyze_innermost): Convert return type from bool to opt_result,
4538 and add a const gimple * param.
4539 (find_data_references_in_stmt): Convert return type from bool to
4541 (runtime_alias_check_p): Likewise.
4542 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
4543 dr_analyze_innermost.
4544 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
4545 Convert return type from bool to opt_result, adding a message for
4546 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
4547 (vect_analyze_data_ref_dependence): Convert return type from bool
4548 to opt_result. Change sense of return type from "false"
4549 effectively meaning "no problems" to "false" meaning a problem,
4550 so that "return false" becomes "return opt_result::success".
4551 Convert "return true" calls to opt_result::failure_at, using
4552 the location of statement A rather than vect_location.
4553 (vect_analyze_data_ref_dependences): Convert return type from bool
4555 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
4556 calls to opt_result::failure_at, using the stmt location rather
4558 (vect_verify_datarefs_alignment): Convert return type from bool
4560 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
4561 into multiple more-tightly-scoped copies.
4562 (vect_analyze_data_refs_alignment): Convert return type from bool
4564 (vect_analyze_data_ref_accesses): Likewise, converting a
4565 "return false" to a "return opt_result::failure_at", adding a
4567 (vect_prune_runtime_alias_test_list): Convert return type from
4568 bool to opt_result, converting dump_printf_loc to
4569 opt_result::failure_at. Add a %G to show the pertinent statement,
4570 and use the stmt's location rather than vect_location.
4571 (vect_find_stmt_data_reference): Convert return type from
4572 bool to opt_result, converting dump_printf_loc to
4573 opt_result::failure_at, using stmt's location.
4574 (vect_analyze_data_refs): Convert return type from bool to
4575 opt_result. Convert "return false" to "return
4576 opt_result::failure_at", adding messages as needed.
4577 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
4578 type from bool to opt_result.
4579 (vect_determine_vf_for_stmt): Likewise.
4580 (vect_determine_vectorization_factor): Likewise, converting
4581 dump_printf_loc to opt_result::failure_at, using location of phi
4582 rather than vect_location.
4583 (vect_analyze_loop_form_1): Convert return type from bool to
4584 opt_result, converting dump_printf_loc calls, retaining the use of
4586 (vect_analyze_loop_form): Convert return type from loop_vec_info
4587 to opt_loop_vec_info.
4588 (vect_analyze_loop_operations): Convert return type from bool to
4589 opt_result, converting dump_printf_loc calls, using the location
4590 of phi/stmt rather than vect_location where available. Convert
4591 various "return false" to "return opt_result::failure_at" with
4592 "unsupported phi" messages.
4593 (vect_get_datarefs_in_loop): Convert return type from bool to
4594 opt_result. Add a message for the
4595 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
4596 (vect_analyze_loop_2): Convert return type from bool to
4597 opt_result. Ensure "ok" is set to a opt_result::failure_at before
4598 each "goto again;", adding new messages where needed.
4599 Add "unsupported grouped {store|load}" messages.
4600 (vect_analyze_loop): Convert return type from loop_vec_info to
4602 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
4604 * tree-vect-stmts.c (process_use): Likewise, converting
4605 dump_printf_loc call and using stmt location, rather than
4607 (vect_mark_stmts_to_be_vectorized): Likeise.
4608 (vect_analyze_stmt): Likewise, adding a %G.
4609 (vect_get_vector_types_for_stmt): Convert return type from bool to
4610 opt_result, converting dump_printf_loc calls and using stmt
4611 location, rather than vect_location.
4612 (vect_get_mask_type_for_stmt): Convert return type from tree to
4613 opt_tree, converting dump_printf_loc calls and using stmt location.
4614 * tree-vectorizer.c: Include "opt-problem.h.
4615 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
4616 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
4617 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
4618 enabled, use it to report at the top level "couldn't vectorize
4619 loop" followed by the problem.
4620 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
4621 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
4623 (vect_analyze_stmt): Likewise.
4624 (vect_get_vector_types_for_stmt): Likewise.
4625 (tree vect_get_mask_type_for_stmt): Likewise.
4626 (vect_analyze_data_ref_dependences): Likewise.
4627 (vect_enhance_data_refs_alignment): Likewise.
4628 (vect_analyze_data_refs_alignment): Likewise.
4629 (vect_verify_datarefs_alignment): Likewise.
4630 (vect_analyze_data_ref_accesses): Likewise.
4631 (vect_prune_runtime_alias_test_list): Likewise.
4632 (vect_find_stmt_data_reference): Likewise.
4633 (vect_analyze_data_refs): Likewise.
4634 (vect_analyze_loop): Convert return type from loop_vec_info to
4636 (vect_analyze_loop_form): Likewise.
4637 (vect_analyze_slp): Convert return type from bool to opt_result.
4639 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4641 * doc/invoke.texi (-fopt-info): Document new "internals"
4643 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
4644 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
4646 (optinfo_verbosity_options): Add "internals".
4647 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
4648 (dump_context::apply_dump_filter_p): New member function.
4649 (dump_context::dump_loc): Use apply_dump_filter_p rather than
4650 explicitly masking the dump_kind.
4651 (dump_context::begin_scope): Increment the scope depth first. Use
4652 apply_dump_filter_p rather than explicitly masking the dump_kind.
4653 (dump_context::emit_item): Use apply_dump_filter_p rather than
4654 explicitly masking the dump_kind.
4655 (dump_dec): Likewise.
4656 (dump_hex): Likewise.
4657 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
4658 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
4659 (opt_info_switch_p): Update handling of default
4660 MSG_OPTIMIZED_LOCATIONS to cope with default of
4661 MSG_PRIORITY_USER_FACING.
4662 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
4663 masking the dump_kind.
4664 (selftest::test_capture_of_dump_calls): Update test_dump_context
4665 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
4666 than MSG_ALL. Generalize scope test to be run at all four
4667 combinations of with/without MSG_PRIORITY_USER_FACING and
4668 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
4669 for each of the two values.
4670 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
4671 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
4672 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
4673 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
4674 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
4675 with MSG_PRIORITY_*.
4676 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
4677 dump messages as MSG_PRIORITY_USER_FACING.
4678 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
4679 about the interaction with MSG_PRIORITY_*.
4681 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4683 * varasm.c (output_constant): Add new parameter merge_strings.
4684 Make strings properly zero terminated in merge string sections.
4685 (mergeable_string_section): Don't fail if the last char is non-zero.
4686 (assemble_variable_contents): Handle merge string sections.
4687 (assemble_variable): Likewise.
4688 (assemble_constant_contents): Likewise.
4689 (output_constant_def_contents): Likewise.
4690 (output_constructor_array_range,
4691 output_constructor_regular_field): Adjust call to output_constant.
4692 (output_object_block): Adjust call to assemble_constant_contents
4693 and assemble_variable_contents.
4695 2018-10-04 Martin Liska <mliska@suse.cz>
4698 * cgraphunit.c (process_function_and_variable_attributes):
4699 Warn about a function with alias attribute and a body.
4701 2018-10-04 Martin Liska <mliska@suse.cz>
4704 * multiple_target.c (redirect_to_specific_clone): New function.
4705 (ipa_target_clone): Use it.
4706 * tree-inline.c: Fix comment.
4708 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4710 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
4712 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
4713 (gcc::dump_manager::register_pass): New member function, adapted
4714 from loop body in gcc::pass_manager::register_pass, adding a
4715 call to update_dfi_for_opt_info.
4716 (gcc::dump_manager::opt_info_enable_passes): Store the
4717 -fopt-info options into the new fields. Move the loop
4719 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
4721 * dumpfile.h (struct opt_pass): New forward decl.
4722 (gcc::dump_manager::register_pass): New decl.
4723 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
4724 (class gcc::dump_manager): Add fields "m_optgroup_flags",
4725 "m_optinfo_flags", and "m_optinfo_filename".
4726 * passes.c (gcc::pass_manager::register_pass): Move all of the
4727 dump-handling code to gcc::dump_manager::register_pass.
4729 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
4731 PR rtl-optimization/87466
4732 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
4733 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
4734 * doc/tm.texi: Regenerate.
4735 * ira-lives.c (process_bb_node_lives): Use the new target hook.
4736 * lra-lives.c (process_bb_lives): Likewise.
4737 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
4740 2018-10-04 Tamar Christina <tamar.christina@arm.com>
4742 * params.c (add_params): Fix initialization.
4744 2018-10-04 Martin Liska <mliska@suse.cz>
4746 PR gcov-profile/84107
4747 * tree-profile.c (init_ic_make_global_vars):
4748 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
4749 Come up with new ic_tuple* variables. Emit
4750 __gcov_indirect_call{,_topn} variables.
4751 (gimple_gen_ic_profiler): Access the variable
4753 (gimple_gen_ic_func_profiler): Access
4754 __gcov_indirect_call.callee field.
4755 (gimple_init_gcov_profiler): Use ptr_type_node.
4756 * value-prof.c (gimple_ic): Use ptr_type_node.
4758 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4760 PR tree-optimization/85787
4761 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
4762 into this function and add support for detecting multiple phis.
4763 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
4765 2018-10-04 Martin Liska <mliska@suse.cz>
4768 * ipa-inline.c (inline_to_all_callers_1):
4769 Call ultimate_alias_target for node being inlined.
4771 2018-10-03 Jeff Law <law@redhat.com>
4773 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
4775 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
4776 * tree.h (get_typenode_from_name): Prototype.
4778 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
4780 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4781 Change operand 2 predicate to nonimmediate_operand.
4782 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4784 2018-10-03 Martin Sebor <msebor@redhat.com>
4785 Jeff Law <law@redhat.com>
4787 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
4789 (get_string_length): Detect unterminated arrays.
4790 (format_string): Same.
4791 (format_directive): Warn about unterminated arrays.
4792 (handle_gimple_call): Mark statements with no_warning as needed.
4794 2018-10-03 Jim Wilson <jimw@sifive.com>
4796 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
4797 also define __riscv_abi_rve. Delete trailing white space.
4799 2018-10-03 Paul Koning <ni1d@arrl.net>
4801 Enable LRA register allocator for PDP11.
4802 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
4805 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
4806 * config/pdp11/pdp11.opt (-mlra): New option.
4807 * doc/invoke.texi (PDP-11 Options): Document -mlra.
4809 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
4811 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
4812 (*<absneg:code>extend<mode>xf2): Ditto.
4814 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
4816 PR tree-optimization/87415
4817 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
4820 2018-10-02 Jeff Law <law@redhat.com>
4822 * gimple-fold.c (get_range_strlen): Only set *nonstr when
4823 an unterminated string is discovered. Bubble up range
4824 even for unterminated strings.
4825 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
4826 indicates the string was not terminated via NONSTR.
4828 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
4830 * tree-vrp.c (extract_range_from_unary_expr): Special case all
4831 pointer conversions.
4832 Do not do anything special for anti-ranges.
4834 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
4836 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
4837 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
4839 2018-10-03 Martin Liska <mliska@suse.cz>
4841 PR gcov-profile/86109
4842 * coverage.c (coverage_begin_function): Do not
4843 mark lambdas as artificial.
4844 * tree-core.h (struct GTY): Remove tm_clone_flag
4845 and introduce new lambda_function.
4846 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
4848 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4851 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
4852 P8_VECTOR and VSX are enabled.
4854 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
4856 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4857 0x3907 as CPU model number.
4859 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
4861 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
4862 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
4863 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
4864 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
4865 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
4866 * config/s390/s390.md: Likewise. Rename also the cpu attribute
4867 value from arch12 to z14.
4869 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
4871 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
4873 (isinf<mode>2): Ditto.
4875 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
4877 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
4878 before emitting fxam. Perform calculations in XFmode.
4880 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
4882 * match.pd (((X /[ex] A) +- B) * A): New transformation.
4884 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
4887 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
4888 * tree.c (signed_or_unsigned_type_for): Handle complex.
4890 2018-10-02 Jeff Law <law@redhat.com>
4892 * gimple-fold.c (get_range_strlen): Remove dead code.
4894 2018-10-02 Martin Sebor <msebor@redhat.com>
4895 Jeff Law <law@redhat.com>
4897 * builtins.c (unterminated_array): Add new arguments.
4898 If argument is not terminated, bubble up size and exact
4900 (expand_builtin_strnlen): Detect, avoid expanding
4901 and diagnose unterminated arrays.
4902 (c_strlen): Fill in offset of start of unterminated strings.
4903 * builtins.h (unterminated_array): Update prototype.
4905 2018-10-02 Richard Biener <rguenther@suse.de>
4907 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
4908 of haddv4df, first reduce to SSE width and exploit the fact
4909 that we only need element zero with the reduction result.
4910 (reduc_plus_scal_v2df): Likewise.
4912 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
4914 * dojump.h (do_jump): Delete.
4915 (do_jump_1): Likewise.
4916 (split_comparison): Move around.
4917 * dojump.c (do_jump): Make static.
4918 (do_jump_1): Likewise.
4919 (jumpifnot): Move around.
4920 (jumpifnot_1): Likewise.
4922 (jumpif_1): Likewise.
4923 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
4925 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
4927 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
4928 insns in the delay slot and add_insn_after for the jump insn.
4930 2018-10-02 Richard Biener <rguenther@suse.de>
4932 * tree-inline.c (expand_call_inline): Use the location of
4933 the callee declaration for the inline-entry marker.
4934 * final.c (notice_source_line): Remove special-casing of
4935 NOTE_INSN_INLINE_ENTRY.
4937 2018-10-01 Carl Love <cel@us.ibm.com>
4940 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
4941 (__builtin_mtfsb0): New.
4942 (__builtin_mtfsb1): New.
4943 ( __builtin_set_fpscr_rn): New.
4944 (__builtin_set_fpscr_drn): New.
4945 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
4946 (rs6000_expand_set_fpscr_rn_builtin): Add.
4947 (rs6000_expand_set_fpscr_drn_builtin): Add.
4948 (rs6000_expand_builtin): Add case statement entries for
4949 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
4950 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
4951 RS6000_BUILTIN_MFFSL.
4952 (rs6000_init_builtins): Add ftype initialization and def_builtin
4953 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
4954 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
4955 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
4956 rs6000_mffscdrn): Add define_insn.
4957 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
4958 * doc/extend.texi: Add documentation for the builtins.
4960 2018-10-01 Richard Biener <rguenther@suse.de>
4962 PR tree-optimization/87465
4963 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
4964 causing branch miscounts.
4966 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4968 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
4969 aarch64_option_default_param): New.
4970 (params.h): Include.
4971 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
4972 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
4973 stack-clash protection validation code.
4975 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4977 * params.c (validate_param): New.
4978 (add_params): Use it.
4979 (set_param_value): Refactor param validation into validate_param.
4980 (diagnostic.h): Include.
4981 * diagnostic.h (diagnostic_ready_p): New.
4983 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4985 * params.c (set_param_value):
4986 Add index of parameter being validated.
4987 * common/common-target.def (option_validate_param): New.
4988 * common/common-targhooks.h (default_option_validate_param): New.
4989 * common/common-targhooks.c (default_option_validate_param): New.
4990 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
4991 * doc/tm.texi: Regenerate.
4993 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4996 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4997 Add validation for stack-clash parameters and set defaults.
4999 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5002 * configure.ac: Add stack-clash-protection-guard-size.
5003 * doc/install.texi: Document it.
5004 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
5005 * params.def: Update comment for guard-size.
5006 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
5007 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
5008 * configure: Regenerate.
5010 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5013 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
5014 STACK_DYNAMIC_OFFSET): New.
5015 * config/aarch64/aarch64.c (aarch64_layout_frame):
5016 Update outgoing args size.
5017 (aarch64_stack_clash_protection_alloca_probe_range,
5018 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5020 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5023 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
5025 * target.def (stack_clash_protection_alloca_probe_range): New.
5026 (stack_clash_protection_final_dynamic_probe): Remove.
5027 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
5028 (default_stack_clash_protection_final_dynamic_probe): Remove.
5029 * targhooks.c: Likewise.
5030 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5031 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
5032 * doc/tm.texi: Regenerate.
5034 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5037 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
5038 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
5039 aarch64_clamp_to_uimm12_shift): New.
5040 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
5041 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
5043 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5046 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
5048 2018-10-01 Jeff Law <law@redhat.com>
5049 Richard Sandiford <richard.sandiford@linaro.org>
5050 Tamar Christina <tamar.christina@arm.com>
5053 * config/aarch64/aarch64.md
5054 (probe_stack_range): Add k (SP) constraint.
5055 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
5056 STACK_CLASH_MAX_UNROLL_PAGES): New.
5057 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
5058 stack probes for stack clash.
5059 (aarch64_allocate_and_probe_stack_space): New.
5060 (aarch64_expand_prologue): Use it.
5061 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
5062 (aarch64_sub_sp): Add emit_move_imm optional param.
5064 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
5066 PR tree-optimization/87261
5067 * match.pd: Remove trailing whitespace.
5068 Add (x & y) | ~(x | y) -> ~(x ^ y),
5069 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
5071 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
5073 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
5075 (ashlsi3_insn): Update instruction constraints.
5076 (ashrsi3_insn): Likewise.
5077 (rotrsi3): Likewise.
5078 (add_shift): Likewise.
5079 * config/arc/constraints.md (Csz): New 32 bit constraint. It
5080 avoids placing in the limm field small constants which, otherwise,
5081 could end into a small instruction.
5083 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
5085 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
5086 destination register is not odd-even.
5087 (umaddsidi4_split): Likewise.
5089 2018-10-01 Richard Biener <rguenther@suse.de>
5091 * tree-inline.c (expand_call_inline): Store origin of fn
5092 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
5093 * tree.c (block_ultimate_origin): Simplify and do some
5096 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
5098 * config/i386/mmx.md (EMMS): New int iterator.
5099 (emms): New int attribute.
5100 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
5101 EMMS int iterator. Explicitly declare clobbers.
5102 (mmx_emms): Remove expander.
5104 * config/i386/predicates.md (emms_operation): Remove predicate.
5105 (vzeroall_pattern): New predicate.
5106 (vzeroupper_pattern): Rename from vzeroupper_operation.
5107 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
5108 vzeroupper_pattern and vzeroall_pattern predicates.
5110 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
5112 PR rtl-optimization/86939
5113 * ira-lives.c (make_hard_regno_born): Rename from this...
5114 (make_hard_regno_live): ... to this. Remove update to conflict
5115 information. Update function comment.
5116 (make_hard_regno_dead): Add conflict information update. Update
5118 (make_object_born): Rename from this...
5119 (make_object_live): ... to this. Remove update to conflict information.
5120 Update function comment.
5121 (make_object_dead): Add conflict information update. Update function
5123 (mark_pseudo_regno_live): Call make_object_live.
5124 (mark_pseudo_regno_subword_live): Likewise.
5125 (mark_hard_reg_dead): Update function comment.
5126 (mark_hard_reg_live): Call make_hard_regno_live.
5127 (process_bb_node_lives): Likewise.
5128 * lra-lives.c (make_hard_regno_born): Rename from this...
5129 (make_hard_regno_live): ... to this. Remove update to conflict
5130 information. Remove now uneeded check_pic_pseudo_p argument.
5131 Update function comment.
5132 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
5133 to conflict information. Update function comment.
5134 (mark_pseudo_live): Remove update to conflict information. Update
5136 (mark_pseudo_dead): Add conflict information update.
5137 (mark_regno_live): Call make_hard_regno_live.
5138 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
5139 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
5141 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
5144 * config/i386/i386.c (construct_container): Use TImode for
5145 BLKmode values in 2 integer registers.
5147 2018-09-29 Jeff Law <law@redhat.com>
5149 * builtins.c (unterminated_array): Pass in c_strlen_data * to
5150 c_strlen rather than just a tree *.
5151 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
5152 Update recursive calls appropriately. If caller did not provide a
5153 suitable data pointer, create a local one. When a non-terminated
5154 string is discovered, bubble up information about the string via the
5155 c_strlen_data object.
5156 * builtins.h (c_strlen): Update prototype.
5157 (c_strlen_data): New structure.
5158 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
5159 For a type 2 call, if c_strlen indicates a non-terminated string
5160 use the length of the non-terminated string.
5161 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
5163 2018-09-29 Jakub Jelinek <jakub@redhat.com>
5166 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
5167 __m512d type for __A argument rather than __m512.
5169 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
5171 * match.pd (simple_comparison): Don't optimize if either operand is
5172 a function pointer when target needs function pointer canonicalization.
5174 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5176 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
5177 power5 .. power9 to remove indirection.
5178 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
5179 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
5180 ASM_CPU_476_SPEC): Delete.
5181 (ASM_CPU_SPEC): Adjust.
5182 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
5183 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
5185 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5187 * config.in: Delete HAVE_AS_DCI.
5188 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
5189 * config/rs6000/rs6000.h: Ditto.
5190 * configure.ac: Delete HAVE_AS_DCI.
5191 * configure: Regenerate.
5193 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5195 * config.in (HAVE_AS_LWSYNC): Delete.
5196 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
5197 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
5199 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
5200 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
5202 * configure.ac: Delete HAVE_AS_LWSYNC.
5203 * configure: Regenerate.
5205 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
5206 Pierre-Marie de Rodat <derodat@adacore.com>
5208 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
5209 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
5210 (cgraph_node::create_thunk): Add indirect_offset parameter.
5211 (thunk_adjust): Likewise.
5212 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
5213 and initialize the corresponding field with it.
5214 (cgraph_node::dump): Dump indirect_offset field.
5215 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
5216 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
5217 (thunk_adjust): Add indirect_offset parameter and deal with it.
5218 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
5219 pass it to thunk_adjust. Do not call the target hook if it's non-zero
5220 or if the thunk is external or local. Fix formatting. Do not chain
5221 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
5222 if any, in the GIMPLE representation.
5223 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
5224 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
5225 (input_node): Read indirect_offset field.
5226 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
5227 call to thunk_adjust.
5228 * tree-nested.c (struct nesting_info): Add thunk_p field.
5229 (create_nesting_tree): Set it.
5230 (convert_all_function_calls): Copy static chain from targets to thunks.
5231 (finalize_nesting_tree_1): Return early for thunks.
5232 (unnest_nesting_tree_1): Do not finalize thunks.
5233 (gimplify_all_functions): Do not gimplify thunks.
5235 2018-09-28 David Malcolm <dmalcolm@redhat.com>
5237 * opt-suggestions.c (option_proposer::build_option_suggestions):
5238 Release "option_values".
5240 2018-09-28 David Malcolm <dmalcolm@redhat.com>
5242 * coverage.c (get_coverage_counts): Convert problem-reporting dump
5243 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
5244 * dumpfile.c (kind_as_string): New function.
5245 (dump_loc): Rather than a hardcoded prefix of "note: ", use
5246 kind_as_string to vary the prefix based on dump_kind.
5247 (selftest::test_capture_of_dump_calls): Update for above.
5249 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
5251 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
5252 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
5254 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
5256 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
5257 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
5258 INVALID_REGNUM instead of FPSR_REG.
5259 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
5260 * config/i386/i386.md: Update comment of FP compares.
5261 (fldenv): Do not clobber FPSR_REG.
5263 2018-09-28 Richard Biener <rguenther@suse.de>
5265 * tree.h (BLOCK_ORIGIN): New.
5266 * omp-expand.c (grid_expand_target_grid_body): Assign
5267 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
5268 * tree-inline.c (remap_block): Likewise.
5269 * auto-profile.c (get_function_decl_from_block): Simplify
5270 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
5271 * langhooks.c (lhd_print_error_function): Likewise.
5272 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
5274 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
5275 * tree.c (block_nonartificial_location): Likewise.
5276 (block_ultimate_origin): Likewise.
5277 * tree-pretty-print.c (percent_K_format): Likewise. Remove
5278 no longer needed LTO case.
5280 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
5281 Jan Hubicka <jh@suse.cz>
5282 Martin Jambor <mjambor@suse.cz>
5284 * simplify-rtx.c (simplify_merge_mask): New function.
5285 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
5286 same masks are used in op1 or op2.
5287 (test_vec_merge): New function.
5288 (test_vector_ops): Call test_vec_merge.
5290 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
5292 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
5293 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
5294 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
5295 (cypress_costs): Set it.
5296 (supersparc_costs): Likewise.
5297 (hypersparc_costs): Likewise.
5298 (leon_cost): Likewise.
5299 (leon3_costs): Likewise.
5300 (sparclet_costs): Likewise.
5301 (ultrasparc_costs): Likewise.
5302 (ultrasparc_costs): Likewise.
5303 (niagara_costs): Likewise.
5304 (niagara2_costs): Likewise.
5305 (niagara3_costs): Likewise.
5306 (niagara4_costs): Likewise.
5307 (niagara7_costs): Likewise.
5308 (m8_costs): Likewise.
5309 (TARGET_CAN_FOLLOW_JUMP): Define.
5310 (pass_work_around_errata::gate): Minor tweak.
5311 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
5312 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
5313 Automaitcally clear MASK_FSMULD mask for V7 processors.
5314 (sparc_can_follow_jump): New static function.
5315 (output_ubranch): Deal with CROSSING_JUMP_P.
5316 (sparc_use_sched_lookahead): Rewrite using switch statement.
5317 (sparc_issue_rate): Reorder.
5318 (sparc_branch_cost): New function.
5320 2018-09-27 Martin Sebor <msebor@redhat.com>
5322 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
5323 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
5324 (int_fits_type_p): Same.
5326 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
5328 * config/i386/i386.md (FPCR_REG): Remove.
5329 (UNSPEC_FLDCW): Remove.
5330 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
5331 (x86_fldcw_1): Remove insn pattern.
5332 (fnstenv): Do not clobber FPCR_REG.
5334 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
5335 (CALL_USED_REGISTERS): Ditto.
5336 (REG_ALLOC_ORDER): Ditto.
5337 (REG_CLASS_CONTENTS): Ditto.
5338 (HI_REGISTER_NAMES): Ditto.
5339 (ADDITIONAL_REGISTER_NAMES): Use defines instead
5340 of numerical constants.
5341 * config/i386/i386.c (regclass_map): Remove fpsr register.
5342 (dbx_register_map): Ditto.
5343 (dbx64_register_map): Ditto.
5344 (svr4_dbx_register_map): Ditto.
5345 (print_reg): Do not handle FPCR_REG.
5347 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
5350 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
5351 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
5352 Delete, always treat as true.
5353 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
5354 Ditto. Simplify remaining code.
5355 * config/powerpcspe/powerpcspe.h: Ditto.
5356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
5357 Simplify remaining code.
5358 (rs6000_expand_builtin): Ditto.
5359 * config/rs6000/rs6000.h: Ditto.
5360 * configure.ac: Ditto.
5361 * configure: Regenerate.
5363 2018-09-27 Martin Liska <mliska@suse.cz>
5365 * coverage.c (get_coverage_counts): Revert the formatting
5366 of missing profile opt info.
5368 2018-09-27 Richard Biener <rguenther@suse.de>
5372 * dwarf2out.c (set_block_origin_self): Do not mark outermost
5373 block as we do not output that.
5374 (gen_inlined_subroutine_die): Elide the originally outermost
5375 block, matching what we do for concrete instances.
5376 (decls_for_scope): Add parameter specifying whether to recurse
5379 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
5380 Tom de Vries <tom@codesourcery.com>
5384 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
5385 STORE_FLAG_VALUE == 1.
5387 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
5389 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
5390 constant definitions.
5391 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
5392 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
5393 ("speculation_barrier"): New expander definition.
5395 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
5397 PR gcov-profile/86957
5398 * common.opt: New warning option -Wmissing-profile.
5399 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
5400 * doc/invoke.texi: Document -Wmissing-profile.
5402 2018-09-26 Jim Wilson <jimw@sifive.com>
5404 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
5405 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
5407 2018-09-26 Martin Sebor <msebor@redhat.com>
5409 * tree.c (zerop): Change return type to bool.
5410 (integer_zerop, integer_onep, integer_each_onep): Same.
5411 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
5412 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
5413 (real_onep, real_minus_onep, chain_index): Same.
5414 (print_type_hash_statistics, type_list_equal): Same.
5415 * tree.h (zerop): Same.
5416 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
5417 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
5418 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
5419 (real_onep, real_minus_onep, chain_index): Same.
5420 (print_type_hash_statistics, type_list_equal): Same.
5422 2018-09-26 Jim Wilson <jimw@sifive.com>
5424 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
5426 2018-09-26 Jakub Jelinek <jakub@redhat.com>
5429 * config/i386/i386.c: Include debug.h and dwarf2out.h.
5430 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
5433 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
5435 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
5437 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
5439 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
5440 and FP_SECOND_SSE_REGS.
5441 (REG_CLASS_NAMES): Ditto.
5442 (REG_CLASS_CONTENTS): Ditto.
5443 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
5444 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
5445 (ix86_preferred_output_reload_class): Ditto.
5446 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
5447 clobber constraint to "=&f".
5448 (fix_truncdi_i387): Ditto.
5449 (lrintxfdi2): Ditto.
5450 (fistdi2_<rounding>): Ditto.
5451 (fpremxf4_i387): Change "=u" constraint to "=f".
5452 (fprem1xf4_i387): Ditto.
5454 (fptanxf4_i387): Ditto.
5455 (fxtractxf3_i387): Ditto.
5456 (fscalexf4_i387): Ditto.
5457 (atan2xf3): Change "u" constraint to "f".
5458 (fyl2xxf3_i387): Ditto.
5459 (fyl2xp1xf3_i387): Ditto.
5461 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
5464 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
5465 for removed I387_MASK_PM entity.
5468 2018-09-26 Jeff Law <law@redhat.com>
5470 2018-09-26 Alexey Neyman <stilor@att.net>
5472 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
5473 headers are no longer pulled in by <isl/val.h>.
5475 2018-09-26 Richard Biener <rguenther@suse.de>
5478 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
5479 or concrete instance DIE to the tree. Create abstract origin
5480 attributes also for concrete instances.
5482 2018-09-26 Alexey Neyman <stilor@att.net>
5484 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
5485 headers are no longer pulled in by <isl/val.h>.
5487 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
5489 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
5490 Use new helper functions.
5491 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
5492 Use new helper functions.
5493 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
5494 aarch_mm_needs_release): New declarations.
5495 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
5496 aarch_mm_needs_release): New.
5498 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
5500 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
5501 (arm32_output_mi_thunk): Deal with long calls.
5503 2018-09-26 Richard Biener <rguenther@suse.de>
5507 * tree-inline.c (expand_call_inline): When the location
5508 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
5509 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
5510 the inserted BLOCK to make inlined_function_outer_scope_p
5512 * dwarf2out.c (add_call_src_coords_attributes): Do not add
5513 coords for reserved locations.
5515 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5517 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
5518 (*call_indirect_nonlocal_sysv<mode>): Ditto.
5519 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
5520 (*sibcall_nonlocal_sysv<mode>): Ditto.
5521 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
5522 (<bd>_<mode>): Ditto.
5523 (<bd>tf_<mode>): Ditto.
5525 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5527 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
5528 control string as a list of templates instead of as C code.
5529 (*altivec_movti): Ditto.
5530 * config/rs6000/darwin.md (movdf_low_di): Ditto.
5532 2018-09-25 Jim Wilson <jimw@sifive.com>
5534 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
5535 when target symbol is weak.
5537 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
5540 * builtins.c (unterminated_array): Simplify.
5541 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
5542 where pointer arithmetic is safe.
5544 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5547 * config/rs6000/altivec.md (altivec_vspltb): Use
5548 const_0_to_15_operand instead of u5bit_cint_operand.
5549 (*altivec_vspltb_internal): Ditto.
5550 (altivec_vspltb_direct): Ditto.
5551 (altivec_vsplth): Use const_0_to_7_operand instead of
5553 (*altivec_vsplth_internal): Ditto.
5554 (altivec_vsplth_direct): Ditto.
5555 (altivec_vspltw): Use const_0_to_3_operand instead of
5557 (*altivec_vspltw_internal): Ditto.
5558 (altivec_vspltw_direct): Ditto.
5559 (altivec_vspltsf): Ditto.
5560 (*altivec_vspltsf_internal): Ditto.
5561 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
5562 various splats with the proper size immediate. Reorder the various
5563 cases by ascending size of immediate, and put all such together.
5565 2018-09-25 Richard Biener <rguenther@suse.de>
5568 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
5570 (maybe_create_die_with_external_ref): Do not create
5571 DW_TAG_imported_unit here.
5572 (add_abstract_origin_attribute): Handle external BLOCK refs.
5573 (dwarf2out_abstract_function): Simplify LTO case.
5574 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
5575 rather than using maybe_create_die_with_external_ref.
5577 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
5580 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
5581 (frndintxf2_mask_pm_i387): Ditto.
5582 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
5583 Enable for !flag_trapping_math.
5584 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
5585 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
5586 Change operand 1 predicate to nonimmediate_operand.
5587 (attr "i387_cw"): Remove mask_pm.
5588 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
5589 (enum ix86_entity): Remove I387_MASK_PM.
5590 * config/i386/i386.c (ix86_i387_mode_needed): Do not
5591 handle I387_MASK_PM.
5592 (ix86_mode_needed): Ditto.
5593 (ix86_mode_after): Ditto.
5594 (ix86_mode_entry): Ditto.
5595 (ix86_mode_exit): Ditto.
5596 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
5598 2018-09-25 Jakub Jelinek <jakub@redhat.com>
5600 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
5601 to_update_switch_stmts to vNULL instead of calling create on them
5604 2018-09-25 Richard Biener <rguenther@suse.de>
5606 PR tree-optimization/87402
5607 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
5608 (visit_phi): Re-instantiate handling of supposed to be VARYING
5609 but non-VARYING backedge value.
5611 2018-09-25 Richard Biener <rguenther@suse.de>
5614 * dwarf2out.c (struct sym_off_pair): New.
5615 (external_die_map): New global.
5616 (lookup_decl_die): When in LTO create DIEs lazily from the
5618 (lookup_block_die): New function, create DIEs lazily in LTO.
5619 (equate_block_to_die): New function.
5620 (dwarf2out_die_ref_for_decl): During WPA get the association
5621 from the external DIE map.
5622 (dwarf2out_register_external_die): Record mapping into the
5624 (maybe_create_die_with_external_ref): New function split out from
5625 DIE generation part of old dwarf2out_register_external_die.
5626 (add_abstract_origin_attribute): Do not return the DIE. When
5627 in LTO reference externals directly.
5628 (dwarf2out_abstract_function): When in LTO ignore calls for
5629 decls with external DIEs (already present abstract instances).
5630 (gen_call_site_die): Adjust.
5631 (add_high_low_attributes): Likewise.
5632 (gen_lexical_block_die): Likewise.
5633 (gen_inlined_subroutine_die): Likewie.
5634 (gen_block_die): Likewise.
5635 (dwarf2out_inline_entry): Likewise.
5636 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
5639 2018-09-25 Martin Liska <mliska@suse.cz>
5641 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
5642 integers and not by a float value.
5644 2018-09-25 Martin Liska <mliska@suse.cz>
5647 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
5649 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
5650 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
5651 * godump.c (go_format_type): Remove extra parenthesis.
5653 2018-09-25 Martin Liska <mliska@suse.cz>
5655 * alias.c (set_dest_equal_p): Remove unused function.
5656 * config/i386/i386.c (def_builtin_pure2): Likewise.
5657 * diagnostic-show-locus.c (class layout): Remove
5659 (layout::layout): Likewise here.
5660 * dump-context.h (class temp_dump_context): Likewise.
5661 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
5662 (add_AT_loclistsptr): Likewise.
5663 (add_AT_offset): Likewise.
5664 (get_AT_hi_pc): Likewise.
5665 (is_comdat_die): Likewise.
5666 (type_is_enum): Likewise.
5667 (ceiling): Likewise.
5668 (add_AT_vms_delta): Likewise.
5669 (is_class_die): Likewise.
5670 * edit-context.c (class line_event): Remove unused field.
5671 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
5673 * ipa-cp.c (ipa_get_vr_lat): Likewise.
5674 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
5675 (ok_for_base_p_nonstrict): Likewise.
5676 * tree-chrec.c (is_not_constant_evolution): Likewise.
5677 (chrec_fold_poly_cst): Likewise.
5678 * tree-if-conv.c (has_pred_critical_p): Likewise.
5679 * tree-ssa-coalesce.c (print_exprs): Likewise.
5680 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
5681 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
5682 * tree-vrp.c (value_ranges_intersect_p): Likewise.
5683 (value_range_nonnegative_p): Likewise.
5685 2018-09-25 Martin Liska <mliska@suse.cz>
5687 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
5688 Do not handle "GNU Pascal".
5689 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
5691 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
5692 from documentation. Likewise.
5693 * dbxout.c (dbxout_range_type): Likewise.
5694 * doc/cpp.texi: Likewise.
5695 * doc/extend.texi: Likewise.
5696 * doc/frontends.texi: Likewise.
5697 * doc/invoke.texi: Remove Pascal entry.
5698 * tree.def (CLEANUP_POINT_EXPR): Likewise.
5699 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
5701 2018-09-25 Martin Liska <mliska@suse.cz>
5704 * doc/invoke.texi: Document all parameters and remove default
5707 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
5710 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
5711 contains HOST_WIDE_INTs when computing its size.
5713 2018-09-24 Jim Wilson <jimw@sifive.com>
5716 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
5718 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
5720 2018-09-24 Andrew Pinski <apinski@marvell.com>
5722 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
5723 access prev before checking it for NULLness in the
5724 AARCH64_FUSE_CMP_BRANCH case.
5726 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
5729 * config/i386/i386.c (rest_of_insert_endbranch): Set
5730 endbr_queued_at_entrance to true and don't insert ENDBR if
5731 x86_function_profiler will be called.
5732 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
5734 * config/i386/i386.h (machine_function): Add
5735 endbr_queued_at_entrance.
5737 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
5739 * genattrtab.c (mk_attr_alt): Use alternative_mask.
5740 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
5742 (check_attr_test): Use alternative_mask.
5743 (get_attr_value): Likewise.
5744 (compute_alternative_mask): Use alternative_mask and XWINT.
5745 (make_alternative_compare): Use alternative_mask.
5746 (attr_alt_subset_p): Use XWINT.
5747 (attr_alt_subset_of_compl_p): Likewise.
5748 (attr_alt_intersection): Use alternative_mask and XWINT.
5749 (attr_alt_union): Likewise.
5750 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
5751 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
5752 (simplify_test_exp): Use alternative_mask and XWINT.
5753 (write_test_expr): Use alternative_mask and XWINT, adjust bit
5754 number calculation to support 64 bits. Generate code that
5755 checks 64-bit masks.
5756 (main): Use alternative_mask.
5757 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
5759 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
5762 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
5763 RETURN+USE when returning via %r14.
5765 2018-09-24 Martin Liska <mliska@suse.cz>
5767 * gcov.c (output_lines): Print colorization legend
5768 for both flag_use_colors and flag_use_hotness_colors.
5771 2018-09-24 Martin Liska <mliska@suse.cz>
5773 * coverage.c (get_coverage_counts): Use warning_at
5774 with current_function_decl location. Use %qD in warning
5777 2018-09-24 Martin Liska <mliska@suse.cz>
5779 * memory-block.h (memory_block_pool::release): Annotate with
5780 valgrind that the memory is not accessible.
5782 2018-09-24 Martin Liska <mliska@suse.cz>
5785 * asan.c: Make asan_handled_variables extern.
5787 * cfgexpand.c (expand_stack_vars): Make sure
5788 a representative is unpoison if another
5789 variable in the partition is handled by
5790 use-after-scope sanitization.
5792 2018-09-24 Richard Biener <rguenther@suse.de>
5794 PR tree-optimization/63155
5795 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
5796 the worklist when the edge of the respective argument isn't
5799 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
5801 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
5802 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
5803 (MASK_CLASS_P): Update for rename.
5804 (MAYBE_MASK_CLASS_P): Ditto.
5805 (REG_CLASS_NAMES): Update.
5806 (REG_CLASS_CONTENT): Update.
5807 * config/i386/i386.c (regclass_map): Update for MASK_REG
5808 and ALL_MASK_REGS rename.
5809 * config/i386/constraints.md (Yk): Update for rename.
5812 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
5814 * config/i386/i386.h (enum reg_class): Remove
5815 EVEX_SSE_REGS and MOD4_SSE_REGS.
5816 (REG_CLASS_NAMES): Update.
5817 (REG_CLASS_CONTENT): Update.
5818 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
5819 registers as ALL_SSE_REGS.
5820 (ix86_additional_allocno_class_p): Remove.
5821 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
5822 (ix86_register_priority): Lower priority of EVEX SSE registers.
5823 Use IN_RANGE macro where appropriate.
5824 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
5825 AVX-5124VNNIW checks.
5826 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
5827 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
5828 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
5829 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
5830 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
5831 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
5832 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
5833 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
5834 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
5835 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
5836 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
5837 * config/i386/constraints.md (Yh): Remove.
5839 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
5841 * config/i386/i386.c (regclass_map): Declare integer REX registers
5844 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
5846 * doc/service.texi (Service): Switch the fsf.org link to https.
5848 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
5851 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5852 Define to speculation_safe_value_not_needed.
5854 2018-09-21 Florian Weimer <fweimer@redhat.com>
5857 * doc/extend.texi (Common Function Attributes): Mention that
5858 noreturn suppresses tail call optimization.
5860 2018-09-21 Jeff Law <law@redhat.com>
5862 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
5863 vr_values::cleanup_edges_and_switches.
5864 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
5866 (identify_jump_threads): Remove EDGE_IGNORE handling.
5867 (execute_vrp): Move handling of to_remove_edges and
5868 to_update_switch_stmts into vr_values class member functions.
5869 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
5870 (to_update_switch_stmts): Likewise.
5871 * vr-values.c: Include cfghooks.h.
5872 (vr_values::vr_values): Initialize to_remove_edges and
5873 to_update_switch_stmts.
5874 (vr_values::~vr_values): Verify to_remove_edges and
5875 to_update_switch_stmts are empty.
5876 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
5877 (vr_values::cleanup_edges_and_switches): New member function.
5878 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
5879 function. Add new data members.
5881 2018-09-21 David Malcolm <dmalcolm@redhat.com>
5883 PR tree-optimization/87309
5884 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
5885 calls with pflags and alt_flags.
5886 (selftest::test_capture_of_dump_calls): Add test of interaction of
5887 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
5889 2018-09-21 Olivier Hainque <hainque@adacore.com>
5891 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
5893 2018-09-21 Olivier Hainque <hainque@adacore.com>
5895 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
5897 2018-09-21 Olivier Hainque <hainque@adacore.com>
5899 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
5900 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
5901 (VXWORKS_LIBS_RTP): Minor reordering.
5903 2018-09-21 Olivier Hainque <hainque@adacore.com>
5905 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
5906 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
5908 2018-09-21 Olivier Hainque <hainque@adacore.com>
5910 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
5911 (PTRDIFF_TYPE): Likewise.
5913 2018-09-21 Olivier Hainque <hainque@adacore.com>
5915 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
5916 triplet, similar to support for VxWorks7.
5917 * config/vxworks-dummy.h: Provide a default definition
5918 of TARGET_VXWORKS64 to 0.
5920 2018-09-21 Olivier Hainque <hainque@adacore.com>
5922 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
5923 * config/vxworks-dummy.h: here.
5925 2018-09-21 Olivier Hainque <hainque@adacore.com>
5927 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
5929 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
5930 Bo Zhou <zbo.zhou@hisilicon.com>
5932 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
5933 * config/aarch64/aarch64-tune.md: Regenerated.
5934 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
5935 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
5936 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
5938 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
5939 Julian Brown <julian@codesourcery.com>
5941 * builtins.c (get_builtin_sync_mem): Handle address spaces.
5943 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
5945 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
5946 if the call takes a static chain.
5948 2018-09-21 Martin Liska <mliska@suse.cz>
5950 * auto-profile.c (autofdo_source_profile::read): Do not
5952 (read_profile): Do not add working sets.
5953 (read_autofdo_file): Remove sum_all.
5954 (afdo_callsite_hot_enough_for_early_inline): Remove const
5956 * coverage.c (struct counts_entry): Remove gcov_summary.
5957 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
5958 do not support GCOV_TAG_PROGRAM_SUMMARY.
5959 (get_coverage_counts): Remove summary and expected
5961 * coverage.h (get_coverage_counts): Likewise.
5962 * doc/gcov-dump.texi: Remove -w option.
5963 * gcov-dump.c (dump_working_sets): Remove.
5964 (main): Do not support '-w' option.
5965 (print_usage): Likewise.
5966 (tag_summary): Likewise.
5967 * gcov-io.c (gcov_write_summary): Do not dump
5969 (gcov_read_summary): Likewise.
5970 (gcov_histo_index): Remove.
5971 (gcov_histogram_merge): Likewise.
5972 (compute_working_sets): Likewise.
5973 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
5975 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
5976 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
5977 (GCOV_HISTOGRAM_SIZE): Remove.
5978 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
5979 (struct gcov_summary): Simplify rapidly just
5980 to runs and sum_max fields.
5981 (gcov_histo_index): Remove.
5982 (NUM_GCOV_WORKING_SETS): Likewise.
5983 (compute_working_sets): Likewise.
5984 * gcov-tool.c (print_overlap_usage_message): Remove
5985 trailing empty line.
5986 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
5987 (output_lines): Remove program related line.
5988 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
5989 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
5991 (input_profile_summary): Do not read it.
5992 (merge_profile_summaries): And do not merge it.
5993 (input_symtab): Do not call removed function.
5994 * modulo-sched.c (sms_schedule): Do not print sum_max.
5995 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
5996 removed when histogram method was invented.
5997 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
5999 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
6001 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
6002 and dump selected value.
6003 * profile.c (add_working_set): Remove.
6004 (get_working_sets): Likewise.
6005 (find_working_set): Likewise.
6006 (get_exec_counts): Do not work with working sets.
6007 (read_profile_edge_counts): Do not inform as sum_max is removed.
6008 (compute_branch_probabilities): Likewise.
6009 (compute_value_histograms): Remove argument for call of
6010 get_coverage_counts.
6011 * profile.h: Do not make gcov_summary const.
6013 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
6015 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
6017 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6019 PR tree-optimization/86990
6020 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
6021 Check that the entire merged store group is made of constants only for
6024 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
6026 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
6027 (VTABLE_VERIFICATION_SPEC): Likewise.
6028 (SANITIZER_EARLY_SPEC): Likewise.
6029 (SANITIZER_SPEC): Likewise.
6030 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
6031 * doc/invoke.texi (Link Options): Document -r.
6033 2018-09-20 Richard Biener <rguenther@suse.de>
6036 * gimplify.c (gimplify_expr): Retain alignment of
6037 addressable lvalue in dereference.
6039 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
6042 * configure.ac: Check for .loc is_stmt support.
6043 * configure, config.in: Rebuilt.
6044 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
6047 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6049 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
6050 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
6053 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6055 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
6057 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
6060 2018-09-20 Marek Polacek <polacek@redhat.com>
6062 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
6064 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
6066 PR tree-optimization/87288
6067 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
6068 into account when determining PEELING_FOR_NITERS.
6070 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
6072 PR tree-optimization/86877
6073 * tree-vect-loop.c (vect_analyze_loop_2): Call
6074 vect_verify_datarefs_alignment.
6076 2018-09-19 Marek Polacek <polacek@redhat.com>
6078 * doc/invoke.texi: Document -Wclass-conversion.
6080 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
6082 * config/pa/pa.c (pa_adjust_priority): Delete.
6083 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
6085 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
6086 (atomic_storehi): Likewise.
6087 (atomic_storesi): Likewise.
6088 (atomic_loaddi): Restore compare and swap exchange loop code.
6090 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6092 PR rtl-optimization/86902
6093 * combine.c (try_combine): When changing the CC mode used, don't change
6094 an unrelated mode in other_insn to that new CC mode.
6096 2018-09-19 David Malcolm <dmalcolm@redhat.com>
6098 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
6099 with %T in place of calls to dump_generic_expr.
6100 (prune_runtime_alias_test_list): Likewise.
6101 (create_runtime_alias_checks): Likewise.
6102 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
6103 (vect_analyze_data_ref_dependence): Likewise.
6104 (vect_slp_analyze_data_ref_dependence): Likewise.
6105 (vect_record_base_alignment): Likewise. Use %G in place of call
6106 to dump_gimple_stmt.
6107 (vect_compute_data_ref_alignment): Likewise.
6108 (verify_data_ref_alignment): Likewise.
6109 (vect_find_same_alignment_drs): Likewise.
6110 (vect_analyze_group_access_1): Likewise.
6111 (vect_analyze_data_ref_accesses): Likewise.
6112 (dependence_distance_ge_vf): Likewise.
6113 (dump_lower_bound): Likewise.
6114 (vect_prune_runtime_alias_test_list): Likewise.
6115 (vect_find_stmt_data_reference): Likewise.
6116 (vect_analyze_data_refs): Likewise.
6117 (vect_create_addr_base_for_vector_ref): Likewise.
6118 (vect_create_data_ref_ptr): Likewise.
6119 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
6120 (vect_can_advance_ivs_p): Likewise.
6121 (vect_update_ivs_after_vectorizer): Likewise.
6122 (vect_gen_prolog_loop_niters): Likewise.
6123 (vect_prepare_for_masked_peels): Likewise.
6124 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
6125 (vect_determine_vectorization_factor): Likewise.
6126 (vect_is_simple_iv_evolution): Likewise.
6127 (vect_analyze_scalar_cycles_1): Likewise.
6128 (vect_analyze_loop_operations): Likewise.
6129 (report_vect_op): Likewise.
6130 (vect_is_slp_reduction): Likewise.
6131 (check_reduction_path): Likewise.
6132 (vect_is_simple_reduction): Likewise.
6133 (vect_create_epilog_for_reduction): Likewise.
6134 (vect_finalize_reduction:): Likewise.
6135 (vectorizable_induction): Likewise.
6136 (vect_transform_loop_stmt): Likewise.
6137 (vect_transform_loop): Likewise.
6138 (optimize_mask_stores): Likewise.
6139 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
6140 (vect_split_statement): Likewise.
6141 (vect_recog_over_widening_pattern): Likewise.
6142 (vect_recog_average_pattern): Likewise.
6143 (vect_determine_min_output_precision_1): Likewise.
6144 (vect_determine_precisions_from_range): Likewise.
6145 (vect_determine_precisions_from_users): Likewise.
6146 (vect_mark_pattern_stmts): Likewise.
6147 (vect_pattern_recog_1): Likewise.
6148 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
6149 (vect_record_max_nunits): Likewise.
6150 (vect_build_slp_tree_1): Likewise.
6151 (vect_build_slp_tree_2): Likewise.
6152 (vect_print_slp_tree): Likewise.
6153 (vect_analyze_slp_instance): Likewise.
6154 (vect_detect_hybrid_slp_stmts): Likewise.
6155 (vect_detect_hybrid_slp_1): Likewise.
6156 (vect_slp_analyze_operations): Likewise.
6157 (vect_slp_analyze_bb_1): Likewise.
6158 (vect_transform_slp_perm_load): Likewise.
6159 (vect_schedule_slp_instance): Likewise.
6160 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
6161 (vect_mark_stmts_to_be_vectorized): Likewise.
6162 (vect_init_vector_1): Likewise.
6163 (vect_get_vec_def_for_operand): Likewise.
6164 (vect_finish_stmt_generation_1): Likewise.
6165 (vect_check_load_store_mask): Likewise.
6166 (vectorizable_call): Likewise.
6167 (vectorizable_conversion): Likewise.
6168 (vectorizable_operation): Likewise.
6169 (vectorizable_load): Likewise.
6170 (vect_analyze_stmt): Likewise.
6171 (vect_is_simple_use): Likewise.
6172 (vect_get_vector_types_for_stmt): Likewise.
6173 (vect_get_mask_type_for_stmt): Likewise.
6174 * tree-vectorizer.c (increase_alignment): Likewise.
6176 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
6178 * doc/rtl.texi: Adjust vec_select description.
6179 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
6180 non-constant selectors.
6182 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
6184 * config/aarch64/aarch64-protos.h
6185 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
6186 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
6187 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
6188 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
6189 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
6190 (AARCH64_FL_PROFILE): Move index so flags are ordered.
6191 (AARCH64_ISA_RCPC8_4): New flag.
6192 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
6193 to aarch64_offset_9bit_signed_unscaled_p.
6194 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
6196 * config/aarch64/constraints.md (Ust): New constraint.
6197 * config/aarch64/predicates.md.
6198 (aarch64_9bit_offset_memory_operand): New predicate.
6199 (aarch64_rcpc_memory_operand): New predicate.
6201 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
6203 PR rtl-optimization/87361
6204 * rtlanal.c (nonzero_bits1): Revert accidental change.
6206 2018-09-19 Richard Biener <rguenther@suse.de>
6208 PR tree-optimization/87349
6209 PR tree-optimization/87342
6210 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
6212 2018-09-18 Marek Polacek <polacek@redhat.com>
6214 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
6215 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
6217 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
6219 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
6221 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
6223 PR rtl-optimization/86882
6224 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
6226 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
6228 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
6229 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
6231 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
6234 * doc/invoke.texi (Link Options): Fix formatting and grammar.
6236 2018-09-18 Richard Biener <rguenther@suse.de>
6239 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
6240 (compute_samebase_partition_bases): Likewise.
6241 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
6242 (gimple_can_coalesce_p): Simplify.
6244 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
6246 Handle a library implementation of ffs calling __builtin_ffs.
6247 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
6248 (mmix_init_libfuncs): New function: make __builtin_ffs expand
6251 2018-09-17 David Malcolm <dmalcolm@redhat.com>
6253 * diagnostic-show-locus.c (class layout_range): Add field
6255 (layout_range::layout_range): Add "original_idx" param and use it
6256 to initialize new field.
6257 (make_range): Use 0 for original_idx.
6258 (layout::layout): Pass in index to calls to
6259 maybe_add_location_range.
6260 (layout::maybe_add_location_range): Add param "original_idx" and
6261 pass it on to layout_range.
6262 (layout::print_any_labels): Pass on range->m_original_idx to
6264 (gcc_rich_location::add_location_if_nearby): Use 0 for
6266 * gcc-rich-location.h (text_range_label::get_text): Update for new
6268 (range_label_for_type_mismatch::get_text): Likewise.
6270 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
6272 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
6274 2018-09-17 David Malcolm <dmalcolm@redhat.com>
6276 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
6277 format_string_diagnostic_t.
6278 (fmtwarn_n): Likewise.
6279 * substring-locations.c
6280 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
6281 (format_warning_n_va): Convert to...
6282 (format_string_diagnostic_t::emit_warning_n_va): ...this.
6283 (format_warning_va): Convert to...
6284 (format_string_diagnostic_t::emit_warning_va): ...this.
6285 (format_warning_at_substring): Convert to...
6286 (format_string_diagnostic_t::emit_warning): ...this.
6287 (format_warning_at_substring_n): Convert to...
6288 (format_string_diagnostic_t::emit_warning_n): ...this.
6289 * substring-locations.h (class format_string_diagnostic_t): New
6291 (format_warning_va): Convert to
6292 format_string_diagnostic_t::emit_warning_va.
6293 (format_warning_n_va): Convert to
6294 format_string_diagnostic_t::emit_warning_n_va.
6295 (format_warning_at_substring): Convert to
6296 format_string_diagnostic_t::emit_warning.
6297 (format_warning_at_substring_n): Convert to
6298 format_string_diagnostic_t::emit_warning_n.
6300 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
6301 Bernd Schmidt <bernds_cb1@t-online.de>
6303 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
6306 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
6308 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
6309 operand 0 predicate to nonimmediate operand.
6310 (rint<mode>2_frndint): Remove insn pattern.
6311 (rint<mode>2): Change operand 1 predicate to general_operand.
6312 Extend operand 1 to XFmode and generate rintxf2 insn.
6313 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
6314 Do not use X87MODEF mode macro.
6315 (frndintxf2_<rounding>_i387): Rename from
6316 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
6317 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
6318 to XFmode and generate significandxf3 insn.
6320 2018-09-17 Richard Biener <rguenther@suse.de>
6322 PR tree-optimization/87328
6323 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
6324 visiting unexecutable backedges when not iterating.
6325 (do_rpo_vn): Mark all edges not executable even when not
6328 2018-09-17 Martin Jambor <mjambor@suse.cz>
6331 * doc/invoke.texi (Warning Options): Likewise.
6333 2018-09-17 Richard Biener <rguenther@suse.de>
6335 PR tree-optimization/87301
6336 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
6337 clean EH info from leftover copy assignments.
6339 2018-09-17 Martin Liska <mliska@suse.cz>
6341 PR gcov-profile/85871
6342 * gcov.c (output_intermediate_file): Fix out of bounds
6345 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
6347 * config/arc/arc.c: Object attributes for core4 not reflected
6349 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
6352 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
6354 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
6356 2018-09-17 Martin Liska <mliska@suse.cz>
6358 * doc/gcov.texi: Document new option --use-hotness-colors.
6359 * gcov.c (struct source_info): Declare new field.
6360 (source_info::source_info): Set default for maximum_count.
6361 (print_usage): Add new -q option.
6362 (process_args): Process it.
6363 (accumulate_line_info): Save src->maximum_count.
6364 (output_line_beginning): Make color line number if
6365 flag_use_hotness_colors is set.
6366 (output_line_details): Pass default argument value.
6367 (output_lines): Pass src->maximum_count.
6369 2018-09-17 Martin Liska <mliska@suse.cz>
6371 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
6372 Use processor_names table.
6373 * config/i386/i386.c (ix86_default_align): Use
6374 processor_cost_table for alignment values.
6375 (ix86_option_override_internal): Use processor_names.
6376 (ix86_function_specific_print): Likewise.
6377 * config/i386/i386.h (struct processor_costs):
6378 Add alignment values.
6379 (struct ptt): Remove and replace with const char *.
6380 * config/i386/x86-tune-costs.h (struct processor_costs):
6381 Declare default alignments for all costs.
6383 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
6385 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
6386 symbolics or VR_VARYING ranges for ABS_EXPR.
6387 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
6388 when range will wrap.
6390 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
6393 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
6394 before and after a JUMP_TABLE_DATA.
6396 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
6399 * dojump.c (do_compare_and_jump): Canonicalize function pointers
6400 when one operand is a function pointer. Use POINTER_TYPE_P and
6401 FUNC_OR_METHOD_TYPE_P.
6402 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
6403 * fold-const.c (build_range_check): Likewise.
6404 * match.pd (simple_comparison): Likewise.
6406 2018-09-14 David Malcolm <dmalcolm@redhat.com>
6409 * spellcheck.c (get_edit_distance_cutoff): New function.
6410 (selftest::test_edit_distance_unit_test_oneway): Rename to...
6411 (selftest::test_get_edit_distance_one_way): ...this.
6412 (selftest::test_get_edit_distance_unit): Rename to...
6413 (selftest::test_get_edit_distance_both_ways): ...this.
6414 (selftest::test_edit_distances): Move tests to this new function,
6415 and test some more pairs of strings. Update for above renaming.
6416 (selftest::get_old_cutoff): New function.
6417 (selftest::test_get_edit_distance_cutoff): New function.
6418 (selftest::assert_suggested_for): New function.
6419 (ASSERT_SUGGESTED_FOR): New macro.
6420 (selftest::assert_not_suggested_for): New function.
6421 (ASSERT_NOT_SUGGESTED_FOR): New macro.
6422 (selftest::test_suggestions): New function.
6423 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
6424 tests to selftest::test_edit_distances and call it. Add calls to
6425 selftest::test_get_edit_distance_cutoff and
6426 selftest::test_suggestions.
6427 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
6428 (best_match::consider): Replace hard-coded cutoff calculation with
6430 (best_match::get_cutoff): New declaration.
6431 (best_match::get_best_meaningful_candidate): Likewise.
6433 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
6435 * builtins.c (fold_builtin_strlen): Remove TODO comment.
6437 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
6440 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
6442 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
6443 terminated string literal.
6445 2018-09-14 Martin Sebor <msebor@redhat.com>
6447 * builtins.c (unterminated_array): Handle ARRAY_REF.
6448 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
6449 * builtins.h (unterminated_array): Declare extern.
6450 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
6452 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
6455 2018-09-14 Martin Sebor <msebor@redhat.com>
6456 Jeff Law <law@redhat.com>
6458 * builtins.c (unterminated_array): New.
6459 (expand_builtin_strcpy): Adjust.
6460 (expand_builtin_strcpy_args): Detect unterminated arrays.
6461 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
6462 unterminated arrays.
6463 * gimple-fold.h (get_maxval_strlen): Add argument.
6464 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
6466 * gimple-fold.c (get_range_strlen): Add argument.
6467 (get_maxval_strlen): Adjust.
6468 * gimple-fold.h (get_range_strlen): Add argument.
6470 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
6472 * config/i386/movdirintrin.h: Fix copyright year.
6474 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
6476 * reg-stack.c: Include regs.h.
6477 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
6478 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
6479 FIRST_STACK_REG, not DFmode.
6480 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
6481 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
6482 (change stack): Default register mode to the reg_raw_mode of
6483 FIRST_STACK_REG, not DFmode.
6484 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
6485 (*swapxf): Rename from swapxf.
6487 2018-09-14 Carl Love <cel@us.ibm.com>
6489 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
6490 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
6492 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
6495 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
6498 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
6501 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
6503 2018-09-14 Jason Merrill <jason@redhat.com>
6505 Fix --enable-gather-detailed-mem-stats.
6506 * hash-table.c (hash_table_usage): Change from variable to function.
6507 * hash-table.h: Adjust.
6508 * Makefile.in: Add missing dependencies on hash-table.h.
6510 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6512 PR tree-optimization/87259
6514 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
6515 execute_cse_reciprocals_1 has tried transforming.
6517 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
6519 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
6520 VR_VARYING for PLUS/MINUS_EXPR.
6522 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
6524 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
6527 2018-09-14 Richard Biener <rguenther@suse.de>
6530 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
6531 bits for the merged partition.
6533 2018-09-13 Martin Sebor <msebor@redhat.com>
6534 Bernd Edlinger <bernd.edlinger@hotmail.de>
6536 * builtins.h (c_srlen): Add argument.
6537 * builtins.c (warn_string_no_nul): New function.
6538 (c_strlen): Add argument and use it. Update recursive calls.
6539 Pass DECL argument to string_constant to get info on non
6540 terminated strings. Update *NONSTR as needed.
6541 (fold_builtin_strlen): Add argument to calls to c_strlen.
6542 Warn for unterminated arrays.
6543 (warn_string_no_null): Add prototype.
6544 * expr.c (string_constant): Update arguments. Update recursive
6545 calls appropriately. Detect missing NUL terminator and outermost
6546 declaration its missing in.
6547 Improve checks for arrays with nonzero lower bound or elements
6548 that are not a single byte. Simplify offset computation.
6549 Simplify checks for non-NUL terminated strings.
6550 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
6551 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
6553 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
6555 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
6557 * fold-const.c (c_getstr): Fix function comment. Remove unused third
6558 argument. Fix range checks.
6559 * fold-const.h (c_getstr): Adjust protoype.
6560 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
6561 string is constant but contains no NUL byte.
6563 * expr.c (string_constant): Adjust function comment.
6564 Remove bogus check for zero termination.
6566 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
6568 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
6569 (get_constant_size): Don't make STRING_CSTs larger than they are.
6570 (check_string_literal): New check function for STRING_CSTs.
6571 (output_constant): Use it.
6573 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
6576 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
6578 2018-09-13 Richard Biener <rguenther@suse.de>
6580 PR tree-optimization/87263
6581 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
6582 (struct unwind_state): Add max_rpo field.
6583 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
6584 Compute max_rpo, the max RPO number a block can be backwards reached
6585 from. Re-write non-iterating mode to a RPO ordered worklist approach,
6586 separating it from the iterating mode.
6588 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
6590 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
6591 (rfs_decision): New scheduling decision.
6593 2018-09-13 Richard Biener <rguenther@suse.de>
6596 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
6597 (vn_nary_op_insert_pieces_predicated): Do not write useless
6598 valid_dominated_by_p entry outside of the allocated storage.
6600 2018-09-13 Omar Sandoval <osandov@osandov.com>
6601 Tom de Vries <tdevries@suse.de>
6604 * dwarf2out.c (is_c): New function.
6605 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
6607 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
6610 * config/aarch64/aarch64.md (*aarch64_bfxil):
6612 * config/aarch64/constraints.md (Ulc): Define.
6613 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
6615 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
6618 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
6620 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
6621 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
6622 aarch64_layout_frame call.
6623 (aarch64_expand_epilogue): Likewise.
6624 (aarch64_initial_elimination_offset): Likewise.
6625 (aarch64_get_separate_components): Likewise.
6626 (aarch64_use_return_insn_p): Likewise.
6627 (aarch64_layout_frame): Remove unneeded check.
6629 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6631 * configure.ac: Only append
6632 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
6633 gcc_config_arguments if it was never reconfigured or last reconfigure
6634 was with different arguments.
6635 * configure: Regenerated.
6637 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6638 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6641 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
6642 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
6644 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6646 PR tree-optimization/87287
6647 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
6648 X % C == 0 to X % (unsigned) C == 0 optimization to ...
6649 * match.pd (X % C == 0): ... here. New optimization.
6651 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6654 * expr.h (maybe_optimize_mod_cmp): Declare.
6655 * expr.c (mod_inv): New function.
6656 (maybe_optimize_mod_cmp): New function.
6657 (do_store_flag): Use it.
6658 * cfgexpand.c (expand_gimple_cond): Likewise.
6660 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
6661 Julian Brown <julian@codesourcery.com>
6664 * gimplify.c (gimplify_scan_omp_clauses): Set
6665 target_firstprivatize_array_bases in OpenACC parallel and kernels
6666 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
6667 OpenACC data regions.
6669 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
6671 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
6672 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
6674 2018-09-12 Richard Biener <rguenther@suse.de>
6676 PR tree-optimization/87280
6677 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
6678 edge but unreachable target.
6679 (do_rpo_vn): For conservatively handling a PHI only mark
6680 the backedge executable but not the block reachable.
6682 2018-09-12 Richard Biener <rguenther@suse.de>
6684 PR tree-optimization/87266
6685 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
6688 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
6690 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
6692 ("trunc<BFP:mode><DFP_ALL:mode>2")
6693 ("trunc<DFP_ALL:mode><BFP:mode>2")
6694 ("extend<BFP:mode><DFP_ALL:mode>2")
6695 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
6696 according to the target operand type.
6698 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6699 Andreas Krebbel <krebbel@linux.ibm.com>
6701 PR tree-optimization/86844
6702 * gimple-ssa-store-merging.c
6703 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
6704 there are any overlapping stores in between them, make sure they are
6705 also coalesced or we give up completely.
6707 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6710 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
6711 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
6713 2018-09-12 Tom de Vries <tdevries@suse.de>
6715 * common.opt (gdescribe-dies): Add option.
6716 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
6717 attribute for artifical and nameless decls.
6718 (dwarf2out_register_external_die): Add description attribute to
6719 external reference die.
6720 (add_desc_attribute): New functions.
6721 (gen_subprogram_die): Add description attribute to
6722 DW_TAG_call_site_parameter.
6723 * tree-pretty-print.c (print_generic_expr_to_str): New function.
6724 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
6725 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
6727 (@item -gdescribe-dies): Add.
6729 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
6731 * tree-vrp.c (vrp_shift_undefined_p): Remove.
6732 (extract_range_from_binary_expr_1: Call
6733 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
6734 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
6737 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
6739 * gimple-ssa-warn-alloca.c
6740 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
6741 field for ALLOCA_BOUND_*_LARGE.
6743 2018-09-11 Nathan Sidwell <nathan@acm.org>
6745 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
6747 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
6749 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
6750 for clobbers. Remove obsolete comment.
6752 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
6754 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
6755 mpxchk, mpxld and mpxst types.
6756 (define_attr length_immediate): Remove all processing of mpx types.
6757 (define_attr prefix_0f): Ditto.
6758 (define_attr memory): Ditto.
6760 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
6762 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
6763 (log<mode>2): Change operand 1 predicate to general_operand.
6764 Extend operand 1 to XFmode and generate logxf3 insn.
6765 (log10<mode>2): Change operand 1 predicate to general_operand.
6766 Extend operand 1 to XFmode and generate log10xf3 insn.
6767 (log2<mode>2): Change operand 1 predicate to general_operand.
6768 Extend operand 1 to XFmode and generate log2xf3 insn.
6769 (fyl2xp1_extend<mode>xf3_i387): Remove.
6770 (log1p<mode>2): Change operand 1 predicate to general_operand.
6771 Extend operand 1 to XFmode and generate log1pxf3 insn.
6772 (fxtract_extend<mode>xf3_i387): Remove.
6773 (logb<mode>2): Change operand 1 predicate to general_operand.
6774 Extend operand 1 to XFmode and generate logbxf3 insn.
6775 (ilogb<mode>2): Change operand 1 predicate to general_operand.
6776 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
6777 (significand<mode>2): Change operand 1 predicate to general_operand.
6778 Extend operand 1 to XFmode and generate significandxf3 insn.
6780 2018-09-11 Nathan Sidwell <nathan@acm.org>
6782 * gcc.c (perror_with_name, pfatal_with_name): Delete.
6783 (load_specs): Use fatal_error.
6784 (DELETE_IF_ORDINARY, process_command): Use error.
6785 (execute, run_attempt): Use fatal_error.
6787 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
6789 * diagnostic-core.h (sorry_at): New prototype.
6790 * diagnostic.c (sorry_at): New function.
6792 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
6794 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
6795 by zero as VR_UNDEFINED.
6797 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
6799 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
6800 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
6801 (<sincos>mode2): New expander.
6802 (sincos_extend<mode>xf3_i387): Remove insn pattern.
6803 (sincos -> sin, cos splitters): Remove splitter patterns.
6804 (sincos<mode>3): Change operand 2 predicate to general_operand.
6805 Extend operand 2 to XFmode and generate sincosxf3 insn.
6806 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
6807 Change operand 3 predicate to const1_operand.
6808 (fptan_extend<mode>xf4_i387): Remove insn pattern.
6809 (tanxf2): Update operands in the call to fptanxf4_i387.
6810 (tan<mode>2): Change operand 1 predicate to general_operand.
6811 Extend operand 1 to XFmode and generate tanxf3 insn.
6812 (atan2xf3): Rename from *fpatanxf3_i387.
6813 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
6814 (atan2xf3): Remove expander.
6815 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
6816 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
6817 (atan<mode>2): Change operand 1 predicate to general_operand.
6818 Extend operand 1 to XFmode and generate atanxf3 insn.
6820 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
6822 * config/i386/i386.md (x87/SSE constant load splitter): Use
6823 memory_operand instead of nonimmediate_operand for input operand
6826 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
6828 * config/i386/i386.md (float partial SSE register stall splitter): Move
6829 splitter near its instruction pattern.
6830 (float_extend partial SSE register stall splitter): Ditto.
6831 (float_truncate partial SSE register stall splitter): Ditto.
6833 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
6836 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
6837 to speculation_safe_value_not_needed.
6840 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
6841 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
6842 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
6843 leaf_function_p, instead use has_hard_reg_initial_val.
6845 2018-09-09 Nathan Sidwell <nathan@acm.org>
6847 * gcc.h (pfatal_with_name): Don't declare here.
6848 * gcc.c (pfatal_with_name): Make static.
6850 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
6852 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
6855 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
6857 PR rtl-optimization/85458
6858 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
6859 priority hook to reduce the priority of EXPR.
6861 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
6863 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
6864 DImode for x87 on 32bit targets. Conditionally disable x87 modes
6865 with X87_ENABLE_FLOAT. Remove preparation code.
6866 (*float<SWI48:mode><MODEF:mode>2): Rename from
6867 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
6868 math using "enabled" attribute.
6869 (*floatdi<MODEF:mode>2_i387): Rename from
6870 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
6871 enable for 32bit targets only.
6872 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
6874 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
6875 as operand 1 predicate. Rewrite as define_insn_and_split.
6876 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
6878 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
6880 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
6881 to fallthru to FLOAT case.
6883 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
6886 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
6887 around folding of vec_sl to handle out of range shift values.
6889 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
6891 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
6892 Update callers to gen_fix_trunc<mode>_i387_fisttp
6893 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
6894 nonimmediate_operand.
6895 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
6896 and corresponding splitters.
6897 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
6898 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
6899 (fix_truncdi_i387_with_temp): Remove insn pattern
6900 and corresponding splitters.
6901 (fix_trunc<mode>_i387): Change operand 0 predicate to
6902 nonimmediate_operand.
6903 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
6904 and corresponding splitters.
6905 (*fistdi2_1): Remove.
6907 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
6908 (lrintxfdi2): Remove expander. Reimplement as define_insn.
6909 (*fist<mode>2_1): Remove.
6910 (fist<mode>2): Ditto.
6911 (fist<mode>2_with_temp): Remove insn pattern and corresponding
6913 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
6914 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
6915 (fistdi2_<rounding>): Change operand 0 predicate to
6916 nonimmediate_operand.
6917 (fistdi2_<rounding>_with_temp): Remove insn pattern
6918 and corresponding splitters.
6919 (fist<mode>2_<rounding>): Change operand 0 predicate to
6920 nonimmediate_operand.
6921 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
6922 and corresponding splitters.
6924 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
6926 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6928 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
6931 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
6933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6934 early gimple folding of vec_splat().
6935 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
6936 * gimple-fold.h: Add an extern define for tree_vec_extract().
6938 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
6940 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
6941 wrappers around TREE_TYPE comparisons.
6943 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
6946 * config/s390/predicates.md: Add nonsym_memory_operand.
6947 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
6948 contains a SYMBOL_REF, load it into an intermediate pseudo.
6949 (s390_emit_compare_and_swap): Legitimize operand.
6950 * config/s390/s390.md: Use the new nonsym_memory_operand
6951 with UNSPECV_CAS patterns.
6953 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
6956 * config/s390/s390-passes.def: New file.
6957 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
6959 (make_pass_s390_early_mach): Add declaration.
6960 * config/s390/s390.c (make_pass_s390_early_mach):
6961 (s390_option_override): Remove dynamic registration.
6962 * config/s390/t-s390: Add s390-passes.def.
6964 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
6966 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6967 Remove UNSPEC_LTREL_BASE check.
6968 (annotate_constant_pool_refs): Likewise.
6969 (find_constant_pool_ref): Likewise.
6970 (find_ltrel_base): Removed.
6971 (replace_ltrel_base): Removed.
6972 (s390_mainpool_finish): Remove replace_ltrel_base call.
6973 (s390_chunkify_start): Remove pending LTREL_BASE logic.
6974 (s390_chunkify_finish): Remove replace_ltrel_base call.
6975 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
6977 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
6980 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
6981 to speculation_safe_value_not_needed.
6983 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
6984 Bernd Schmidt <bernds_cb1@t-online.de>
6986 * config/nvptx/nvptx-opts.h: New file.
6987 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
6988 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
6990 (TARGET_SM35): New macro.
6991 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
6993 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
6995 (misa=): New option.
6996 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
6998 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
7000 * config/i386/i386.md (truncdfsf2): Remove expander.
7001 (truncdfsf2_with_temp): Ditto.
7002 (truncxf<mode>2): Ditto.
7003 (*truncdfsf_fast_mixed): Remove insn pattern.
7004 (*truncdfsf_fast_i387): Ditto.
7005 (*truncdfsf_mixed): Ditto.
7006 (*truncdfsf_i387): Ditto.
7007 (*truncdfsf2_i387_1): Ditto.
7008 (*truncxfsf2_mixed): Ditto.
7009 (*truncxfdf2_mixed): Ditto.
7010 (*truncxf<mode>2_i387_noop): Ditto. Update callers
7011 to call gen_truncxf<mode>2 instead.
7012 (*truncxf<mode>2_i387): Remove.
7013 (reg->reg splitters): Remove splitter pattern.
7014 (reg->mem splitters): Ditto.
7016 (truncdfsf2): New insn pattern.
7017 (truncxf<mode>2): Ditto.
7019 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7021 * tree-ssa-math-opts.c (is_mult_by): New function.
7022 (is_square_of): Use the above.
7023 (optimize_recip_sqrt): New function.
7024 (pass_cse_reciprocals::execute): Use the above.
7026 2018-09-05 Richard Biener <rguenther@suse.de>
7029 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
7030 to zero-init the emplaced vec.
7032 2018-09-05 Martin Liska <mliska@suse.cz>
7034 PR tree-optimization/87205
7035 * tree-switch-conversion.c (pass_lower_switch::execute):
7036 Group cases for switch statements.
7038 2018-09-05 Richard Biener <rguenther@suse.de>
7040 PR tree-optimization/87217
7041 * tree-ssa-sccvn.c (vuse_valueize): New.
7042 (vn_reference_lookup_pieces): Use it.
7043 (vn_reference_lookup): Likewise.
7045 2018-09-05 Nathan Sidwell <nathan@acm.org>
7048 * stor-layout.c (place_field): Scan forwards to check last
7049 bitfield when ms_bitfield_placement is in effect.
7051 2018-09-05 Richard Biener <rguenther@suse.de>
7054 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
7057 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
7058 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
7060 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
7061 * config.gcc (extra_objs): Build it.
7062 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
7064 * config/aarch64/aarch64-passes.def
7065 (pass_tag_collision_avoidance): New pass.
7066 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
7067 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
7068 (aarch64_classify_address): Remove static qualifier.
7069 (aarch64_address_info, aarch64_address_type): Move to...
7070 * config/aarch64/aarch64-protos.h: ... here.
7071 (make_pass_tag_collision_avoidance): New function.
7072 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
7075 2018-09-05 Martin Liska <mliska@suse.cz>
7077 * doc/gcov.texi: Update documentation of humar
7079 * gcov.c (format_count): Print one decimal place, it provides
7080 more fine number of situations like '1G' vs. '1.4G'.
7082 2018-09-05 Martin Liska <mliska@suse.cz>
7085 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
7086 * optc-gen.awk: Allow 'Var' for Deprecated options in order
7087 to generate a MASK value.
7089 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
7092 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
7093 if frame pointer isn't used.
7094 (compute_frame_pointer_to_fb_displacement): Likewise.
7096 2018-09-04 Jakub Jelinek <jakub@redhat.com>
7099 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
7100 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
7101 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
7102 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
7103 and OPTION_MASK_ISA_XSAVEC_UNSET.
7105 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
7107 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
7108 XOR operations in NAND case.
7110 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
7112 * wide-int-range.cc (wide_int_range_convert): New.
7113 * wide-int-range.h (wide_int_range_convert): New.
7114 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
7115 code into wide_int_range_convert.
7116 (extract_range_into_wide_ints): Do not munge anti range constants
7117 into the entire domain. Just return the range back.
7119 2018-09-04 Martin Liska <mliska@suse.cz>
7121 * genmatch.c (output_line_directive): Add new argument
7123 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
7125 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
7127 * doc/invoke.texi (Option Summary): Add whitespace.
7129 * doc/invoke.texi (Option Summary): Add -Waligned-new.
7131 2018-09-04 Richard Biener <rguenther@suse.de>
7133 PR tree-optimization/87211
7134 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
7135 backedge value we're supposed to treat as VARYING also number
7136 the PHI to VARYING in case it got a different value-number already.
7138 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
7140 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
7141 (extract_range_from_binary_expr_1): Do not call
7142 vrp_can_optimize_bit_op.
7143 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
7145 (wide_int_range_get_mask_and_bounds): New.
7146 (wide_int_range_optimize_bit_op): New.
7147 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
7148 (wide_int_range_bit_and): Same.
7149 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
7150 (wide_int_range_optimize_bit_op): New.
7151 (wide_int_range_get_mask_and_bounds): New.
7153 2018-09-04 Richard Biener <rguenther@suse.de>
7155 PR tree-optimization/87176
7156 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
7157 variable. When value-numbering a virtual PHI node make sure
7158 to not value-number to the backedge value.
7160 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
7162 * doc/extend.texi (Long Long, Hex Floats): Document support for
7163 long long and hex floats in more recent versions of ISO C++.
7165 2018-09-03 Richard Biener <rguenther@suse.de>
7167 PR tree-optimization/87177
7168 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
7171 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7173 * bb-reorder.c (edge_order): Convert to C-qsort-style
7174 tri-state comparator.
7175 (reorder_basic_blocks_simple): Change std::stable_sort to
7178 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7180 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
7181 tri-state comparator.
7182 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
7184 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7186 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
7187 (mergesort): ... here as maximum count for using netsort.
7188 (gcc_qsort): Set nlim to 3 if stable sort is requested.
7189 (gcc_stablesort): New.
7190 * system.h (gcc_stablesort): Declare.
7192 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7194 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
7195 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
7196 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
7198 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
7200 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
7201 lxsdx and stxsdx alternatives.
7202 (*mov<mode>_hardfloat64): Ditto.
7203 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
7205 2018-09-03 Richard Biener <rguenther@suse.de>
7207 PR tree-optimization/87200
7208 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
7211 2018-09-03 Martin Liska <mliska@suse.cz>
7213 PR tree-optimization/87201
7214 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
7215 Fix parenthesis in an expression.
7217 2018-09-03 Richard Biener <rguenther@suse.de>
7219 PR tree-optimization/87197
7220 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
7221 visited. CSE the VN_INFO hashtable lookup.
7223 PR tree-optimization/87169
7224 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
7225 iterating make sure there's no extra backedges from irreducible
7226 regions feeding the header. Mark the destination block
7229 2018-09-03 Martin Liska <mliska@suse.cz>
7232 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
7233 * common/common-targhooks.c (default_get_valid_option_values):
7235 * common/common-targhooks.h (default_get_valid_option_values):
7237 * common/config/i386/i386-common.c: Move processor_target_table
7239 (ix86_get_valid_option_values): New function.
7240 (TARGET_GET_VALID_OPTION_VALUES): New macro.
7241 * config/i386/i386.c (struct ptt): Move to i386-common.c.
7242 (PTA_*): Move all defined masks into i386-common.c.
7243 (ix86_function_specific_restore): Use new processor_cost_table.
7244 * config/i386/i386.h (struct ptt): Moved from i386.c.
7245 (struct pta): Likewise.
7246 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
7247 * doc/tm.texi.in: Likewise.
7248 * opt-suggestions.c (option_proposer::suggest_option):
7249 Pass prefix to build_option_suggestions.
7250 (option_proposer::get_completions): Likewise.
7251 (option_proposer::build_option_suggestions): Use the new target
7253 * opts.c (struct option_help_tuple): New struct.
7254 (print_filtered_help): Use the new target hook.
7256 2018-09-03 Martin Liska <mliska@suse.cz>
7259 * predict.c (set_even_probabilities): Add likely_edges
7260 argument and handle cases where we have precisely one
7262 (combine_predictions_for_bb): Catch also likely_edges.
7263 (tree_predict_by_opcode): Handle gswitch statements.
7264 * tree-cfg.h (find_case_label_for_value): New declaration.
7265 (find_taken_edge_switch_expr): Likewise.
7266 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
7267 Find pivot in decision tree based on probabily, not by number of
7270 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
7272 * doc/standards.texi (Standards): Update Objective-C reference.
7274 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7276 * doc/install.texi (Prerequisites): Update link for MPC.
7278 2018-09-01 Michael Matz <matz@suse.de>
7280 PR tree-optimization/87074
7281 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
7282 PHIs for outer-loop uses.
7284 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7286 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
7287 * doc/invoke.texi (C Dialect Options): Ditto.
7289 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7291 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
7293 2018-08-31 Richard Biener <rguenther@suse.de>
7295 PR tree-optimization/87168
7296 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
7297 (rpo_elim::eliminate_avail): When OP was not visited it must
7300 2018-08-31 David Malcolm <dmalcolm@redhat.com>
7302 * tree-vrp.c (copy_value_range): Convert param "from" from
7303 "value_range *" to "const value_range *".
7304 (range_is_null): Likewise for param "vr".
7305 (range_int_cst_p): Likewise.
7306 (range_int_cst_singleton_p): Likewise.
7307 (symbolic_range_p): Likewise.
7308 (value_ranges_intersect_p): Likewise for both params.
7309 (value_range_nonnegative_p): Likewise for param "vr".
7310 (value_range_constant_singleton): Likewise.
7311 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
7312 (extract_range_into_wide_ints): Likewise for param "vr".
7313 (extract_range_from_multiplicative_op): Likewise for params "vr0"
7315 (vrp_can_optimize_bit_op): Likewise.
7316 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
7318 (extract_range_from_unary_expr): Likewise.
7319 (debug_value_range): Likewise for param "vr".
7320 (value_range::dump): Add "const" qualifier.
7321 (vrp_prop::check_array_ref): Convert local "vr" from
7322 "value_range *" to "const value_range *".
7323 (vrp_prop::check_mem_ref): Likewise.
7324 (vrp_prop::visit_stmt): Likewise for local "old_vr".
7325 (vrp_intersect_ranges_1): Likewise for param "vr_1".
7326 (vrp_intersect_ranges): Likewise.
7327 (simplify_stmt_for_jump_threading): Likewise for local "vr".
7328 (vrp_prop::vrp_finalize): Likewise.
7329 * tree-vrp.h (value_range::dump): Add "const" qualifier.
7330 (vrp_intersect_ranges): Add "const" qualifier to params as above.
7331 (extract_range_from_unary_expr): Likewise.
7332 (value_range_constant_singleton): Likewise.
7333 (symbolic_range_p): Likewise.
7334 (copy_value_range): Likewise.
7335 (extract_range_from_binary_expr_1): Likewise.
7336 (range_int_cst_p): Likewise.
7337 (vrp_set_zero_nonzero_bits): Likewise.
7338 (range_int_cst_singleton_p): Likewise.
7340 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
7342 * config/aarch64/arm_neon.h (vabsd_s64): New.
7343 (vnegd_s64): Likewise.
7345 2018-08-31 Martin Jambor <mjambor@suse.cz>
7347 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
7349 2018-08-31 Martin Liska <mliska@suse.cz>
7351 * ipa-icf.c (sem_item::add_type): Use
7352 sem_item::m_type_hash_cache.
7353 * ipa-icf.h: Move the cache from sem_item_optimizer
7356 2018-08-31 Nathan Sidwell <nathan@acm.org>
7358 * doc/extend.texi (Backwards Compatibility): Remove implicit
7359 extern C leeway of () being (...).
7361 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7363 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
7365 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
7369 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
7371 2018-08-31 Jakub Jelinek <jakub@redhat.com>
7374 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
7375 gen_int_mode. Formatting fixes.
7377 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
7379 * target.def (custom_function_descriptors): Improve documentation.
7380 * doc/tm.texi.in (Trampolines): Expand discussion of function
7381 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
7382 beginning of the section.
7383 * doc/tm.texi: Regenerated.
7385 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
7387 * cfg.h (class auto_edge_flag): Spell out the template-id of the
7388 base class in the initializer list. This is a workaround for
7389 building with older GCC.
7390 (class auto_bb_flag): Likewise.
7392 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
7394 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
7395 (altivec_vcmpequ<VI_char>_p): Remove star.
7396 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
7398 (expand_strncmp_vec_sequence): New function.
7399 (emit_final_str_compare_vec): New function.
7400 (expand_strn_compare): Add support for vector strncmp.
7401 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
7402 length specification to bytes.
7403 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
7404 (vcmpnezb_p): New pattern.
7405 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
7406 for option -mstring-compare-inline-limit.
7408 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
7410 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
7411 (PTA_SKYLAKE): Add PTA_AES.
7412 (PTA_GOLDMONT): Likewise.
7414 2018-08-29 Jan Hubicka <jh@suse.cz>
7417 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
7418 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
7420 2018-08-29 Jan Hubicka <jh@suse.cz>
7422 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
7424 (hash_tree): Do not visit TYPE_STUB_DECL.
7425 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
7426 stream TYPE_STUB_DECL.
7427 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
7428 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
7429 after free_lang_data.
7430 (type_in_anonymous_namespace_p): Likewise.
7432 2018-08-29 Jan Hubicka <jh@suse.cz>
7434 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
7435 comment that it has to be even number.
7436 (class sreal): Change m_sig type to int32_t.
7437 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
7438 int64_t for temporary calculations.
7439 (sreal_verify_basics): Drop one bit from minimum and maximum.
7441 2018-08-30 Richard Biener <rguenther@suse.de>
7443 PR tree-optimization/87147
7444 * tree-ssa-sccvn.c (SSA_VISITED): New function.
7445 (visit_phi): When the degenerate result is from the backedge and
7446 we didn't visit its definition yet drop to VARYING.
7447 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
7449 2018-08-29 Jan Hubicka <jh@suse.cz>
7451 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
7453 (hash_tree): Likewise.
7455 2018-08-29 Jan Hubicka <jh@suse.cz>
7457 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
7458 and TYPE_NEXT_REF_TO.
7460 2018-08-29 Jan Hubicka <jh@suse.cz>
7462 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
7463 comment that it has to be even number.
7464 (class sreal): Change m_sig type to int32_t.
7465 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
7466 int64_t for temporary calculations.
7467 (sreal_verify_basics): Drop one bit from minimum and maximum.
7469 2018-08-30 Tamar Christina <tamar.christina@arm.com>
7471 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
7473 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
7476 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
7477 if to_add is negative.
7479 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
7482 * builtins.c (c_strlen): Improve range checks.
7484 2018-08-29 Martin Sebor <msebor@redhat.com>
7485 Jeff Law <law@redhat.com>
7487 PR tree-optimization/86714
7488 PR tree-optimization/86711
7489 * builtins.c (c_strlen): Add arguments to call to string_constant.
7490 * expr.c (string_constant): Add argument. Detect missing nul
7491 terminator and outermost declaration it's missing in.
7492 * expr.h (string_constant): Add argument.
7493 * fold-const.c (read_from_constant_string): Add arguments to call to
7495 (c_getstr): Likewise.
7496 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
7498 * tree-ssa-strlen.c (get_stridx): Likewise.
7500 2018-08-29 Jan Hubicka <jh@suse.cz>
7502 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
7503 Do not stream DECL_VINDEX.
7504 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
7505 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
7506 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
7508 2018-08-29 Richard Biener <rguenther@suse.de>
7510 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
7511 virtual operands that are not default defs to honor region
7513 (rpo_vn_valueize): Remove ineffective code here.
7515 2018-08-29 Richard Biener <rguenther@suse.de>
7517 PR tree-optimization/87132
7518 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
7519 when skipping defs reachable over backedges.
7521 2018-08-29 Richard Biener <rguenther@suse.de>
7523 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
7524 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
7525 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
7526 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
7527 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
7528 (vn_reference_lookup_3): Remove use of const_parms.
7529 (free_rpo_vn): Do not free const_parms.
7530 (do_rpo_vn): Do not call init_const_parms.
7531 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
7532 SSA_NAME_POINTS_TO_READONLY_MEMORY.
7533 (call_may_clobber_ref_p_1): Likewise.
7535 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
7538 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
7540 (-ftree-scev-cprop): Document.
7542 2018-08-29 Jan Hubicka <jh@suse.cz>
7544 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
7546 (sreal constructor): Update.
7547 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
7548 sreal:operator/): Update.
7550 2018-08-29 Martin Liska <mliska@suse.cz>
7552 * tree-switch-conversion.c (switch_conversion::expand):
7553 Strenghten assumption about gswitch statements.
7555 2018-08-29 Richard Biener <rguenther@suse.de>
7557 PR tree-optimization/87117
7558 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
7559 re-value-number released SSA VDEFs.
7561 2018-08-29 Richard Biener <rguenther@suse.de>
7563 PR tree-optimization/87126
7564 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
7566 2018-08-28 Jim Wilson <jimw@sifive.com>
7568 * config/riscv/pic.md: Rewrite.
7569 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
7571 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
7572 (SOFTF, default_load, softload, softstore): New.
7574 2018-08-28 Jeff Law <law@redhat.com>
7576 * fold-const.c (fold_binary_loc): Remove recently added assert.
7578 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
7580 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
7581 to OP parmeter of generated function.
7583 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
7585 PR tree-optimization/87009
7586 * match.pd: Add boolean optimizations.
7588 2018-08-28 Martin Sebor <msebor@redhat.com>
7591 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
7592 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
7593 (pass_walloca::gate): Use it.
7594 (alloca_call_type): Same.
7595 (pass_walloca::execute): Same.
7596 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
7598 2018-08-28 David Malcolm <dmalcolm@redhat.com>
7600 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
7601 GCC_VERSION for usage of "__gcc_dump_printf__" format from
7604 2018-08-28 Richard Biener <rguenther@suse.de>
7606 PR tree-optimization/87124
7607 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
7608 constants before looking up avail.
7610 2018-08-28 Jakub Jelinek <jakub@redhat.com>
7613 * calls.c (maybe_warn_nonstring_arg): Punt early if
7614 warn_stringop_overflow is zero. Don't call get_range_strlen
7615 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
7616 Swap comparison operands to have constants on rhs. Only use
7617 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
7618 increment lenrng[0].
7620 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
7622 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
7623 use of tree_to_shwi. Remove duplicated test for the size being
7624 a whole number of bytes.
7626 2018-08-28 Richard Biener <rguenther@suse.de>
7628 PR tree-optimization/87117
7629 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
7630 Handle removed stmt without LHS (GIMPLE_NOP).
7632 2018-08-28 Richard Biener <rguenther@suse.de>
7634 PR tree-optimization/87117
7635 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
7636 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
7638 2018-08-28 Richard Biener <rguenther@suse.de>
7640 PR tree-optimization/87117
7641 * tree-ssa-pre.c (compute_avail): Do not make expressions
7642 with predicated values available.
7643 (get_expr_value_id): Assert we do not run into predicated value
7646 2018-08-28 Richard Biener <rguenther@suse.de>
7648 PR tree-optimization/87117
7649 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
7650 get virtual operands.
7651 (get_expr_operands): Handle STRING_CST like other decls.
7653 2018-08-28 Martin Liska <mliska@suse.cz>
7655 * tree.h: Update documentation of fndecl_built_in_p
7659 2018-08-27 Jeff Law <law@redhat.com>
7660 PR tree-optimization/87110
7661 * tree-ssa-dse.c (compute_trims): Handle non-constant
7664 2018-08-27 Martin Sebor <msebor@redhat.com>
7666 PR tree-optimization/86914
7667 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
7669 2018-08-27 Martin Sebor <msebor@redhat.com>
7671 PR tree-optimization/87112
7672 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
7673 the type of the bound argument.
7675 2018-08-27 Jeff Law <law@redhat.com>
7677 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
7678 type does not have a TYPE_SIZE_UNIT.
7680 2018-08-27 Steve Ellcey <sellcey@cavium.com>
7682 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
7683 with include of backend.h.
7685 2018-08-27 Richard Biener <rguenther@suse.de>
7687 PR tree-optimization/86927
7688 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
7689 use const cond reduction code.
7691 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
7693 PR tree-optimization/85758
7694 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
7696 2018-08-27 David Malcolm <dmalcolm@redhat.com>
7699 * diagnostic-show-locus.c (class layout_range): Update for
7700 conversion of show_caret_p to a tri-state.
7701 (layout_range::layout_range): Likewise.
7702 (make_range): Likewise.
7703 (layout::maybe_add_location_range): Likewise.
7704 (layout::should_print_annotation_line_p): Don't show annotation
7705 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
7706 (layout::get_state_at_point): Update for conversion of
7707 show_caret_p to a tri-state. Bail out early for
7708 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
7709 underlining or source colorization.
7710 (gcc_rich_location::add_location_if_nearby): Update for conversion
7711 of show_caret_p to a tri-state.
7712 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
7713 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
7715 (selftest::test_one_liner_labels): Likewise.
7716 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
7717 conversion of show_caret_p to a tri-state.
7718 * pretty-print.c (text_info::set_location): Likewise.
7719 * pretty-print.h (text_info::set_location): Likewise.
7720 * substring-locations.c (format_warning_n_va): Likewise.
7721 * tree-diagnostic.c (default_tree_printer): Likewise.
7722 * tree-pretty-print.c (newline_and_indent): Likewise.
7724 2018-08-27 David Malcolm <dmalcolm@redhat.com>
7727 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
7728 line above for line-insertion fix-it hints.
7729 (selftest::test_fixit_insert_containing_newline): Update the
7730 expected results, and add a test with line-numbering enabled.
7732 2018-08-27 Martin Liska <mliska@suse.cz>
7735 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
7736 params with DECL_HAS_VALUE_EXPR_P.
7738 2018-08-27 Martin Liska <mliska@suse.cz>
7740 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
7741 selected expansion strategy.
7743 2018-08-27 Martin Liska <mliska@suse.cz>
7745 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
7746 * builtins.c (is_builtin_fn): Likewise.
7747 * attribs.c (diag_attr_exclusions): Use new function
7748 fndecl_built_in_p and remove check for FUNCTION_DECL if
7750 (builtin_mathfn_code): Likewise.
7751 (fold_builtin_expect): Likewise.
7752 (fold_call_expr): Likewise.
7753 (fold_builtin_call_array): Likewise.
7754 (fold_call_stmt): Likewise.
7755 (set_builtin_user_assembler_name): Likewise.
7756 (is_simple_builtin): Likewise.
7757 * calls.c (gimple_alloca_call_p): Likewise.
7758 (maybe_warn_nonstring_arg): Likewise.
7759 * cfgexpand.c (expand_call_stmt): Likewise.
7760 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
7761 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
7762 (cgraph_node::verify_node): Likewise.
7763 * cgraphclones.c (build_function_decl_skip_args): Likewise.
7764 (cgraph_node::create_clone): Likewise.
7765 * config/arm/arm.c (arm_insert_attributes): Likewise.
7766 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
7767 * dse.c (scan_insn): Likewise.
7768 * expr.c (expand_expr_real_1): Likewise.
7769 * fold-const.c (operand_equal_p): Likewise.
7770 (fold_binary_loc): Likewise.
7771 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7772 * gimple-low.c (lower_stmt): Likewise.
7773 * gimple-pretty-print.c (dump_gimple_call): Likewise.
7774 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
7776 * gimple.c (gimple_build_call_from_tree): Likewise.
7777 (gimple_call_builtin_p): Likewise.
7778 (gimple_call_combined_fn): Likewise.
7779 * gimplify.c (gimplify_call_expr): Likewise.
7780 (gimple_boolify): Likewise.
7781 (gimplify_modify_expr): Likewise.
7782 (gimplify_addr_expr): Likewise.
7783 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
7784 * ipa-cp.c (determine_versionability): Likewise.
7785 * ipa-fnsummary.c (compute_fn_summary): Likewise.
7786 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
7787 * ipa-split.c (visit_bb): Likewise.
7788 (split_function): Likewise.
7789 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
7790 * lto-cgraph.c (input_node): Likewise.
7791 * lto-streamer-out.c (write_symbol): Likewise.
7792 * omp-low.c (setjmp_or_longjmp_p): Likewise.
7793 (lower_omp_1): Likewise.
7794 * predict.c (strip_predict_hints): Likewise.
7795 * print-tree.c (print_node): Likewise.
7796 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
7797 * trans-mem.c (is_tm_irrevocable): Likewise.
7798 (is_tm_load): Likewise.
7799 (is_tm_simple_load): Likewise.
7800 (is_tm_store): Likewise.
7801 (is_tm_simple_store): Likewise.
7802 (is_tm_abort): Likewise.
7803 (tm_region_init_1): Likewise.
7804 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
7805 * tree-cfg.c (verify_gimple_call): Likewise.
7806 (move_stmt_r): Likewise.
7807 (stmt_can_terminate_bb_p): Likewise.
7808 * tree-eh.c (lower_eh_constructs_2): Likewise.
7809 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
7810 * tree-inline.c (remap_gimple_stmt): Likewise.
7811 (copy_bb): Likewise.
7812 (estimate_num_insns): Likewise.
7813 (fold_marked_statements): Likewise.
7814 * tree-sra.c (scan_function): Likewise.
7815 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
7816 (optimize_stack_restore): Likewise.
7817 (pass_fold_builtins::execute): Likewise.
7818 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7819 (mark_all_reaching_defs_necessary_1): Likewise.
7820 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
7821 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
7822 (pass_forwprop::execute): Likewise.
7823 * tree-ssa-loop-im.c (stmt_cost): Likewise.
7824 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7825 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
7826 * tree-ssa-strlen.c (get_string_length): Likewise.
7827 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
7828 (find_func_aliases_for_call): Likewise.
7829 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
7830 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
7831 * tree-tailcall.c (find_tail_calls): Likewise.
7832 * tree.c (need_assembler_name_p): Likewise.
7833 (free_lang_data_in_decl): Likewise.
7834 (get_call_combined_fn): Likewise.
7835 * ubsan.c (is_ubsan_builtin_p): Likewise.
7836 * varasm.c (incorporeal_function_p): Likewise.
7837 * tree.h (DECL_BUILT_IN): Remove and replace with
7839 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
7840 (fndecl_built_in_p): New.
7842 2018-08-27 Martin Liska <mliska@suse.cz>
7844 PR tree-optimization/86847
7845 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
7846 Dump also subtree probability.
7847 (switch_decision_tree::do_jump_if_equal): New function.
7848 (switch_decision_tree::emit_case_nodes): Handle special
7849 situations in balanced tree that can be emitted much simpler.
7850 Fix calculation of probabilities that happen in tree expansion.
7851 * tree-switch-conversion.h (struct cluster): Add
7853 (struct simple_cluster): Likewise.
7854 (struct case_tree_node): Add new function has_child.
7855 (do_jump_if_equal): New.
7857 2018-08-27 Martin Liska <mliska@suse.cz>
7859 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
7860 Add new argument to bit_test_cluster constructor.
7861 (bit_test_cluster::emit): Set bits really number of values
7863 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
7864 probability argument.
7865 * tree-switch-conversion.h (struct bit_test_cluster):
7866 Add m_handles_entire_switch.
7868 2018-08-27 Martin Liska <mliska@suse.cz>
7870 PR tree-optimization/86702
7871 * tree-switch-conversion.c (jump_table_cluster::emit):
7872 Make probabilities even for values in jump table
7873 according to number of cases handled.
7874 (switch_decision_tree::compute_cases_per_edge): Pass
7875 argument to reset_out_edges_aux function.
7876 (switch_decision_tree::analyze_switch_statement): Likewise.
7877 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
7880 2018-08-27 Martin Liska <mliska@suse.cz>
7882 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
7883 cfun argument explicitly.
7884 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
7885 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
7886 function gimple_switch_default_bb.
7887 (convert_switch_statements):
7889 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
7890 * stmt.c (label_to_block_fn): Use label_to_block and pass
7891 cfun argument explicitly and use gimple_switch_label_bb.
7892 (expand_case): Likewise.
7893 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
7894 cfun argument explicitly. Likewise.
7895 (make_edges_bb): Likewise.
7896 (make_cond_expr_edges): Likewise.
7897 (get_cases_for_edge): Likewise.
7898 (make_gimple_switch_edges): Likewise.
7899 (label_to_block_fn): Likewise.
7900 (label_to_block): Likewise.
7901 (make_goto_expr_edges): Likewise.
7902 (make_gimple_asm_edges): Likewise.
7903 (main_block_label): Likewise.
7904 (group_case_labels_stmt): Likewise.
7905 (find_taken_edge_computed_goto): Likewise.
7906 (find_taken_edge_switch_expr): Likewise.
7907 (gimple_verify_flow_info): Likewise.
7908 (gimple_redirect_edge_and_branch): Likewise.
7909 (gimple_switch_label_bb): New function.
7910 (gimple_switch_default_bb): Likewise.
7911 (gimple_switch_edge): Likewise.
7912 (gimple_switch_default_edge): Likewise.
7913 * tree-cfg.h (label_to_block_fn): Remove and replace ...
7914 (label_to_block): ... with this.
7915 (gimple_switch_label_bb): New.
7916 (gimple_switch_default_bb): Likewise.
7917 (gimple_switch_edge): Likewise.
7918 (gimple_switch_default_edge): Likewise.
7919 * tree-cfgcleanup.c (convert_single_case_switch): Use
7920 new gimple functions and pass new argument to label_to_block.
7921 (cleanup_control_flow_bb):
7922 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
7923 cfun argument explicitly.
7924 (make_eh_edges): Likewise.
7925 (redirect_eh_dispatch_edge): Likewise.
7926 (lower_resx): Likewise.
7927 (lower_eh_dispatch): Likewise.
7928 (maybe_remove_unreachable_handlers): Likewise.
7929 (unsplit_eh): Likewise.
7930 (cleanup_empty_eh): Likewise.
7931 (verify_eh_edges): Likewise.
7932 (verify_eh_dispatch_edge): Likewise.
7933 * tree-ssa-dom.c (record_edge_info): Likewise.
7934 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
7935 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
7936 (thread_through_normal_block): Likewise.
7937 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
7938 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
7939 * tree-switch-conversion.c (switch_conversion::collect): Use new
7941 (switch_conversion::check_final_bb): Likewise.
7942 (switch_conversion::gather_default_values): Pass new argument
7944 (switch_conversion::build_constructors): Likewise.
7945 (switch_decision_tree::compute_cases_per_edge): Use new
7946 gimple_switch_edge function.
7947 (switch_decision_tree::analyze_switch_statement): Pass new argument
7949 (switch_decision_tree::try_switch_expansion): Use
7950 gimple_switch_default_edge.
7951 * tree-vrp.c (find_switch_asserts): Pass new argument
7953 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
7954 (vr_values::simplify_switch_using_ranges): Likewise.
7956 2018-08-27 Richard Biener <rguenther@suse.de>
7958 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
7959 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
7961 * tree-ssa-sccvn.h (struct vn_pval): New structure.
7962 (struct vn_nary_op_s): Add unwind_to member. Add
7963 predicated_values flag and put result into a union together
7964 with a linked list of vn_pval.
7965 (struct vn_ssa_aux): Add name member to make maintaining
7966 a map of SSA name to vn_ssa_aux possible. Remove no longer
7967 needed info, dfsnum, low, visited, on_sccstack, use_processed
7968 and range_info_anti_range_p members.
7969 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
7970 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
7972 (vn_valueize): New global.
7973 (vn_context_bb): Likewise.
7974 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
7975 VN_INFO_PTR_INFO): Remove.
7976 * tree-ssa-sccvn.c: ... (rewrite)
7977 (pass_fre::execute): For -O2+ initialize loops and run
7978 RPO VN in optimistic mode (iterating). For -O1 and -Og
7979 run RPO VN in non-optimistic mode.
7980 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
7981 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
7982 * doc/invoke.texi (sccvn-max-scc-size): Remove.
7983 (rpo-vn-max-loop-depth): Document.
7984 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
7985 when valuezing the VUSE signals we walked out of the region.
7986 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
7987 (phi_translate): Set VN context block to use for availability
7989 (compute_avail): Likewise.
7990 (pre_valueize): New function.
7991 (pass_pre::execute): Adjust to the RPO VN API.
7993 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
7994 (propagate_constants_for_unrolling): Remove.
7995 (tree_unroll_loops_completely): Perform value-numbering
7996 on the unrolled bodies loop parent.
7998 2018-08-27 Richard Biener <rguenther@suse.de>
8000 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
8001 for partial antic compute.
8003 2018-08-27 Jakub Jelinek <jakub@redhat.com>
8005 PR rtl-optimization/87065
8006 * combine.c (simplify_if_then_else): Formatting fix.
8007 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
8009 (known_cond): Don't return const_true_rtx for vector modes. Use
8010 CONST0_RTX instead of const0_rtx. Formatting fixes.
8012 2018-08-27 Martin Liska <mliska@suse.cz>
8014 PR gcov-profile/87069
8015 * gcov.c (process_file): Record files already processed
8016 and warn about a file being processed multiple times.
8018 2018-08-27 Martin Liska <mliska@suse.cz>
8021 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8022 Set default values for x_aarch64_*_string strings.
8023 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
8024 prefix. For -mabi do not print '=ABI' in help and use
8025 <option_value> format for -msve-vector-bits and -moverride
8028 2018-08-26 Jeff Law <law@redhat.com>
8030 * config/mips/frame-header-opt.c: Include "backend.h" rather than
8033 2018-08-26 Marek Polacek <polacek@redhat.com>
8035 PR c++/87029, Implement -Wredundant-move.
8036 * doc/invoke.texi: Document -Wredundant-move.
8038 2018-08-25 Martin Sebor <msebor@redhat.com>
8040 PR tree-optimization/87059
8041 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
8042 to the same type as the other.
8043 * fold-const.c (fold_binary_loc): Assert expectation.
8045 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
8047 * config/darwin.c (machopic_legitimize_pic_address): Clean up
8048 extraneous parentheses, dead code section and formatting.
8050 2018-08-24 David Malcolm <dmalcolm@redhat.com>
8053 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
8054 wide enough for jumps in the line-numbering to be visible.
8055 (layout::print_gap_in_line_numbering): New member function.
8056 (layout::calculate_line_spans): When using line numbering, merge
8057 line spans that are only 1 line apart.
8058 (diagnostic_show_locus): When printing line numbers, show gaps in
8059 line numbering directly, rather than printing headers.
8060 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
8061 line-numbering with multiple line spans.
8062 (selftest::test_fixit_insert_containing_newline_2): Add test of
8063 line-numbering, in which the spans are close enough to be merged.
8065 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
8067 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
8068 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
8069 * tree-vrp.c (range_is_nonnull): Remove.
8070 (range_includes_zero_p): Accept value_range instead of min/max.
8071 (extract_range_from_binary_expr_1): Do not early bail on
8073 Use range_includes_zero_p instead of range_is_nonnull.
8074 (extract_range_from_unary_expr): Use range_includes_zero_p instead
8075 of range_is_nonnull.
8076 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
8077 special case VR_ANTI_RANGE.
8078 (vrp_finalize): Same.
8079 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
8081 (range_is_nonnull): Remove.
8082 * vr-values.c (vrp_stmt_computes_nonzero): Use
8083 range_includes_zero_p instead of range_is_nonnull.
8084 (extract_range_basic): Pass value_range to range_includes_zero_p
8085 instead of range_is_nonnull.
8087 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
8089 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
8090 * emit-rtl.h (rtl_data): Remove return_bnd.
8091 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
8092 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
8093 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
8094 (POINTER_BOUNDS_MODE): Remove definition.
8095 (make_pointer_bounds_mode): Remove.
8096 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
8097 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
8098 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
8099 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
8100 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
8101 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
8102 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
8104 * config/i386/i386-modes.def (BND32, BND64): Remove.
8105 * config/i386/i386.c (dbx_register_map): Remove bound registers.
8106 (dbx64_register_map): Ditto.
8107 (svr4_dbx_register_map): Ditto.
8108 (indirect_thunk_bnd_needed): Remove.
8109 (indirect_thunks_bnd_used): Ditto.
8110 (indirect_return_bnd_needed): Ditto.
8111 (indirect_return_via_cx_bnd): Ditto.
8112 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
8113 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
8114 (output_indirect_thunk): Ditto. Remove need_prefix argument.
8115 (output_indirect_thunk_function): Remove handling of
8116 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
8117 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
8118 (ix86_save_reg): Remove handling of crtl->return_bnd.
8119 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
8120 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
8121 and UNSPEC_BNDLX_ADDR.
8122 (ix86_output_indirect_branch_via_reg): Remove handling of
8123 indirect_thunk_prefix_bnd.
8124 (ix86_output_indirect_branch_via_push): Ditto.
8125 (ix86_output_function_return): Ditto.
8126 (ix86_output_indirect_function_return): Ditto.
8127 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
8128 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
8129 (CALL_USED_REGISTERS): Ditto.
8130 (REG_ALLOC_ORDER): Update for removal of bound registers.
8131 (HI_REGISTER_NAMES): Ditto.
8132 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
8133 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
8134 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
8135 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
8136 (FIRST_PSEUDO_REG): Update.
8137 (BND): Remove mode iterator.
8138 * config/i386/predicates.md (bnd_mem_operator): Remove.
8140 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
8142 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
8145 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
8147 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
8148 the case in which the permute needs only a single element and
8149 repeats for every vector of the result. Extend that case to
8150 handle variable-length vectors.
8151 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8153 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
8156 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
8157 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
8159 2018-08-24 Richard Biener <rguenther@suse.de>
8161 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
8162 bb_flags_allocated members.
8163 (auto_flag): New RAII class for allocating flags.
8164 (auto_edge_flag): New RAII class for allocating edge flags.
8165 (auto_bb_flag): New RAII class for allocating bb flags.
8166 * cfgloop.c (verify_loop_structure): Allocate temporary edge
8168 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
8169 in favor of temporarily allocated BB flag.
8170 * hsa-brig.c: Re-order includes.
8171 * hsa-dump.c: Likewise.
8172 * hsa-regalloc.c: Likewise.
8173 * print-rtl.c: Likewise.
8174 * profile-count.c: Likewise.
8176 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
8179 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
8182 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
8185 * wide-int-range.cc (wide_int_range_div): Do not ignore result
8186 from wide_int_range_multiplicative_op.
8188 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8190 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
8193 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
8195 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
8198 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
8200 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
8201 full GENERIC expression used for replacement.
8203 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
8205 * tree-vrp.c (abs_extent_range): Remove.
8206 (extract_range_into_wide_ints): Pass wide ints by reference.
8207 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
8208 Pass wide ints by reference in all calls to
8209 extract_range_into_wide_ints.
8210 * wide-int-range.cc (wide_int_range_div): New.
8211 * wide-int-range.h (wide_int_range_div): New.
8212 (wide_int_range_includes_zero_p): New.
8213 (wide_int_range_zero_p): New.
8215 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
8217 * config/aarch64/aarch64.md (arches): New enum.
8218 (arch): New enum attr.
8219 (arch_enabled): New attr.
8220 (enabled): Now uses arch_enabled only.
8221 (simd, sve, fp16): Removed attribute.
8222 (fp): Attr now defined in terms of 'arch'.
8223 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
8224 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
8225 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
8226 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
8227 attributes into 'arch'.
8228 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
8229 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
8230 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
8231 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
8232 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
8233 'simd' attribute into 'arch'.
8234 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
8235 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
8236 Convert use of 'fp' attribute to 'arch'.
8237 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
8238 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
8240 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
8241 (different modes) Merge 'fp' and 'simd' into 'arch'.
8242 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
8245 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
8247 PR rtl-optimization/87026
8248 * expmed.c (canonicalize_comparison): If we can no longer create
8249 pseudoregisters, don't.
8251 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
8254 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
8256 * config/arm/arm.c (speculation_barrier_libfunc): New static
8258 (arm_init_libfuncs): Initialize it.
8259 (arm_emit_speculation_barrier): New function.
8260 * config/arm/arm.md (speculation_barrier): Call
8261 arm_emit_speculation_barrier for architectures that do not have
8263 (speculation_barrier_insn): Only match on Armv7 or later.
8265 2018-08-23 Richard Biener <rguenther@suse.de>
8268 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
8271 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
8273 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
8274 of single-vector TBLs.
8275 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
8278 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
8281 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
8282 aarch64_evpc_tbl guard.
8284 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
8286 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
8289 2018-08-22 Martin Sebor <msebor@redhat.com>
8292 * tree-pretty-print.c (pretty_print_string): Add argument.
8293 (dump_generic_node): Call to pretty_print_string with string size.
8295 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
8297 PR rtl-optimization/86771
8298 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
8299 of two SETs into those two SETs, one to be placed at i2, if that SETs
8300 destination is modified between i2 and i3.
8302 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8304 PR tree-optimization/86725
8305 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
8307 (vect_analyze_scalar_cycles_1): Check it.
8309 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8311 PR tree-optimization/86725
8312 * tree-vect-loop.c (vect_is_simple_reduction): When treating
8313 an outer loop phi as a double reduction, make sure that the
8314 single user of the phi result is an inner loop phi.
8316 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8318 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
8319 grouped stores with gaps to a strided group.
8321 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8323 * tree-vect-stmts.c (get_group_load_store_type)
8324 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
8325 first statement in a group.
8327 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8329 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
8330 the sequence used in gcc/gcc.c.
8332 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8335 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
8338 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8340 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
8341 Darwin10-specific unwinder-shim.
8342 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
8343 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
8344 New to cater for Darwin10 Rosetta.
8346 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8348 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
8349 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
8351 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8356 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
8357 (dwarf2out_switch_text_section): Generate a local label for the
8358 second function sub-section and apply it as the second FDE start
8360 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
8361 second sub-section start.
8363 2018-08-22 Richard Biener <rguenther@suse.de>
8365 PR tree-optimization/86988
8366 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
8368 2018-08-22 Richard Biener <rguenther@suse.de>
8370 PR tree-optimization/86945
8371 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
8373 2018-08-22 Alexandre Oliva <oliva@adacore.com>
8375 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
8376 a comment about how uses of r2 for .sdata2 come about.
8378 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
8380 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
8382 2018-08-21 Marek Polacek <polacek@redhat.com>
8384 PR c++/86981, Implement -Wpessimizing-move.
8385 * doc/invoke.texi: Document -Wpessimizing-move.
8387 2018-08-21 Jan Hubicka <jh@suse.cz>
8389 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
8390 * tree.h (is_redundant_typedef): Remove.
8391 * dwarf2out.c (is_redundant_typedef): Turn into static function.
8393 2018-08-21 Jan Hubicka <jh@suse.cz>
8395 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
8398 2018-08-21 Tamar Christina <tamar.christina@arm.com>
8400 * expmed.c (extract_low_bits): Reject invalid subregs early.
8402 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
8405 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
8408 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8410 * config/vxworks.h: Guard vxworks_asm_out_constructor and
8411 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
8412 * config/vxworks.c: Likewise.
8414 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8416 * config/vxworks.c: Set targetm.have_ctors_dtors
8417 if HAVE_INITFINI_ARRAY_SUPPORT.
8418 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
8419 if HAVE_INITFINI_ARRAY_SUPPORT.
8421 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8423 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
8424 default search path for VxWorks < 7.
8426 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
8428 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
8429 (get_format_string): Refer to c_getstr.
8431 2018-08-21 Tom de Vries <tdevries@suse.de>
8433 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
8434 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
8435 (debuginfo_early_stop): Declare.
8436 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
8437 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
8438 (debuginfo_early_stop): New function.
8439 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
8440 and debuginfo_early_stop.
8441 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
8442 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
8443 (general_init): Call debuginfo_early_init.
8444 (finalize): Call debuginfo_fini.
8445 (do_compile): Call debuginfo_init.
8446 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
8448 (@item -fdump-debug, @item -fdump-earlydebug): Add.
8450 2018-08-21 Tom de Vries <tdevries@suse.de>
8452 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
8453 flag_dump_noaddr and flag_dump_unnumbered.
8455 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
8457 * wide-int-range.cc (wide_int_range_abs): New.
8458 (wide_int_range_order_set): Rename from wide_int_range_min_max.
8459 * wide-int-range.h (wide_int_range_abs): New.
8460 (wide_int_range_min_max): New.
8461 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
8462 case to call wide_int_range_abs.
8463 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
8464 (extract_range_from_abs_expr): Delete.
8466 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
8469 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
8470 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
8473 2018-08-20 Nathan Sidwell <nathan@acm.org>
8474 Jeff Law <law@redhat.com>
8476 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
8477 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
8479 2018-08-20 David Malcolm <dmalcolm@redhat.com>
8482 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
8483 (decl_attributes): Likewise.
8484 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
8486 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
8487 * diagnostic-core.h (class auto_diagnostic_group): New class.
8488 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
8489 (diagnostic_report_diagnostic): Handle the first diagnostics within
8491 (emit_diagnostic): Add auto_diagnostic_group instance.
8493 (inform_n): Likewise.
8494 (warning): Likewise.
8495 (warning_at): Likewise.
8496 (warning_n): Likewise.
8497 (pedwarn): Likewise.
8498 (permerror): Likewise.
8500 (error_n): Likewise.
8501 (error_at): Likewise.
8503 (fatal_error): Likewise.
8504 (internal_error): Likewise.
8505 (internal_error_no_backtrace): Likewise.
8506 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
8507 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
8508 * diagnostic.h (struct diagnostic_context): Add fields
8509 "diagnostic_group_nesting_depth",
8510 "diagnostic_group_emission_count", "begin_group_cb",
8512 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
8513 Add auto_diagnostic_group instance(s).
8514 (find_explicit_erroneous_behavior): Likewise.
8515 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
8516 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
8517 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
8518 (gimplify_va_arg_expr): Likewise.
8519 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
8520 (HSA_SORRY_AT): Likewise.
8521 * ipa-devirt.c (compare_virtual_tables): Likewise.
8522 (warn_odr): Likewise.
8523 * multiple_target.c (expand_target_clones): Likewise.
8524 * opts-common.c (cmdline_handle_error): Likewise.
8525 * reginfo.c (globalize_reg): Likewise.
8526 * substring-locations.c (format_warning_n_va): Likewise.
8527 * tree-inline.c (expand_call_inline): Likewise.
8528 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
8529 * tree-ssa-loop-niter.c
8530 (do_warn_aggressive_loop_optimizations): Likewise.
8531 * tree-ssa-uninit.c (warn_uninit): Likewise.
8532 * tree.c (warn_deprecated_use): Likewise.
8534 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
8537 * config/i386/i386.md (eh_return): Always update EH return
8538 address in word_mode.
8540 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
8542 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
8543 TARGET_SPLIT_COMPLEX_ARG is defined.
8545 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
8547 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
8549 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
8552 * expr.c (expand_assignment): Assert that bitpos is positive.
8553 (store_field): Likewise
8554 (expand_expr_real_1): Make sure that bitpos is positive.
8555 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
8558 2018-08-20 Nathan Sidwell <nathan@acm.org>
8560 * Makefile.in (CPP_ID_DATA_H): Delete.
8561 (CPP_INTERNAL_H): Don't add it.
8562 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
8563 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
8565 2018-08-20 Richard Biener <rguenther@suse.de>
8567 PR tree-optimization/78655
8568 * tree-vrp.c (extract_range_from_binary_expr_1): Make
8569 pointer + offset nonnull if either operand is nonnull work.
8571 2018-08-20 Tom de Vries <tdevries@suse.de>
8573 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
8574 unless the referenced die describes the added property using
8575 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
8576 Otherwise, add a DW_AT_location to the referenced die.
8578 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
8581 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
8582 register_operand when calling ix86_set_reg_reg_cost.
8583 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
8584 Set *total to 0 for operands that satisfy x86_64_immediate_operand
8585 predicate and to 1 otherwise.
8587 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
8589 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
8590 emit a diagnostic that it is not supported and reset the option.
8591 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
8592 supported and consume the option. (ASM_FINAL_SPEC): New.
8594 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
8596 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
8599 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
8601 C-SKY port: Documentation
8603 * doc/extend.texi (C-SKY Function Attributes): New section.
8604 * doc/invoke.texi (Option Summary): Add C-SKY options.
8605 (C-SKY Options): New section.
8606 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
8608 2018-08-17 Jojo <jijie_rong@c-sky.com>
8609 Huibin Wang <huibin_wang@c-sky.com>
8610 Sandra Loosemore <sandra@codesourcery.com>
8611 Chung-Lin Tang <cltang@codesourcery.com>
8613 C-SKY port: Backend implementation
8615 * config/csky/*: New.
8616 * common/config/csky/*: New.
8618 2018-08-17 Jojo <jijie_rong@c-sky.com>
8619 Huibin Wang <huibin_wang@c-sky.com>
8620 Sandra Loosemore <sandra@codesourcery.com>
8621 Chung-Lin Tang <cltang@codesourcery.com>
8622 Andrew Jenner <andrew@codesourcery.com>
8624 C-SKY port: Configury
8626 * config.gcc (csky-*-*): New.
8627 * configure.ac: Add csky to targets for dwarf2 debug_line support.
8628 * configure: Regenerated.
8630 2018-08-17 David Malcolm <dmalcolm@redhat.com>
8632 * dump-context.h: Include "dumpfile.h".
8633 (dump_context::dump_printf_va): Convert final param from va_list
8634 to va_list *. Convert from ATTRIBUTE_PRINTF to
8635 ATTRIBUTE_GCC_DUMP_PRINTF.
8636 (dump_context::dump_printf_loc_va): Likewise.
8637 * dumpfile.c: Include "stringpool.h".
8638 (make_item_for_dump_printf_va): Delete.
8639 (make_item_for_dump_printf): Delete.
8640 (class dump_pretty_printer): New class.
8641 (dump_pretty_printer::dump_pretty_printer): New ctor.
8642 (dump_pretty_printer::emit_items): New member function.
8643 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
8645 (dump_pretty_printer::emit_item): New member function.
8646 (dump_pretty_printer::stash_item): New member function.
8647 (dump_pretty_printer::format_decoder_cb): New member function.
8648 (dump_pretty_printer::decode_format): New member function.
8649 (dump_context::dump_printf_va): Reimplement in terms of
8650 dump_pretty_printer.
8651 (dump_context::dump_printf_loc_va): Convert final param from va_list
8653 (dump_context::begin_scope): Reimplement call to
8654 make_item_for_dump_printf.
8655 (dump_printf): Update for change to dump_printf_va.
8656 (dump_printf_loc): Likewise.
8657 (selftest::test_capture_of_dump_calls): Convert "stmt" from
8658 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
8659 with %T, %E, and %G.
8660 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
8661 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
8662 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
8663 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
8664 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
8665 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
8666 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
8667 within a dump_printf_loc call to "%wu".
8668 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
8669 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
8670 missing space after "=".
8671 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
8672 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
8673 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
8674 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
8675 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
8676 duplicate "vectorized" from message.
8678 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
8680 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
8681 polyNxK_t element's TYPE_STRING_FLAG.
8683 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
8685 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
8686 (they were unnamed before). Fix comments.
8688 2018-08-17 Nathan Sidwell <nathan@acm.org>
8690 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
8692 2018-08-17 Richard Biener <rguenther@suse.de>
8694 PR tree-optimization/86841
8695 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
8697 2018-08-17 Martin Liska <mliska@suse.cz>
8699 * common.opt: Remove Warn, Init and Report for options with
8700 Ignore/Deprecated flag. Warning is done automatically for
8702 * config/i386/i386.opt: Likewise.
8703 * config/ia64/ia64.opt: Likewise.
8704 * config/rs6000/rs6000.opt: Likewise.
8705 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
8706 Remove usage of flag_check_pointer_bounds.
8707 * lto-wrapper.c (merge_and_complain): Do not handle
8708 OPT_fcheck_pointer_bounds.
8709 (append_compiler_options): Likewise.
8710 * opt-functions.awk: Do not handle Deprecated.
8711 * optc-gen.awk: Check that Var, Report and Init are not
8712 used for an option with Ignore/Deprecated flag.
8713 * opts-common.c (decode_cmdline_option): Do not report
8715 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
8717 * opts.h (struct cl_option): Remove cl_deprecated flag.
8718 (CL_ERR_DEPRECATED): Remove error enum value.
8720 2018-08-17 Richard Biener <rguenther@suse.de>
8723 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
8724 across a va-arg-pack using call adjust its return value accordingly.
8726 2018-08-16 Martin Sebor <msebor@redhat.com>
8728 PR tree-optimization/86853
8729 * gimple-ssa-sprintf.c (struct format_result): Rename member.
8730 (struct fmtresult): Add member and initialize it in ctors.
8731 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
8732 (format_string): Handle %S the same as %ls. Set MAYFAIL.
8733 (format_directive): Set POSUNDER4K when MAYFAIL is set.
8734 (parse_directive): Handle %C same as %c.
8735 (sprintf_dom_walker::compute_format_length): Adjust.
8736 (is_call_safe): Adjust.
8738 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
8740 * builtins.c (c_strlen): Add new parameter eltsize. Use it
8741 for determining how to count the elements.
8742 * builtins.h (c_strlen): Adjust prototype.
8743 * expr.c (string_constant): Add new parameter mem_size.
8744 Set *mem_size appropriately.
8745 * expr.h (string_constant): Adjust protoype.
8746 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
8747 * gimple-fold.h (get_range_strlen): Adjust prototype.
8748 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
8749 (format_string): Call get_string_length with eltsize.
8751 2018-08-16 David Malcolm <dmalcolm@redhat.com>
8753 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
8754 to emit the span, rather than setting it as the prefix.
8756 2018-08-16 David Malcolm <dmalcolm@redhat.com>
8758 * diagnostic-show-locus.c (layout::start_annotation_line): Add
8759 "margin_char" parameter, defaulting to space. Use it in place
8760 of pp_space for the initial part of the margin.
8761 (layout::print_leading_fixits): Use '+' when filling the margin
8762 of line-insertion fix-it hints.
8764 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8766 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
8769 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8771 * config/rs6000/altivec.md: Don't set length attribute to the default
8773 * config/rs6000/darwin.md: Ditto.
8774 * config/rs6000/dfp.md: Ditto.
8775 * config/rs6000/htm.md: Ditto.
8776 * config/rs6000/rs6000.md: Ditto.
8777 * config/rs6000/sync.md: Ditto.
8778 * config/rs6000/vsx.md: Ditto.
8780 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8782 * config/rs6000/altivec.md: Don't set length attribute to the default
8783 value, for branch instructions.
8784 * config/rs6000/darwin.md: Ditto.
8785 * config/rs6000/rs6000.md: Ditto.
8787 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8789 * config/rs6000/rs6000.md (length): Always define as const_int 4.
8790 (unnamed conditional branch define_insn): Set length to 4 or 8
8791 depending on offset.
8792 (<bd>_<mode>): Similar, for alternative 0.
8793 (<bd>tf_<mode>): Ditto.
8795 2018-08-16 Tamar Christina <tamar.christina@arm.com>
8797 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
8799 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
8801 * doc/rtl.texi: Replace old RTX class names with new names.
8804 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
8806 * expmed.h (canonicalize_comparison): New declaration.
8807 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
8808 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
8809 * optabs.c (prepare_cmp_insn): Likewise.
8810 * rtl.h (unsigned_condition_p): New function which checks if a
8811 comparison operator is unsigned.
8813 2018-08-16 Nathan Sidwell <nathan@acm.org>
8815 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
8816 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
8818 2018-08-16 Tamar Christina <tamar.christina@arm.com>
8821 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
8822 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
8823 (mov<mov>): ..this and enable unconditionally.
8825 2018-08-16 Tamar Christina <tamar.christina@arm.com>
8827 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
8829 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
8831 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
8832 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
8833 "Common" with "Target".
8835 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
8837 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
8838 * doc/invoke.texi (mmitigate-rop): Remove.
8839 * config/i386/i386.c: Do not include "regrename.h".
8840 (ix86_rop_should_change_byte_p, reg_encoded_number)
8841 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
8843 (ix86_reorg): Remove call to ix86_mitigate_rop.
8844 * config/i386/i386.md (attr "modrm_class"): Remove.
8845 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
8846 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
8847 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
8849 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
8851 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
8852 allow folding of mergeh() and mergel() for the float and double types.
8853 (fold_mergehl_helper): Rework to handle building a permute tree
8856 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
8858 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
8861 2018-08-15 David Malcolm <dmalcolm@redhat.com>
8863 * common.opt (fdiagnostics-show-labels): New option.
8864 * diagnostic-show-locus.c (class layout_range): Add field
8866 (class layout): Add field "m_show_labels_p".
8867 (layout_range::layout_range): Add param "label" and use it to
8869 (make_range): Pass in NULL for new "label" param of layout_range's
8871 (layout::layout): Initialize m_show_labels_p.
8872 (layout::maybe_add_location_range): Pass in loc_range->m_label
8873 when constructing layout_range instances.
8874 (struct line_label): New struct.
8875 (layout::print_any_labels): New member function.
8876 (layout::print_line): Call it if label-printing is enabled.
8877 (selftest::test_one_liner_labels): New test.
8878 (selftest::test_diagnostic_show_locus_one_liner): Call it.
8879 * diagnostic.c (diagnostic_initialize): Initialize
8880 context->show_labels_p.
8881 * diagnostic.h (struct diagnostic_context): Add field
8883 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
8884 -fno-diagnostics-show-labels.
8885 * dwarf2out.c (gen_producer_string): Add
8886 OPT_fdiagnostics_show_labels to the ignored options.
8887 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
8889 (gcc_rich_location::maybe_add_expr): Likewise.
8890 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
8891 label" param, defaulting to NULL.
8892 (gcc_rich_location::add_expr): Add "label" param.
8893 (gcc_rich_location::maybe_add_expr): Likewise.
8894 (class text_range_label): New class.
8895 (class range_label_for_type_mismatch): New class.
8896 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
8897 of format_warning_va.
8898 (fmtwarn_n): Likewise for new params of format_warning_n_va.
8899 * lto-wrapper.c (merge_and_complain): Add
8900 OPT_fdiagnostics_show_labels to the "pick one setting" options.
8901 (append_compiler_options): Likewise to the dropped options.
8902 (append_diag_options): Likewise to the passed-on options.
8903 * opts.c (common_handle_option): Handle the new option.
8904 * selftest-diagnostic.c
8905 (test_diagnostic_context::test_diagnostic_context): Enable
8907 * substring-locations.c: Include "gcc-rich-location.h".
8908 (format_warning_n_va): Add "fmt_label" and "param_label" params
8909 and use them as appropriate.
8910 (format_warning_va): Add "fmt_label" and "param_label" params,
8911 passing them on to format_warning_n_va.
8912 (format_warning_at_substring): Likewise.
8913 (format_warning_at_substring_n): Likewise.
8914 * substring-locations.h (format_warning_va): Add "fmt_label" and
8915 "param_label" params.
8916 (format_warning_n_va): Likewise.
8917 (format_warning_at_substring): Likewise.
8918 (format_warning_at_substring_n): Likewise.
8919 * toplev.c (general_init): Initialize global_dc->show_labels_p.
8921 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
8924 * builtins.c (expand_builtin_memcmp): Do not expand the call
8925 when overflow is detected.
8927 2018-08-15 Martin Sebor <msebor@redhat.com>
8929 PR tree-optimization/71625
8930 * config/aarch64/aarch64-builtins.c
8931 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
8933 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
8935 * config/s390/s390.c (s390_reorg): Remove loop.
8937 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
8940 (darwin_function_switched_text_sections): Delete.
8941 * gcc/config/darwin.h
8942 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
8944 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
8947 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
8948 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
8949 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
8951 2018-08-15 Martin Liska <mliska@suse.cz>
8953 PR tree-optimization/86925
8954 * predict.c (expr_expected_value_1): When taking
8955 later predictor, assign also probability.
8956 Use fold_build2_initializer_loc in order to fold
8957 the expression in -frounding-math.
8959 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
8961 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
8963 (expand_vec_perm_1): Try the new method.
8965 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
8968 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
8969 Check whether lra_live_max_point is 0 before dividing.
8971 2018-08-14 Martin Sebor <msebor@redhat.com>
8973 PR tree-optimization/86650
8974 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
8975 (vrp_prop::check_mem_ref): Same.
8977 2018-08-13 Liu Hao <lh_mouse@126.com>
8979 * pretty-print.c (eat_esc_sequence): Swap the foreground and
8980 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
8981 and clear it thereafter, as it only works for DBCS.
8983 2018-08-13 Liu Hao <lh_mouse@126.com>
8985 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
8986 handle returned by _get_osf_handle().
8988 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8990 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
8991 for folding vec_perm.
8993 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8995 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
8996 Add support for gimple-folding of vec_pack() and vec_unpack()
8999 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9001 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
9002 vec_xst variants to the list.
9003 (rs6000_gimple_fold_builtin): Add support for folding unaligned
9004 vector loads and stores.
9006 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
9008 * config.gcc (rs6000-ibm-aix4.x): Delete.
9009 (rs6000-ibm-aix5.1): Delete.
9010 (rs6000-ibm-aix5.2): Delete.
9011 (rs6000-ibm-aix5.3): Delete.
9012 * config/rs6000/aix43.h: Delete.
9013 * config/rs6000/aix51.h: Delete.
9014 * config/rs6000/aix52.h: Delete.
9015 * config/rs6000/t-aix43: Delete.
9017 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
9019 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9021 (s390_decompose_address): Factor out constant pool ref
9024 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9026 * config/nds32/nds32-predicates.c
9027 (nds32_can_use_bclr_p): Change return type as bool.
9028 (nds32_can_use_bset_p): Ditto.
9029 (nds32_can_use_btgl_p): Ditto.
9030 (nds32_can_use_bitci_p): Ditto.
9031 * config/nds32/nds32-protos.h
9032 (nds32_can_use_bclr_p): Change declaration.
9033 (nds32_can_use_bset_p): Ditto.
9034 (nds32_can_use_btgl_p): Ditto.
9035 (nds32_can_use_bitci_p): Ditto.
9037 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9039 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
9040 Support -msched-prolog-epilog option.
9041 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
9043 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9045 * common/config/nds32/nds32-common.c
9046 (nds32_option_optimization_table): Enalbe -malways-align.
9048 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9050 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
9052 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
9053 OPT_misr_secure_ case.
9054 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
9055 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
9056 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
9058 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
9059 (nds32_isr_info): New field security_level.
9060 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
9061 * config/nds32/nds32.md (return_internal): Consider critical attribute.
9062 * config/nds32/nds32.opt (misr-secure): New option.
9063 * config/nds32/nds32_init.inc: New file.
9064 * config/nds32/nds32_isr.h: New file.
9066 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
9068 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
9069 Update comment for atomic instructions.
9070 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
9071 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
9073 (atomic_loaddi): Revise fence expansion to only emit fence prior to
9074 load for __ATOMIC_SEQ_CST model.
9075 (atomic_loaddi_1): Remove float register target.
9076 (atomic_storedi): Handle CONST_INT values.
9077 (atomic_storedi_1): Remove float register source. Add special case
9079 (memory_barrier): New expander and insn.
9081 2018-08-11 Jakub Jelinek <jakub@redhat.com>
9083 PR tree-optimization/86835
9084 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
9085 new_stmt after def_gsi, make sure to insert new_square_stmt after
9086 that stmt, not 2 stmts before it.
9088 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
9091 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
9094 2018-08-10 Martin Liska <mliska@suse.cz>
9097 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
9099 * builtins.c (expand_builtin_expect_with_probability):
9101 (expand_builtin_expect_with_probability): New function.
9102 (build_builtin_expect_predicate): Add new argumnet probability
9103 for BUILT_IN_EXPECT_WITH_PROBABILITY.
9104 (fold_builtin_expect):
9107 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
9108 * builtins.h (fold_builtin_expect): Set new argument.
9109 * doc/extend.texi: Document __builtin_expect_with_probability.
9110 * doc/invoke.texi: Likewise.
9111 * gimple-fold.c (gimple_fold_call): Pass new argument.
9112 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
9113 also BUILT_IN_EXPECT_WITH_PROBABILITY.
9114 * predict.c (get_predictor_value): New function.
9115 (expr_expected_value): Add new argument probability. Assume
9116 that predictor and probability are always non-null.
9117 (expr_expected_value_1): Likewise. For __builtin_expect and
9118 __builtin_expect_with_probability set probability. Handle
9119 combination in binary expressions.
9120 (tree_predict_by_opcode): Simplify code by simply calling
9121 get_predictor_value.
9122 (pass_strip_predict_hints::execute): Add handling of
9123 BUILT_IN_EXPECT_WITH_PROBABILITY.
9124 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
9126 * tree.h (DECL_BUILT_IN_P): New function.
9128 2018-08-10 Martin Liska <mliska@suse.cz>
9130 PR tree-optimization/85799
9131 * passes.def: Add argument for pass_strip_predict_hints.
9132 * predict.c (class pass_strip_predict_hints): Add new argument
9134 (strip_predictor_early): New function.
9135 (pass_strip_predict_hints::execute): Call the function to
9137 (strip_predict_hints): New function.
9138 * predict.def: Fix comment.
9140 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
9142 * Makefile.in: Clarify which tm.texi to copy over to assert the
9143 right to grant a GFDL license for all.
9145 2018-08-09 Jeff Law <law@redhat.com>
9147 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
9150 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
9152 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
9155 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9157 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
9158 reductions for variable-length vectors.
9160 2018-08-09 David Malcolm <dmalcolm@redhat.com>
9163 * common.opt (fdiagnostics-show-line-numbers): New option.
9164 * diagnostic-show-locus.c (class layout): Add fields
9165 "m_show_line_numbers_p" and "m_linenum_width";
9166 (num_digits): New function.
9167 (test_num_digits): New function.
9168 (layout::layout): Initialize new fields. Update m_x_offset
9169 logic to handle any left margin.
9170 (layout::print_source_line): Print line number when requested.
9171 (layout::start_annotation_line): New member function.
9172 (layout::print_annotation_line): Call it.
9173 (layout::print_leading_fixits): Likewise.
9174 (layout::print_trailing_fixits): Likewise. Update calls to
9175 move_to_column for new parameter.
9176 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
9177 it to potentially call start_annotation_line.
9178 (layout::show_ruler): Call start_annotation_line.
9179 (selftest::test_line_numbers_multiline_range): New selftest.
9180 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
9181 and selftest::test_line_numbers_multiline_range.
9182 * diagnostic.c (diagnostic_initialize): Initialize
9183 show_line_numbers_p.
9184 * diagnostic.h (struct diagnostic_context): Add field
9185 "show_line_numbers_p".
9186 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
9187 -fno-diagnostics-show-line-numbers.
9188 * dwarf2out.c (gen_producer_string): Add
9189 OPT_fdiagnostics_show_line_numbers to the ignored options.
9190 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
9191 one setting" options.
9192 (append_compiler_options): Likewise to the dropped options.
9193 (append_diag_options): Likewise to the passed-on options.
9194 * opts.c (common_handle_option): Handle the new option.
9195 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
9197 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
9199 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
9200 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
9201 third argument of type "const signed char" to descriptions of
9202 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
9203 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
9204 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
9205 __builtin_bcdsub_ov functions.
9207 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9209 PR tree-optimization/86858
9210 * tree-vect-loop.c (vect_is_simple_reduction): Restore
9211 flow_bb_inside_loop_p calls.
9213 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9215 PR tree-optimization/86871
9216 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
9217 instead of gimple_assign_lhs.
9219 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
9222 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
9223 register constraint to operand 0.
9224 (add<mode>3_carryinC): Likewise.
9225 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
9227 2018-08-09 Martin Liska <mliska@suse.cz>
9230 * common.opt: Remove extra line.
9232 2018-08-09 Martin Liska <mliska@suse.cz>
9234 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
9235 at the end of a line, make first letter capital and end up
9236 a sentence with a dot.
9237 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
9238 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
9239 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
9240 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
9241 (PARAM_MAX_ISL_OPERATIONS): Likewise.
9242 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
9243 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
9244 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
9245 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
9246 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
9247 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
9248 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
9249 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
9250 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
9251 (PARAM_TREE_REASSOC_WIDTH): Likewise.
9252 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
9253 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
9254 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
9256 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
9259 * config/s390/s390.c (s390_option_override_internal): Reduce the
9260 stack-clash-protection-probe-interval param if it would be too big
9263 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
9266 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
9267 * config/m68k/m68k.c (handle_move_double): Don't call
9268 m68k_final_prescan_insn.
9269 (m68k_adjust_decorated_operand): Renamed from
9270 m68k_final_prescan_insn, remove first and third operand and
9272 (print_operand): Call it.
9273 (print_operand_address): Call it.
9275 2018-08-08 Nathan Sidwell <nathan@acm.org>
9277 * diagnostic.c (diagnostic_report_current_module): Use
9278 linemap_included_from & linemap_included_from_linemap.
9280 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
9282 * config/aarch64/aarch64-cores.def: Add phecda core.
9283 * config/aarch64/aarch64-tune.md: Regenerate.
9284 * doc/invoke.texi: Add phecda core.
9286 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
9289 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
9291 * config/s390/s390.md ("movti"): Add more alternatives for
9292 constant to GPR copies.
9294 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
9296 * config/s390/s390.c: Fix whitespace damage throughout the file.
9297 * config/s390/s390.h: Likewise.
9298 * config/s390/tpf.h: Likewise.
9300 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
9302 * config/s390/s390.c (s390_loadrelative_operand_p):
9303 Remove TARGET_CPU_ZARCH usages.
9304 (s390_rtx_costs): Likewise.
9305 (s390_legitimate_constant_p): Likewise.
9306 (s390_cannot_force_const_mem): Likewise.
9307 (legitimate_reload_constant_p): Likewise.
9308 (s390_preferred_reload_class): Likewise.
9309 (legitimize_pic_address): Likewise.
9310 (legitimize_tls_address): Likewise.
9311 (s390_split_branches): Removed.
9312 (s390_add_execute): Removed.
9313 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
9314 (s390_mainpool_start): Likewise.
9315 (s390_mainpool_finish): Likewise.
9316 (s390_mainpool_cancel): Removed.
9317 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
9318 (s390_chunkify_cancel): Likewise.
9319 (s390_return_addr_rtx): Likewise.
9320 (s390_register_info): Remove split_branches_pending_p uages.
9321 (s390_optimize_register_info): Likewise.
9322 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
9323 split_branches_pending_p usages.
9324 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
9325 (s390_load_got): Likewise.
9326 (s390_expand_split_stack_prologue): Likewise.
9327 (output_asm_nops): Likewise.
9328 (s390_function_profiler): Likewise.
9329 (s390_emit_call): Likewise.
9330 (s390_conditional_register_usage): Likewise.
9331 (s390_optimize_prologue): Likewise.
9332 (s390_reorg): Remove TARGET_CPU_ZARCH and
9333 split_branches_pending_p usages.
9334 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
9336 (s390_output_indirect_thunk_function): Likewise.
9337 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
9338 (TARGET_CPU_ZARCH_P): Removed.
9339 (struct machine_function): Remove split_branches_pending_p.
9340 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
9342 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
9344 * common/config/s390/s390-common.c (processor_flags_table):
9346 * config.gcc: Remove with_arch/with_tune support.
9347 * config/s390/2064.md: Remove cpu attribute comparisons.
9348 * config/s390/driver-native.c (s390_host_detect_local_cpu):
9350 * config/s390/linux.h (ASM_SPEC):
9351 Remove -march support.
9352 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
9353 Use a table to get an arch level.
9354 * config/s390/s390-opts.h (enum processor_type):
9356 * config/s390/s390.c
9357 (processor_table): Remove entries, add arch_level values.
9358 (s390_issue_rate): Remove cases.
9359 (s390_option_override): Adjust
9360 s390_option_override_internal() call.
9361 (s390_option_override_internal): Remove deprecation warning.
9362 (s390_valid_target_attribute_tree): Adjust
9363 s390_option_override_internal() call.
9364 * config/s390/s390.h (struct s390_processor):
9365 Share with s390-c.c, add arch_level field.
9366 * config/s390/s390.md:
9367 Remove occurrences in cpu attribute.
9368 * config/s390/s390.opt: Remove -march/-mtune support.
9369 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
9370 * doc/invoke.texi: Remove deprecation warning.
9372 2018-08-08 Luis Machado <luis.machado@linaro.org>
9374 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
9376 (qdf24xx_tunings): Set vector cost structure to
9377 qdf24xx_vector_cost.
9379 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
9380 <register_sextend>: Set to 3.
9382 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
9385 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
9386 * config/aarch64/aarch64-simd.md
9387 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
9388 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
9389 (aarch64_frecpx<mode>): ...this new pattern.
9390 * config/aarch64/aarch64-simd-builtins.def: Remove comment
9391 about aarch64_frecp<FRECP:frecp_suffix><mode>.
9393 2018-08-07 Martin Liska <mliska@suse.cz>
9396 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
9397 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
9398 * predict.def (PRED_MALLOC_NONNULL): New predictor.
9399 * doc/extend.texi: Document that malloc attribute adds
9402 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
9405 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9406 Define to speculation_safe_value_not_needed.
9408 2018-08-06 Jeff Law <law@redhat.com>
9410 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
9411 the vr_values instance to cprop_into_stmt.
9412 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
9413 (cprop_operand): Also query EVRP to determine if OP is a constant.
9415 2018-08-06 Nathan Sidwell <nathan@acm.org>
9417 * diagnostic.c (diagnostic_report_current_module): Reroll
9418 included-at loop. Translate text.
9420 2018-08-06 David Malcolm <dmalcolm@redhat.com>
9422 * function-tests.c (selftest::test_expansion_to_rtl): Call
9423 free_after_compilation.
9425 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9427 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
9429 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
9431 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
9432 loops with memory block operations from getting unrolled.
9434 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
9437 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9438 Define to speculation_safe_value_not_needed.
9440 2018-08-06 Jeff Law <law@redhat.com>
9442 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
9445 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9448 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
9449 with all enabled __intN types.
9451 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
9453 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9455 * alias.c (record_set): Check for clobber high.
9456 * cfgexpand.c (expand_gimple_stmt): Likewise.
9457 * combine-stack-adj.c (single_set_for_csa): Likewise.
9458 * combine.c (find_single_use_1): Likewise.
9459 (set_nonzero_bits_and_sign_copies): Likewise.
9460 (get_combine_src_dest): Likewise.
9461 (is_parallel_of_n_reg_sets): Likewise.
9462 (try_combine): Likewise.
9463 (record_dead_and_set_regs_1): Likewise.
9464 (reg_dead_at_p_1): Likewise.
9465 (reg_dead_at_p): Likewise.
9466 * dce.c (deletable_insn_p): Likewise.
9467 (mark_nonreg_stores_1): Likewise.
9468 (mark_nonreg_stores_2): Likewise.
9469 * df-scan.c (df_find_hard_reg_defs): Likewise.
9470 (df_uses_record): Likewise.
9471 (df_get_call_refs): Likewise.
9472 * dwarf2out.c (mem_loc_descriptor): Likewise.
9473 * haifa-sched.c (haifa_classify_rtx): Likewise.
9474 * ira-build.c (create_insn_allocnos): Likewise.
9475 * ira-costs.c (scan_one_insn): Likewise.
9476 * ira.c (equiv_init_movable_p): Likewise.
9477 (rtx_moveable_p): Likewise.
9478 (interesting_dest_for_shprep): Likewise.
9479 * jump.c (mark_jump_label_1): Likewise.
9480 * postreload-gcse.c (record_opr_changes): Likewise.
9481 * postreload.c (reload_cse_simplify): Likewise.
9482 (struct reg_use): Add source expr.
9483 (reload_combine): Check for clobber high.
9484 (reload_combine_note_use): Likewise.
9485 (reload_cse_move2add): Likewise.
9486 (move2add_note_store): Likewise.
9487 * print-rtl.c (print_pattern): Likewise.
9488 * recog.c (decode_asm_operands): Likewise.
9489 (store_data_bypass_p): Likewise.
9490 (if_test_bypass_p): Likewise.
9491 * regcprop.c (kill_clobbered_value): Likewise.
9492 (kill_set_value): Likewise.
9493 * reginfo.c (reg_scan_mark_refs): Likewise.
9494 * reload1.c (maybe_fix_stack_asms): Likewise.
9495 (eliminate_regs_1): Likewise.
9496 (elimination_effects): Likewise.
9497 (mark_not_eliminable): Likewise.
9498 (scan_paradoxical_subregs): Likewise.
9499 (forget_old_reloads_1): Likewise.
9500 * reorg.c (find_end_label): Likewise.
9501 (try_merge_delay_insns): Likewise.
9502 (redundant_insn): Likewise.
9503 (own_thread_p): Likewise.
9504 (fill_simple_delay_slots): Likewise.
9505 (fill_slots_from_thread): Likewise.
9506 (dbr_schedule): Likewise.
9507 * resource.c (update_live_status): Likewise.
9508 (mark_referenced_resources): Likewise.
9509 (mark_set_resources): Likewise.
9510 * rtl.c (copy_rtx): Likewise.
9511 * rtlanal.c (reg_referenced_p): Likewise.
9512 (single_set_2): Likewise.
9513 (noop_move_p): Likewise.
9514 (note_stores): Likewise.
9515 * sched-deps.c (sched_analyze_reg): Likewise.
9516 (sched_analyze_insn): Likewise.
9518 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9520 * cse.c (invalidate_reg): New function extracted from...
9521 (invalidate): ...here.
9522 (canonicalize_insn): Check for clobber high.
9523 (invalidate_from_clobbers): invalidate clobber highs.
9524 (invalidate_from_sets_and_clobbers): Likewise.
9525 (count_reg_usage): Check for clobber high.
9526 (insn_live_p): Likewise.
9527 * cselib.c (cselib_expand_value_rtx_1):Likewise.
9528 (cselib_invalidate_regno): Check for clobber in setter.
9529 (cselib_invalidate_rtx): Pass through setter.
9530 (cselib_invalidate_rtx_note_stores):
9531 (cselib_process_insn): Check for clobber high.
9532 * cselib.h (cselib_invalidate_rtx): Add operand.
9534 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9536 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
9537 (mark_not_eliminable): Likewise.
9538 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
9539 * lra-lives.c (process_bb_lives): Check for clobber high.
9540 * lra.c (new_insn_reg): Remember clobber highs.
9541 (collect_non_operand_hard_regs): Check for clobber high.
9542 (lra_set_insn_recog_data): Likewise.
9543 (add_regs_to_insn_regno_info): Likewise.
9544 (lra_update_insn_regno_info): Likewise.
9546 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9548 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
9549 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
9551 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9553 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
9554 (copy_insn_1): Likewise.
9555 (gen_hard_reg_clobber_high): New gen function.
9556 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
9557 * genemit.c (gen_exp): Likewise.
9558 (gen_emit_seq): Pass through info.
9559 (gen_insn): Check for CLOBBER_HIGH.
9560 (gen_expand): Pass through info.
9561 (gen_split): Likewise.
9562 (output_add_clobbers): Likewise.
9563 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
9564 (remove_clobbers): Likewise.
9565 * rtl.h (gen_hard_reg_clobber_high): New declaration.
9567 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9569 * doc/rtl.texi (clobber_high): Add.
9570 (parallel): Add in clobber high
9571 * rtl.c (rtl_check_failed_code3): Add function.
9572 * rtl.def (CLOBBER_HIGH): Add expression.
9573 * rtl.h (RTL_CHECKC3): Add macro.
9574 (rtl_check_failed_code3): Add declaration.
9575 (XC3EXP): Add macro.
9577 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
9580 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
9581 cfun->machine->max_used_stack_alignment if needed.
9583 2018-08-04 Martin Sebor <msebor@redhat.com>
9585 PR tree-optimization/86571
9586 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
9589 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
9592 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9595 2018-08-03 Jeff Law <law@redhat.com>
9598 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9599 Define to speculation_safe_value_not_needed.
9601 2018-08-03 David Malcolm <dmalcolm@redhat.com>
9603 * doc/gcov.texi (-x): Remove duplicate "to".
9604 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
9605 (-Wif-not-aligned): Remove duplicate "is".
9606 (-flto): Remove duplicate "the".
9607 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
9608 duplicate "v5.00.b".
9609 (MSP430 Options): Remove duplicate "and" from the description
9610 of "-mgprel-sec=regexp".
9611 (x86 Options): Remove duplicate copies of "vmldLog102" and
9612 vmlsLog104 from description of "-mveclibabi=type".
9614 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
9616 * internal-fn.h (first_commutative_argument): Declare.
9617 * internal-fn.c (first_commutative_argument): New function.
9618 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
9619 restrictions for pattern statements. Use first_commutative_argument
9620 to look for commutative operands in calls to internal functions.
9622 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
9624 * Makefile.in (wide-int-range.o): New.
9625 * tree-vrp.c: Move all the wide_int_* functions to...
9626 * wide-int-range.cc: ...here.
9627 * tree-vrp.h: Move all the wide_int_* prototypes to...
9628 * wide-int-range.h: ...here.
9630 2018-08-03 Tom de Vries <tdevries@suse.de>
9632 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
9634 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
9635 * except.c (output_function_exception_table): Do early exit if
9636 targetm_common.except_unwind_info (&global_options) == UI_NONE.
9638 2018-08-03 Martin Liska <mliska@suse.cz>
9640 * predict.c (dump_prediction): Change to 2 digits
9643 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
9645 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
9647 (falkor_am_1_gtov_gtov): ... a new insn reservation.
9649 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
9651 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
9652 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
9653 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
9654 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
9655 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
9657 2018-08-02 David Malcolm <dmalcolm@redhat.com>
9659 * diagnostic-show-locus.c (diagnostic_show_locus): Use
9660 pp_take_prefix when saving the existing prefix.
9661 * diagnostic.c (diagnostic_append_note): Likewise.
9662 * langhooks.c (lhd_print_error_function): Likewise.
9663 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
9664 param's type. Free the existing prefix.
9665 (pp_take_prefix): New function.
9666 (pretty_printer::pretty_printer): Drop the prefix parameter.
9667 Rename the length parameter to match the comment.
9668 (pretty_printer::~pretty_printer): Free the prefix.
9669 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
9671 (struct pretty_printer): Drop the "const" from "prefix" field's
9672 type and clarify memory management.
9673 (pp_set_prefix): Drop the "const" from the 2nd param.
9674 (pp_take_prefix): New decl.
9676 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
9678 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
9679 for word_mode_ok here instead of passing as argument.
9680 (expand_block_compare): Change select_block_compare_mode() call.
9681 (expand_strncmp_gpr_sequence): New function.
9682 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
9684 2018-08-02 Jeff Law <law@redhat.com>
9687 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9688 Define to speculation_safe_value_not_needed.
9691 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9692 Define to speculation_safe_value_not_needed.
9694 2018-08-02 Tom de Vries <tdevries@suse.de>
9697 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
9698 function. Return UI_TARGET unconditionally.
9699 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
9700 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
9702 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
9704 * genemit.c (print_overload_test): Fix typo.
9706 2018-08-02 Richard Biener <rguenther@suse.de>
9708 PR tree-optimization/86816
9709 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
9710 which checks for value availability before querying it.
9711 (gvn_uses_equal): Use it.
9712 (same_succ_hash): Likewise.
9713 (gimple_equal_p): Likewise.
9715 2018-08-02 Nick Clifton <nickc@redhat.com>
9718 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9719 Define to speculation_safe_value_not_needed.
9722 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9723 Define to speculation_safe_value_not_needed.
9726 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9727 Define to speculation_safe_value_not_needed.
9730 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9731 Define to speculation_safe_value_not_needed.
9734 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9735 Define to speculation_safe_value_not_needed.
9738 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9739 Define to speculation_safe_value_not_needed.
9742 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9743 Define to speculation_safe_value_not_needed.
9746 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9747 Define to speculation_safe_value_not_needed.
9750 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
9751 speculation_safe_value_not_needed.
9754 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
9755 to speculation_safe_value_not_needed.
9757 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
9759 * doc/md.texi: Expand the documentation of instruction names
9760 to mention port-local uses. Document '@' in pattern names.
9761 * read-md.h (overloaded_instance, overloaded_name): New structs.
9763 (md_reader::handle_overloaded_name): New member function.
9764 (md_reader::get_overloads): Likewise.
9765 (md_reader::m_first_overload): New member variable.
9766 (md_reader::m_next_overload_ptr): Likewise.
9767 (md_reader::m_overloads_htab): Likewise.
9768 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
9769 m_next_overload_ptr and m_overloads_htab.
9770 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
9771 (get_mode_token, get_code_token, get_int_token): New functions.
9772 (map_attr_string): Add an optional argument that passes back
9773 the associated iterator.
9774 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
9775 (md_reader::handle_overloaded_name, add_overload_instance): New
9777 (apply_iterators): Handle '@' names. Report an error if '@'
9778 is used without iterators.
9779 (initialize_iterators): Initialize the new iterator_group fields.
9780 * genopinit.c (handle_overloaded_code_for)
9781 (handle_overloaded_gen): New functions.
9782 (main): Use them to print declarations of maybe_code_for_* and
9783 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
9784 * genemit.c (print_overload_arguments, print_overload_test)
9785 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
9786 (main): Use it to print definitions of maybe_code_for_* and
9787 maybe_gen_* functions.
9788 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
9789 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
9790 instead of explicit mode checks.
9791 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
9792 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
9793 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
9794 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
9795 (aarch64_expand_compare_and_swap): Likewise
9796 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
9797 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
9798 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
9799 (aarch64_constant_pool_reload_icode): Delete.
9800 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
9801 instead of aarch64_constant_pool_reload_icode. Use
9802 code_for_aarch64_reload_mov instead of explicit mode checks.
9803 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
9804 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
9805 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
9806 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
9807 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
9808 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
9809 (aarch64_atomic_load_op_code): Delete.
9810 (aarch64_emit_atomic_load_op): Likewise.
9811 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
9812 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
9813 instead of aarch64_emit_atomic_load_op.
9814 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
9815 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
9816 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
9817 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
9818 character before the pattern name.
9819 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
9820 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
9821 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
9822 (aarch64_frecps<mode>): Likewise.
9823 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
9824 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
9825 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
9826 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
9827 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
9829 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
9831 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
9832 Allow HFmode constants if TARGET_FP_F16INST.
9834 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
9837 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
9838 No longer check last store for clobber of address register.
9840 2018-08-02 Martin Liska <mliska@suse.cz>
9842 PR gcov-profile/86817
9843 * gcov.c (process_all_functions): New function.
9845 (process_file): Move functions processing to
9846 process_all_functions.
9848 2018-08-02 David Malcolm <dmalcolm@redhat.com>
9850 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
9851 "const" to the "gimple *" and "rtx_insn *" parameters.
9852 * dumpfile.h (dump_user_location_t::dump_user_location_t):
9854 (dump_location_t::dump_location_t): Likewise.
9856 2018-08-01 Martin Sebor <msebor@redhat.com>
9858 PR tree-optimization/86650
9859 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
9860 rather than a "gcall *". Directly pass the data of interest
9861 to percent_K_format, rather than building a temporary CALL_EXPR
9863 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
9864 (gimple_fold_builtin_strncat): Adjust.
9865 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
9866 gcall* argument with gimple*.
9867 * gimple-ssa-warn-restrict.c (check_call): Same.
9868 (wrestrict_dom_walker::before_dom_children): Same.
9869 (builtin_access::builtin_access): Same.
9870 (check_bounds_or_overlap): Same
9871 (maybe_diag_overlap): Same.
9872 (maybe_diag_offset_bounds): Same.
9873 * tree-diagnostic.c (default_tree_printer): Move usage of
9874 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
9876 * tree-pretty-print.c (percent_K_format): Add argument.
9877 * tree-pretty-print.h: Add argument.
9878 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
9879 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
9880 (maybe_diag_stxncpy_trunc): Same.
9881 (handle_builtin_stxncpy): Same.
9882 (handle_builtin_strcat): Same.
9884 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9886 * match.pd: Optimise pointer range checks.
9888 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9890 PR tree-optimization/86758
9891 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
9892 to remove pattern statements.
9894 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9896 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
9897 result of dfs_enumerate_from when constructing stmt_vec_infos,
9898 instead of additionally calling get_loop_body.
9900 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9902 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
9904 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
9905 When creating an iv, assert that the step is not known to be zero.
9906 (vect_setup_realignment): Update call accordingly.
9907 * tree-vect-stmts.c (vectorizable_store): Likewise.
9908 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
9910 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9912 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
9913 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
9914 (vectorizable_reduction): Likewise.
9915 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9916 (vect_detect_hybrid_slp_stmts): Likewise.
9917 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9919 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
9921 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
9922 (wide_int_set_zero_nonzero_bits): ...this.
9923 (zero_nonzero_bits_from_vr): Rename to...
9924 (vrp_set_zero_nonzero_bits): ...this.
9925 (extract_range_from_multiplicative_op_1): Abstract wide int
9927 (wide_int_range_multiplicative_op): ...here.
9928 (extract_range_from_binary_expr_1): Extract wide int binary
9929 operations into their own functions.
9930 (wide_int_range_lshift): New.
9931 (wide_int_range_can_optimize_bit_op): New.
9932 (wide_int_range_shift_undefined_p): New.
9933 (wide_int_range_bit_xor): New.
9934 (wide_int_range_bit_ior): New.
9935 (wide_int_range_bit_and): New.
9936 (wide_int_range_trunc_mod): New.
9937 (extract_range_into_wide_ints): New.
9938 (vrp_shift_undefined_p): New.
9939 (extract_range_from_multiplicative_op): New.
9940 (vrp_can_optimize_bit_op): New.
9941 * tree-vrp.h (value_range::dump): New.
9942 (wide_int_range_multiplicative_op): New.
9943 (wide_int_range_lshift):New.
9944 (wide_int_range_shift_undefined_p): New.
9945 (wide_int_range_bit_xor): New.
9946 (wide_int_range_bit_ior): New.
9947 (wide_int_range_bit_and): New.
9948 (wide_int_range_trunc_mod): New.
9949 (zero_nonzero_bits_from_bounds): Rename to...
9950 (wide_int_set_zero_nonzero_bits): ...this.
9951 (zero_nonzero_bits_from_vr): Rename to...
9952 (vrp_set_zero_nonzero_bits): ...this.
9953 (range_easy_mask_min_max): Rename to...
9954 (wide_int_range_can_optimize_bit_op): this.
9955 * vr-values.c (simplify_bit_ops_using_ranges): Rename
9956 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
9958 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9960 * tree-vectorizer.h (vect_orig_stmt): New function.
9961 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
9962 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
9963 (vect_create_epilog_for_reduction): Likewise.
9964 (vectorizable_live_operation): Likewise.
9965 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
9966 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
9967 * tree-vect-stmts.c (vectorizable_call): Likewise.
9968 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
9970 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9972 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
9974 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
9975 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
9976 (vect_transform_loop): Likewise.
9977 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9979 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9981 * tree-vectorizer.h (vect_schedule_slp): Return void.
9982 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9983 (vect_schedule_slp): Likewise.
9985 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9987 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
9989 (vect_transform_loop): Update calls accordingly. Schedule SLP
9990 instances before the main loop, if any exist.
9992 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9994 PR tree-optimization/86749
9995 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
9996 If the lhs is used in a COND_EXPR, check that it is being used
9997 as the "then" or "else" value.
9999 2018-08-01 Tom de Vries <tdevries@suse.de>
10002 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10003 speculation_safe_value_not_needed.
10005 2018-08-01 Richard Biener <rguenther@suse.de>
10007 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
10008 as base and offset.
10010 2018-08-01 Martin Liska <mliska@suse.cz>
10012 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
10013 format how successful transformation is dumped.
10014 (gimple_mod_pow2_value_transform): Likewise.
10015 (gimple_mod_subtract_transform): Likewise.
10016 (gimple_stringops_transform): Likewise.
10018 2018-08-01 Martin Liska <mliska@suse.cz>
10020 PR value-prof/35543
10021 * value-prof.c (interesting_stringop_to_profile_p):
10022 Simplify the code and add BUILT_IN_MEMMOVE.
10023 (gimple_stringops_transform): Likewise.
10025 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
10027 * config/aarch64/aarch64-simd.md
10028 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
10029 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
10030 use GPI iterator instead of SI mode.
10032 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10034 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
10035 rs6000_speculation_barrier.
10036 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
10037 new barrier pattern name.
10039 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10041 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
10042 (speculation_barrier): New insn.
10044 2018-08-01 Richard Biener <rguenther@suse.de>
10047 * graphite.h: Include isl/id.h and isl/space.h to allow build
10050 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
10053 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
10054 mode for COFF targets.
10055 * defaults.h (TARGET_COFF): Define.
10056 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
10057 TARGET_COFF): Define.
10058 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
10059 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
10060 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
10062 2018-07-31 Alexandre Oliva <oliva@adacore.com>
10064 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
10065 * gimple-streamer-out.c (output_bb): Save it.
10066 * lto-streamer-in.c (input_struct_function_base): Restore
10067 instance discriminator if available. Create map on demand.
10068 * lto-streamer-out.c (output_struct_function_base): Save it if
10070 * final.c (decl_to_instance_map): Document LTO strategy.
10072 2018-07-31 Alexandre Oliva <oliva@adacore.com>
10073 Olivier Hainque <hainque@adacore.com>
10075 * debug.h (decl_to_instance_map_t): New type.
10076 (decl_to_instance_map): Declare.
10077 (maybe_create_decl_to_instance_map): New inline function.
10078 * final.c (bb_discriminator, last_bb_discriminator): New statics,
10079 to track basic block discriminators.
10080 (final_start_function_1): Initialize them.
10081 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
10083 (decl_to_instance_map): New variable.
10084 (map_decl_to_instance, maybe_set_discriminator): New functions.
10085 (notice_source_line): Set discriminator.
10087 2018-07-31 Ian Lance Taylor <iant@golang.org>
10089 * targhooks.c (default_have_speculation_safe_value): Add
10092 2018-07-31 David Malcolm <dmalcolm@redhat.com>
10094 * dump-context.h: Include "pretty-print.h".
10095 (dump_context::refresh_dumps_are_enabled): New decl.
10096 (dump_context::emit_item): New decl.
10097 (class dump_context): Add fields "m_test_pp" and
10099 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
10100 (temp_dump_context::get_dumped_text): New decl.
10101 (class temp_dump_context): Add field "m_pp".
10102 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
10103 (dump_context::refresh_dumps_are_enabled): ...and add a test for
10105 (set_dump_file): Update for above change.
10106 (set_alt_dump_file): Likewise.
10107 (dump_loc): New overload, taking a pretty_printer *.
10108 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
10109 to any test pretty-printer.
10110 (make_item_for_dump_gimple_stmt): New function, adapted from
10111 optinfo::add_gimple_stmt.
10112 (dump_context::dump_gimple_stmt): Call it, and use the result,
10113 eliminating the direct usage of dump_file and alt_dump_file in
10114 favor of indirectly using them via emit_item.
10115 (make_item_for_dump_gimple_expr): New function, adapted from
10116 optinfo::add_gimple_expr.
10117 (dump_context::dump_gimple_expr): Call it, and use the result,
10118 eliminating the direct usage of dump_file and alt_dump_file in
10119 favor of indirectly using them via emit_item.
10120 (make_item_for_dump_generic_expr): New function, adapted from
10122 (dump_context::dump_generic_expr): Call it, and use the result,
10123 eliminating the direct usage of dump_file and alt_dump_file in
10124 favor of indirectly using them via emit_item.
10125 (make_item_for_dump_printf_va): New function, adapted from
10126 optinfo::add_printf_va.
10127 (make_item_for_dump_printf): New function.
10128 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
10129 and use the result, eliminating the direct usage of dump_file and
10130 alt_dump_file in favor of indirectly using them via emit_item.
10131 (make_item_for_dump_dec): New function.
10132 (dump_context::dump_dec): Call it, and use the result,
10133 eliminating the direct usage of dump_file and alt_dump_file in
10134 favor of indirectly using them via emit_item.
10135 (make_item_for_dump_symtab_node): New function, adapted from
10136 optinfo::add_symtab_node.
10137 (dump_context::dump_symtab_node): Call it, and use the result,
10138 eliminating the direct usage of dump_file and alt_dump_file in
10139 favor of indirectly using them via emit_item.
10140 (dump_context::begin_scope): Reimplement, avoiding direct usage
10141 of dump_file and alt_dump_file in favor of indirectly using them
10143 (dump_context::emit_item): New member function.
10144 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
10145 Set up test pretty-printer on the underlying context. Call
10146 refresh_dumps_are_enabled.
10147 (temp_dump_context::~temp_dump_context): Call
10148 refresh_dumps_are_enabled.
10149 (temp_dump_context::get_dumped_text): New member function.
10150 (selftest::verify_dumped_text): New function.
10151 (ASSERT_DUMPED_TEXT_EQ): New macro.
10152 (selftest::test_capture_of_dump_calls): Run all tests twice, with
10153 and then without optinfo enabled. Add uses of
10154 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
10155 * dumpfile.h: Update comment for the dump_* API.
10156 * optinfo-emit-json.cc
10157 (selftest::test_building_json_from_dump_calls): Update for new
10158 param for temp_dump_context ctor.
10159 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
10160 and "m_owned" field.
10161 (optinfo_item::~optinfo_item): Likewise.
10162 (optinfo::add_item): New member function.
10163 (optinfo::emit): Update comment.
10164 (optinfo::add_string): Delete.
10165 (optinfo::add_printf): Delete.
10166 (optinfo::add_printf_va): Delete.
10167 (optinfo::add_gimple_stmt): Delete.
10168 (optinfo::add_gimple_expr): Delete.
10169 (optinfo::add_tree): Delete.
10170 (optinfo::add_symtab_node): Delete.
10171 (optinfo::add_dec): Delete.
10172 * optinfo.h (class dump_context): New forward decl.
10173 (optinfo::add_item): New decl.
10174 (optinfo::add_string): Delete.
10175 (optinfo::add_printf): Delete.
10176 (optinfo::add_printf_va): Delete.
10177 (optinfo::add_gimple_stmt): Delete.
10178 (optinfo::add_gimple_expr): Delete.
10179 (optinfo::add_tree): Delete.
10180 (optinfo::add_symtab_node): Delete.
10181 (optinfo::add_dec): Delete.
10182 (optinfo::add_poly_int): Delete.
10183 (optinfo_item::optinfo_item): Remove "owned" param.
10184 (class optinfo_item): Remove field "m_owned".
10186 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10188 PR middle-end/86705
10189 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
10190 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
10191 requested variable alignment.
10192 (expand_one_ssa_partition): Likewise.
10193 (expand_one_var): Likewise.
10195 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10197 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
10198 to speculation_safe_value_not_needed.
10200 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10202 * targhooks.h (speculation_safe_value_not_needed): New prototype.
10203 * targhooks.c (speculation_safe_value_not_needed): New function.
10204 * target.def (have_speculation_safe_value): Update documentation.
10205 * doc/tm.texi: Regenerated.
10207 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10209 * config/aarch64/iterators.md (ALLI_TI): New iterator.
10210 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
10212 (despeculate_copy<ALLI:mode>_insn): New insn.
10213 (despeculate_copyti_insn): New insn.
10214 (despeculate_simple<ALLI:mode>): New insn
10215 (despeculate_simpleti): New insn.
10216 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
10218 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
10219 aarch64_speculation_safe_value.
10220 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
10222 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10224 * config/aarch64/aarch64-speculation.cc: New file.
10225 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
10226 before pass_reorder_blocks.
10227 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
10229 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
10230 X14 and X15 when tracking speculation.
10231 * config/aarch64/aarch64.md (register name constants): Add
10232 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
10233 (unspec): Add UNSPEC_SPECULATION_TRACKER.
10234 (speculation_barrier): New insn attribute.
10235 (cmp<mode>): Allow SP in comparisons.
10236 (speculation_tracker): New insn.
10237 (speculation_barrier): Add speculation_barrier attribute.
10238 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
10239 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
10240 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
10242 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10244 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
10245 aarch64_track_speculation is true.
10246 (tb<optab><mode>1): Likewise.
10247 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
10248 generate CB[N]Z when tracking speculation.
10249 (aarch64_split_compare_and_swap): Likewise.
10250 (aarch64_split_atomic_op): Likewise.
10252 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10254 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
10256 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10258 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
10259 (speculation_barrier): New insn.
10261 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10263 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
10264 * config/arm/arm.md (speculation_barrier): New expand.
10265 (speculation_barrier_insn): New pattern.
10267 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10269 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
10270 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
10271 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
10272 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
10274 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
10275 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
10276 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
10277 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
10278 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
10279 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
10280 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
10281 * builtins.c (expand_speculation_safe_value): New function.
10282 (expand_builtin): Call it.
10283 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
10284 * doc/extend.texi: Document __builtin_speculation_safe_value.
10285 * doc/md.texi: Document "speculation_barrier" pattern.
10286 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
10287 TARGET_HAVE_SPECULATION_SAFE_VALUE.
10288 * doc/tm.texi: Regenerated.
10289 * target.def (have_speculation_safe_value, speculation_safe_value): New
10291 * targhooks.c (default_have_speculation_safe_value): New function.
10292 (default_speculation_safe_value): New function.
10293 * targhooks.h (default_have_speculation_safe_value): Add prototype.
10294 (default_speculation_safe_value): Add prototype.
10296 2018-07-31 David Malcolm <dmalcolm@redhat.com>
10298 * dump-context.h (dump_context::dump_loc): New decl.
10299 * dumpfile.c (dump_context::dump_loc): New member function.
10300 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
10301 and dump_gimple_stmt.
10302 (dump_context::dump_gimple_expr_loc): Likewise, using
10304 (dump_context::dump_generic_expr_loc): Likewise, using
10306 (dump_context::dump_printf_loc_va): Likewise, using
10308 (dump_context::begin_scope): Explicitly using the global function
10309 "dump_loc", rather than the member function.
10311 2018-07-31 Martin Sebor <msebor@redhat.com>
10313 PR tree-optimization/86741
10314 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
10316 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
10318 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
10319 depend on whether prefetch instructions will be emitted or not.
10320 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
10321 will be emitted or not.
10322 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
10323 (TARGET_SETMEM_PFD): New macros.
10325 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10327 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
10328 (NULL_STMT_VEC_INFO): Delete.
10329 (stmt_vec_info::operator*): Likewise.
10330 (stmt_vec_info::operator gimple *): Likewise.
10331 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
10332 of NULL_STMT_VEC_INFO.
10333 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10334 (vect_reassociating_reduction_p): Likewise.
10335 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
10336 (vectorizable_store): Likewise.
10337 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
10338 (vec_info::free_stmt_vec_infos): Likewise.
10340 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10342 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
10343 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
10345 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10347 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
10348 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
10349 (vec_info::free_stmt_vec_info): New private member functions.
10350 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
10351 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
10352 * tree-parloops.c (gather_scalar_reductions): Remove calls to
10353 set_stmt_vec_info_vec and free_stmt_vec_infos.
10354 * tree-vect-loop.c (_loop_vec_info): Remove call to
10355 set_stmt_vec_info_vec.
10356 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
10357 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
10358 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
10359 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
10360 (vec_info::free_stmt_vec_info): ...these new functions. Remove
10361 assignments in {vec_info::,}new_stmt_vec_info that are redundant
10362 with the clearing in the xcalloc.
10363 (stmt_vec_info_vec): Delete.
10364 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
10365 (vectorize_loops): Likewise.
10366 (vec_info::~vec_info): Remove argument from call to
10367 free_stmt_vec_infos.
10368 (vec_info::add_stmt): Remove vinfo argument from call to
10371 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10373 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
10374 rather than a gimple stmt.
10375 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
10376 information for pattern statements when passed the original
10377 statement; instead wait to be passed the pattern statement itself.
10378 Don't call set_vinfo_for_stmt here.
10379 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
10380 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
10381 stmt_vec_infos here.
10382 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
10383 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
10384 stmt_vec_infos entry.
10386 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10388 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
10389 * tree-vectorizer.c (vec_info::replace_stmt): New function.
10390 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
10391 * tree-vect-stmts.c (vectorizable_call): Likewise.
10392 (vectorizable_simd_clone_call): Likewise.
10394 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10396 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
10397 * tree-vectorizer.c (vec_info::remove_stmt): New function.
10398 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
10399 * tree-vect-loop.c (vect_transform_loop): Likewise.
10400 * tree-vect-slp.c (vect_schedule_slp): Likewise.
10401 * tree-vect-stmts.c (vect_remove_stores): Likewise.
10403 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10405 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
10406 (vect_dr_stmt): Delete.
10407 * tree-vectorizer.c (vec_info::lookup_dr): New function.
10408 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
10410 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
10411 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
10412 (vect_verify_datarefs_alignment, vect_peeling_supportable)
10413 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
10414 (vect_analyze_data_refs): Likewise.
10415 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
10417 (vect_find_same_alignment_drs): Likewise.
10418 (vect_slp_analyze_node_dependences): Update calls accordingly.
10419 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
10420 instead of DR_VECT_AUX.
10421 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
10422 of a vector data references. Use vec_info::lookup_dr instead of
10424 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
10425 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
10426 instead of DR_VECT_AUX.
10428 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10430 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
10432 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
10434 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10435 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10436 (vect_gen_prolog_loop_niters): Likewise.
10438 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10440 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
10441 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
10442 (vect_known_alignment_in_bytes, vect_dr_behavior)
10443 (vect_get_scalar_dr_size): Take references as dr_vec_infos
10444 instead of data_references. Update calls to other routines for
10445 which the same change has been made.
10446 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
10447 dr_vec_infos instead of stmt_vec_infos.
10448 (vect_analyze_data_ref_dependence): Update call accordingly.
10449 (vect_slp_analyze_data_ref_dependence)
10450 (vect_record_base_alignments): Use DR_VECT_AUX.
10451 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
10452 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
10453 (vector_alignment_reachable_p, vect_get_data_access_cost)
10454 (vect_peeling_supportable, vect_analyze_group_access_1)
10455 (vect_analyze_group_access, vect_analyze_data_ref_access)
10456 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
10457 (vect_compile_time_alias, vect_small_gap_p)
10458 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
10459 (vect_supportable_dr_alignment): Take references as dr_vec_infos
10460 instead of data_references. Update calls to other routines for
10461 which the same change has been made.
10462 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
10463 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
10464 (vect_slp_analyze_and_verify_node_alignment)
10465 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
10466 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
10467 (vect_setup_realignment): Use dr_vec_infos. Update calls after
10469 (_vect_peel_info::dr): Replace with...
10470 (_vect_peel_info::dr_info): ...this new field.
10471 (vect_peeling_hash_get_most_frequent)
10472 (vect_peeling_hash_choose_best_peeling): Update accordingly.
10473 (vect_peeling_hash_get_lowest_cost):
10474 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
10475 routines for which the same change has been made.
10476 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
10478 * tree-vect-loop-manip.c (get_misalign_in_elems)
10479 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
10481 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10482 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
10483 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
10484 (get_group_load_store_type, get_negative_load_store_type)
10485 (vect_get_data_ptr_increment, vectorizable_store)
10486 (vectorizable_load): Likewise.
10487 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
10488 Update calls to other routines for which the same change has been made.
10490 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10492 * tree-vectorizer.h (vec_info::move_dr): New member function.
10493 (dataref_aux): Rename to...
10494 (dr_vec_info): ...this and add "dr" and "stmt" fields.
10495 (_stmt_vec_info::dr_aux): Update accordingly.
10496 (_stmt_vec_info::data_ref_info): Delete.
10497 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
10498 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
10499 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
10500 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
10502 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
10503 (STMT_VINFO_DR_INFO): New macro.
10504 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
10505 (set_dr_misalignment): Update after rename of dataref_aux.
10506 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
10507 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
10508 initialization of STMT_VINFO_DATA_REF.
10509 * tree-vectorizer.c (vec_info::move_dr): New function.
10510 * tree-vect-patterns.c (vect_recog_bool_pattern)
10511 (vect_recog_mask_conversion_pattern)
10512 (vect_recog_gather_scatter_pattern): Use it.
10513 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
10514 the "dr" and "stmt" fields of dr_vec_info instead of
10515 STMT_VINFO_DATA_REF.
10517 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10519 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
10520 (is_pattern_stmt_p): Use it.
10521 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
10522 on pattern statements.
10524 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10526 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
10527 original stmt as a stmt_vec_info rather than a gimple stmt.
10528 (vect_pattern_recog_1): Take the statement directly as a
10529 stmt_vec_info, rather than via a gimple_stmt_iterator.
10530 Update call to vect_mark_pattern_stmts.
10531 (vect_pattern_recog): Update calls accordingly.
10533 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10535 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
10536 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
10537 a vect_def_type for the first argument.
10538 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
10539 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
10540 operand if it isn't defined by a vectorized statement.
10541 (vect_build_gather_load_calls): Remove the mask_dt argument and
10542 update calls to vect_get_vec_def_for_stmt_copy.
10543 (vectorizable_bswap): Likewise the dt argument.
10544 (vectorizable_call): Update calls to vectorizable_bswap and
10545 vect_get_vec_def_for_stmt_copy.
10546 (vectorizable_simd_clone_call, vectorizable_assignment)
10547 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
10548 (vectorizable_comparison): Update calls to
10549 vect_get_vec_def_for_stmt_copy.
10550 (vectorizable_store): Likewise. Remove now-unnecessary calls to
10551 vect_is_simple_use.
10552 (vect_get_loop_based_defs): Remove dt argument and update call
10553 to vect_get_vec_def_for_stmt_copy.
10554 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
10555 and vect_get_vec_def_for_stmt_copy.
10556 (vectorizable_load): Update calls to vect_build_gather_load_calls
10557 and vect_get_vec_def_for_stmt_copy.
10558 * tree-vect-loop.c (vect_create_epilog_for_reduction)
10559 (vectorizable_reduction, vectorizable_live_operation): Update calls
10560 to vect_get_vec_def_for_stmt_copy.
10562 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10564 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
10565 and gimple stmt arguments with a stmt_vec_info.
10566 (vect_record_base_alignments): Update calls accordingly.
10567 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
10568 and gimple stmt arguments with a stmt_vec_info.
10569 (vect_build_slp_tree_1): Remove vinfo argument and update call
10570 to vect_record_max_nunits.
10571 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
10572 and vect_record_max_nunits.
10574 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10576 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
10577 file and take a stmt_vec_info instead of a gimple stmt.
10578 (supportable_widening_operation, vect_finish_replace_stmt)
10579 (vect_finish_stmt_generation, vect_get_store_rhs)
10580 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
10581 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
10582 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
10583 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
10584 (vect_create_data_ref_ptr, bump_vector_ptr)
10585 (vect_permute_store_chain, vect_setup_realignment)
10586 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
10587 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
10588 (vectorizable_reduction, vectorizable_induction)
10589 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
10590 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
10591 than gimple stmts as arguments.
10592 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
10593 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
10594 (can_group_stmts_p, vect_check_gather_scatter)
10595 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
10596 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
10597 (vect_permute_load_chain, vect_shift_permute_load_chain)
10598 (vect_transform_grouped_load)
10599 (vect_record_grouped_load_vectors): Likewise.
10600 * tree-vect-loop.c (vect_fixup_reduc_chain)
10601 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
10602 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
10603 (vectorizable_reduction, vectorizable_induction)
10604 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
10605 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
10606 (vect_get_load_store_mask): Likewise.
10607 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
10608 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
10609 * tree-vect-stmts.c (vect_mark_relevant)
10610 (is_simple_and_all_uses_invariant)
10611 (exist_non_indexing_operands_for_use_p, process_use)
10612 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
10613 (vect_get_vec_def_for_operand, vect_get_vec_defs)
10614 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
10615 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
10616 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
10617 (get_negative_load_store_type, get_load_store_type)
10618 (vect_check_load_store_mask, vect_check_store_rhs)
10619 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
10620 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10621 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10622 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10623 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
10624 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
10625 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
10626 (supportable_widening_operation): Likewise.
10628 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10630 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
10631 a stmt_vec_info instead of a gcall.
10632 (vect_check_gather_scatter): Update call accordingly.
10633 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
10635 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
10636 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
10637 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
10638 instead of a gimple stmt.
10639 (vect_transform_loop): Update calls accordingly.
10640 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
10641 stmt_vec_infos instead of gimple stmts.
10642 (vect_analyze_slp_instance): Update use accordingly.
10643 * tree-vect-stmts.c (read_vector_array, write_vector_array)
10644 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
10645 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
10646 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
10647 (vect_gen_widened_results_half, vect_get_loop_based_defs)
10648 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
10649 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
10650 down to subroutines.
10652 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10654 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
10655 of the worklist from a vector of gimple stmts to a vector of
10657 * tree-vect-stmts.c (vect_mark_relevant, process_use)
10658 (vect_mark_stmts_to_be_vectorized): Likewise
10660 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10662 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
10663 statement before passing it to vect_analyze_stmt.
10664 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
10665 the chain of phi vector definitions. Track the exit phi via its
10667 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
10668 STMT_VINFO_REDUC_DEF.
10669 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
10670 stmt_vec_infos to handle the statement chains.
10671 (vect_get_slp_defs): Record the first statement in the node
10672 using a stmt_vec_info.
10673 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
10674 statements here and pass their stmt_vec_info down to subroutines.
10675 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
10676 down to vect_finish_stmt_generation.
10677 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
10678 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
10679 stmt_vec_infos to subroutines.
10680 (vect_remove_stores): Use stmt_vec_infos to handle the statement
10683 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10685 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
10686 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
10687 (vect_permute_store_chain, vect_setup_realignment)
10688 (vect_permute_load_chain, vect_shift_permute_load_chain)
10689 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
10690 stmts internally, and when passing values to other vectorizer routines.
10691 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
10692 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
10693 (vect_analyze_loop_operations, get_initial_def_for_reduction)
10694 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
10695 (vectorizable_reduction, vectorizable_induction)
10696 (vectorizable_live_operation, vect_transform_loop_stmt)
10697 (vect_transform_loop): Likewise.
10698 * tree-vect-patterns.c (vect_reassociating_reduction_p)
10699 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
10700 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
10701 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10702 (vect_slp_analyze_node_operations_1): Likewise.
10703 * tree-vect-stmts.c (vect_mark_relevant, process_use)
10704 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
10705 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
10706 (vect_finish_stmt_generation_1, get_group_load_store_type)
10707 (get_load_store_type, vect_build_gather_load_calls)
10708 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10709 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10710 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10711 (vectorizable_store, vectorizable_load, vectorizable_condition)
10712 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
10713 (supportable_widening_operation): Likewise.
10714 (vect_get_vector_types_for_stmt): Likewise.
10715 * tree-vectorizer.h (vect_dr_behavior): Likewise.
10717 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10719 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
10720 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
10721 (vect_permute_store_chain, vect_permute_load_chain)
10722 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
10723 repeated stmt_vec_info lookups.
10724 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
10725 (vect_update_ivs_after_vectorizer): Likewise.
10726 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
10727 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
10728 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
10729 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10730 (vect_bb_slp_scalar_cost): Likewise.
10731 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
10733 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10735 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
10736 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
10737 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
10738 to get gassigns and gcalls, rather than operating on generc gimple
10740 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
10741 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
10742 (vectorizable_load, vect_analyze_stmt): Likewise.
10743 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
10745 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10747 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
10748 return stmt_vec_infos rather than gimple stmts. Do not accept
10750 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
10752 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
10753 Update use of get_later_stmt.
10754 (vect_get_constant_vectors): Update call accordingly.
10755 (vect_schedule_slp_instance): Likewise
10756 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
10757 (vect_slp_analyze_instance_dependence): Likewise.
10758 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
10760 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10762 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
10763 (stmt_info_for_cost::stmt_info): ...this new field.
10764 (add_stmt_costs): Update accordingly.
10765 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
10766 (vect_get_known_peeling_cost): Likewise.
10767 (vect_estimate_min_profitable_iters): Likewise.
10768 * tree-vect-stmts.c (record_stmt_cost): Likewise.
10770 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10772 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
10773 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
10774 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
10776 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
10778 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10780 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
10781 a gimple stmt to a stmt_vec_info.
10782 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
10784 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10786 * tree-vectorizer.h (vec_info::grouped_stores): Change from
10787 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
10788 (_loop_vec_info::reduction_chains): Likewise.
10789 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
10791 * tree-vect-slp.c (vect_analyze_slp): Likewise.
10793 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10795 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
10796 a gimple stmt to a stmt_vec_info.
10797 (_stmt_vec_info::next_element): Likewise.
10798 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
10799 (vect_slp_analyze_and_verify_node_alignment)
10800 (vect_analyze_group_access_1, vect_analyze_group_access)
10801 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
10802 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
10803 (vect_supportable_dr_alignment): Update accordingly.
10804 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
10805 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
10806 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
10807 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
10808 * tree-vect-slp.c (vect_build_slp_tree_1)
10809 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
10810 (vect_split_slp_store_group, vect_analyze_slp_instance)
10811 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
10812 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
10813 (get_group_load_store_type, get_load_store_type)
10814 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
10815 (vect_transform_stmt, vect_remove_stores): Likewise.
10817 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10819 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
10820 than a gimple stmt.
10821 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
10822 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
10823 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
10824 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
10825 (vector_alignment_reachable_p, vect_get_data_access_cost)
10826 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
10827 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
10828 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
10829 (vect_analyze_group_access_1, vect_analyze_group_access)
10830 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
10831 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
10832 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
10833 result of vect_dr_stmt and use the stmt_vec_info instead of
10834 the associated gimple stmt.
10835 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10836 (vect_gen_prolog_loop_niters): Likewise.
10837 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10839 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10841 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
10842 to a vec<stmt_vec_info>.
10843 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
10844 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
10845 vec<stmt_vec_info>.
10846 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
10847 to a vec<stmt_vec_info>.
10848 (bst_traits::value_type, bst_traits::value_type): Likewise.
10849 (bst_traits::hash): Update accordingly.
10850 (vect_get_and_check_slp_defs): Change the stmts parameter from
10851 a vec<gimple *> to a vec<stmt_vec_info>.
10852 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
10853 (vect_build_slp_tree): Likewise.
10854 (vect_build_slp_tree_2): Likewise. Update uses of
10855 SLP_TREE_SCALAR_STMTS.
10856 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
10857 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
10858 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
10859 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
10860 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
10861 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
10862 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
10863 (vect_get_constant_vectors, vect_get_slp_defs)
10864 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
10865 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
10866 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
10867 instead of gimple stmts.
10868 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
10869 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
10870 (vect_slp_analyze_instance_dependence): Update uses of
10871 SLP_TREE_SCALAR_STMTS.
10872 (vect_slp_analyze_and_verify_node_alignment): Likewise.
10873 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
10874 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
10875 (get_initial_defs_for_reduction): Likewise.
10876 (vect_create_epilog_for_reduction): Likewise.
10877 (vectorize_fold_left_reduction): Likewise.
10878 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
10879 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
10880 (can_vectorize_live_stmts): Likewise.
10882 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10884 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
10885 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
10886 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
10888 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
10890 (gather_scalar_reductions): Update after above interface changes.
10891 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
10892 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
10894 (vect_force_simple_reduction): Likewise.
10895 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
10896 LOOP_VINFO_REDUCTIONS.
10897 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10899 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10901 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
10902 a gimple stmt to a stmt_vec_info.
10903 * tree-vect-loop.c (vect_active_double_reduction_p)
10904 (vect_force_simple_reduction, vectorizable_reduction): Update
10907 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10909 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
10910 vec<gimple *> to a vec<stmt_vec_info>.
10911 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
10912 the reduction_phis argument from a vec<gimple *> to a
10913 vec<stmt_vec_info>.
10914 (vectorizable_reduction): Likewise the phis local variable that
10915 is passed to vect_create_epilog_for_reduction. Update for new type
10916 of SLP_TREE_VEC_STMTS.
10917 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
10918 (vectorizable_live_operation): Likewise.
10919 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
10920 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
10922 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10924 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
10925 a gimple stmt to a stmt_vec_info.
10926 (vectorizable_condition, vectorizable_live_operation)
10927 (vectorizable_reduction, vectorizable_induction): Pass back the
10928 vectorized statement as a stmt_vec_info.
10929 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
10930 use of STMT_VINFO_VEC_STMT.
10931 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
10932 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
10933 as stmt_vec_infos rather than gimple stmts.
10934 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
10935 to a stmt_vec_info.
10936 (vectorizable_live_operation): Likewise.
10937 (vectorizable_reduction, vectorizable_induction): Likewise,
10938 updating use of STMT_VINFO_VEC_STMT.
10939 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
10940 of STMT_VINFO_VEC_STMT.
10941 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
10942 (vectorizable_simd_clone_call, vectorizable_conversion)
10943 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10944 (vectorizable_store, vectorizable_load, vectorizable_condition)
10945 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
10946 from a gimple stmt to a stmt_vec_info.
10947 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
10948 pointer to a stmt_vec_info to the vectorizable_* routines.
10950 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10952 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
10953 a gimple stmt to a stmt_vec_info.
10954 (is_pattern_stmt_p): Update accordingly.
10955 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
10956 (vect_record_grouped_load_vectors): Likewise.
10957 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
10958 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
10959 (vect_model_reduction_cost): Likewise.
10960 (vect_create_epilog_for_reduction): Likewise.
10961 (vectorizable_reduction, vectorizable_induction): Likewise.
10962 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10963 Return the stmt_vec_info for the pattern statement.
10964 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
10965 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
10966 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
10967 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
10968 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
10969 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
10970 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
10971 (free_stmt_vec_info, vect_is_simple_use): Likewise.
10973 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10975 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
10976 (vect_finish_stmt_generation): Likewise.
10977 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
10978 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
10979 (vect_build_gather_load_calls): Use the return value of the above
10980 functions instead of a separate call to vinfo_for_stmt. Use narrow
10981 scopes for the input gimple stmt and wider scopes for the associated
10982 stmt_vec_info. Use vec_info::lookup_def when setting these
10983 stmt_vec_infos from an SSA_NAME definition.
10984 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10985 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10986 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10987 (vectorizable_store, vectorizable_load, vectorizable_condition)
10988 (vectorizable_comparison): Likewise.
10989 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
10990 (vectorizable_reduction): Likewise.
10992 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10994 * tree-vectorizer.h (vect_is_simple_use): Add an optional
10995 stmt_vec_info * parameter before the optional gimple **.
10996 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
10997 (process_use, vect_get_vec_def_for_operand_1): Update callers.
10998 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
10999 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11000 (vectorizable_live_operation): Likewise.
11001 * tree-vect-patterns.c (type_conversion_p): Likewise.
11002 (vect_look_through_possible_promotion): Likewise.
11003 (vect_recog_rotate_pattern): Likewise.
11004 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11006 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11008 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
11009 a typedef to a wrapper class.
11010 (NULL_STMT_VEC_INFO): New macro.
11011 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
11012 (stmt_vec_info::operator*): New function.
11013 (stmt_vec_info::operator gimple *): Likewise.
11014 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
11015 (add_stmt_costs): Likewise.
11016 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
11017 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11018 (vect_get_known_peeling_cost): Likewise.
11019 (vect_estimate_min_profitable_iters): Likewise.
11020 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11021 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
11022 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
11023 (vectorizable_store, free_stmt_vec_infos): Likewise.
11024 (new_stmt_vec_info): Change return type of xcalloc to
11027 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11029 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
11030 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
11031 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
11032 a single_imm_use-based sequence.
11033 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
11035 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11037 * tree-vectorizer.h (vec_info::lookup_def): Declare.
11038 * tree-vectorizer.c (vec_info::lookup_def): New function.
11039 * tree-vect-patterns.c (vect_get_internal_def): Use it.
11040 (vect_widened_op_tree): Likewise.
11041 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11042 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
11043 (vectorizable_reduction): Likewise.
11044 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
11046 (vect_is_slp_reduction): Update calls accordingly. Use
11047 vec_info::lookup_def.
11048 (vect_is_simple_reduction): Likewise
11049 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
11051 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11053 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
11054 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
11055 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
11057 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
11058 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
11059 (vect_update_vf_for_slp, vect_analyze_loop_operations)
11060 (vect_is_slp_reduction, vectorizable_induction)
11061 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
11062 * tree-vect-patterns.c (vect_init_pattern_stmt):
11063 (vect_determine_min_output_precision_1, vect_determine_precisions)
11064 (vect_pattern_recog): Likewise.
11065 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
11066 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
11067 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
11068 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
11069 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
11070 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
11071 info field from a loop to a loop_vec_info.
11073 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11075 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
11076 (vec_info::add_stmt): Declare.
11077 * tree-vectorizer.c (vec_info::add_stmt): New function.
11078 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
11079 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
11080 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
11081 (vectorizable_induction): Likewise.
11082 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
11083 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
11084 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
11085 (vectorizable_load): Likewise.
11086 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11087 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
11088 (vect_recog_gather_scatter_pattern): Likewise.
11089 (append_pattern_def_seq): Likewise. Remove a check that is
11090 performed by add_stmt itself.
11092 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11094 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
11095 which make_ssa_name was called with new_stmt before new_stmt
11098 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11100 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
11102 (vect_is_slp_reduction): ...here...
11103 (vect_is_simple_reduction): ...and here. Remove repetition of tests
11104 that are already known to be false.
11106 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11108 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
11109 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
11110 STMT_VINFO_NUM_SLP_USES when it's true.
11111 (vect_free_slp_instance): Add a final_p parameter and pass it to
11112 vect_free_slp_tree.
11113 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
11114 (vect_analyze_slp_instance): Likewise.
11115 (vect_slp_analyze_operations): Likewise.
11116 (vect_slp_analyze_bb_1): Likewise.
11117 * tree-vectorizer.c (vec_info): Likewise.
11118 * tree-vect-loop.c (vect_transform_loop): Likewise.
11120 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11122 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11123 function is not called for second and subsequent members of
11126 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11128 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
11129 cases for nested loops from here to ...
11130 (vect_create_epilog_for_reduction): ...here. Only call
11131 vect_is_simple_use for inner-loop reductions.
11133 2018-07-31 Martin Liska <mliska@suse.cz>
11135 PR gcov-profile/85338
11136 PR gcov-profile/85350
11137 PR gcov-profile/85372
11138 * profile.c (struct location_triplet): New.
11139 (struct location_triplet_hash): Likewise.
11140 (output_location): Do not output a BB that
11141 is already recorded for a line.
11142 (branch_prob): Use streamed_locations.
11144 2018-07-31 Martin Liska <mliska@suse.cz>
11146 PR gcov-profile/85370
11147 * coverage.c (coverage_begin_function): Do not mark target
11148 clones as artificial functions.
11150 2018-07-31 Martin Liska <mliska@suse.cz>
11152 PR gcov-profile/83813
11153 PR gcov-profile/84758
11154 PR gcov-profile/85217
11155 PR gcov-profile/85332
11156 * profile.c (branch_prob): Do not record GOTO expressions
11157 for GIMPLE statements which locations are already streamed.
11159 2018-07-31 Olivier Hainque <hainque@adacore.com>
11161 * gcc.c (handle_spec_function): Accept a soft_matched_part
11162 argument, as do_spec_1. Pass it down to ...
11163 (eval_spec_function): Accept a soft_matched_part argument,
11164 and pass it down to ...
11165 (do_spec_2): Accept a soft_matched_part argument, and pass
11166 it down to do_spec_1.
11167 (do_spec_1): Pass soft_matched_part to handle_spec_function.
11168 (handle_braces): Update call to handle_spec_function.
11169 (driver::set_up_specs): Update calls to do_spec_2.
11170 (compare_debug_dump_opt_spec_function): Likewise.
11171 (compare_debug_self_opt_spec_function): Likewise.
11173 2018-07-31 Olivier Hainque <hainque@adacore.com>
11175 * common.opt (nolibc): New option.
11176 * doc/invoke.texi (Link Options): Document it.
11177 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
11178 * config/alpha/linux.h: Likewise.
11179 * config/arc/elf.h: Likewise.
11180 * config/arm/uclinux-elf.h: Likewise.
11181 * config/arm/unknown-elf.h: Likewise.
11182 * config/avr/avrlibc.h: Likewise.
11183 * config/bfin/bfin.h: Likewise.
11184 * config/bfin/linux.h: Likewise.
11185 * config/bfin/uclinux.h: Likewise.
11186 * config/darwin.h: Likewise.
11187 * config/darwin10.h: Likewise.
11188 * config/darwin12.h: Likewise.
11189 * config/gnu-user.h: Likewise.
11190 * config/lm32/uclinux-elf.h: Likewise.
11191 * config/pa/pa-hpux11.h: Likewise.
11192 * config/pa/pa64-hpux.h: Likewise.
11193 * config/sparc/sparc.h: Likewise.
11195 2018-07-31 Olivier Hainque <hainque@adacore.com>
11197 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
11198 undefined variables.
11200 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
11203 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
11204 instead of GEN_INT.
11206 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11208 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
11209 terminated string literal.
11211 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
11213 PR rtl-optimization/85160
11214 * combine.c (is_just_move): New function.
11215 (try_combine): Allow combining two instructions into two if neither of
11216 the original instructions was a move.
11218 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
11221 * doc/extend.texi (Global Register Variables): Discourage use of type
11223 (Local Register Variables): Likewise.
11225 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
11227 PR tree-optimization/86506
11228 * hwint.h (ceil_log2): Resync with hwint.c implementation.
11230 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
11233 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
11234 hard_regno, make sure no insn between `from` and `to` clobbers it.
11236 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
11237 Tom de Vries <tdevries@suse.de>
11239 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
11240 (PTX_DEFAULT_RUNTIME_DIM): ... this.
11241 (nvptx_goacc_validate_dims): Set default worker and gang dims to
11242 PTX_DEFAULT_RUNTIME_DIM.
11243 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
11245 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
11247 * config/pa/pa.c (pa_output_addr_vec): Align address table.
11248 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
11249 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
11251 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
11253 * config/rs6000/constraints.md (wG constraint): Delete, no longer
11255 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
11256 predicate to reflect toc fusion has been deleted.
11257 (toc_fusion_mem_raw): Delete, no longer used.
11258 (toc_fusion_mem_wrapped): Likewise.
11259 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
11261 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
11262 Delete, no longer used.
11263 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
11264 meant to be used for toc fusion.
11265 (rs6000_debug_print_mode): Delete toc fusion debugging.
11266 (rs6000_debug_reg_global): Likewise.
11267 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
11268 fusion and secondary reload support that were never used.
11269 (rs6000_option_override_internal): Delete TOC fusion, that was only
11270 partially defined, and it did not work unless you also used the
11272 (rs6000_legitimate_address_p): Delete TOC fusion support.
11273 (rs6000_opt_masks): Likewise.
11274 (fusion_wrap_memory_address): Delete function, no longer used.
11275 (fusion_split_address); Delete TOC fusion support.
11276 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
11277 longer used with toc fusion being deleted.
11278 (TARGET_TOC_FUSION_FP): Likewise.
11279 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
11281 (toc fusion spliter): Delete TOC fusion support.
11282 (toc_fusionload_<mode>): Likewise.
11283 (toc_fusionload_di): Likewise.
11284 (fusion_gpr_load_<mode>): Delete generator function, this insn no
11285 longer needs to be named. Rename predicate to delete TOC fusion.
11286 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
11287 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
11288 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
11289 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
11290 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
11292 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11294 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
11295 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
11296 __int128 in built-in function prototypes.
11297 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
11298 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
11300 2018-07-27 Martin Sebor <msebor@redhat.com>
11302 PR tree-optimization/86696
11303 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
11304 types, including enums.
11305 (handle_char_store): Be prepared for the above function to fail.
11307 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
11309 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
11310 when optimization level is lower than 2 or optimize for size.
11312 2018-07-26 Martin Sebor <msebor@redhat.com>
11314 PR tree-optimization/86043
11315 PR tree-optimization/86042
11316 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
11317 (get_string_cst_length): Rename...
11318 (get_min_string_length): ...to this. Add argument.
11319 (handle_char_store): Extend to handle multi-character stores by
11321 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
11322 * tree.h (initializer_zerop): Add argument.
11324 2018-07-26 Jakub Jelinek <jakub@redhat.com>
11326 PR middle-end/86660
11327 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
11328 declare target to variables if they have always,{to,from,tofrom} map
11331 2018-07-26 Martin Liska <mliska@suse.cz>
11334 * lto-wrapper.c: Add linker_output as prefix
11335 for ltrans_output_file.
11337 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
11339 PR rtl-optimization/85805
11340 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
11341 value for hard registers if that was written in the same mode.
11343 2018-07-26 Martin Liska <mliska@suse.cz>
11345 PR gcov-profile/86536
11346 * gcov.c (format_gcov): Use printf format %.*f directly
11347 and do not handle special values.
11349 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11351 * common/config/arc/arc-common.c (arc_option_optimization_table):
11352 Update default optimizations for size.
11354 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11356 * config/arc/arc.md (movsf_insn): Add short instruction selection.
11357 * config/arc/constraints.md (CfZ): New constraint.
11358 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
11359 (subsf3_fpu): Likewise.
11360 (cmpsf_fpu): Likewise.
11361 (cmpsf_fpu_uneq): Likewise.
11363 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11365 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
11367 (arc_is_uncached_mem_p): uncached applies to both the variable and
11370 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11372 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
11375 2018-07-25 David Malcolm <dmalcolm@redhat.com>
11377 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
11378 field "m_scopes" from vec to auto_vec.
11380 2018-07-25 Martin Liska <mliska@suse.cz>
11382 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
11385 2018-07-25 Richard Biener <rguenther@suse.de>
11388 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
11389 special wrt context_die late.
11390 (gen_subprogram_die): Re-use DIEs in local scope.
11392 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
11394 PR tree-optimization/86644
11395 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
11397 2018-07-25 Martin Liska <mliska@suse.cz>
11399 PR middle-end/86645
11400 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
11401 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
11403 2018-07-25 Martin Liska <mliska@suse.cz>
11406 * params.def: Explain ASan abbreviation and provide
11407 a documentation link.
11409 2018-07-24 Martin Sebor <msebor@redhat.com>
11411 PR tree-optimization/86622
11412 PR tree-optimization/86532
11413 * builtins.h (string_length): Declare.
11414 * builtins.c (c_strlen): Correct handling of non-constant offsets.
11415 (check_access): Be prepared for non-constant length ranges.
11416 (string_length): Make extern.
11417 * expr.c (string_constant): Only handle the minor non-constant
11418 array index. Use string_constant to compute the length of
11419 a generic string constant.
11421 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
11423 PR tree-optimization/86618
11424 * tree-vect-stmts.c (vectorizable_call): Don't take the address
11425 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
11427 2018-07-24 David Malcolm <dmalcolm@redhat.com>
11429 PR tree-optimization/86636
11430 * json.cc (json::object::set): Fix comment. Add assertions.
11431 (json::array::append): Move here from json.h. Add comment and an
11433 (json::string::string): Likewise.
11434 * json.h (json::array::append): Move to json.cc.
11435 (json::string::string): Likewise.
11436 * optinfo-emit-json.cc
11437 (optrecord_json_writer::impl_location_to_json): Assert that we
11438 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
11439 wrapper around it. Expand the location once, rather than three
11441 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
11442 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
11444 (optrecord_json_writer::optinfo_to_json): Likewise, in four
11445 places. Fix some overlong lines.
11447 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
11449 * config/aarch64/aarch64-simd.md
11450 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
11451 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
11452 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
11453 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
11454 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
11455 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
11456 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
11457 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
11458 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
11460 2018-07-24 Jakub Jelinek <jakub@redhat.com>
11462 PR middle-end/86627
11463 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
11464 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
11465 and abs_d == d, do the power of two handling if profitable.
11467 2018-07-24 Richard Biener <rguenther@suse.de>
11469 * match.pd: Add BIT_FIELD_REF canonicalizations.
11471 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11474 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
11476 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11478 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
11479 terminated STRING_CST object.
11481 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11483 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
11485 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
11487 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
11488 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
11489 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
11490 the elements into a register.
11491 (rs6000_split_v4si_init_di_reg): Delete.
11492 (rs6000_split_v4si_init): Delete.
11493 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
11494 (vsx_init_v4si): Rewrite as a define_expand.
11496 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
11498 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
11499 zero_extend argument from memory): New.
11501 2018-07-22 Martin Sebor <msebor@redhat.com>
11504 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
11505 diagnosing calls with unknown arguments unless -Walloca-larger-than
11506 is restricted to less than PTRDIFF_MAX bytes.
11508 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
11510 * doc/gcov.texi (Invoking Gcov): Editorial changes.
11512 2018-07-20 David Malcolm <dmalcolm@redhat.com>
11514 * pretty-print.c (text_info::set_location): Remove redundant
11515 "line_table" parameter from call to rich_location::set_range.
11517 2018-07-20 Martin Sebor <msebor@redhat.com>
11519 PR middle-end/82063
11520 * builtins.c (expand_builtin_alloca): Adjust.
11521 * calls.c (alloc_max_size): Simplify.
11522 * cgraphunit.c (cgraph_node::expand): Adjust.
11523 * common.opt (larger_than_size, warn_frame_larger_than): Remove
11525 (frame_larger_than_size): Same.
11526 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
11527 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
11529 * doc/invoke.texi (GCC Command Options): Document option arguments.
11530 Explain byte-size arguments and suffixes.
11531 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
11532 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
11533 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
11534 * doc/options.texi (UInteger): Expand.
11535 (Host_Wide_Int, ByteSize): Document new properties.
11536 * final.c (final_start_function_1): Include sizes in an error message.
11537 * function.c (frame_offset_overflow): Same.
11538 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
11539 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
11540 Diagnose unbounded alloca calls only for limits of less than
11542 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
11543 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
11545 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
11546 only for limits of less than PTRDIFF_MAX.
11547 * langhooks-def.h (lhd_handle_option): Change function argument
11549 * langhooks.c (lhd_handle_option): Same.
11550 * langhooks.h (handle_option): Same.
11551 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
11553 (var_type, var_type_struct): Same.
11554 (var_set): Handle ByteSize flag.
11555 * optc-gen.awk: Add comments to output to ease debugging. Make
11556 use of HOST_WIDE_INT where appropriate.
11557 * opts-gen-save.awk: Use %lx to format unsigned long.
11558 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
11559 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
11560 arguments. Parse bytes-size suffixes.
11561 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
11562 (enum_value_to_arg): Same.
11563 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
11564 (handle_option): Adjust.
11565 (generate_option): Change function argument to HOST_WIDE_INT.
11566 (cmdline_handle_error): Adjust.
11567 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
11568 (set_option): Change function argument to HOST_WIDE_INT.
11569 (option_enabled): Handle cl_host_wide_int.
11570 (get_option_state): Handle CLVC_SIZE.
11571 (control_warning_option): Same.
11572 * opts.c (common_handle_option): Change function argument to
11573 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
11574 OPT_Wvla_larger_than_.
11575 * opts.h (enum cl_var_type): Add an enumerator.
11576 * stor-layout.c (layout_decl): Print a more meaningful warning.
11577 * toplev.c (output_stack_usage): Adjust.
11579 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
11581 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
11582 call to inline_expand_builtin_string_cmp.
11583 (expand_builtin_strcmp): Likewise.
11584 (expand_builtin_strncmp): Likewise.
11585 (inline_string_cmp): Delete the last parameter, change char_type_node
11586 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
11588 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
11589 the inlining expansion on target where the type of the call has same or
11590 narrower precision than unsigned char.
11592 2018-07-20 David Malcolm <dmalcolm@redhat.com>
11594 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
11595 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
11596 * common.opt (fsave-optimization-record): New option.
11597 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
11598 * doc/invoke.texi (-fsave-optimization-record): New option.
11599 * dumpfile.c: Include "optinfo-emit-json.h".
11600 (struct kv_pair): Move to coretypes.h.
11601 (optgroup_options): Make non-static.
11602 (dump_context::end_scope): Call
11603 optimization_records_maybe_pop_dump_scope.
11604 * dumpfile.h (optgroup_options): New decl.
11605 * json.cc: New file.
11606 * json.h: New file.
11607 * optinfo-emit-json.cc: New file.
11608 * optinfo-emit-json.h: New file.
11609 * optinfo.cc: Include "optinfo-emit-json.h".
11610 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
11611 (optinfo_enabled_p): Check optimization_records_enabled_p.
11612 (optinfo_wants_inlining_info_p): Likewise.
11613 * optinfo.h: Update comment.
11614 * profile-count.c (profile_quality_as_string): New function.
11615 * profile-count.h (profile_quality_as_string): New decl.
11616 (profile_count::quality): New accessor.
11617 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
11618 and optinfo_emit_json_cc_tests.
11619 * selftest.h (selftest::json_cc_tests): New decl.
11620 (selftest::optinfo_emit_json_cc_tests): New decl.
11621 * toplev.c: Include "optinfo-emit-json.h".
11622 (compile_file): Call optimization_records_finish.
11623 (do_compile): Call optimization_records_start.
11624 * tree-ssa-live.c: Include optinfo.h.
11625 (remove_unused_scope_block_p): Retain inlining information if
11626 optinfo_wants_inlining_info_p returns true.
11628 2018-07-20 Richard Biener <rguenther@suse.de>
11631 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
11632 to cover -flto-partition=none.
11634 2018-07-20 Martin Liska <mliska@suse.cz>
11636 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
11637 (get_decl_source_range): Remove unused function.
11639 2018-07-20 Richard Biener <rguenther@suse.de>
11641 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
11642 (struct vn_phi_s): Likewise.
11643 (struct vn_reference_s): Likewise.
11644 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
11645 for searching the slot of an entry known to be in the hash itself.
11646 (vn_phi_hasher::equal): Likewise.
11647 (vn_reference_hasher::equal): Likewise.
11648 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
11650 (optimistic_info, current_info): Remove, keeping only valid_info.
11651 (vn_reference_lookup_1): Remove fallback lookup.
11652 (vn_reference_lookup_2): Likewise.
11653 (vn_nary_op_lookup_1): Likewise.
11654 (vn_phi_lookup): Likewise.
11655 (vn_nary_build_or_lookup_1): Make sure to not chain the built
11657 (vn_reference_insert): Adjust, chain the inserted hash element
11658 at last_inserted_ref.
11659 (vn_reference_insert_pieces): Likewise.
11660 (visit_reference_op_call): Likewise.
11661 (vn_nary_op_insert_into): Chain the inserted hash element at
11662 last_inserted_nary.
11663 (vn_nary_op_insert_pieces): Adjust.
11664 (vn_nary_op_insert): Likewise.
11665 (vn_nary_op_insert_stmt): Likewise.
11666 (vn_phi_insert): Adjust, chain the inserted hash element at
11668 (process_scc): Remove clearing and copying the optimistic
11669 table. Instead remove elements inserted during an optimistic
11670 iteration from the single table we maintain.
11671 (init_scc_vn): Adjust.
11672 (free_scc_vn): Likewise.
11673 (sccvn_dom_walker::record_cond): Likewise.
11674 (sccvn_dom_walker::after_dom_children): Likewise.
11676 2018-07-19 Martin Sebor <msebor@redhat.com>
11678 PR tree-optimization/84047
11679 PR tree-optimization/83776
11680 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
11681 (check_array_bounds): Call it.
11683 2018-07-19 Martin Sebor <msebor@redhat.com>
11685 * align.h (align_flags): Use member initialization.
11687 2018-07-19 David Malcolm <dmalcolm@redhat.com>
11689 * Makefile.in (OBJS): Add optinfo.o.
11690 * coretypes.h (class symtab_node): New forward decl.
11691 (struct cgraph_node): New forward decl.
11692 (class varpool_node): New forward decl.
11693 * dump-context.h: New file.
11694 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
11696 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
11697 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
11698 (set_alt_dump_file): Likewise.
11699 (dump_context::~dump_context): New dtor.
11700 (dump_gimple_stmt): Move implementation to...
11701 (dump_context::dump_gimple_stmt): ...this new member function.
11702 Add the stmt to any pending optinfo, creating one if need be.
11703 (dump_gimple_stmt_loc): Move implementation to...
11704 (dump_context::dump_gimple_stmt_loc): ...this new member function.
11705 Start a new optinfo and add the stmt to it.
11706 (dump_gimple_expr): Move implementation to...
11707 (dump_context::dump_gimple_expr): ...this new member function.
11708 Add the stmt to any pending optinfo, creating one if need be.
11709 (dump_gimple_expr_loc): Move implementation to...
11710 (dump_context::dump_gimple_expr_loc): ...this new member function.
11711 Start a new optinfo and add the stmt to it.
11712 (dump_generic_expr): Move implementation to...
11713 (dump_context::dump_generic_expr): ...this new member function.
11714 Add the tree to any pending optinfo, creating one if need be.
11715 (dump_generic_expr_loc): Move implementation to...
11716 (dump_context::dump_generic_expr_loc): ...this new member
11717 function. Add the tree to any pending optinfo, creating one if
11719 (dump_printf): Move implementation to...
11720 (dump_context::dump_printf_va): ...this new member function. Add
11721 the text to any pending optinfo, creating one if need be.
11722 (dump_printf_loc): Move implementation to...
11723 (dump_context::dump_printf_loc_va): ...this new member function.
11724 Start a new optinfo and add the stmt to it.
11725 (dump_dec): Move implementation to...
11726 (dump_context::dump_dec): ...this new member function. Add the
11727 value to any pending optinfo, creating one if need be.
11728 (dump_context::dump_symtab_node): New member function.
11729 (dump_context::get_scope_depth): New member function.
11730 (dump_context::begin_scope): New member function.
11731 (dump_context::end_scope): New member function.
11732 (dump_context::ensure_pending_optinfo): New member function.
11733 (dump_context::begin_next_optinfo): New member function.
11734 (dump_context::end_any_optinfo): New member function.
11735 (dump_context::s_current): New global.
11736 (dump_context::s_default): New global.
11737 (dump_scope_depth): Delete global.
11738 (dumpfile_ensure_any_optinfo_are_flushed): New function.
11739 (dump_symtab_node): New function.
11740 (get_dump_scope_depth): Reimplement in terms of dump_context.
11741 (dump_begin_scope): Likewise.
11742 (dump_end_scope): Likewise.
11743 (selftest::temp_dump_context::temp_dump_context): New ctor.
11744 (selftest::temp_dump_context::~temp_dump_context): New dtor.
11745 (selftest::verify_item): New function.
11746 (ASSERT_IS_TEXT): New macro.
11747 (ASSERT_IS_TREE): New macro.
11748 (ASSERT_IS_GIMPLE): New macro.
11749 (selftest::test_capture_of_dump_calls): New test.
11750 (selftest::dumpfile_c_tests): Call it.
11751 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
11752 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
11753 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
11754 descriptive comment.
11755 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
11756 (dump_node, dump_bb): Move these unrelated decls.
11757 (class dump_manager): Add leading comment.
11758 * optinfo.cc: New file.
11759 * optinfo.h: New file.
11761 2018-07-19 Michael Collison <michael.collison@arm.com>
11762 Richard Henderson <rth@redhat.com>
11764 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
11765 (subti): Handle op1 zero.
11766 (subvti4, usub4ti4): New.
11767 (*sub<GPI>3_compare1_imm): New.
11768 (sub<GPI>3_carryinCV): New.
11769 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
11770 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
11772 2018-07-19 Michael Collison <michael.collison@arm.com>
11773 Richard Henderson <rth@redhat.com>
11775 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
11776 (addti3): Create simpler code if low part is already known to be 0.
11777 (addvti4, uaddvti4): New.
11778 (*add<GPI>3_compareC_cconly_imm): New.
11779 (*add<GPI>3_compareC_cconly): New.
11780 (*add<GPI>3_compareC_imm): New.
11781 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
11782 handle constants within this pattern..
11783 (*add<GPI>3_compareV_cconly_imm): New.
11784 (*add<GPI>3_compareV_cconly): New.
11785 (*add<GPI>3_compareV_imm): New.
11786 (add<GPI>3_compareV): New.
11787 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
11788 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
11789 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
11790 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
11791 with 'comparison' operator.
11792 (*add<GPI>3_compareV_cconly_imm): Ditto.
11793 (*add<GPI>3_compareV_cconly): Ditto.
11794 (*add<GPI>3_compareV_imm): Ditto.
11795 (add<GPI>3_compareV): Ditto.
11796 (add<mode>3_carryinC): Ditto.
11797 (*add<mode>3_carryinC_zero): Ditto.
11798 (*add<mode>3_carryinC): Ditto.
11799 (add<mode>3_carryinV): Ditto.
11800 (*add<mode>3_carryinV_zero): Ditto.
11801 (*add<mode>3_carryinV): Ditto.
11803 2018-07-19 Michael Collison <michael.collison@arm.com>
11804 Richard Henderson <rth@redhat.com>
11806 * config/aarch64/aarch64-modes.def (CC_V): New.
11807 * config/aarch64/aarch64-protos.h
11808 (aarch64_addti_scratch_regs): Declare
11809 (aarch64_subvti_scratch_regs): Declare.
11810 (aarch64_expand_subvti): Declare.
11811 (aarch64_gen_unlikely_cbranch): Declare
11812 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
11813 for signed overflow using CC_Vmode.
11814 (aarch64_get_condition_code_1): Handle CC_Vmode.
11815 (aarch64_gen_unlikely_cbranch): New function.
11816 (aarch64_addti_scratch_regs): New function.
11817 (aarch64_subvti_scratch_regs): New function.
11818 (aarch64_expand_subvti): New function.
11820 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
11822 * config/aarch64/aarch64-option-extensions.def: New entry for profile
11824 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
11825 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
11828 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
11831 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11832 address check not strict.
11834 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
11836 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
11838 (store_pair_lanes<mode>): Likewise.
11839 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
11840 enum value 'ADDR_QUERY_LDP_STP_N'.
11841 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
11842 (aarch64_print_address_internal): Add declaration.
11843 (aarch64_print_ldpstp_address): Remove.
11844 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
11845 (aarch64_print_operand): Change printing of 'y'.
11846 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
11847 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
11848 'true' rather than '1'.
11849 * config/aarch64/constraints.md (Uml): Likewise.
11850 (Uml): Rename to Umn.
11853 2018-07-19 Richard Biener <rguenther@suse.de>
11855 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
11857 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
11858 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
11859 (vn_reference_hasher): Likewise.
11860 (struct vn_tables_s): Remove obstack and alloc-pool members.
11861 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
11862 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
11863 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
11864 (vn_reference_insert_pieces): Likewise.
11865 (alloc_vn_nary_op_noinit): Adjust.
11866 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
11867 (vn_phi_eq): Adjust.
11868 (shared_lookup_phiargs): Remove.
11869 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
11870 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
11871 (visit_reference_op_call): Likewise.
11872 (copy_nary, copy_phi, copy_reference): Remove.
11873 (process_scc): Rewind the obstack when iterating. Do not
11874 copy the elements to valid_info but just move them from one
11875 hashtable to the other.
11876 (allocate_vn_table): Adjust.
11877 (free_vn_table): Likewise.
11878 (init_scc_vn): Likewise.
11879 (free_scc_vn): Likewise.
11881 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
11884 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
11885 indirect_return as function type attribute.
11886 (ix86_attribute_table): Change indirect_return to function
11888 * doc/extend.texi: Update indirect_return attribute.
11890 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
11892 * wide-int.h (widest2_int): New.
11893 * gimple-fold.c (arith_overflowed_p): Use it.
11894 * tree.h (widest2_int_cst): New.
11895 * tree-vrp.c (wide_int_binop_overflow): Rename from
11896 vrp_int_const_binop.
11897 Rewrite to work on trees.
11898 (extract_range_from_multiplicative_op_1): Abstract code to...
11899 (wide_int_range_min_max): ...here.
11900 (wide_int_range_cross_product): ...and here.
11901 (extract_range_from_binary_expr_1): Abstract overflow code to...
11902 (wide_int_range_mult_wrapping): ...here.
11903 * tree-vrp.h (wide_int_range_cross_product): New.
11904 (wide_int_range_mult_wrapping): New.
11906 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
11907 Julia Koval <julia.koval@intel.com>
11909 * config/i386/x86-tune-costs.h (skylake_memcpy,
11910 skylake_memset): Replace rep_prefix with unrolling for size 512.
11912 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
11914 PR middle-end/86544
11915 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
11916 comparision with EQ_EXPR in last stmt.
11918 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
11920 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
11921 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
11922 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
11923 previously known as "PowerPC AltiVec Built-in Functions". Move
11924 some material to new subsubsections "PowerPC AltiVec Built-in
11925 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
11927 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
11928 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
11929 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
11930 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
11932 2018-07-18 Richard Biener <rguenther@suse.de>
11934 PR tree-optimization/86557
11935 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
11938 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
11940 * config/s390/s390.c (s390_function_profiler): Generate CFI.
11942 2018-07-17 Jeff Law <law@redhat.com>
11944 * config/arm/arm.c (get_label_padding): Update for recent
11945 changes to label_to_alignment.
11947 PR tree-optimization/86010
11948 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
11950 * config/mips/mips.c (vr4130_align_insns): Update for recent
11951 changes to label_to_alignment.
11953 * config/frv/frv.c (frv_label_align): Update for recent changes
11954 to label_to_alignment.
11956 * config/nios2/nios2.c (nios2_label_align): Update for recent
11957 changes which dropped ALIGN_LABELS_LOG.
11959 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
11961 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
11962 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
11964 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11966 * config/arc/arc.c (arc_label_align): Use align_labels instead of
11967 deprecated align_labels_log.
11969 2018-07-17 Richard Biener <rguenther@suse.de>
11972 * dwarf2out.c (init_sections_and_labels): Always generate
11973 a debug_line_str_section for early LTO debug.
11974 (dwarf2out_finish): Reset debug_line_str_hash output early.
11975 Bump counter for extra dwarf5 .debug_loc labels to not conflict
11977 (dwarf2out_early_finish): Output debug_line_str.
11979 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
11981 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
11982 index register on z196 or later.
11984 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
11986 * config/s390/s390.c (s390_default_align): Set default function
11988 (s390_override_options_after_change): Call s390_default align.
11989 (s390_option_override_internal): Call s390_default_align.
11990 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
11992 2018-07-17 Jakub Jelinek <jakub@redhat.com>
11994 PR middle-end/86542
11995 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
11996 to _looptemp_ clauses, other than the first two.
11998 2018-07-17 Martin Liska <mliska@suse.cz>
12000 * opts.c: Do not enable OPT_falign_* for -Os.
12002 2018-07-17 Martin Liska <mliska@suse.cz>
12004 * align.h (MAX_CODE_ALIGN): New.
12005 (MAX_CODE_ALIGN_VALUE): New.
12006 * common/config/i386/i386-common.c (ix86_handle_option):
12007 (MAX_CODE_ALIGN): Moved to align.h.
12008 * final.c (MAX_CODE_ALIGN): Likewise.
12009 * opts.c (parse_and_check_align_values):
12010 (MAX_CODE_ALIGN): Likewise.
12011 (MAX_CODE_ALIGN_VALUE): Likewise.
12013 2018-07-17 Martin Liska <mliska@suse.cz>
12015 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
12016 in order to fulfil coding style.
12017 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
12018 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12019 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
12020 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
12021 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
12022 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
12023 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
12024 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
12025 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12027 2018-07-17 Martin Liska <mliska@suse.cz>
12029 * align.h: New file.
12030 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
12032 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
12033 align_flags of label_to_alignment.
12034 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
12036 * config/m68k/m68k.c: Do not use removed align_labels_value and
12038 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
12039 (LOOP_ALIGN): Likewise.
12040 (LABEL_ALIGN): Likewise.
12041 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12042 Remove not used macro.
12043 (rs6000_loop_align): Change return type to align_flags.
12044 (rs6000_loop_align_max_skip): Remove.
12045 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
12046 Change return type to align_flags.
12047 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12048 Remove not used macro.
12049 (rs6000_loop_align): Change return type to align_flags.
12050 (rs6000_loop_align_max_skip): Remove.
12051 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
12052 * config/rx/rx-protos.h (rx_align_for_label): Make it
12054 * config/rx/rx.c (rx_align_for_label): Change return type
12056 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
12058 into align_flags class.
12059 (LABEL_ALIGN): Likewise.
12060 (LOOP_ALIGN): Likewise.
12061 * config/s390/s390.c (s390_label_align): Use align_flags
12063 (s390_asm_output_function_label): Likewise.
12064 * config/sh/sh.c (sh_override_options_after_change):
12065 Use align_flags class directly without macros.
12066 (find_barrier): Likewise.
12067 (barrier_align): Likewise.
12068 (sh_loop_align): Likewise.
12069 * config/spu/spu.c (spu_option_override):
12070 Use align_flags_tuple::get_value instead of removed macros.
12071 (spu_sched_init): Likewise.
12072 * config/spu/spu.h (GTY): Likewise.
12073 * config/visium/visium.c (visium_option_override):
12074 Set "8" as default secondary alignment.
12075 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
12076 in order to guarantee secondary alignment of 8.
12077 * coretypes.h: Include align.h header file.
12078 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
12079 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
12080 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
12081 * doc/tm.texi.in: Likewise.
12082 * final.c (struct label_alignment): Remove not used structure.
12083 (LABEL_ALIGN): Change type to align_flags.
12084 (LOOP_ALIGN): Likewise.
12085 (JUMP_ALIGN): Likewise.
12086 (default_loop_align_max_skip): Remove.
12087 (default_label_align_max_skip): Likewise.
12088 (default_jump_align_max_skip): Likewise.
12089 (default_label_align_after_barrier_max_skip):
12090 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
12091 (LABEL_TO_MAX_SKIP): Remove.
12092 (label_to_alignment): Return align_flags type instead of integer.
12093 (label_to_max_skip): Remove.
12094 (align_fuzz): Use align_flags type.
12095 (compute_alignments): Use align_flags type and use align_flags::max
12096 to combine multiple alignments.
12097 (grow_label_align): Grow vec instead of C array.
12098 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
12099 (shorten_branches): Use align_flags type and use align_flags::max
12100 to combine multiple alignments.
12101 (final_scan_insn_1): Remove usage of secondary alignment that comes
12102 from label alignment, but instead use proper secondary alignment
12103 which is computed in grow_label_align.
12104 * flags.h (struct align_flags_tuple): Move to align.h.
12105 (struct align_flags): Likewise.
12106 (state_align_loops): Rename to align_loops.
12107 (state_align_jumps): Rename to align_jumps.
12108 (state_align_labels): Rename to align_labels.
12109 (state_align_functions): Rename to align_functions.
12110 (align_loops_log): Remove.
12111 (align_jumps_log): Remove.
12112 (align_labels_log): Remove.
12113 (align_functions_log): Remove.
12114 (align_loops_max_skip): Remove.
12115 (align_jumps_max_skip): Remove.
12116 (align_labels_max_skip): Remove.
12117 (align_functions_max_skip): Remove.
12118 (align_loops_value): Remove.
12119 (align_jumps_value): Remove.
12120 (align_labels_value): Remove.
12121 (align_functions_value): Remove.
12122 * output.h (label_to_alignment): Change return type to align_flags.
12123 (label_to_max_skip): Remove.
12124 * target.def: Remove loop_align_max_skip, label_align_max_skip,
12125 jump_align_max_skip macros.
12126 * targhooks.h (default_loop_align_max_skip): Remove.
12127 (default_label_align_max_skip): Likewise.
12128 (default_jump_align_max_skip): Likewise.
12129 (default_label_align_after_barrier_max_skip): Remove.
12130 * toplev.c (read_log_maxskip): Use ::normalize function.
12131 (parse_N_M): Remove not used argument and also call ::normalize.
12132 (parse_alignment_opts): Do not pass unused arguments.
12133 * varasm.c (assemble_start_function): Use directly align_functions
12134 instead of removed macros.
12135 * system.h: Do not poison removed macros.
12137 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12139 PR middle-end/86539
12140 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
12141 and cond temporaries don't have reference type if iterator has
12142 pointer type. For init use &for_pre_body instead of pre_p if
12143 for_pre_body is non-empty.
12145 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12147 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
12148 double-double modes to SFmode directly directly.
12149 (trunc<mode>sf2_fprs): Delete.
12151 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12153 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
12154 for conversions between IFmode and the decimal floating point modes.
12155 (init_float128_ieee): Use the correct names for conversions between
12156 KFmode and the decimal floating point modes.
12158 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12160 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
12161 for the conversions between TDmode and IFmode.
12162 (init_float128_ieee): Use more correct names for the conversions
12163 between TDmode and KFmode.
12165 2018-07-16 Jakub Jelinek <jakub@redhat.com>
12167 PR tree-optimization/86526
12168 * builtins.c (expand_builtin_memcmp): Formatting fixes.
12169 (inline_expand_builtin_string_cmp): Likewise.
12170 (inline_string_cmp): Likewise. Use c_readstr instead of
12171 builtin_memcpy_read_str. Add unit_mode temporary.
12173 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
12175 PR middle-end/86528
12176 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
12177 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
12179 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
12181 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
12182 Alphabetize prototypes of built-in functions, separating out
12183 built-in functions that are listed in this section but should be
12184 described elsewhere.
12186 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
12189 * expmed.c (emit_store_flag): Do not emit setcc followed by a
12190 conditional move when trapping comparison was split to a
12191 non-trapping one (and vice versa).
12193 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12195 * config/s390/s390.c (s390_function_profiler): Generate nops
12196 instead of profiler call sequences.
12197 * config/s390/s390.opt: Add the new option.
12199 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12201 * config/s390/s390.c (s390_function_profiler): Generate
12202 __mcount_loc section.
12203 * config/s390/s390.opt: Add the new option.
12205 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12207 * common.opt: Add the new warning.
12208 * config/s390/s390.c (s390_function_profiler): Emit "brasl
12209 %r0,__fentry__" when -mfentry is specified.
12210 (s390_option_override_internal): Disallow -mfentry for 31-bit
12212 * config/s390/s390.opt: Add the new option.
12214 2018-07-16 Richard Biener <rguenther@suse.de>
12217 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
12218 for function-local FUNCTION_DECL and RESULT_DECL immediately.
12220 2018-07-16 Martin Liska <mliska@suse.cz>
12223 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
12226 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
12228 * config/arc/arcHS.md: Update ARCHS scheduling rules.
12230 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
12232 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
12234 * config/arc/arc-cpus.def (hs4x): New cpu.
12236 * config/arc/arc-tables.opt: Regenerate.
12237 * config/arc/arc.c (arc_sched_issue_rate): New function.
12238 (TARGET_SCHED_ISSUE_RATE): Define.
12239 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
12240 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
12242 (attr tune): Add ARCHS4x tune values.
12243 (attr tune_dspmpy): Define.
12244 (*tst): Correct instruction type.
12245 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
12246 * config/arc/arcHS4x.md: New file.
12247 * config/arc/fpu.md: Update instruction type attributes.
12248 * config/arc/t-multilib: Regenerate.
12250 2018-07-16 Tom de Vries <tdevries@suse.de>
12253 * var-tracking.c (vt_initialize): Fix pre_dec handling.
12255 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12257 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
12260 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
12262 PR tree-optimization/86514
12263 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
12264 conversion to a boolean type from a type with greater precision.
12266 2018-07-16 Tom de Vries <tdevries@suse.de>
12268 * var-tracking.c (vt_initialize): Print adjusted insn slim if
12269 dump_flags request TDF_SLIM.
12271 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
12273 * fold-const.c (int_const_binop_1): Abstract...
12274 (wide_int_binop): ...wide int code here.
12275 (poly_int_binop): ...poly int code here.
12276 Abstract the rest of int_const_binop_1 into int_const_binop.
12277 * fold-const.h (wide_int_binop): New.
12278 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
12279 Remove useless PLUS/MINUS_EXPR case.
12280 (zero_nonzero_bits_from_vr): Move wide int code...
12281 (zero_nonzero_bits_from_bounds): ...here.
12282 (extract_range_from_binary_expr_1): Move mask optimization code...
12283 (range_easy_mask_min_max): ...here.
12284 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
12285 (range_easy_mask_min_max): New.
12287 2018-07-15 Jeff Law <law@redhat.com>
12290 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
12293 2018-07-14 Jim Wilson <jimw@sifive.com>
12295 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
12297 2018-07-14 Paul Koning <ni1d@arrl.net>
12299 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
12301 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
12303 * lto-streamer-out.c (copy_function_or_variable): Dump info about
12306 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
12307 Steve Munroe <munroesj52@gmail.com>
12309 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
12310 (_mm_andnot_si128): Likewise.
12311 (_mm_or_si128): Likewise.
12312 (_mm_xor_si128): Likewise.
12314 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
12316 PR middle-end/78809
12317 * builtins.c (expand_builtin_memcmp): Inline the calls first
12318 when result_eq is false.
12319 (expand_builtin_strcmp): Inline the calls first.
12320 (expand_builtin_strncmp): Likewise.
12321 (inline_string_cmp): New routine. Expand a string compare
12322 call by using a sequence of char comparison.
12323 (inline_expand_builtin_string_cmp): New routine. Inline expansion
12324 a call to str(n)cmp/memcmp.
12325 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
12327 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
12329 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
12331 * config/arm/driver-arm.c: Include arm-native.h.
12332 (host_detect_local_cpu): Use auto-generated data tables.
12333 (vendors, arm_cpu_table): Delete. Move part information to ...
12334 * config/arm/arm-cpus.in: ... here.
12335 * config/arm/parsecpu.awk (gen_native): New function.
12336 (vendor, part): New CPU fields.
12337 (END): Add support for building the native CPU detection tables.
12338 * config/arm/t-arm (arm-native.h): Add build rule.
12339 (driver-arm.o): Add dependency on arm-native.h.
12341 2018-07-13 Richard Biener <rguenther@suse.de>
12343 PR middle-end/85974
12344 * match.pd (addr1 - addr2): Allow either of the operand to
12347 2018-07-13 Tom de Vries <tdevries@suse.de>
12349 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
12352 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
12354 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
12355 arrays instead of numbered variables.
12357 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
12359 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
12360 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
12362 2018-07-13 Richard Biener <rguenther@suse.de>
12365 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
12366 instead of get_context_die.
12368 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12369 Richard Biener <rguenther@suse.de>
12371 PR middle-end/86489
12372 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
12373 that the loop latch destination where phi is defined.
12375 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
12377 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
12378 (riscv_expand_epilogue): Add assertion to check interrupt mode.
12379 (riscv_set_current_function): Extract getting interrupt type to new
12381 (riscv_get_interrupt_type): New function.
12382 (riscv_merge_decl_attributes): New function, checking interrupt type is
12384 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
12386 2018-07-12 Paul Koning <ni1d@arrl.net>
12388 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
12391 2018-07-12 Paul Koning <ni1d@arrl.net>
12393 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
12394 zero reference, add doloop_end instead.
12395 * doc/md.texi (decrement_and_branch_until_zero): Remove.
12396 (Looping patterns): Remove decrement_and_branch_until_zero. Add
12397 detail for doloop_end.
12399 2018-07-12 Martin Sebor <msebor@redhat.com>
12402 * attribs.c (decl_attributes): Reject conflicting attributes before
12403 calling attribute handlers.
12405 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
12407 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
12409 (gcc::dump_manager::get_dump_file_name): likewise.
12410 (dump_begin): Likewise.
12411 * dumpfile.h (dump_begin): Update prototype.
12412 (gcc::dump_manager::get_dump_file_name,
12413 gcc::dump_manager::get_dump_file_name): Update prototype.
12415 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12417 * internal-fn.h (vectorizable_internal_fn_p): New function.
12418 * tree-vect-slp.c (compatible_calls_p): Likewise.
12419 (vect_build_slp_tree_1): Remove nops argument. Handle calls
12420 to internal functions.
12421 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
12423 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12425 * fold-const.h (inverse_conditions_p): Declare.
12426 * fold-const.c (inverse_conditions_p): New function.
12427 * match.pd: Use inverse_conditions_p. Add folds of view_converts
12428 that test the inverse condition of a conditional internal function.
12429 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
12430 * internal-fn.c (internal_fn_mask_index): Handle conditional
12431 internal functions.
12432 (vectorized_internal_fn_supported_p): New function.
12433 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
12434 (any_pred_load_store): Replace with...
12435 (need_to_predicate): ...this new variable.
12436 (redundant_ssa_names): New variable.
12437 (ifcvt_can_use_mask_load_store): Move initial checks to...
12438 (ifcvt_can_predicate): ...this new function. Handle tree codes
12439 for which a conditional internal function exists.
12440 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
12441 instead of ifcvt_can_use_mask_load_store. Update after variable
12443 (predicate_load_or_store): New function, split out from
12444 predicate_mem_writes.
12445 (check_redundant_cond_expr): New function.
12446 (value_available_p): Likewise.
12447 (predicate_rhs_code): Likewise.
12448 (predicate_mem_writes): Rename to...
12449 (predicate_statements): ...this. Use predicate_load_or_store
12450 and predicate_rhs_code.
12451 (combine_blocks, tree_if_conversion): Update after above name changes.
12452 (ifcvt_local_dce): Handle redundant_ssa_names.
12453 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
12454 general conditional functions.
12455 * tree-vect-stmts.c (vectorizable_call): Likewise.
12457 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12458 Alan Hayward <alan.hayward@arm.com>
12459 David Sherwood <david.sherwood@arm.com>
12461 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
12462 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
12463 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
12464 plus and minus and convert them into IFN_COND_FMA-based sequences.
12465 (convert_mult_to_fma): Handle conditional plus and minus.
12467 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12469 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
12470 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
12471 (cond_fnms_optab): New optabs.
12472 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
12473 internal functions.
12474 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
12475 * internal-fn.h (get_conditional_internal_fn): Declare.
12476 (get_unconditional_internal_fn): Likewise.
12477 * internal-fn.c (cond_ternary_direct): New macro.
12478 (expand_cond_ternary_optab_fn): Likewise.
12479 (direct_cond_ternary_optab_supported_p): Likewise.
12480 (FOR_EACH_COND_FN_PAIR): Likewise.
12481 (get_conditional_internal_fn): New function.
12482 (get_unconditional_internal_fn): Likewise.
12483 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
12484 (gimple_match_op::gimple_match_op): Add a new overload for 5
12486 (gimple_match_op::set_op): Likewise.
12487 (gimple_resimplify5): Declare.
12488 * genmatch.c (decision_tree::gen): Generate simplifications for
12490 * gimple-match-head.c (gimple_simplify): Define an overload for
12491 5 operands. Handle calls with 5 arguments in the top-level overload.
12492 (convert_conditional_op): Handle conversions from unconditional
12493 internal functions to conditional ones.
12494 (gimple_resimplify5): New function.
12495 (build_call_internal): Pass a fifth operand.
12496 (maybe_push_res_to_seq): Likewise.
12497 (try_conditional_simplification): Try converting conditional
12498 internal functions to unconditional internal functions.
12499 Handle 3-operand unconditional forms.
12500 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
12501 Define ternary equivalents of the current rules for binary conditional
12502 internal functions.
12503 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
12504 ternary operations.
12505 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
12506 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
12507 (optab): Handle them.
12508 (SVE_COND_FP_TERNARY): New int iterator.
12509 (sve_fmla_op, sve_fmad_op): New int attributes.
12510 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
12511 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
12512 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
12514 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12516 * target.def (preferred_else_value): New target hook.
12517 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
12518 * doc/tm.texi: Regenerate.
12519 * targhooks.h (default_preferred_else_value): Declare.
12520 * targhooks.c (default_preferred_else_value): New function.
12521 * internal-fn.h (conditional_internal_fn_code): Declare.
12522 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
12523 (get_conditional_internal_fn): Use it.
12524 (conditional_internal_fn_code): New function.
12525 * gimple-match.h (gimple_match_cond): New struct.
12526 (gimple_match_op): Add a cond member function.
12527 (gimple_match_op::gimple_match_op): Update all forms to take a
12529 * genmatch.c (expr::gen_transform): Use the same condition as res_op
12530 for the suboperation, but don't specify a particular else_value.
12531 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
12532 (visit_nary_op, visit_reference_op_load): Pass
12533 gimple_match_cond::UNCOND to the gimple_match_op constructor.
12534 * gimple-match-head.c: Include tree-eh.h
12535 (convert_conditional_op): New function.
12536 (maybe_resimplify_conditional_op): Likewise.
12537 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
12538 (gimple_resimplify2): Likewise.
12539 (gimple_resimplify3): Likewise.
12540 (gimple_resimplify4): Likewise.
12541 (maybe_push_res_to_seq): Return null for conditional operations.
12542 (try_conditional_simplification): New function.
12543 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
12545 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
12547 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
12549 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
12551 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
12553 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12555 (hash_tree): Do not hash DECL_FCONTEXT
12556 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
12557 Do not stream DECL_FCONTEXT.
12558 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
12559 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
12561 2018-07-12 Richard Biener <rguenther@suse.de>
12564 * dwarf2out.c (gen_block_die): Only output blocks when they have
12565 at least one !DECL_IGNORED_P variable.
12567 2018-07-12 Richard Biener <rguenther@suse.de>
12570 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
12571 Remove -mieee-fp handling.
12573 2018-07-12 Richard Biener <rguenther@suse.de>
12575 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
12576 left-over from last patch.
12578 2018-07-12 Jakub Jelinek <jakub@redhat.com>
12580 PR tree-optimization/86492
12581 * gimple-ssa-store-merging.c
12582 (imm_store_chain_info::coalesce_immediate_stores): Call
12583 check_no_overlap even for the merge_overlapping case. Formatting fix.
12585 2018-07-12 Richard Biener <rguenther@suse.de>
12587 PR middle-end/86479
12588 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
12589 move possibly trapping operations into the conditional.
12591 2018-07-12 Richard Biener <rguenther@suse.de>
12593 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
12594 (vn_lookup_simplify_result): Remove recursion limit applied
12596 (vn_nary_build_or_lookup_1): Adjust.
12597 (try_to_simplify): Likewise.
12598 * gimple-match-head.c (gimple_resimplify1): Instead apply one
12600 (gimple_resimplify2): Likewise.
12601 (gimple_resimplify3): Likewise.
12602 (gimple_resimplify4): Likewise.
12604 2018-07-11 Jakub Jelinek <jakub@redhat.com>
12606 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
12607 Use __mmask64 type instead of __mmask8 for __M argument.
12608 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
12609 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
12611 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
12612 __mmask16 for __M argument.
12613 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
12614 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
12615 to __mmask16 instead of __mmask8.
12616 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
12617 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
12618 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
12619 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
12620 instead of __mmask16 for __U argument.
12621 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
12622 __mmask16 instead of __mmask8 for __U argument.
12623 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
12625 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
12627 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
12629 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
12631 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
12633 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
12635 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
12637 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
12639 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
12641 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
12643 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
12644 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
12645 return type as well as __M argument type and all casts from __mmask8
12647 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
12648 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
12649 return type as well as __M argument type and all casts from __mmask8
12651 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
12652 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
12653 return type as well as __M argument type and all casts from __mmask8
12655 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
12656 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
12657 return type as well as __M argument type and all casts from __mmask8
12659 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
12660 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
12663 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
12665 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
12666 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
12669 2018-07-11 Paul Koning <ni1d@arrl.net>
12671 * doc/md.texi (define_subst): Document how multiple occurrences of
12672 the same argument in the replacement pattern are handled.
12674 2018-07-11 Paul Koning <ni1d@arrl.net>
12676 * doc/extend.texi (Common Variable Attributes): Move "mode" into
12677 alphabetical order.
12678 (Common Type Attributes): Add "mode" attribute.
12680 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
12682 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
12683 stream DECL_ORIGINAL_TYPE.
12684 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
12685 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
12686 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
12687 Do not walk original type.
12688 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
12690 (write_ts_decl_non_common_tree_pointers): Do not stream
12692 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
12693 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
12695 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
12697 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
12698 thread twice from the same starting edge.
12700 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
12702 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
12703 * gimple.c (gimple_call_nonnull_result_p): ...here...
12704 (gimple_call_nonnull_arg): ...and here.
12705 * gimple.h (gimple_call_nonnull_result_p): New.
12706 (gimple_call_nonnull_arg): New.
12708 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
12710 * config/arm/arm-cpus.in: Move information from fpu field of each
12711 cpu definition to the isa field.
12712 * config/arm/parsecpu.awk (fpu): Delete match rule.
12713 (gen_comm_data): Don't add bits from the CPU's FPU entry.
12715 2018-07-11 Richard Biener <rguenther@suse.de>
12718 * dwarf2out.c (init_sections_and_labels): Use
12719 output_asm_line_debug_info consistently.
12720 (dwarf2out_early_finish): Likewise.
12721 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
12724 2018-07-11 Richard Biener <rguenther@suse.de>
12726 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
12727 Rework father_bb setting in a way to avoid propagating constants
12728 multiple times on a loop body.
12730 2018-07-10 Mark Wielaard <mark@klomp.org>
12733 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
12735 2018-07-10 Richard Biener <rguenther@suse.de>
12737 * hash-map.h (hash_map::iterator::operator*): Return
12738 references to key and value.
12740 2018-07-10 Jakub Jelinek <jakub@redhat.com>
12743 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
12744 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
12745 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
12746 (gimplify_omp_for): For composite loops, move outer
12747 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
12748 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
12749 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
12750 TREE_LIST for both the original class iterator and the "last" helper
12751 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
12752 loop, remember has_decl_expr from outer composite loops for the
12753 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
12755 2018-07-09 Martin Sebor <msebor@redhat.com>
12757 PR middle-end/77357
12758 PR middle-end/86428
12759 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
12760 accessing implicitly initialized array elements.
12761 * expr.c (string_constant): Handle string initializers of
12762 character arrays within aggregates.
12763 * gimple-fold.c (fold_array_ctor_reference): Add argument.
12764 Store element offset. As a special case, handle zero size.
12765 (fold_nonarray_ctor_reference): Same.
12766 (fold_ctor_reference): Add argument. Store subobject offset.
12767 * gimple-fold.h (fold_ctor_reference): Add argument.
12769 2018-07-09 Paul Koning <ni1d@arrl.net>
12771 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
12772 (pdp11_insn_cost): New function.
12773 (pdp11_md_asm_adjust): New function.
12774 (TARGET_INVALID_WITHIN_DOLOOP): Define.
12775 (pdp11_rtx_costs): Update to match machine better.
12776 (output_addr_const_pdp11): Correct format mismatch warnings.
12777 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
12778 * config/pdp11/pdp11.md: General change to add base_cost and/or
12779 length attributes for use by new pdp11_insn_cost function.
12780 (MIN_BRANCH): Correct definition.
12782 (doloop_end): Use standard pattern name for looping pattern.
12783 (doloop_end_nocc): New.
12784 (movsf): Add another constraint alternative.
12785 (zero_extendqihi2): Add constraint alternatives for not in place
12787 (zero_extendhisi2): Remove.
12788 (shift patterns): Add CC handling variants.
12792 (define_peephole2): New peephole to recognize mov that sets CC for
12795 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12797 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
12798 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
12801 2018-07-09 Jakub Jelinek <jakub@redhat.com>
12804 * real.c (real_nextafter): Return true if result is denormal.
12806 2018-07-09 Martin Liska <mliska@suse.cz>
12808 * common.opt: Add back wrongly removed attribute.
12810 2018-07-09 Richard Biener <rguenther@suse.de>
12813 * dwarf2out.c (gen_block_die): For an early generated DIE
12814 always output high/low PC attributes.
12816 2018-07-09 Tom de Vries <tdevries@suse.de>
12818 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
12819 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
12820 onto VAR_DECL with abstract origin.
12822 2018-07-07 Jim Wilson <jimw@sifive.com>
12824 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
12826 2018-07-07 Tom de Vries <tdevries@suse.de>
12828 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
12830 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
12832 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
12834 (combine_bound): Use wide-int overflow calculation instead of
12836 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
12838 * fold-const.c (int_const_binop_2): Same.
12839 (extract_muldiv_1): Same.
12840 (fold_div_compare): Same.
12841 (fold_abs_const): Same.
12843 * poly-int.h (add): Same.
12847 * predict.c (predict_iv_comparison): Same.
12848 * profile-count.c (slow_safe_scale_64bit): Same.
12849 * simplify-rtx.c (simplify_const_binary_operation): Same.
12850 * tree-chrec.c (tree_fold_binomial): Same.
12851 * tree-data-ref.c (split_constant_offset_1): Same.
12852 * tree-if-conv.c (idx_within_array_bound): Same.
12853 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
12854 * tree-ssa-phiopt.c (minmax_replacement): Same.
12855 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
12856 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
12857 * vr-values.c (vr_values::adjust_range_with_scev): Same.
12858 * wide-int.cc (wi::add_large): Same.
12859 (wi::mul_internal): Same.
12860 (wi::sub_large): Same.
12861 (wi::divmod_internal): Same.
12862 * wide-int.h: Change overflow type to overflow_type for neg, add,
12863 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
12864 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
12865 mul_internal, divmod_internal.
12866 (overflow_type): New enum.
12867 (accumulate_overflow): New.
12869 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
12871 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
12872 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
12874 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
12876 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
12877 argument is checked for zero before entering loop, avoid checking again.
12879 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
12881 * gimplify.h (generic_expr_could_trap_p): Set as global function.
12882 * gimplify.h (generic_expr_could_trap_p): Likwise.
12883 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
12885 2018-07-06 Jakub Jelinek <jakub@redhat.com>
12887 PR tree-optimization/86401
12888 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
12889 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
12890 (fold_bit_and_mask): ... here. New helper function for match.pd.
12891 * fold-const.h (fold_bit_and_mask): Declare.
12892 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
12894 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
12897 * target.def (translate_mode_attribute): New hook.
12898 * targhooks.h (default_translate_mode_attribute): Declare.
12899 * targhooks.c (default_translate_mode_attribute): New function.
12900 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
12901 * doc/tm.texi: Regenerate.
12902 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
12903 (rs6000_translate_mode_attribute): New function.
12905 2018-07-06 Paul Koning <ni1d@arrl.net>
12907 * doc/md.texi (define_split): Document DONE and FAIL.
12908 (define_peephole2): Ditto.
12910 2018-07-05 Jeff Law <law@redhat.com>
12912 PR tree-optimization/86010
12913 * tree-ssa-dse.c (compute_trims): More aggressively trim at
12914 both the head and tail of mem* and str* calls.
12916 2018-07-05 Jim Wilson <jimw@sifive.com>
12918 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
12920 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
12922 * config/aarch64/aarch64-simd.md: correct flags text for
12923 MIN_EXPR replacement.
12925 2018-07-05 James Clarke <jrtc27@jrtc27.com>
12927 * configure: Regenerated.
12929 2018-07-05 Carl Love <cel@us.ibm.com>
12931 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
12932 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
12933 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
12934 VSX_BUILTIN_DOUBLEL_V4SF.
12936 2018-07-05 Martin Sebor <msebor@redhat.com>
12939 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
12940 than its domain to compute its the upper bound of a char array.
12942 2018-07-05 Nathan Sidwell <nathan@acm.org>
12944 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
12945 * doc/cpp.texi: Update comment.
12946 * doc/tm.texi: Rebuilt.
12947 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
12948 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
12949 * doc/extend.texi (Backwards Compatibility): Clarify it is system
12950 headers affected by extern "C".
12951 * system.h: Poison NO_IMPLICIT_EXTERN_C.
12952 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
12953 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
12954 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
12955 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
12956 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
12957 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
12958 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
12959 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
12960 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
12961 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
12962 config/sparc/sp64-elf.h, config/spu/spu.h,
12963 config/stormy16/stormy16.h, config/v850/v850.h,
12964 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
12965 define NO_IMPLICIT_EXTERN_C.
12966 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
12968 2018-07-05 Tamar Christina <tamar.christina@arm.com>
12971 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
12972 instead of GET_MODE_SIZE when comparing Units.
12974 2018-07-05 Tamar Christina <tamar.christina@arm.com>
12977 * rtlanal.c (set_noop_p): Constrain on mode change,
12978 include hard-reg-set.h
12980 2018-07-05 Tamar Christina <tamar.christina@arm.com>
12982 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
12984 2018-07-05 Jakub Jelinek <jakub@redhat.com>
12987 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
12990 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
12992 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
12994 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
12997 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
12999 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13001 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
13003 PR middle-end/86380
13004 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
13006 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
13008 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
13009 neg_*_op* variables.
13011 2018-07-04 Martin Liska <mliska@suse.cz>
13013 * tree-switch-conversion.c: Define
13014 max_ratio_for_speed and max_ratio_for_size constants.
13016 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13017 Martin Liska <mliska@suse.cz>
13019 PR middle-end/66240
13022 * common.opt: Rename align options with 'str_' prefix.
13023 * common/config/i386/i386-common.c (set_malign_value): New
13025 (ix86_handle_option): Use it to set -falign-* options/
13026 * config/aarch64/aarch64-protos.h (struct tune_params): Change
13027 type from int to string.
13028 * config/aarch64/aarch64.c: Update default values from int
13030 * config/alpha/alpha.c (alpha_override_options_after_change):
13032 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
13033 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13034 max skip conditionally.
13035 * config/i386/freebsd.h (SUBALIGN_LOG): New.
13036 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13037 max skip conditionally.
13038 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13039 max skip conditionally.
13040 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
13041 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13042 max skip conditionally.
13043 * config/i386/i386.c (struct ptt): Change type from int to
13045 (ix86_default_align): Set default values.
13046 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
13047 max skip conditionally.
13048 * config/i386/iamcu.h (SUBALIGN_LOG): New.
13049 (ASM_OUTPUT_MAX_SKIP_ALIGN):
13050 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
13051 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13052 max skip conditionally.
13053 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
13054 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
13055 * config/i386/x86-64.h (SUBALIGN_LOG): New.
13056 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13057 max skip conditionally.
13058 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13059 * config/ia64/ia64.c (ia64_option_override): Set default values
13060 for alignment options.
13061 * config/m68k/m68k.c: Handle new str_align_* options.
13062 * config/mips/mips.c (mips_set_compression_mode): Change
13064 (mips_option_override): Set default values for options.
13065 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
13067 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13069 * config/rx/rx.c (rx_option_override): Likewise.
13070 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
13071 (LABEL_ALIGN): Use align_labels_log.
13072 (LOOP_ALIGN): Use align_loops_align.
13073 * config/s390/s390.c (s390_asm_output_function_label): Use new
13075 * config/sh/sh.c (sh_override_options_after_change):
13076 Change type of constants.
13077 * config/spu/spu.c (spu_sched_init): Likewise.
13078 * config/sparc/sparc.c (sparc_option_override): Set default
13079 values for options.
13080 * config/visium/visium.c (visium_option_override): Likewise.
13081 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
13082 emit p2align format with last argument if it's not needed.
13083 * doc/invoke.texi: Document extended format of -falign-*.
13084 * final.c: Use align_labels alignment.
13085 * flags.h (struct target_flag_state): Change type to use
13087 (struct align_flags_tuple): New.
13088 (struct align_flags): Likewise.
13089 (align_loops_log): Redefine macro to use new types.
13090 (align_loops_max_skip): Redefine macro to use new types.
13091 (align_jumps_log): Redefine macro to use new types.
13092 (align_jumps_max_skip): Redefine macro to use new types.
13093 (align_labels_log): Redefine macro to use new types.
13094 (align_labels_max_skip): Redefine macro to use new types.
13095 (align_functions_log): Redefine macro to use new types.
13096 (align_loops): Redefine macro to use new types.
13097 (align_jumps): Redefine macro to use new types.
13098 (align_labels): Redefine macro to use new types.
13099 (align_functions): Redefine macro to use new types.
13100 (align_functions_max_skip): Redefine macro to use new types.
13101 (align_loops_value): New macro.
13102 (align_jumps_value): New macro.
13103 (align_labels_value): New macro.
13104 (align_functions_value): New macro.
13105 * function.c (invoke_set_current_function_hook): Propagate
13106 alignment values from flags to global variables default in
13108 * ipa-icf.c (sem_function::equals_wpa): Use
13109 cl_optimization_option_eq instead of memcmp.
13110 * lto-streamer.h (cl_optimization_stream_out): Support streaming
13112 (cl_optimization_stream_in): Likewise.
13113 * optc-save-gen.awk: Support strings in cl_optimization.
13114 * opth-gen.awk: Likewise.
13115 * opts.c (finish_options): Remove error checking of invalid
13117 (MAX_CODE_ALIGN): Remove.
13118 (MAX_CODE_ALIGN_VALUE): Likewise.
13119 (parse_and_check_align_values): New function.
13120 (check_alignment_argument): Likewise.
13121 (common_handle_option): Use check_alignment_argument.
13122 * opts.h (parse_and_check_align_values): Declare.
13123 * toplev.c (init_alignments): Remove.
13124 (read_log_maxskip): New.
13125 (parse_N_M): Likewise.
13126 (parse_alignment_opts): Likewise.
13127 (backend_init_target): Remove usage of init_alignments.
13128 * toplev.h (parse_alignment_opts): Declare.
13129 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
13131 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
13132 * tree.c (cl_option_hasher::equal): New.
13133 * varasm.c: Use new global macros.
13135 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13137 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
13138 Use a simpler align directive also if MAXSKIP = ALIGN-1.
13139 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13140 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13141 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13142 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13143 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
13144 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
13145 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
13146 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13147 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13148 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13149 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13151 2018-07-04 Martin Liska <mliska@suse.cz>
13152 Jonathan Wakely <jwakely@redhat.com>
13154 * coverage.c: Use correct type.
13155 * doc/invoke.texi: Language correction.
13157 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
13160 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
13161 ENDBRANCH for non-tail call which may return via indirect branch.
13162 * doc/extend.texi: Document indirect_return attribute.
13164 2018-07-03 Martin Sebor <msebor@redhat.com>
13166 PR tree-optimization/86274
13167 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
13169 (format_floating): Correct handling of infinities and NaNs.
13171 2018-07-03 Martin Sebor <msebor@redhat.com>
13173 * print-tree.c (print_real_cst): New function.
13174 (print_node_brief): Call it.
13175 (print_node): Ditto.
13177 2018-07-03 Jeff Law <law@redhat.com>
13179 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
13180 into a single pattern.
13182 * config/h8300/h8300.md (ors code_iterator): New.
13183 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
13184 a single pattern and single splitter.
13185 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
13186 (iorqi3_1, xorqi3_1): Likewise.
13187 (iorqi3, xorqi3 expanders): Similarly.
13189 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
13190 (movmd_internal) into a single pattern using the P mode iterator.
13191 (movmd splitters): Similarly.
13192 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
13193 (movsd splitters): Similarly.
13195 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
13196 ADDB, ADDW and ADDL into a single ADD attribute which selects the
13197 right table based on the size of the operand.
13198 * config/h8300/h8300.md (length_table): Corresponding changes. All
13199 references to "addb", "addw" and "addl" changed to "add".
13200 (btst patterns): Merge two variants into a single pattern.
13201 (tstqi, tsthi): Likewise.
13202 (addhi3_incdec, addsi3_incdec): Likewise.
13203 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
13204 (mulhi3, mulsi3): Likewise.
13205 (udivhi3, udivsi3): Likewise.
13206 (divhi3, divsi3): Likewise.
13207 (andorqi3, andorhi3, andorsi3): Likewise.
13209 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
13212 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
13213 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
13215 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13217 PR tree-optimization/85694
13218 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
13219 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
13221 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
13222 (<u>avg<mode>3_ceil): New patterns.
13224 2018-07-03 David Malcolm <dmalcolm@redhat.com>
13226 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
13227 scan-tree-dump directive.
13228 * gcc.dg/vect/slp-perm-2.c: Likewise.
13229 * gcc.dg/vect/slp-perm-3.c: Likewise.
13230 * gcc.dg/vect/slp-perm-5.c: Likewise.
13231 * gcc.dg/vect/slp-perm-6.c: Likewise.
13232 * gcc.dg/vect/slp-perm-7.c: Likewise.
13233 * gcc.dg/vect/slp-perm-8.c: Likewise.
13235 2018-07-03 Marek Polacek <polacek@redhat.com>
13237 PR middle-end/86202
13238 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
13240 2018-07-03 Richard Biener <rguenther@suse.de>
13243 * tree-ssa-structalias.c (find_func_clobbers): Properly
13244 handle indirect calls.
13246 2018-07-03 Jeff Law <law@redhat.com>
13248 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
13249 (shifts): New code iterator.
13250 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
13251 expander. Fix HImode handling on H8/SX.
13252 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
13253 (subqi3, subhi3, subsi3 expanders): Likewise.
13254 (andqi3, andhi3, andsi3 expanders): Likewise.
13255 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
13256 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
13257 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
13258 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
13259 (zero_extendqihi2, zero_extendqisi2): Likewise.
13260 (extendqihi2, extendqisi2): Likewise.
13261 (rotlqi3, rotlhi3, rotlsi3): Likewise.
13262 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
13263 (rotlqi3_1, rotlhi3_1): Likewise.
13264 (logicalhi3_sn, logicalsi3_sn): Likewise.
13265 (logicalhi3, logicalsi3): Likewise.
13267 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13269 * tree-vect-patterns.c (vect_recog_rotate_pattern)
13270 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
13271 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
13272 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
13273 type to append_pattern_def_seq instead of creating a stmt_vec_info
13275 (build_mask_conversion): Likewise. Remove vinfo argument.
13276 (vect_add_conversion_to_patterm): Likewise, renaming to...
13277 (vect_add_conversion_to_pattern): ...this.
13278 (vect_recog_mask_conversion_pattern): Update call to
13279 build_mask_conversion. Pass the vector type to
13280 append_pattern_def_seq here too.
13281 (vect_recog_gather_scatter_pattern): Update call to
13282 vect_add_conversion_to_pattern.
13284 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13286 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
13287 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13288 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
13289 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
13290 STMT_VINFO_PATTERN_DEF_SEQ to null here.
13291 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
13292 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
13293 append_pattern_def_seq instead of new_pattern_def_seq.
13294 (vect_recog_divmod_pattern): Do both of the above.
13295 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
13298 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13300 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
13301 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
13302 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
13303 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
13304 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
13305 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
13306 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
13307 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
13308 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
13309 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
13310 parameter with a single stmt_vec_info.
13311 (vect_recog_func_ptr): Likewise.
13312 (vect_recog_gather_scatter_pattern): Likewise, folding in...
13313 (vect_try_gather_scatter_pattern): ...this.
13314 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
13315 the stmt_vec_info of the statement to be matched. Don't clear
13316 STMT_VINFO_RELATED_STMT.
13317 (vect_pattern_recog): Update call accordingly.
13319 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13321 PR tree-optimization/85694
13322 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
13323 (uavgM3_ceil): Document new optabs.
13324 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
13325 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
13327 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
13328 (savg_ceil_optab): New optabs.
13329 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
13330 (vect_vect_recog_func_ptrs): Add it.
13331 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
13332 constant directly from the associated lhs.
13334 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13336 * tree-vect-patterns.c (vect_split_statement): New function.
13337 (vect_convert_input): Use it to try to split an existing cast.
13339 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13341 * poly-int.h (print_hex): New function.
13342 * dumpfile.h (dump_dec, dump_hex): Declare.
13343 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
13344 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
13345 min_input_precision, operation_precision and operation_sign.
13346 * tree-vect-patterns.c (vect_get_range_info): New function.
13347 (vect_same_loop_or_bb_p, vect_single_imm_use)
13348 (vect_operation_fits_smaller_type): Delete.
13349 (vect_look_through_possible_promotion): Add an optional
13350 single_use_p parameter.
13351 (vect_recog_over_widening_pattern): Rewrite to use new
13352 stmt_vec_info infomration. Handle one operation at a time.
13353 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
13354 (vect_truncatable_operation_p, vect_set_operation_type)
13355 (vect_set_min_input_precision): New functions.
13356 (vect_determine_min_output_precision_1): Likewise.
13357 (vect_determine_min_output_precision): Likewise.
13358 (vect_determine_precisions_from_range): Likewise.
13359 (vect_determine_precisions_from_users): Likewise.
13360 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
13361 (vect_vect_recog_func_ptrs): Put over_widening first.
13363 (vect_pattern_recog): Call vect_determine_precisions.
13365 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13367 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
13368 statements that have been replaced by further pattern statements.
13369 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
13371 2018-07-03 Richard Biener <rguenther@suse.de>
13373 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
13374 always set *dt. Dump vectype in vectype overload.
13375 * dumpfile.h (dump_gimple_expr): New function.
13376 (dump_gimple_expr_loc): Likewise.
13377 * dumpfile.c (dump_gimple_expr): New function.
13378 (dump_gimple_expr_loc): Likewise.
13380 2018-07-02 Jeff Law <law@redhat.com>
13382 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
13383 the H8/300, H8/300H and H8/S variants into a single pattern.
13384 (movhi_h8300, movqi_h8300hs): Similarly.
13385 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
13386 (QHI mode iterator): New.
13388 * config/h8300/h8300.md: Remove trailing whitespace.
13390 2018-07-02 Jim Wilson <jimw@sifive.com>
13392 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
13393 instead of emit_insn for interrupt returns.
13394 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
13395 (riscv_sret, riscv_uret): Likewise.
13397 2018-07-02 David Malcolm <dmalcolm@redhat.com>
13399 * pretty-print.c (selftest::test_pp_format): Move save and restore
13400 of quotes to class auto_fix_quotes, and add an instance.
13401 * selftest.c: Include "intl.h".
13402 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
13403 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
13404 * selftest.h (selftest::auto_fix_quotes): New class.
13406 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
13408 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
13409 (aarch64_sve_prepare_conditional_op): Remove.
13410 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
13411 Allow aarch64_simd_reg_or_zero as select operand; remove
13412 the aarch64_sve_prepare_conditional_op call.
13413 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
13414 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
13415 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
13416 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
13417 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
13418 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
13419 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
13420 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
13421 and a splitters to match all of the *_any patterns.
13422 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
13424 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
13425 (SVE_COND_FP_BINARY_REV): Remove.
13426 (sve_int_op_rev, sve_fp_op_rev): New.
13427 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
13428 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
13429 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
13430 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
13431 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
13432 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
13433 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
13434 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
13435 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
13437 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
13438 Remove match_dup 1 from the inner unspec.
13439 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
13441 * config/aarch64/aarch64.md (movprfx): New attr.
13442 (length): Default movprfx to 8.
13443 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
13444 (*madd<SVE_I>, *msub<SVE_I): Likewise.
13445 (*<su>mul<SVE_I>3_highpart): Likewise.
13446 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
13447 (*v<ASHIFT><SVE_I>3): Likewise.
13448 (*<su><MAXMIN><SVE_I>3): Likewise.
13449 (*<su><MAXMIN><SVE_F>3): Likewise.
13450 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
13451 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
13452 (*div<SVE_F>4): Likewise.
13454 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
13456 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
13459 2018-07-02 Richard Biener <rguenther@suse.de>
13461 PR tree-optimization/86363
13462 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
13463 memset argument refers to a non-variable address.
13465 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
13467 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
13468 {PLUS,MINUS}_EXPR code to...
13469 (adjust_symbolic_bound): ...here,
13470 (combine_bound): ...here,
13471 (set_value_range_with_overflow): ...and here.
13473 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
13475 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
13477 (extract_range_from_abs_expr): ...here.
13479 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
13481 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
13482 -fno-omit-frame-pointer when not optimizing.
13484 2018-07-02 Martin Liska <mliska@suse.cz>
13487 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
13488 (propagate_nothrow): Likewise.
13490 2018-07-02 Martin Liska <mliska@suse.cz>
13493 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
13495 2018-07-02 David Malcolm <dmalcolm@redhat.com>
13497 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
13498 function in r262149, changing "loc" param from source_location to
13499 const dump_location_t &.
13500 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
13501 declaration, as above.
13503 2018-07-01 Paul Koning <ni1d@arrl.net>
13505 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
13506 -munit-asm, -mgnu-asm, -mdec-asm.
13507 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
13508 (pdp11_output_labelref): New.
13509 (pdp11_output_def): New.
13510 (pdp11_output_addr_vec_elt): New.
13511 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
13512 %# and %@ format codes.
13513 (pdp11_option_override): New.
13514 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
13515 (pdp11_output_ident): New.
13516 (pdp11_asm_named_section): New.
13517 (pdp11_asm_init_sections): New.
13518 (pdp11_file_start): New.
13519 (pdp11_file_end): New.
13520 (output_ascii): Use .ascii/.asciz for -mdec-asm.
13521 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
13522 %o, like %c but octal.
13523 (pdp11_option_override): New.
13524 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
13526 (DATA_SECTION_ASM_OP): Ditto.
13527 (READONLY_DATA_SECTION_ASM_OP): New.
13528 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
13529 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
13530 (ASM_OUTPUT_LABELREF): Ditto.
13531 (ASM_OUTPUT_DEF): Ditto.
13532 (ASM_OUTPUT_EXTERNAL): New.
13533 (ASM_OUTPUT_SOURCE_FILENAME): New.
13534 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
13535 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
13536 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
13537 %# and %@ format codes.
13538 * config/pdp11/pdp11.opt (mgnu-asm): New.
13539 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
13540 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
13541 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
13543 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
13545 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
13546 dereferencing path[] beyond its length.
13548 (debug_all_paths): New.
13549 (rewire_first_differing_edge): New.
13550 (adjust_paths_after_duplication): New.
13551 (duplicate_thread_path): Call adjust_paths_after_duplication.
13553 (thread_through_all_blocks): Add new argument to
13554 duplicate_thread_path.
13556 2018-06-30 Jim Wilson <jimw@sifive.com>
13558 * config/riscv/predicates.md (p2m1_shift_operand): New.
13559 (high_mask_shift_operand): New.
13560 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
13561 pattern using p2m1_shift_operand.
13562 (lshsi3_zero_extend_3+2): New combiner pattern using
13563 high_mask_shift_operand.
13565 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13567 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
13569 (vect_recog_rotate_pattern): ...here.
13570 (vect_convert_input): Try to insert casts of invariants in the
13572 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
13573 preheader to be empty.
13575 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13577 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
13578 vector type. If given, install it in the new statement's
13579 STMT_VINFO_VECTYPE.
13580 (vect_element_precision): New function.
13581 (vect_unpromoted_value): New struct.
13582 (vect_unpromoted_value::vect_unpromoted_value): New function.
13583 (vect_unpromoted_value::set_op): Likewise.
13584 (vect_look_through_possible_promotion): Likewise.
13585 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
13586 (vect_widened_op_tree, vect_convert_input): Likewise.
13587 (vect_convert_inputs, vect_convert_output): Likewise.
13588 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
13589 to handle the optional cast of the multiplication result and
13590 vect_widened_op_tree to detect the widened multiplication itself.
13591 Do not require the input and output of promotion casts to have
13592 the same sign, but base the signedness of the operation on the
13593 input rather than the result. If the pattern includes two
13594 promotions, check that those promotions have the same sign.
13595 Do not restrict the MULT_EXPR handling to a double-width result;
13596 handle quadruple-width results and wider. Use vect_convert_inputs
13597 to convert the inputs to the common type.
13598 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
13599 to handle the optional cast of the ABS result. Also allow a sign
13600 change or a sign extension between the ABS and MINUS.
13601 Use vect_widened_op_tree to detect the widened subtraction and use
13602 vect_convert_inputs to convert the inputs to the common type.
13603 (vect_handle_widen_op_by_const): Delete.
13604 (vect_recog_widen_op_pattern): New function.
13605 (vect_recog_widen_mult_pattern): Use it.
13606 (vect_recog_widen_shift_pattern): Likewise.
13607 (vect_recog_widen_sum_pattern): Use
13608 vect_look_through_possible_promotion to handle the promoted
13611 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13613 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
13614 the containing gimple_seq *.
13615 * gimple-iterator.h (gsi_for_stmt): Declare it.
13616 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
13617 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
13618 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
13619 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
13620 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
13622 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
13624 (vect_mark_pattern_stmts): ...here. Handle cases in which the
13625 statement being replaced is part of an existing pattern
13626 definition sequence, inserting the new pattern statements before
13628 (vect_pattern_recog_1): Don't return a bool. If the statement
13629 is already part of a pattern, instead apply pattern matching
13630 to the pattern definition statements. Don't clear the
13631 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
13632 (vect_pattern_recog): Don't break after the first match;
13633 continue processing the pattern definition statements instead.
13634 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
13636 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13638 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
13639 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13640 (vect_recog_widen_sum_pattern): Use it.
13642 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13644 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13645 phi is not a pattern statement and has not been replaced by
13646 a pattern statement.
13647 * tree-vect-patterns.c (type_conversion_p): Don't check
13648 STMT_VINFO_IN_PATTERN_P.
13649 (vect_recog_vector_vector_shift_pattern): Likewise.
13650 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
13651 the pattern statement rather than the original statement; check
13652 directly for a WIDEN_MULT_EXPR here.
13653 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
13654 vect_is_simple_use to return the pattern statement rather
13655 than the original statement; use is_pattern_stmt_p to check
13656 for such a pattern statement.
13657 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
13658 to return the pattern statement rather than the original statement;
13659 don't do the same transformation here.
13660 (vect_is_simple_use): If the defining statement has been replaced
13661 by a pattern statement, return the pattern statement instead.
13662 Remove the corresponding (local) transformation from the vectype
13665 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13667 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
13668 end and default to null.
13669 * tree-vect-loop.c (vect_create_epilog_for_reduction)
13670 (vectorizable_reduction): Update calls accordingly, dropping the
13671 gimple ** argument if the passed-back statement isn't needed.
13672 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
13673 (vect_recog_rotate_pattern): Likewise.
13674 (vect_recog_mask_conversion_pattern): Likewise.
13675 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
13676 (vect_mask_constant_operand_p): Likewise.
13677 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
13678 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
13679 (get_group_load_store_type, get_load_store_type): Likewise.
13680 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
13681 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
13682 (vectorizable_conversion, vectorizable_assignment): Likewise.
13683 (vectorizable_shift, vectorizable_operation): Likewise.
13684 (vectorizable_store, vect_is_simple_cond): Likewise.
13685 (vectorizable_condition, vectorizable_comparison): Likewise.
13686 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
13687 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
13688 and move it to the end. Cope with null def_stmt_outs.
13690 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
13692 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
13694 2018-06-29 Jeff Law <law@redhat.com>
13696 * config/v850/v850.c (v850_legitimate_address_p): Handle large
13697 displacements for TARGET_V850E2V3 and newer.
13698 (TARGET_LRA_P): Remove. Defaults to LRA now.
13699 * config/v850/v850.md (sign23byte_load): Remove.
13700 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
13701 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
13703 2018-06-29 Martin Liska <mliska@suse.cz>
13706 * coverage.c (coverage_init): Mangle full path name.
13707 * doc/invoke.texi: Document the change.
13708 * gcov-io.c (mangle_path): New.
13709 * gcov-io.h (mangle_path): Likewise.
13710 * gcov.c (mangle_name): Use mangle_path for path mangling.
13712 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13714 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
13715 if starting source register is not even.
13717 2018-06-29 Martin Liska <mliska@suse.cz>
13719 PR tree-optimization/86263
13720 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
13721 Make edge redirection.
13723 2018-06-29 David Malcolm <dmalcolm@redhat.com>
13725 * dumpfile.c (dump_loc): Add indentation based on scope depth.
13726 (dump_scope_depth): New variable.
13727 (get_dump_scope_depth): New function.
13728 (dump_begin_scope): New function.
13729 (dump_end_scope): New function.
13730 * dumpfile.h (get_dump_scope_depth): New declaration.
13731 (dump_begin_scope): New declaration.
13732 (dump_end_scope): New declaration.
13733 (class auto_dump_scope): New class.
13734 (AUTO_DUMP_SCOPE): New macro.
13735 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
13738 2018-06-29 Richard Biener <rguenther@suse.de>
13740 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
13741 compute_all_dependences succeeds.
13742 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
13743 exceed --param loop-max-datarefs-for-datadeps.
13745 2018-06-29 Jakub Jelinek <jakub@redhat.com>
13747 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
13749 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
13752 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
13753 alternative 0 in preferred_for_speed attribute.
13755 2018-06-28 Paul Koning <ni1d@arrl.net>
13757 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
13758 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
13759 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
13760 * config/pdp11/pdp11.md: Correct "length" attribute calculation
13761 for shift insn patterns.
13763 2018-06-28 David Malcolm <dmalcolm@redhat.com>
13765 * cgraph.c (cgraph_node::get_body): Replace assignments to
13766 "dump_file" with calls to set_dump_file.
13767 * dumpfile.c (alt_dump_file): Make static, and group with...
13768 (alt_flags): ...this definition.
13769 (dumps_are_enabled): New variable.
13770 (refresh_dumps_are_enabled): New function.
13771 (set_dump_file): New function.
13772 (set_alt_dump_file): New function.
13773 (gcc::dump_manager::dump_start): Replace assignments to
13774 "dump_file" and "alt_dump_file" with calls to set_dump_file and
13776 (gcc::dump_manager::dump_finish): Likewise.
13777 * dumpfile.h (alt_dump_file): Delete decl.
13778 (dumps_are_enabled): New variable decl.
13779 (set_dump_file): New function decl.
13780 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
13782 * tree-nested.c (lower_nested_functions): Replace assignments to
13783 "dump_file" with calls to set_dump_file.
13785 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
13787 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
13788 goto_locus of each outgoing edge of each basic block.
13790 2018-06-28 Richard Biener <rguenther@suse.de>
13792 * dwarf2out.c (decl_scope_table): Remove.
13793 (push_decl_scope): Likewise.
13794 (pop_decl_scope): Likewise.
13795 (gen_type_die_for_member): Do not call push/pop_decl_scope.
13796 (gen_struct_or_union_type_die): Likewise.
13797 (gen_tagged_type_die): Likewise.
13798 (dwarf2out_init): Do not initialize decl_scope_table.
13799 (dwarf2out_c_finalize): Do not free it.
13801 2018-06-28 Richard Biener <rguenther@suse.de>
13803 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
13804 deciding whether to not re-use a DIE.
13806 2018-06-28 Richard Biener <rguenther@suse.de>
13808 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
13809 DW_AT_abstract_origin attribute.
13811 2018-06-28 Martin Liska <mliska@suse.cz>
13813 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
13814 Use newly introduced constants.
13815 * tree-switch-conversion.h (struct jump_table_cluster):
13816 Define max_ratio_for_size and max_ratio_for_speed.
13818 2018-06-28 Martin Liska <mliska@suse.cz>
13820 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
13821 Add new checking assert to catch invalid state.
13822 (jump_table_cluster::can_be_handled): Handle single case
13824 (jump_table_cluster::is_beneficial): Bail out for such case.
13825 (bit_test_cluster::find_bit_tests):
13826 Add new checking assert to catch invalid state.
13827 (bit_test_cluster::can_be_handled): Handle single case
13829 (bit_test_cluster::is_beneficial): Bail out for such case.
13830 (switch_decision_tree::analyze_switch_statement):
13833 2018-06-28 Martin Liska <mliska@suse.cz>
13835 * common.opt: Introduce -completion option.
13836 * gcc.c (driver_handle_option): Handle it.
13837 (driver::main): Print completions if completion
13839 * opt-suggestions.c (option_proposer::get_completions):
13841 (option_proposer::suggest_completion): Likewise.
13842 (option_proposer::find_param_completions): Likewise.
13843 (verify_autocompletions): Likewise.
13844 (test_completion_valid_options): Likewise.
13845 (test_completion_valid_params): Likewise.
13846 (in_completion_p): Likewise.
13847 (empty_completion_p): Likewise.
13848 (test_completion_partial_match): Likewise.
13849 (test_completion_garbage): Likewise.
13850 (opt_proposer_c_tests): Likewise.
13851 * opt-suggestions.h: Declare new functions.
13852 * opts.c (common_handle_option): Handle OPT__completion_.
13853 * selftest-run-tests.c (selftest::run_tests): Add
13854 opt_proposer_c_tests.
13855 * selftest.c (assert_str_startswith): New.
13856 * selftest.h (assert_str_startswith): Likewise.
13857 (opt_proposer_c_tests): New.
13858 (ASSERT_STR_STARTSWITH): Likewise.
13860 2018-06-28 Martin Liska <mliska@suse.cz>
13862 * Makefile.in: Add opt-suggestions.o.
13863 * gcc-main.c: Include opt-suggestions.h.
13864 * gcc.c (driver::driver): Likewise.
13865 (driver::~driver): Remove m_option_suggestions.
13866 (driver::build_option_suggestions): Moved to option_proposer.
13867 (driver::suggest_option): Likewise.
13868 (driver::handle_unrecognized_options): Use option_proposer.
13869 * gcc.h (class driver): Add new memver m_option_proposer.
13870 * opt-suggestions.c: New file.
13871 * opt-suggestions.h: New file.
13873 2018-06-28 Martin Liska <mliska@suse.cz>
13875 * vec.h (class auto_string_vec): New (moved from auto_argvec).
13876 (auto_string_vec::~auto_string_vec): Likewise.
13878 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
13880 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
13881 prevent_decl_creation_for_types fields up and add reset_location field.
13882 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
13883 statement if id->reset_location is true.
13884 (copy_edges_for_bb): Do not set goto_locus on the new edges if
13885 id->reset_location is true.
13886 (copy_phis_for_bb): Force input_location on the arguments if
13887 id->reset_location is true.
13888 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
13889 is set on the function to be inlined.
13891 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
13893 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
13895 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
13897 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
13898 registers for Pmode.
13899 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
13900 hard registers for the clobbered pseudo.
13902 2018-06-27 Paul Koning <ni1d@arrl.net>
13904 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
13905 mutually exclusive options.
13906 * config/pdp11/constraints.md (h): New constraint.
13907 (O): Update definition to match shift code generation.
13908 (D): New constraint.
13909 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
13911 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
13913 (output_jump): Change arguments.
13914 (pdp11_fixed_cc_regs): New function.
13915 (pdp11_cc_mode): Ditto.
13916 (pdp11_expand_shift): Ditto.
13917 (pdp11_assemble_shift): Ditto.
13918 (pdp11_small_shift): Ditto.
13919 (pdp11_branch_cost): Remove.
13920 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
13922 (pdp11_register_move_cost): Update for CC registers.
13923 (pdp11_rtx_costs): Add case for LSHIFTRT.
13924 (pdp11_output_jump): Add CCNZ mode conditional branches.
13925 (notice_update_cc_on_set): Remove.
13926 (pdp11_cc_mode): New function.
13927 (simple_memory_operand): Correct pre/post decrement case.
13928 (no_side_effect_operand): New function.
13929 (pdp11_regno_reg_class): Add CC_REGS class.
13930 (pdp11_fixed_cc_regs): New function.
13931 (pdp11_small_shift): New function.
13932 (pdp11_expand_shift): New function to expand shift insns.
13933 (pdp11_assemble_shift): New function to output shifts.
13934 (pdp11_branch_cost): Remove.
13935 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
13936 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
13937 (WCHAR_TYPE): Ditto.
13938 (PTRDIFF_TYPE): Ditto.
13939 (ADJUST_INSN_LENGTH): New macro.
13940 (FIXED_REGISTERS): Add CC registers.
13941 (CALL_USED_REGISTERS): Ditto.
13942 (reg_class): Ditto.
13943 (REG_CLASS_NAMES): Ditto.
13944 (REG_CLASS_CONTENTS): Ditto.
13945 (SELECT_CC_MODE): Use new function.
13946 (TARGET_FLAGS_REGNUM): New macro.
13947 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
13948 (cc0_reg_rtx): Remove.
13949 (CC_STATUS_MDEP): Remove.
13950 (CC_STATUS_MDEFP_INIT): Remove.
13951 (CC_IN_FPU): Remove.
13952 (NOTICE_UPDATE_CC): Remove.
13953 (REGISTER_NAMES): Add CC registers.
13954 (BRANCH_COST): Change to constant 1.
13955 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
13957 * config/pdp11/pdp11.opt (mbcopy): Remove.
13958 (mbcopy-builtin): Remove.
13959 (mbranch-cheap): Remove.
13960 (mbranch-expensive): Remove.
13961 * config/pdp11/predicates.md (expand_shift_operand): Update to
13962 match shift code generation.
13963 (ccnz_operator): New predicate.
13964 * doc/invoke.texi (PDP-11 Options): Remove deleted options
13965 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
13966 Remove non-existent option -mabshi, -mno-abshi. Document mutually
13968 * doc/md.texi (PDP-11): Document new D and h constraints. Update
13969 description of O constraint.
13971 2018-06-27 Jeff Law <law@redhat.com>
13972 Austin Law <austinklaw@gmail.com>
13974 * config/v850/v850.md (addsi3_set_flags): New pattern.
13975 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
13976 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
13977 (zero_extendhisi2_v850_set_flags): Likewise.
13978 (zero_extendqisi2_v850_set_flags): Likewise.
13979 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
13980 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
13981 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
13983 * config/v850/v850-protos.h (notice_update_cc): Remove.
13984 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
13985 (v850_print_operand): Handle 'D' and "d".
13986 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
13987 Add handling of arithmetic/logical operations compared against zero.
13988 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
13989 Do not look at v850_compare_op, instead get mode from last argument.
13990 (v850_gen_compare): Remove
13991 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
13992 after reload for prologue insns.
13993 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
13995 (construct_save_jarl): Likewise.
13996 (TARGET_FLAGS_REGNUM): Define.
13997 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
13998 (NOTICE_UPDATE_CC): Remove.
13999 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
14000 than cc0. Conditionalize on reload_completed.
14001 (cmpsi_insn, setfcc_insn): Likewise.
14002 (tst1 splitter): Turn into define_and_split which sets the flags
14004 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
14005 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
14006 (cstoresf4, cstoredf4): Clobber the flags.
14007 (cmpsi, cmpsf, cmpdf): Remove expanders.
14008 (setf_insn): Remove pattern.
14009 (addsi3): Turn into define_and_split which clobbers the flags after
14010 reload and a suitable pattern (addsi3_clobber_flags) for use after
14012 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
14013 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
14014 (ashrsi3, ashrsi3_v850e2): Likewise.
14015 (bins): Clobber the flags.
14016 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
14017 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
14018 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
14019 (call_value_internal_short, call_value_internal_long): Likewise.
14020 (callt_save_interrupt, callt_return_interrupt): Likewise.
14021 (save_interrupt, return_interrupt): Likewise.
14022 (callt_save_all_interrupt, save_all_interrupt): Likewise.
14023 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
14024 (restore_all_interrupt, _restore_all_interrupt): Likewise.
14025 (All FP comparisons): Only allow after reload has completed.
14027 (divh, divhu): Tweak output template.
14028 (branch_z_normal, branch_z_invert): Remove
14029 (branch_nz_normal, branch_nz_invert): Likewise.
14030 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
14032 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
14033 * config/v850/v850.c (notice_update_cc): Remove.
14034 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
14035 (CC_NO_CARRY): Likewise.
14036 (NOTICE_UPDATE_CC): Define to nothing.
14037 * config/v850/v850.md: Remove block comment on cc0 handling
14038 Remove "cc" attribute from all patterns. Remove cc_status handling
14039 from all patterns. Minor formatting fixes.
14041 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14043 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
14044 (cortex-a76.cortex-a55): Likewise.
14045 * config/aarch64/aarch64-tune.md: Regenerate.
14046 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
14047 cortex-a76.cortex-a55.
14049 2018-06-27 Jeff Law <law@redhat.com>
14051 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
14052 (MULTILIB_DIRNAMES): Similarly.
14054 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
14056 * gimple.h (gimple_return_retbnd): Delete.
14057 (gimple_return_set_retbnd): Likewise.
14058 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
14059 gimple_return_set_retbnd.
14060 * gimple-pretty-print.c (dump_gimple_return): Remove call to
14061 gimple_return_retbnd and adjust.
14062 * tree-inline.h (struct copy_body_data): Remove retbnd field.
14063 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
14064 Explicitly return NULL in a couple more cases. Move assertion
14065 on debug statements and remove unreachable code.
14066 (reset_debug_binding): Do not test id->retbnd.
14067 (expand_call_inline): Do not set it.
14069 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
14071 * configure.ac: Add --disable-gcov option.
14072 * configure: Regenerate.
14073 * Makefile.in: Honour @enable_gcov@.
14074 * doc/install.texi: Document --disable-gcov.
14076 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14078 * config/arm/arm-cpus.in (cortex-a76): New entry.
14079 (cortex-a76.cortex-a55): Likewise.
14080 * config/arm/arm-tables.opt: Regenerate.
14081 * config/arm/arm-tune.md: Likewise.
14082 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
14083 * doc/invoke.texi (ARM Options): Document cortex-a76 and
14084 cortex-a76.cortex-a55.
14086 2018-06-27 Tamar Christina <tamar.christina@arm.com>
14089 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
14091 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
14093 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
14095 (EPILOGUE_USES): Likewise.
14097 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
14099 * tree-inline.c (remap_location): New function extracted from...
14100 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
14101 (copy_phis_for_bb): ...here. Call remap_location.
14102 (copy_cfg_body): Adjust call to copy_edges_for_bb.
14104 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
14106 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
14107 unaligned vsx for 16B memset.
14109 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
14112 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
14113 ieee128_float_type_node to long_double_type_node unless
14114 TARGET_LONG_DOUBLE_128 is set.
14116 2018-06-26 David Malcolm <dmalcolm@redhat.com>
14118 * cfgloop.c (get_loop_location): Convert return type from
14119 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
14120 by implicit construction from rtx_insn *, and using
14121 dump_user_location_t::from_function_decl for the fallback case.
14122 * cfgloop.h (get_loop_location): Convert return type from
14123 location_t to dump_user_location_t.
14124 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
14125 dump_printf_loc to pass in a dump_location_t rather than a
14126 location_t, via the gimple stmt.
14127 * coverage.c (get_coverage_counts): Update calls to
14128 dump_printf_loc to pass in dump_location_t rather than a
14130 * doc/optinfo.texi (Dump types): Convert example of
14131 dump_printf_loc from taking "locus" to taking "insn". Update
14132 description of the "_loc" calls to cover dump_location_t.
14133 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
14135 (dump_user_location_t::dump_user_location_t): New constructors,
14136 from gimple *stmt and rtx_insn *.
14137 (dump_user_location_t::from_function_decl): New function.
14138 (dump_loc): Make static.
14139 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
14140 const dump_location_t &.
14141 (dump_generic_expr_loc): Delete.
14142 (dump_printf_loc): Convert param "loc" from location_t to
14143 const dump_location_t &.
14144 (selftest::test_impl_location): New function.
14145 (selftest::dumpfile_c_tests): New function.
14146 * dumpfile.h: Include "profile-count.h".
14147 (class dump_user_location_t): New class.
14148 (struct dump_impl_location_t): New struct.
14149 (class dump_location_t): New class.
14150 (dump_printf_loc): Convert 2nd param from source_location to
14151 const dump_location_t &.
14152 (dump_generic_expr_loc): Delete.
14153 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
14154 const dump_location_t &.
14155 * gimple-fold.c (fold_gimple_assign): Update call to
14156 dump_printf_loc to pass in a dump_location_t rather than a
14157 location_t, via the gimple stmt.
14158 (gimple_fold_call): Likewise.
14159 * gimple-loop-interchange.cc
14160 (loop_cand::analyze_iloop_reduction_var): Update for change to
14161 check_reduction_path.
14162 (tree_loop_interchange::interchange): Update for change to
14163 find_loop_location.
14164 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
14165 change in return-type of find_loop_location.
14166 (graphite_regenerate_ast_isl): Likewise.
14167 * graphite-optimize-isl.c (optimize_isl): Likewise.
14168 * graphite.c (graphite_transform_loops): Likewise.
14169 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
14170 pass in a dump_location_t rather than a location_t, via the
14172 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14173 * ipa.c (walk_polymorphic_call_targets): Likewise.
14174 * loop-unroll.c (report_unroll): Convert "locus" param from
14175 location_t to dump_location_t.
14176 (decide_unrolling): Update for change to get_loop_location's
14178 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
14179 location_t to dump_user_location_t.
14180 (grid_find_single_omp_among_assignments_1): Updates calls to
14181 dump_printf_loc to pass in a dump_location_t rather than a
14182 location_t, via the gimple stmt.
14183 (grid_parallel_clauses_gridifiable): Convert "tloc" from
14184 location_t to dump_location_t. Updates calls to dump_printf_loc
14185 to pass in a dump_location_t rather than a location_t, via the
14187 (grid_inner_loop_gridifiable_p): Likewise.
14188 (grid_dist_follows_simple_pattern): Likewise.
14189 (grid_gfor_follows_tiling_pattern): Likewise.
14190 (grid_target_follows_gridifiable_pattern): Likewise.
14191 (grid_attempt_target_gridification): Convert initialization
14192 of local "grid" from memset to zero-initialization; FIXME: does
14193 this require C++11? Update call to dump_printf_loc to pass in a
14194 optinfo_location rather than a location_t, via the gimple stmt.
14195 * profile.c (read_profile_edge_counts): Updates call to
14196 dump_printf_loc to pass in a dump_location_t rather than a
14198 (compute_branch_probabilities): Likewise.
14199 * selftest-run-tests.c (selftest::run_tests): Call
14201 * selftest.h (dumpfile_c_tests): New decl.
14202 * tree-loop-distribution.c (pass_loop_distribution::execute):
14203 Update for change in return type of find_loop_location.
14204 * tree-parloops.c (parallelize_loops): Likewise.
14205 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
14206 "locus" from location_t to dump_user_location_t.
14207 (canonicalize_loop_induction_variables): Likewise.
14208 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
14209 for change in return type of find_loop_location.
14210 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
14211 to dump_printf_loc to pass in a dump_location_t rather than a
14212 location_t, via the stmt.
14213 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
14215 * tree-vect-loop-manip.c (find_loop_location): Convert return
14216 type from source_location to dump_user_location_t.
14217 (vect_do_peeling): Update for above change.
14218 (vect_loop_versioning): Update for change in type of
14220 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
14221 from location_t to dump_user_location_t.
14222 (vect_estimate_min_profitable_iters): Update for change in type
14224 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
14225 location_t to dump_location_t.
14226 (vect_slp_bb): Update for change in type of vect_location.
14227 * tree-vectorizer.c (vect_location): Convert from source_location
14228 to dump_user_location_t.
14229 (try_vectorize_loop_1): Update for change in vect_location's type.
14230 (vectorize_loops): Likewise.
14231 (increase_alignment): Likewise.
14232 * tree-vectorizer.h (vect_location): Convert from source_location
14233 to dump_user_location_t.
14234 (find_loop_location): Convert return type from source_location to
14235 dump_user_location_t.
14236 (check_reduction_path): Convert 1st param from location_t to
14237 dump_user_location_t.
14238 * value-prof.c (check_counter): Update call to dump_printf_loc to
14239 pass in a dump_user_location_t rather than a location_t; update
14240 call to error_at for change in type of "locus".
14241 (check_ic_target): Update call to dump_printf_loc to
14242 pass in a dump_user_location_t rather than a location_t, via the
14245 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
14247 * config/s390/s390.h (enum processor_flags): Do not use
14248 default tune parameter when -march was specified.
14250 2018-06-26 Jakub Jelinek <jakub@redhat.com>
14253 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
14254 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
14257 2018-06-26 Richard Biener <rguenther@suse.de>
14259 PR tree-optimization/86287
14261 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
14262 (vect_analyze_loop): Initialize n_stmts.
14264 2018-06-26 Richard Biener <rguenther@suse.de>
14266 PR middle-end/86271
14267 * fold-const.c (fold_convertible_p): Pointer extension
14270 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
14273 * dwarf2out.c (loc_list_has_views): Adjust comments.
14274 (dw_loc_list): Split single cross-partition range with
14277 2018-06-25 Jeff Law <law@redhat.com>
14279 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
14280 on -mbig-switch by default.
14282 * config/v850/predicates.md (const_float_1_operand): Fix match_code
14284 (const_float_0_operand): Remove unused predicate.
14285 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
14286 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
14287 (recipsf2): New expander. Original pattern now called
14289 (recipdf2, recipdf2_insn): Similarly.
14290 (rsqrtsf2, rsqrtsf2_insn): Similarly
14291 (rsqrtdf2, rsqrtdf2_insn): Similarly
14293 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
14295 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
14296 Simplify logic for FreeBSD (twice).
14298 2018-06-25 Martin Sebor <msebor@redhat.com>
14300 PR tree-optimization/86204
14301 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
14302 a strnlen result if it's less than the length of the string.
14304 2018-06-25 Martin Sebor <msebor@redhat.com>
14306 PR tree-optimization/85700
14307 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
14308 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
14309 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
14311 2018-06-25 Martin Sebor <msebor@redhat.com>
14313 * doc/extend.texi (Zero-length arrays): Update and clarify.
14315 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
14317 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
14318 added IEEE/IBM long double multilib support on PowerPC little
14319 endian Linux systems.
14320 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
14321 (MULTILIB_DEFAULTS): Likewise.
14322 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14324 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
14325 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
14326 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
14328 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
14330 PR middle-end/86311
14331 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
14332 (REORDER_45): Likewise.
14334 2018-06-25 Jeff Law <law@redhat.com>
14336 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
14337 dividend to 32 bits. Adjust length.
14338 (udivmodhi4): Cleanup output template. Fix length.
14340 2018-06-25 Carl Love <cel@us.ibm.com>
14342 * config/rs6000/vsx.md: Change word selector to prefered location.
14344 2018-06-25 Richard Biener <rguenther@suse.de>
14346 PR tree-optimization/86304
14347 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
14348 epilogue-if-converted loops as well.
14350 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
14352 * lto-section-out.c (lto_begin_section): Do not print section
14353 name for noaddr and unnumbered dumps.
14355 2018-06-25 Richard Biener <rguenther@suse.de>
14357 * tree-vectorizer.h (struct vec_info_shared): New structure
14358 with parts split out from struct vec_info and loop_nest from
14359 struct _loop_vec_info.
14360 (struct vec_info): Adjust accordingly.
14361 (struct _loop_vec_info): Likewise.
14362 (LOOP_VINFO_LOOP_NEST): Adjust.
14363 (LOOP_VINFO_DATAREFS): Likewise.
14364 (LOOP_VINFO_DDRS): Likewise.
14365 (struct _bb_vec_info): Likewise.
14366 (BB_VINFO_DATAREFS): Likewise.
14367 (BB_VINFO_DDRS): Likewise.
14368 (struct _stmt_vec_info): Add dr_aux member.
14369 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
14370 (DR_MISALIGNMENT_UNINITIALIZED): New.
14371 (set_dr_misalignment): Adjust.
14372 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
14373 (vect_analyze_loop): Adjust prototype.
14374 (vect_analyze_loop_form): Likewise.
14375 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14376 Compute dependences lazily.
14377 (vect_record_base_alignments): Use shared datarefs/ddrs.
14378 (vect_verify_datarefs_alignment): Likewise.
14379 (vect_analyze_data_refs_alignment): Likewise.
14380 (vect_analyze_data_ref_accesses): Likewise.
14381 (vect_analyze_data_refs): Likewise.
14382 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
14383 constructor parameter for shared part.
14384 (vect_analyze_loop_form): Pass in shared part and adjust.
14385 (vect_analyze_loop_2): Pass in storage for the number of
14386 stmts. Move loop nest finding to the caller. Compute
14388 (vect_analyze_loop): Pass in shared part.
14389 (vect_transform_loop): Verify shared datarefs are unchanged.
14390 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
14391 constructor parameter for shared part.
14392 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
14393 (vect_slp_bb): Verify shared datarefs are unchanged before
14395 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
14397 (new_stmt_vec_info): Initialize DR_AUX misalignment to
14398 DR_MISALIGNMENT_UNINITIALIZED.
14399 * tree-vectorizer.c (vec_info::vec_info): Add constructor
14400 parameter for shared part.
14401 (vec_info::~vec_info): Adjust.
14402 (vec_info_shared::vec_info_shared): New.
14403 (vec_info_shared::~vec_info_shared): Likewise.
14404 (vec_info_shared::save_datarefs): Likewise.
14405 (vec_info_shared::check_datarefs): Likewise.
14406 (try_vectorize_loop_1): Construct shared part live for analyses
14407 of a single loop for multiple vector sizes.
14408 * tree-parloops.c (gather_scalar_reductions): Adjust.
14410 2018-06-25 Richard Biener <rguenther@suse.de>
14412 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
14413 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
14414 (vect_analyze_data_refs): Remove similar code from here and
14415 simplify accordingly.
14417 2018-06-25 Richard Biener <rguenther@suse.de>
14419 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
14420 for reverse storage order accesses rather than asserting
14421 they cannot happen here.
14423 2018-06-25 Tom de Vries <tdevries@suse.de>
14426 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
14427 Use data16 instead of .byte for insn prefix.
14429 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
14432 * parser.c (make_char_string_pack): Pass this literal chars
14433 through cpp_interpret_string.
14434 (cp_parser_userdef_numeric_literal): Check the result of
14435 make_char_string_pack.
14437 2018-06-24 Maya Rashish <coypu@sdf.org>
14439 * ginclude/stddef.h: Simplify conditions around avoiding
14440 re-definition of __size_t.
14442 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14444 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
14445 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
14447 2018-06-22 Maya Rashish <coypu@sdf.org>
14449 * doc/invoke.texi (mno-fancy-math-387): Update for changes
14450 made to OpenBSD and NetBSD through the years.
14452 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
14454 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
14455 behavior of vec_pack (vector double, vector double) to match
14456 behavior of vec_float2 (vector double, vector double).
14458 2018-06-22 Olivier Hainque <hainque@adacore.com>
14460 * gimplify.c (gimplify_function_tree): Prevent creation
14461 of a trampoline for the address of the current function
14462 passed to entry/exit instrumentation hooks.
14464 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
14467 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
14470 2018-06-22 Martin Liska <mliska@suse.cz>
14472 PR tree-optimization/86263
14473 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14474 Bail out if is_enabled is false.
14475 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
14477 (jump_table_cluster::is_enabled): New function.
14479 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14481 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
14482 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
14483 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
14484 (lto_input_ts_binfo_tree_pointers): Likewise.
14485 * tree-streamer-out.c (streamer_write_tree_bitfields,
14486 write_ts_binfo_tree_pointers): Likewise.
14487 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
14489 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14491 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
14493 2018-06-22 Martin Liska <mliska@suse.cz>
14495 * symbol-summary.h (get): Make it pure and inline move
14496 functionality from ::get function.
14497 (get): Remove and inline into ::get and ::get_create.
14498 (get_create): Move code from ::get function.
14500 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14503 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
14504 -x assembler-with-cpp.
14506 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14508 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
14509 _FILE_OFFSET_BITS=64 for C++.
14511 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
14513 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
14514 conversion insn that shows up when pr85657-3.c is compiled using
14515 IEEE 128-bit long double.
14516 (neg<mode>2_internal): Use the correct mode to check whether the
14517 mode is IBM extended.
14518 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
14519 multiply and divide external functions from being created more
14522 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14524 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
14526 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
14527 the edge can be forwarded.
14528 (cfg_layout_merge_blocks): Likewise.
14530 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14532 * except.c (finish_eh_generation): Commit edge insertions only after
14533 the EH edges have been redirected from post-landing to landing pads.
14535 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14537 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
14538 create_tmp_var_for to create the FRAME decl.
14539 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
14541 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14543 * tree-inline.c (copy_edges_for_bb): Minor tweak.
14544 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
14545 debug statement when resetting its value.
14546 (expand_call_inline): Copy the locus of the call onto the assignment
14547 of the return value, if any. Use local variable in more cases.
14549 2018-06-21 Martin Liska <mliska@suse.cz>
14551 * ipa-pure-const.c (propagate_nothrow): Use
14552 funct_state_summaries->get.
14553 (dump_malloc_lattice): Likewise.
14554 (propagate_malloc): Likewise.
14556 2018-06-21 Richard Biener <rguenther@suse.de>
14558 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
14559 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
14560 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
14562 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
14563 BLOCK_ABSTRACT_ORIGIN unconditionally.
14565 2018-06-21 David Malcolm <dmalcolm@redhat.com>
14567 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
14569 * ipa-reference.c (ipa_reference_c_finalize): Delete
14570 ipa_ref_opt_sum_summaries and set it to NULL.
14572 2018-06-21 Tom de Vries <tdevries@suse.de>
14574 PR tree-optimization/85859
14575 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
14576 test with comment from bb_no_side_effects_p.
14578 2018-06-21 Richard Biener <rguenther@suse.de>
14580 PR tree-optimization/86232
14581 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
14582 max for constant niter.
14584 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
14586 * config/aarch64/aarch64-simd.md
14587 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
14589 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
14591 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
14592 Make opernads of the unspec commutative.
14594 2018-06-21 Richard Biener <rguenther@suse.de>
14596 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
14597 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14598 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
14599 (vect_analyze_data_ref_dependence): Re-order checks to deal with
14601 (vect_record_base_alignments): Do not record base alignment
14602 for gathers or scatters.
14603 (vect_compute_data_ref_alignment): Drop return value that is always
14604 true. Bail out early for gathers or scatters.
14605 (vect_enhance_data_refs_alignment): Bail out early for gathers
14607 (vect_find_same_alignment_drs): Likewise.
14608 (vect_analyze_data_refs_alignment): Remove dead code.
14609 (vect_slp_analyze_and_verify_node_alignment): Likewise.
14610 (vect_analyze_data_refs): For possible gathers or scatters do
14611 not create an alternate DR, just check their possible validity
14612 and mark them. Adjust DECL_NONALIASED handling to not rely
14613 on DR_BASE_ADDRESS.
14614 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
14615 update inits of gathers or scatters.
14616 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
14617 Also copy gather/scatter flag to pattern vinfo.
14619 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
14621 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
14622 behavior of vec_packsu (vector unsigned long long, vector unsigned
14623 long long) to match behavior of vec_packs with same signature.
14625 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
14626 Thomas Schwinge <thomas@codesourcery.com>
14627 Cesar Philippidis <cesar@codesourcery.com>
14629 * gimplify.c (gimplify_scan_omp_clauses): Add support for
14630 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
14631 (gimplify_adjust_omp_clauses): Likewise.
14632 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
14633 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
14634 (gimplify_omp_target_update): Update handling of acc update and
14636 * omp-low.c (install_var_field): Remove unused parameter
14637 base_pointers_restrict.
14638 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
14639 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
14641 (omp_target_base_pointers_restrict_p): Delete.
14642 (scan_omp_target): Update call to scan_sharing_clauses.
14643 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
14645 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
14646 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
14647 (convert_local_omp_clauses): Likewise.
14648 * tree-pretty-print.c (dump_omp_clause): Likewise.
14649 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
14651 (omp_clause_code_name): Likewise.
14653 2018-06-20 Jakub Jelinek <jakub@redhat.com>
14656 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
14659 PR tree-optimization/86231
14660 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
14661 anti-range don't overwrite *vr0min before using it to compute *vr0max.
14663 2018-06-20 Tom de Vries <tdevries@suse.de>
14665 PR tree-optimization/86097
14666 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
14667 iv type if signedness of iv type is not the same as that of *nit.
14669 2018-06-20 Jakub Jelinek <jakub@redhat.com>
14671 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
14672 EDGE_EH edges, verify they are all EDGE_EH.
14674 2018-06-20 Maya Rashish <coypu@sdf.org>
14676 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
14678 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14680 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
14681 * config/aarch64/aarch64.c (xgene1_tunings): Add
14682 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
14683 (aarch64_mode_valid_for_sched_fusion_p):
14684 Allow 16-byte modes.
14685 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
14686 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
14688 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
14690 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
14691 * config/aarch64/iterators.md (VQ2): New mode iterator.
14693 2018-06-20 Martin Liska <mliska@suse.cz>
14695 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
14696 Change default ratio from 10 to 8.
14698 2018-06-20 Martin Liska <mliska@suse.cz>
14700 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14702 (bit_test_cluster::find_bit_tests): Likewise.
14703 (switch_decision_tree::analyze_switch_statement): Find clusters.
14704 * tree-switch-conversion.h (struct jump_table_cluster): Document
14707 2018-06-20 Martin Liska <mliska@suse.cz>
14709 * tree-switch-conversion.c (switch_conversion::collect):
14710 Record m_uniq property.
14711 (switch_conversion::expand): Bail out for special conditions.
14712 (group_cluster::~group_cluster): New.
14713 (group_cluster::group_cluster): Likewise.
14714 (group_cluster::dump): Likewise.
14715 (jump_table_cluster::emit): New.
14716 (switch_decision_tree::fix_phi_operands_for_edges): New.
14717 (struct case_node): Remove struct.
14718 (jump_table_cluster::can_be_handled): New.
14719 (case_values_threshold): Moved to header.
14720 (reset_out_edges_aux): Likewise.
14721 (jump_table_cluster::is_beneficial): New.
14722 (bit_test_cluster::can_be_handled): Likewise.
14723 (add_case_node): Remove.
14724 (bit_test_cluster::is_beneficial): New.
14725 (case_bit_test::cmp): New.
14726 (bit_test_cluster::emit): New.
14727 (expand_switch_as_decision_tree_p): Remove.
14728 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
14729 (fix_phi_operands_for_edge): Likewise.
14730 (switch_decision_tree::analyze_switch_statement): New.
14731 (compute_cases_per_edge): Move ...
14732 (switch_decision_tree::compute_cases_per_edge): ... here.
14733 (try_switch_expansion): Likewise.
14734 (switch_decision_tree::try_switch_expansion): Likewise.
14735 (record_phi_operand_mapping): Likewise.
14736 (switch_decision_tree::record_phi_operand_mapping): Likewise.
14737 (emit_case_decision_tree): Likewise.
14738 (switch_decision_tree::emit): Likewise.
14739 (balance_case_nodes): Likewise.
14740 (switch_decision_tree::balance_case_nodes): Likewise.
14741 (dump_case_nodes): Likewise.
14742 (switch_decision_tree::dump_case_nodes): Likewise.
14743 (emit_jump): Likewise.
14744 (switch_decision_tree::emit_jump): Likewise.
14745 (emit_cmp_and_jump_insns): Likewise.
14746 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
14747 (emit_case_nodes): Likewise.
14748 (switch_decision_tree::emit_case_nodes): Likewise.
14749 (conditional_probability): Remove.
14750 * tree-switch-conversion.h (enum cluster_type): New.
14752 (struct cluster): Likewise.
14753 (cluster::cluster): Likewise.
14754 (struct simple_cluster): Likewise.
14755 (simple_cluster::simple_cluster): Likewise.
14756 (struct group_cluster): Likewise.
14757 (struct jump_table_cluster): Likewise.
14758 (struct bit_test_cluster): Likewise.
14759 (struct min_cluster_item): Likewise.
14760 (struct case_tree_node): Likewise.
14761 (case_tree_node::case_tree_node): Likewise.
14762 (jump_table_cluster::case_values_threshold): Likewise.
14763 (struct case_bit_test): Likewise.
14764 (struct switch_decision_tree): Likewise.
14765 (struct switch_conversion): Likewise.
14766 (switch_decision_tree::reset_out_edges_aux): Likewise.
14768 2018-06-20 Martin Liska <mliska@suse.cz>
14770 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
14771 (hoist_edge_and_branch_if_true): Likewise.
14772 (expand_switch_using_bit_tests_p): Likewise.
14773 (struct case_bit_test): Likewise.
14774 (case_bit_test_cmp): Likewise.
14775 (emit_case_bit_tests): Likewise.
14776 (switch_conversion::switch_conversion): New class.
14777 (struct switch_conv_info): Remove old struct.
14778 (collect_switch_conv_info): More to ...
14779 (switch_conversion::collect): ... this.
14780 (check_range): Likewise.
14781 (switch_conversion::check_range): Likewise.
14782 (check_all_empty_except_final): Likewise.
14783 (switch_conversion::check_all_empty_except_final): Likewise.
14784 (check_final_bb): Likewise.
14785 (switch_conversion::check_final_bb): Likewise.
14786 (create_temp_arrays): Likewise.
14787 (switch_conversion::create_temp_arrays): Likewise.
14788 (free_temp_arrays): Likewise.
14789 (gather_default_values): Likewise.
14790 (switch_conversion::gather_default_values): Likewise.
14791 (build_constructors): Likewise.
14792 (switch_conversion::build_constructors): Likewise.
14793 (constructor_contains_same_values_p): Likewise.
14794 (switch_conversion::contains_same_values_p): Likewise.
14795 (array_value_type): Likewise.
14796 (switch_conversion::array_value_type): Likewise.
14797 (build_one_array): Likewise.
14798 (switch_conversion::build_one_array): Likewise.
14799 (build_arrays): Likewise.
14800 (switch_conversion::build_arrays): Likewise.
14801 (gen_def_assigns): Likewise.
14802 (switch_conversion::gen_def_assigns): Likewise.
14803 (prune_bbs): Likewise.
14804 (switch_conversion::prune_bbs): Likewise.
14805 (fix_phi_nodes): Likewise.
14806 (switch_conversion::fix_phi_nodes): Likewise.
14807 (gen_inbound_check): Likewise.
14808 (switch_conversion::gen_inbound_check): Likewise.
14809 (process_switch): Use the newly created class.
14810 (switch_conversion::expand): New.
14811 (switch_conversion::~switch_conversion): New.
14812 * tree-switch-conversion.h: New file.
14814 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14816 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
14817 tree-vect-patterns.c.
14818 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
14819 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
14820 (vect_recog_sad_pattern): Likewise.
14821 (vect_recog_widen_sum_pattern): Likewise.
14822 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
14823 (vect_recog_widen_shift_pattern): Remove the type_in argument.
14824 (vect_recog_rotate_pattern): Likewise.
14825 (vect_recog_mult_pattern): Likewise.
14826 (vect_recog_vector_vector_shift_pattern): Likewise.
14827 (vect_recog_divmod_pattern): Likewise.
14828 (vect_recog_mixed_size_cond_pattern): Likewise.
14829 (vect_recog_bool_pattern): Likewise.
14830 (vect_recog_mask_conversion_pattern): Likewise.
14831 (vect_try_gather_scatter_pattern): Likewise.
14832 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
14833 (vect_recog_over_widening_pattern): Likewise.
14834 (vect_recog_gather_scatter_pattern): Likewise.
14835 (vect_recog_func_ptr): Move from tree-vectorizer.h
14836 (vect_vect_recog_func_ptrs): Move further down the file.
14837 (vect_recog_func): Likewise. Remove the third argument.
14838 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
14839 (vect_pattern_recog_1): Expect the pattern function to do any
14840 necessary target tests. Also expect it to provide a vector type.
14841 Remove the type_in handling.
14843 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14845 * tree-vect-patterns.c (vect_pattern_detected): New function.
14846 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
14847 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
14848 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
14849 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
14850 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
14851 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
14852 (vect_recog_mask_conversion_pattern)
14853 (vect_try_gather_scatter_pattern): Likewise.
14855 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14857 * tree-vect-patterns.c (vect_get_internal_def): New function.
14858 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
14859 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
14860 (search_type_for_mask_1): Use it.
14862 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14864 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
14865 redundant WIDEN_SUM_EXPR handling.
14866 (vect_recog_sad_pattern): Likewise.
14868 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14870 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
14871 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
14872 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
14873 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
14874 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
14876 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14878 * tree-vect-stmts.c (vectorizable_call): Make sure that we
14879 use the stmt_vec_info of the original bb statement for the
14880 new zero assignment, even if the call is part of a pattern.
14882 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14884 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
14885 that the sequence is attached to the original statement rather
14886 than the pattern statement.
14887 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
14888 PATTERN_DEF_SEQ from the original statement rather than
14889 the main pattern statement.
14890 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
14891 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
14892 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
14894 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14896 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
14897 definition statements before the early exit for statements that aren't
14899 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
14901 (vect_transform_loop): ...here. Process pattern definition
14902 statements without first checking whether the main pattern
14903 statement is live or relevant.
14905 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
14907 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
14908 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
14910 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
14912 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
14913 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
14914 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
14915 (expand_block_compare): Change select_block_compare_mode call.
14916 (expand_strncmp_align_check): Use new functions, fix comment.
14917 (emit_final_str_compare_gpr): New function.
14918 (expand_strn_compare): Refactor and clean up code.
14919 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
14921 2018-06-19 Tony Reix <tony.reix@atos.com>
14922 Damien Bergamini <damien.bergamini@atos.com>
14923 David Edelsohn <dje.gcc@gmail.com>
14925 * collect2.c (static_obj): New variable.
14926 (static_libs): New variable.
14927 (is_in_list): Uncomment declaration.
14928 (main): Track AIX libraries linked statically.
14929 (is_in_list): Uncomment definition.
14930 (scan_prog_file): Don't add AIX shared libraries initializer
14931 to constructor list if linking statically.
14933 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
14935 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
14937 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
14939 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
14941 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
14944 2018-06-19 Martin Liska <mliska@suse.cz>
14946 * config/i386/i386.c (ix86_can_inline_p): Do not use
14947 ipa_fn_summaries::get_create.
14948 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
14950 (devirtualization_time_bonus): Likewise.
14951 (ipcp_propagate_stage): Likewise.
14952 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
14953 (edge_set_predicate): Likewise.
14954 (evaluate_conditions_for_known_args): Likewise.
14955 (evaluate_properties_for_edge): Likewise.
14956 (ipa_call_summary::reset): Tranform to ...
14957 (ipa_call_summary::~ipa_call_summary): ... this.
14958 (ipa_fn_summary::reset): Transform to ...
14959 (ipa_fn_summary::~ipa_fn_summary): ... this.
14960 (ipa_fn_summary_t::remove): Rename to ...
14961 (ipa_fn_summary_t::remove_callees): ... this.
14962 (ipa_fn_summary_t::duplicate): Use placement new
14963 instead of memory copy.
14964 (ipa_call_summary_t::duplicate): Likewise.
14965 (ipa_call_summary_t::remove): Remove.
14966 (dump_ipa_call_summary): Change get_create to get.
14967 (ipa_dump_fn_summary): Dump only when summary exists.
14968 (analyze_function_body): Use symbol_summary::get instead
14970 (compute_fn_summary): Likewise.
14971 (estimate_edge_devirt_benefit): Likewise.
14972 (estimate_edge_size_and_time): Likewise.
14973 (inline_update_callee_summaries): Likewise.
14974 (remap_edge_change_prob): Likewise.
14975 (remap_edge_summaries): Likewise.
14976 (ipa_merge_fn_summary_after_inlining): Likewise.
14977 (write_ipa_call_summary): Likewise.
14978 (ipa_fn_summary_write): Likewise.
14979 (ipa_free_fn_summary): Likewise.
14980 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
14981 (struct ipa_call_summary): Likewise.
14982 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
14984 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
14985 (estimate_size_after_inlining): Likewise.
14986 (estimate_growth): Likewise.
14987 (growth_likely_positive): Likewise.
14988 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
14989 (inline_call): Likewise.
14990 * ipa-inline.c (caller_growth_limits): Likewise.
14991 (can_inline_edge_p): Likewise.
14992 (can_inline_edge_by_limits_p): Likewise.
14993 (compute_uninlined_call_time): Likewise.
14994 (compute_inlined_call_time): Likewise.
14995 (want_inline_small_function_p): Likewise.
14996 (edge_badness): Likewise.
14997 (update_caller_keys): Likewise.
14998 (update_callee_keys): Likewise.
14999 (inline_small_functions): Likewise.
15000 (inline_to_all_callers_1): Likewise.
15001 (dump_overall_stats): Likewise.
15002 (early_inline_small_functions): Likewise.
15003 (early_inliner): Likewise.
15004 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15005 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15006 * ipa-pure-const.c (malloc_candidate_p): Likewise.
15007 * ipa-split.c (execute_split_functions): Likewise.
15008 * symbol-summary.h: Likewise.
15009 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
15011 2018-06-19 Richard Biener <rguenther@suse.de>
15013 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
15014 (vectorize_loops): ... here. Fix dbgcnt handling.
15015 (try_vectorize_loop): Wrap try_vectorize_loop_1.
15017 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
15020 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
15021 ieee128 argument takes up only one (vector) register, not two (floating
15024 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15026 * gimplify.c (gimplify_init_constructor): Really never clear for an
15027 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
15029 2018-06-19 Richard Biener <rguenther@suse.de>
15031 PR tree-optimization/86179
15032 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
15033 after failed recognition.
15035 2018-06-18 Martin Sebor <msebor@redhat.com>
15037 PR middle-end/85602
15038 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
15039 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
15040 Handle integer subtraction.
15041 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
15042 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
15044 2018-06-18 David Malcolm <dmalcolm@redhat.com>
15046 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
15047 param from rtx to rtx_insn *.
15048 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
15050 (frv_ifcvt_modify_insn): Likwise.
15051 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
15052 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
15053 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
15054 as_a <rtx_insn *> cast to local "unprotected_region" once
15055 it's been established that it's not NULL or pc_rtx.
15056 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
15057 param "sethi" from rtx to rtx_insn *.
15058 (nds32_group_float_insns): Likewise for param "insn".
15059 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
15061 (vax_output_int_subtract): Likewise.
15062 * config/vax/vax.c (vax_output_int_add): Likewise for param
15064 (vax_output_int_subtract): Likewise.
15065 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
15066 (emit_pattern_after): Likewise for param "after".
15067 (emit_insn_after): Likewise.
15068 (emit_jump_insn_after): Likewise.
15069 (emit_call_insn_after): Likewise.
15070 (emit_debug_insn_after): Likewise.
15071 (emit_pattern_before): Likewise for param "before".
15072 (emit_insn_before): Likewise.
15073 (emit_jump_insn_before): Likewise.
15074 * final.c (get_insn_template): Likewise for param "insn", removing
15076 * output.h (get_insn_template): Likewise for 2nd param.
15077 * rtl.h (emit_insn_before): Likewise.
15078 (emit_jump_insn_before): Likewise.
15079 (emit_debug_insn_before_noloc): Likewise.
15080 (emit_insn_after): Likewise.
15081 (emit_jump_insn_after): Likewise.
15082 (emit_call_insn_after): Likewise.
15083 (emit_debug_insn_after): Likewise.
15084 (set_insn_deleted): Likewise for param.
15086 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
15089 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
15090 floating point modes, so that IFmode is numerically greater than
15091 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
15092 to declare the ordering. This prevents IFmode from being
15093 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
15094 machine. Include rs6000-modes.h to share the fractional values
15095 between genmodes* and the rest of the compiler.
15096 (IFmode): Likewise.
15097 (KFmode): Likewise.
15098 (TFmode): Likewise.
15099 * config/rs6000/rs6000-modes.h: New file.
15100 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
15101 meaning of rs6000_long_double_size so that 126..128 selects an
15102 appropriate 128-bit floating point type.
15103 (rs6000_option_override_internal): Likewise.
15104 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
15105 (TARGET_LONG_DOUBLE_128): Change the meaning of
15106 rs6000_long_double_size so that 126..128 selects an appropriate
15107 128-bit floating point type.
15108 (LONG_DOUBLE_TYPE_SIZE): Update comment.
15109 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
15110 source and destination to match the standard usage.
15111 (truncifkf2): Likewise.
15112 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
15113 ISA 2.07 to use an explicit clobber, instead of passing in a
15115 (copysign<mode>3_soft): Likewise.
15117 2018-06-18 David Malcolm <dmalcolm@redhat.com>
15119 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15120 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
15121 (vect_slp_analyze_instance_dependence): Likewise.
15122 (vect_enhance_data_refs_alignment): Likewise.
15123 (vect_analyze_data_refs_alignment): Likewise.
15124 (vect_slp_analyze_and_verify_instance_alignment
15125 (vect_analyze_data_ref_accesses): Likewise.
15126 (vect_prune_runtime_alias_test_list): Likewise.
15127 (vect_analyze_data_refs): Likewise.
15128 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
15129 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15130 (vect_analyze_scalar_cycles_1): Likewise.
15131 (vect_get_loop_niters): Likewise.
15132 (vect_analyze_loop_form_1): Likewise.
15133 (vect_update_vf_for_slp): Likewise.
15134 (vect_analyze_loop_operations): Likewise.
15135 (vect_analyze_loop): Likewise.
15136 (vectorizable_induction): Likewise.
15137 (vect_transform_loop): Likewise.
15138 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
15139 * tree-vect-slp.c (vect_analyze_slp): Likewise.
15140 (vect_make_slp_decision): Likewise.
15141 (vect_detect_hybrid_slp): Likewise.
15142 (vect_slp_analyze_operations): Likewise.
15143 (vect_slp_bb): Likewise.
15144 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
15145 (vectorizable_bswap): Likewise.
15146 (vectorizable_call): Likewise.
15147 (vectorizable_simd_clone_call): Likewise.
15148 (vectorizable_conversion): Likewise.
15149 (vectorizable_assignment): Likewise.
15150 (vectorizable_shift): Likewise.
15151 (vectorizable_operation): Likewise.
15152 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
15154 2018-06-18 Martin Sebor <msebor@redhat.com>
15156 PR tree-optimization/81384
15157 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
15158 * builtins.c (expand_builtin_strnlen): New function.
15159 (expand_builtin): Call it.
15160 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
15161 * builtins.def (BUILT_IN_STRNLEN): New.
15162 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
15163 Warn for bounds in excess of maximum object size.
15164 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
15165 single-value ranges. Handle strnlen.
15166 (handle_builtin_strlen): Handle strnlen.
15167 (strlen_check_and_optimize_stmt): Same.
15168 * doc/extend.texi (Other Builtins): Document strnlen.
15170 2018-06-18 Maya Rashish <coypu@sdf.org>
15172 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
15173 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
15174 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
15176 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
15178 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
15180 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15182 * tree.c (escaped_string::escape): Replace cast to char * by
15183 const_cast<char *> (unescaped).
15185 2018-06-18 Nick Clifton <nickc@redhat.com>
15188 * tree.c (escaped_string): New class. Converts an unescaped
15189 string into its escaped equivalent.
15190 (warn_deprecated_use): Use the new class to convert the
15191 deprecation message, if present.
15192 (test_escaped_strings): New self test.
15193 (test_c_tests): Add test_escaped_strings.
15194 * doc/extend.texi (deprecated): Add a note that the
15195 deprecation message is affected by the -fmessage-length
15196 option, and that control characters will be escaped.
15197 (#pragma GCC error): Document this pragma.
15198 (#pragma GCC warning): Likewise.
15199 * doc/invoke.texi (-fmessage-length): Document this option's
15200 effect on the #warning and #error preprocessor directives and
15201 the deprecated attribute.
15203 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
15205 * tree.c (decl_value_expr_lookup): Revert latest change.
15206 (decl_value_expr_insert): Likewise.
15208 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
15210 * gimplify.c (nonlocal_vlas): Delete.
15211 (nonlocal_vla_vars): Likewise.
15212 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
15214 (gimplify_body): Do not create and destroy nonlocal_vlas.
15215 * tree-nested.c: Include diagnostic.h.
15216 (use_pointer_in_frame): Tweak.
15217 (lookup_field_for_decl): Add assertion and declare the transformation.
15218 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
15219 internal error when the reference is in a wrong context. Do not
15220 create a debug decl by default.
15221 (note_nonlocal_block_vlas): Delete.
15222 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
15223 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
15224 create a debug decl by default.
15225 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
15226 call is in a wrong context.
15227 (fixup_vla_decls): New function.
15228 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
15229 debug variables were created.
15230 * tree.c (decl_value_expr_lookup): Add checking assertion.
15231 (decl_value_expr_insert): Likewise.
15233 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
15235 PR middle-end/82479
15236 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
15237 * tree-scalar-evolution.c (interpret_expr): Likewise.
15238 (expression_expensive_p): Likewise.
15239 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
15240 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
15241 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
15242 (ssa_defined_by_minus_one_stmt_p): New.
15244 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
15246 PR middle-end/64946
15247 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
15248 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
15249 * dojump.c (do_jump): Likewise.
15250 * expr.c (expand_expr_real_2): Check operand type's sign.
15251 * fold-const.c (const_unop): Handle ABSU_EXPR.
15252 (fold_abs_const): Likewise.
15253 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
15254 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
15255 (strip_sign_op_1): Likesise.
15256 * match.pd: Add new pattern to generate ABSU_EXPR.
15257 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
15258 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15259 * tree-eh.c (operation_could_trap_helper_p): Likewise.
15260 * tree-inline.c (estimate_operator_cost): Likewise.
15261 * tree-pretty-print.c (dump_generic_node): Likewise.
15262 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
15263 * tree.def (ABSU_EXPR): New.
15265 2018-06-16 Jakub Jelinek <jakub@redhat.com>
15267 PR middle-end/86095
15268 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
15269 documented as preserved for backward compatibility only.
15270 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
15272 PR rtl-optimization/86108
15273 * bb-reorder.c (create_forwarder_block): Renamed to ...
15274 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
15275 jump from new landing pad to the second part.
15276 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
15279 2018-06-15 Jakub Jelinek <jakub@redhat.com>
15281 PR middle-end/85878
15282 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
15283 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
15284 Only call store_expr for halves if the mode is the same.
15286 PR middle-end/86123
15287 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
15288 Fix up comment formatting.
15290 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15292 * typed-splay-tree.h (typed_splay_tree::remove): New function.
15293 (typed_splay_tree::closure,
15294 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
15295 (typed_splay_tree::typed_splay_tree,
15296 typed_splay_tree::operator =): Declared private.
15297 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
15298 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
15299 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
15300 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
15301 typed_splay_tree::splay_tree_splay,
15302 typed_splay_tree::splay_tree_foreach_helper,
15303 typed_splay_tree::splay_tree_insert,
15304 typed_splay_tree::splay_tree_remove,
15305 typed_splay_tree::splay_tree_lookup,
15306 typed_splay_tree::splay_tree_predecessor,
15307 typed_splay_tree::splay_tree_successor,
15308 typed_splay_tree::splay_tree_min,
15309 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
15310 (typed_splay_tree::root, typed_splay_tree::comp,
15311 typed_splay_tree::delete_key,
15312 typed_splay_tree::delete_value): New data members.
15313 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
15314 typed_splay_tree::remove.
15316 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
15318 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
15319 -mginv and -mno-ginv to the assembler.
15320 * config/mips/mips.opt (-mcrc): New option.
15321 (-mginv): Likewise.
15322 * doc/invoke.text (-mcrc): Document.
15323 (-mginv): Likewise.
15325 2018-06-15 Nick Clifton <nickc@redhat.com>
15328 * tree.c (escaped_string): New class. Converts an unescaped
15329 string into its escaped equivalent.
15330 (warn_deprecated_use): Use the new class to convert the
15331 deprecation message, if present.
15332 (test_escaped_strings): New self test.
15333 (test_c_tests): Add test_escaped_strings.
15334 * doc/extend.texi (deprecated): Add a note that the
15335 deprecation message is affected by the -fmessage-length
15336 option, and that control characters will be escaped.
15337 (#pragma GCC error): Document this pragma.
15338 (#pragma GCC warning): Likewise.
15339 * doc/invoke.texi (-fmessage-length): Document this option's
15340 effect on the #warning and #error preprocessor directives and
15341 the deprecated attribute.
15343 2018-06-15 Richard Biener <rguenther@suse.de>
15345 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
15346 here, also noting vector size used.
15347 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
15348 size used in MSG_OPTIMIZED_LOCATIONS dump.
15349 (pass_slp_vectorize::execute): Adjust.
15351 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
15354 * config/arc/arc.c (arc_return_address_register): Fix
15357 2018-06-15 Richard Biener <rguenther@suse.de>
15359 PR middle-end/86159
15360 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
15361 leave useless conversion stripping to force_gimple_operand_gsi.
15362 (gimplify_build2): Likewise.
15363 (gimplify_build1): Likewise.
15365 2018-06-15 Richard Biener <rguenther@suse.de>
15367 PR middle-end/86076
15368 * tree-cfg.c (move_stmt_op): unshare invariant addresses
15369 before adjusting their block.
15371 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
15373 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
15374 multilibs for *-*-rtems*.
15375 * config/riscv/t-rtems: New file.
15377 2018-06-14 Jakub Jelinek <jakub@redhat.com>
15379 PR middle-end/86122
15380 * match.pd ((A +- CST1) +- CST2): Punt if last resort
15381 unsigned_type_for returns NULL.
15384 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
15385 subregs of multi-word pseudos unless the float mode has word size.
15387 2018-06-14 Richard Biener <rguenther@suse.de>
15389 PR middle-end/86139
15390 * tree-vect-generic.c (build_word_mode_vector_type): Remove
15391 duplicate and harmful type_hash_canon.
15392 * tree.c (type_hash_canon): Assert we didn't find ourselves.
15394 2018-06-14 Richard Biener <rguenther@suse.de>
15397 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
15400 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
15402 * config/rtems.h (STDINT_LONG32): Define.
15404 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
15405 Prachi Godbole <prachi.godbole@imgtec.com>
15407 * config/mips/mips-cpus.def: Define P6600.
15408 * config/mips/mips-tables.opt: Regenerate.
15409 * config/mips/mips.c (mips_ucbranch_type): New enum.
15410 (mips_rtx_cost_data): Add support for P6600.
15411 (mips_issue_rate): Likewise.
15412 (mips_multipass_dfa_lookahead): Likewise.
15413 (mips_avoid_hazard): Likewise.
15414 (mips_reorg_process_insns): Likewise.
15415 (mips_classify_branch_p6600): New function.
15416 * config/mips/mips.h (TUNE_P6600): New define.
15417 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
15418 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
15419 * config/mips/mips.md: Include p6600.md.
15420 (processor): Add p6600.
15421 * config/mips/p6600.md: New file.
15422 * doc/invoke.texi: Add p6600 to supported architectures.
15424 2018-06-13 Martin Sebor <msebor@redhat.com>
15426 PR tree-optimization/86114
15427 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
15429 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
15431 2018-06-13 Richard Biener <rguenther@suse.de>
15433 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
15434 Properly set vector type of the intermediate stmt.
15435 * tree-vect-stmts.c (vectorizable_operation): The destination
15436 var always has vectype_out type.
15438 2018-06-13 Jeff Law <law@redhat.com>
15440 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
15441 integer 0 for argument to print_rtl_with_bb.
15442 (rl78_reorg): Likewise.
15444 2018-06-13 David Malcolm <dmalcolm@redhat.com>
15446 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
15447 from rtx to rtx_insn *.
15448 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
15450 (add_sched_insns_for_speculation): Likewise for local "target",
15451 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
15452 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
15453 from rtx_insn ** to rtx_code_label **.
15454 (reorg_emit_nops): Likewise.
15455 (c6x_reorg): Likewise for local "call_labels".
15456 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
15458 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
15459 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
15460 the loops over LABEL_REFS.
15461 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
15463 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
15464 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
15465 (split_branches): Strengthen local "olabel" from rtx to
15466 rtx_insn *, adding a safe_as_a cast.
15467 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
15469 (add_insn_after): Likewise for first two params.
15470 (add_insn_before): Likewise.
15471 (remove_insn): Likewise for param.
15472 (emit_pattern_before_noloc): Likewise for second and third params.
15473 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
15474 (emit_call_insn_before_noloc): Likewise.
15475 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
15477 (emit_barrier_before): Likewise.
15478 (emit_label_before): Strengthen "label" param from "rtx" to
15479 "rtx_code_label *". Strengthen "before" param from "rtx" to
15481 (emit_insn_after_1): Strengthen "after" param from "rtx" to
15483 (emit_pattern_after_noloc): Likewise.
15484 (emit_insn_after_noloc): Likewise.
15485 (emit_jump_insn_after_noloc): Likewise.
15486 (emit_call_insn_after_noloc): Likewise.
15487 (emit_debug_insn_after_noloc): Likewise.
15488 (emit_barrier_after): Likewise.
15489 (emit_label_after): Likewise for both params.
15490 (emit_pattern_after_setloc): Likewise for "after" param. Convert
15491 "loc" param from "int" to "location_t".
15492 (emit_insn_after_setloc): Likewise.
15493 (emit_jump_insn_after_setloc): Likewise.
15494 (emit_call_insn_after_setloc): Likewise.
15495 (emit_debug_insn_after_setloc): Likewise.
15496 (emit_pattern_before_setloc): Likewise for "before" param. Convert
15497 "loc" param from "int" to "location_t".
15498 (emit_pattern_before): Convert NULL_RTX to NULL.
15499 (emit_insn_before_setloc): Convert "loc" param from "int" to
15501 (emit_jump_insn_before_setloc): Likewise.
15502 (emit_call_insn_before_setloc): Likewise.
15503 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
15504 rtx_insn *. Convert "loc" param from "int" to "location_t".
15505 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
15506 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
15507 Convert 3rd param from "int" to "location_t".
15508 (emit_barrier_before, emit_barrier_after, next_real_insn):
15509 Strengthen param from rtx to rtx_insn *.
15510 (emit_label_before): Strengthen 1st param from "rtx" to
15511 "rtx_code_label *". Strengthen 2nd param from "rtx" to
15513 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
15514 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
15515 Strengthen 2nd param from "rtx" to "rtx_insn *".
15516 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
15517 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
15518 Likewise. Convert 3rd param from "int" to "location_t".
15519 (emit_label_after): Strengthen 1st param from "rtx" to
15520 "rtx_code_label *".
15521 (next_real_insn, remove_insn): Strengthen param from "rtx" to
15523 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
15524 from "rtx" to "rtx_insn *".
15526 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
15528 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
15529 bodies streamed in with -Q.
15530 * dumpfile.c (dump_files): Add lto-stream-out dump file.
15531 * dumpfile.h (tree_dump_index): Add lto_stream_out.
15532 * gimple-streamer-out.c: Include gimple-pretty-print.h
15533 (output_bb): Dump stmts streamed.
15534 * lto-section-out.c: Include print-tree.h
15535 (lto_begin_section): Dump sections created.
15536 (lto_output_decl_index): Dump decl encoded.
15537 * lto-streamer-out.c: Include print-tree.h
15538 (create_output_block): Dump output block created.
15539 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
15540 (output_function): Dump function output.
15541 (output_constructor): Dump constructor streamed.
15542 (write_global_stream): Output indexes encoded.
15543 (produce_asm_for_decls): Dump streams encoded.
15544 * lto-streamer.c (streamer_dump_file): New global var.
15545 * lto-streamer.h (streamer_dump_file): Declare.
15546 * passes.c (ipa_write_summaries): Initialize streamer dump.
15547 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
15550 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
15553 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
15554 offsets for register save directives. Emit a second batch of save
15555 directives, if need be, when the function accesses prior frames.
15557 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15559 * config/arc/fpu.md (fmasf4): Force operand to register.
15560 (fnmasf4): Likewise.
15562 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15564 * config/arc/arc-protos.h (arc_pad_return): Remove.
15565 * config/arc/arc.c (machine_function): Remove force_short_suffix
15567 (arc_print_operand): Adjust printing of '&'.
15568 (arc_verify_short): Remove conditional printing of short suffix.
15569 (arc_final_prescan_insn): Remove reference to size_reason.
15570 (pad_return): New function.
15571 (arc_reorg): Call pad_return.
15572 (arc_pad_return): Remove.
15573 (arc_init_machine_status): Remove reference to force_short_suffix.
15574 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
15575 (attr length): When attribute iscompact is true force to 2
15576 regardless; in the case of maybe check if we want to force the
15577 instruction to have 4 bytes length.
15578 (nopv): Change it to generate 4 byte long nop as well.
15579 (blockage): New pattern.
15580 (simple_return): Remove call to arc_pad_return.
15581 (p_return_i): Likewise.
15583 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15585 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
15587 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15589 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
15592 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15594 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
15595 for ARC700 and ARCv2.
15597 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
15600 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
15601 operands[2] instead of operands[1].
15604 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15606 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
15607 case, check whether the outer register overlaps an unallocatable
15608 register, not just whether it fits the required class.
15610 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15612 * poly-int.h (can_div_trunc_p): Add new overload in which all values
15614 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
15615 (memrefs_conflict_p): Likewise.
15616 (init_alias_analysis): Likewise.
15617 * cfgexpand.c (expand_debug_expr): Likewise.
15618 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
15619 * cse.c (fold_rtx): Likewise.
15620 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
15621 * expr.c (emit_block_move_hints): Likewise.
15622 (clear_storage_hints, push_block, emit_push_insn): Likewise.
15623 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
15624 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
15625 (emit_group_store): Likewise.
15626 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
15627 to read the PRE/POST_MODIFY increment.
15628 * calls.c (store_one_arg): Use strip_offset.
15629 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
15631 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
15633 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
15634 (simplify_binary_operation_1): Extend CONST_INT handling to
15636 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
15637 than a HOST_WIDE_INT.
15638 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
15640 (adjust_mems, add_stores): Update accodingly.
15641 (vt_canonicalize_addr): Track polynomial offsets.
15642 (emit_note_insn_var_location): Likewise.
15643 (vt_add_function_parameter): Likewise.
15644 (vt_initialize): Likewise.
15646 2018-06-12 Jeff Law <law@redhat.com>
15648 * config.gcc (alpha*-*-freebsd*): Remove.
15649 * config/alpha/freebsd.h: Remove.
15651 2018-06-12 David Malcolm <dmalcolm@redhat.com>
15654 * spellcheck-tree.c (levenshtein_distance): Rename to...
15655 (get_edit_distance): ...this, and update for underlying renaming.
15656 * spellcheck-tree.h (levenshtein_distance): Rename to...
15657 (get_edit_distance): ...this.
15658 * spellcheck.c (levenshtein_distance): Rename to...
15659 (get_edit_distance): ...this. Convert from Levenshtein distance
15660 to Damerau-Levenshtein distance by supporting transpositions of
15661 adjacent characters. Rename "v1" to "v_next" and "v0" to
15663 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
15664 (selftest::test_edit_distance_unit_test_oneway): ...this, and
15665 update for underlying renaming.
15666 (selftest::levenshtein_distance_unit_test): Rename to...
15667 (selftest::test_get_edit_distance_unit): ...this, and update for
15668 underlying renaming.
15669 (selftest::test_find_closest_string): Add example from PR 69968
15670 where transposition helps
15671 (selftest::test_metric_conditions): Update for renaming.
15672 (selftest::test_metric_conditions): Likewise.
15673 (selftest::spellcheck_c_tests): Likewise.
15674 * spellcheck.h (levenshtein_distance): Rename both overloads to...
15675 (get_edit_distance): ...this.
15676 (best_match::consider): Update for renaming.
15678 2018-06-12 Martin Sebor <msebor@redhat.com>
15680 PR tree-optimization/85259
15681 * builtins.c (compute_objsize): Handle constant offsets.
15682 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
15683 true iff a warning has been issued.
15684 * gimple.h (gimple_nonartificial_location): New function.
15685 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
15686 gimple_nonartificial_location and handle -Wno-system-headers.
15687 (handle_builtin_stxncpy): Same.
15689 2018-06-12 Martin Sebor <msebor@redhat.com>
15692 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
15694 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15696 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15697 BUILTIN_VEC_XST entries for pointer to double and long long.
15699 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
15702 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
15704 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
15707 2018-06-12 Martin Liska <mliska@suse.cz>
15709 * doc/options.texi: Document IntegerRange.
15711 2018-06-12 Martin Liska <mliska@suse.cz>
15713 * config/i386/i386.opt: Make MPX-related options as Deprecated.
15714 * opt-functions.awk: Handle Deprecated flag.
15715 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
15717 (read_cmdline_option): Report warning for a deprecated option.
15718 * opts.h (struct cl_option): Add new field cl_deprecated.
15719 (CL_ERR_DEPRECATED): New.
15721 2018-06-12 Martin Liska <mliska@suse.cz>
15723 * doc/options.texi: Document Deprecated option flag.
15725 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15727 * config/arc/arc-arch.h (arc_extras): New enum.
15728 (arc_cpu_t):Add field extra.
15729 (arc_cpu_types): Consider the extras.
15730 * config/arc/arc-cpus.def: Add extras info.
15731 * config/arc/arc-opts.h (processor_type): Consider extra field.
15732 * config/arc/arc.c (arc_override_options): Handle extra field.
15734 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15736 * config/arc/arc-arch.h: Update ARC_OPTX macro.
15737 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
15739 * config/arc/arc.c (arc_init): Update pic warning.
15740 (irq_range): Update irq range parsing warnings.
15741 (arc_override_options): Update various warning messages.
15742 (arc_handle_aux_attribute): Likewise.
15744 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
15746 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
15748 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15750 * doc/sourcebuild.texi: Document usage of line number 0 in verify
15751 compiler messages directives.
15753 2018-06-12 Matthew Fortune <mfortune@gmail.com>
15755 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
15756 * config/mips/mips-tables.opt: Regenerate.
15757 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
15759 * doc/invoke.texi: Document -march=i6500.
15761 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
15763 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
15764 (i6400_gpmul): Add cpu_unit.
15765 (i6400_gpdiv): Likewise.
15766 (i6400_msa_add_d): Update reservations.
15767 (i6400_msa_int_add) Likewise.
15768 (i6400_msa_short_logic3) Likewise.
15769 (i6400_msa_short_logic2) Likewise.
15770 (i6400_msa_short_logic) Likewise.
15771 (i6400_msa_move) Likewise.
15772 (i6400_msa_cmp) Likewise.
15773 (i6400_msa_short_float2) Likewise.
15774 (i6400_msa_div_d) Likewise.
15775 (i6400_msa_long_logic1) Likewise.
15776 (i6400_msa_long_logic2) Likewise.
15777 (i6400_msa_mult) Likewise.
15778 (i6400_msa_long_float2) Likewise.
15779 (i6400_msa_long_float4) Likewise.
15780 (i6400_msa_long_float5) Likewise.
15781 (i6400_msa_long_float8) Likewise.
15782 (i6400_fpu_fadd): Include frint type.
15783 (i6400_fpu_store): New define_insn_reservation.
15784 (i6400_fpu_load): Likewise.
15785 (i6400_fpu_move): Likewise.
15786 (i6400_fpu_fcmp): Likewise.
15787 (i6400_fpu_fmadd): Likewise.
15788 (i6400_int_mult): Include imul3nc type and update reservation.
15789 (i6400_int_div): Include idiv3 type and update reservation.
15790 (i6400_int_load): Update to check type not move_type.
15791 (i6400_int_store): Likewise.
15792 (i6400_int_prefetch): Set zero latency.
15794 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
15796 * gcc.c: Document new %@{...} sequence.
15797 (LINK_COMMAND_SPEC): Use it for the -L switches.
15798 (cpp_unique_options): Use it for the -I switches.
15799 (at_file_argbuf): New global variable.
15800 (in_at_file): Likewise.
15801 (alloc_args): Create at_file_argbuf.
15802 (clear_args): Truncate at_file_argbuf.
15803 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
15804 (open_at_file): New function.
15805 (close_at_file): Likewise.
15806 (create_at_file): Delete.
15807 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
15810 (validate_switches_from_spec): Deal with %@{...} sequence.
15811 (validate_switches): Likewise.
15812 (driver::finalize): Call clear_args.
15814 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
15816 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
15818 2018-06-11 Martin Sebor <msebor@redhat.com>
15820 * doc/invoke.texi (-Wall): List -Wc++17-compat.
15821 (Wno-class-memaccess): Add @opindex.
15822 (Wno-templates, Wno-multiple-inheritance): Same.
15823 (Wno-virtual-inheritance, Wno-namespaces): Same.
15824 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
15825 (Wno-format-overflow, Wno-format-truncation): Same.
15826 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
15827 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
15828 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
15829 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
15830 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
15831 (Wno-misspelled-isr): Same.
15833 2018-06-11 Martin Sebor <msebor@redhat.com>
15835 * PR tree-optimization/86083
15836 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
15838 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
15840 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
15842 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
15845 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
15846 on the correct operand.
15847 (*movdi_internal64): Ditto.
15849 2018-06-11 Martin Liska <mliska@suse.cz>
15851 PR tree-optimization/86089
15852 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
15854 2018-06-11 Julia Koval <julia.koval@intel.com>
15856 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
15857 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
15858 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
15860 2018-06-11 Olivier Hainque <hainque@adacore.com>
15862 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
15863 for Ada with strict dwarf2.
15865 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
15868 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
15871 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
15873 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
15875 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
15877 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
15880 2018-06-08 Martin Liska <mliska@suse.cz>
15882 * tree-cfg.h (debug_function): Fix argument type to match
15885 2018-06-08 Martin Liska <mliska@suse.cz>
15887 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
15888 Remove usage of MPX-related (and removed) fields.
15889 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
15891 2018-06-08 David Malcolm <dmalcolm@redhat.com>
15893 * cfg.c (debug): Use TDF_NONE rather than 0.
15894 * cfghooks.c (debug): Likewise.
15895 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
15896 (struct dump_option_value_info): Convert to...
15897 (struct kv_pair): ...this template type.
15898 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
15900 (optinfo_verbosity_options): Likewise.
15901 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
15903 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
15904 than int for "optgroup_flags" param.
15905 (dump_generic_expr_loc): Use dump_flags_t rather than int for
15907 (dump_dec): Likewise.
15908 (dump_finish): Use TDF_NONE rather than 0.
15909 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
15910 rather than int for "optgroup_flags" param. Use TDF_NONE rather
15911 than 0. Update for change to option_ptr.
15912 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
15913 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
15914 0. Update for changes to optinfo_verbosity_options and
15916 (opt_info_switch_p): Convert optgroup_flags from int to
15918 (dump_basic_block): Use dump_flags_t rather than int
15919 for "dump_kind" param.
15920 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
15921 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
15922 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
15923 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
15924 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
15925 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
15926 TDF_NONE): Convert from macros to...
15927 (enum dump_flag): ...this new enum.
15928 (dump_flags_t): Update to use enum.
15929 (operator|, operator&, operator~, operator|=, operator&=):
15930 Implement for dump_flags_t.
15931 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
15932 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
15933 Convert from macros to...
15934 (enum optgroup_flag): ...this new enum.
15935 (optgroup_flags_t): New typedef.
15936 (operator|, operator|=): Implement for optgroup_flags_t.
15937 (struct dump_file_info): Convert field "alt_flags" to
15938 dump_flags_t. Convert field "optgroup_flags" to
15940 (dump_basic_block): Use dump_flags_t rather than int for param.
15941 (dump_generic_expr_loc): Likewise.
15942 (dump_dec): Likewise.
15943 (dump_register): Convert param "optgroup_flags" to
15945 (opt_info_enable_passes): Likewise.
15946 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
15948 * gimple-pretty-print.c (debug): Likewise.
15949 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
15950 (merged_store_group::apply_stores): Likewise.
15951 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
15952 * gimple.c (verify_gimple_pp): Likewise.
15953 * graphite-poly.c (print_pbb_body): Likewise.
15954 * passes.c (pass_manager::register_one_dump_file): Convert
15955 local "optgroup_flags" to optgroup_flags_t.
15956 * print-tree.c (print_node): Use TDF_NONE rather than 0.
15958 (debug_body): Likewise.
15959 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
15960 to optgroup_flags_t.
15961 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
15963 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
15964 (convert_mult_to_fma): Likewise.
15965 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
15966 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
15967 * tree-vect-data-refs.c (dump_lower_bound): Convert param
15968 "dump_kind" to dump_flags_t.
15970 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
15972 * config/rs6000/rs6000.c (min, max): Delete.
15974 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
15976 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
15977 -mabi=spe and -mabi=no-spe.
15979 2018-06-08 Martin Liska <mliska@suse.cz>
15981 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
15982 where we expect an existing summary.
15984 2018-06-08 Martin Liska <mliska@suse.cz>
15986 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
15987 * ipa-inline.h (estimate_edge_growth): Likewise.
15989 2018-06-08 Martin Liska <mliska@suse.cz>
15991 * cgraph.c (function_version_hasher::hash): Use
15992 cgraph_node::get_uid ().
15993 (function_version_hasher::equal):
15994 * cgraph.h (cgraph_node::get_uid): New method.
15995 * ipa-inline.c (update_caller_keys): Use
15996 cgraph_node::get_uid ().
15997 (update_callee_keys): Likewise.
15998 * ipa-utils.c (searchc): Likewise.
15999 (ipa_reduced_postorder): Likewise.
16000 * lto-cgraph.c (input_node): Likewise.
16001 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
16002 * symbol-summary.h (symtab_insertion): Likewise.
16003 (symtab_removal): Likewise.
16004 (symtab_duplication): Likewise.
16005 * tree-pretty-print.c (dump_function_header): Likewise.
16006 * tree-sra.c (convert_callers_for_node): Likewise.
16008 2018-06-08 Martin Liska <mliska@suse.cz>
16010 * cgraph.c (symbol_table::create_edge): Always assign a new
16012 (symbol_table::free_edge): Do not recycle numbers.
16013 * cgraph.h (cgraph_edge::get): New method.
16014 * symbol-summary.h (symtab_removal): Use it.
16015 (symtab_duplication): Likewise.
16016 (call_summary::hashable_uid): Remove.
16018 2018-06-08 Martin Liska <mliska@suse.cz>
16020 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
16021 (initialize_growth_caches): Remove.
16022 (free_growth_caches): Likewise.
16023 (do_estimate_edge_time): Use edge_growth_cache.
16024 (do_estimate_edge_size): Likewise.
16025 (do_estimate_edge_hints): Likewise.
16026 * ipa-inline.c (reset_edge_caches): Likewise.
16027 (recursive_inlining): Likewise.
16028 (inline_small_functions): Likewise.
16029 * ipa-inline.h (initialize_growth_caches): Remove.
16030 (estimate_edge_size): Likewise.
16031 (estimate_edge_time): Likewise.
16032 (estimate_edge_hints): Likewise.
16033 (reset_edge_growth_cache): Likewise.
16034 * symbol-summary.h (call_summary::remove): New method.
16036 2018-06-08 Martin Liska <mliska@suse.cz>
16038 * ipa-cp.c (class edge_clone_summary): New summary.
16039 (grow_edge_clone_vectors): Remove.
16040 (ipcp_edge_duplication_hook): Remove.
16041 (class edge_clone_summary_t): New call_summary class.
16042 (ipcp_edge_removal_hook): Remove.
16043 (edge_clone_summary_t::duplicate): New function.
16044 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
16045 (create_specialized_node): Likewise.
16046 (ipcp_driver): Initialize edge_clone_summaries and do not
16049 2018-06-08 Martin Liska <mliska@suse.cz>
16051 * symbol-summary.h (get): New function.
16052 (call_summary::m_initialize_when_cloning): New class member.
16054 2018-06-08 Martin Liska <mliska@suse.cz>
16056 * cgraph.c (cgraph_node::remove): Do not recycle uid.
16057 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
16058 (symbol_table::allocate_cgraph_symbol): Do not set uid.
16059 * passes.c (uid_hash_t): Record removed_nodes by their uids.
16060 (remove_cgraph_node_from_order): Use the removed_nodes set.
16061 (do_per_function_toporder): Likwise.
16062 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
16063 instead of summary_uid.
16064 (symtab_removal): Likewise.
16065 (symtab_duplication): Likewise.
16067 2018-06-08 Martin Liska <mliska@suse.cz>
16069 * ipa-cp.c (ipcp_store_bits_results): Use
16070 ipcp_transformation_sum.
16071 (ipcp_store_vr_results): Likewise.
16072 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
16074 (ipcp_transformation_initialize): ... this.
16075 (ipa_set_node_agg_value_chain):
16076 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
16077 (write_ipcp_transformation_info): Likewise.
16078 (read_ipcp_transformation_info): Likewise.
16079 (ipcp_update_bits): Likewise.
16080 (ipcp_update_vr): Likewise.
16081 (ipcp_transform_function): Likewise.
16082 * ipa-prop.h: Rename ipcp_transformation_summary to
16083 ipcp_transformation.
16084 (class ipcp_transformation_t): New function summary.
16085 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
16086 (ipa_get_agg_replacements_for_node): Likewise.
16088 2018-06-08 Martin Liska <mliska@suse.cz>
16090 * ipa-pure-const.c (struct funct_state_d): Do it class instead
16092 (class funct_state_summary_t): New function_summary class.
16093 (has_function_state): Remove.
16094 (get_function_state): Likewise.
16095 (set_function_state): Likewise.
16096 (add_new_function): Likewise.
16097 (funct_state_summary_t::insert): New function.
16098 (duplicate_node_data): Remove.
16099 (remove_node_data): Remove.
16100 (funct_state_summary_t::duplicate): New function.
16101 (register_hooks): Create new funct_state_summaries.
16102 (pure_const_generate_summary): Use it.
16103 (pure_const_write_summary): Likewise.
16104 (pure_const_read_summary): Likewise.
16105 (propagate_pure_const): Likewise.
16106 (propagate_nothrow): Likewise.
16107 (dump_malloc_lattice): Likewise.
16108 (propagate_malloc): Likewise.
16109 (execute): Do not register hooks, just remove summary
16111 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
16114 2018-06-08 Martin Liska <mliska@suse.cz>
16116 * ipa-reference.c (remove_node_data): Remove.
16117 (duplicate_node_data): Likewise.
16118 (class ipa_ref_var_info_summary_t): New class.
16119 (class ipa_ref_opt_summary_t): Likewise.
16120 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
16121 (get_reference_optimization_summary): Use
16122 ipa_ref_opt_sum_summaries.
16123 (set_reference_vars_info): Remove.
16124 (set_reference_optimization_summary): Likewise.
16125 (ipa_init): Create summaries.
16126 (init_function_info): Use function summary.
16127 (ipa_ref_opt_summary_t::duplicate): New function.
16128 (ipa_ref_opt_summary_t::remove): New function.
16129 (get_read_write_all_from_node): Fix GNU coding style.
16130 (propagate): Use function summary.
16131 (write_node_summary_p): Fix GNU coding style.
16132 (stream_out_bitmap): Likewise.
16133 (ipa_reference_read_optimization_summary): Use function summary.
16134 (ipa_reference_c_finalize): Do not release hooks.
16136 2018-06-08 Martin Liska <mliska@suse.cz>
16138 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
16139 (analyze_function_body): Extract multiple calls of get_create.
16140 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
16141 * ipa-inline.c (recursive_inlining): Use ::get method.
16142 * ipa-inline.h (estimate_edge_growth): Likewise.
16144 2018-06-08 Martin Liska <mliska@suse.cz>
16146 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
16148 (hsa_function_summary::hsa_function_summary): Use the new enum
16150 (hsa_gpu_implementation_p): Use hsa_summaries::get.
16151 * hsa-gen.c (hsa_get_host_function): Likewise.
16152 (get_brig_function_name): Likewise.
16153 * ipa-hsa.c (process_hsa_functions): Likewise.
16154 (ipa_hsa_write_summary): Likewise.
16155 * symbol-summary.h (symtab_duplication): Use ::get function/
16156 (get): New function.
16158 2018-06-08 Martin Liska <mliska@suse.cz>
16160 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
16162 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
16163 (hsa_register_kernel): Likewise.
16164 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
16165 * hsa-gen.c (hsa_get_host_function): Likewise.
16166 (get_brig_function_name): Likewise.
16167 (generate_hsa): Likewise.
16168 (pass_gen_hsail::execute): Likewise.
16169 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
16170 (devirtualization_time_bonus): Likewise.
16171 (ipcp_propagate_stage): Likewise.
16172 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
16173 (edge_set_predicate): Likewise.
16174 (evaluate_conditions_for_known_args): Likewise.
16175 (evaluate_properties_for_edge): Likewise.
16176 (ipa_fn_summary::reset): Likewise.
16177 (ipa_fn_summary_t::duplicate): Likewise.
16178 (dump_ipa_call_summary): Likewise.
16179 (ipa_dump_fn_summary): Likewise.
16180 (analyze_function_body): Likewise.
16181 (compute_fn_summary): Likewise.
16182 (estimate_edge_devirt_benefit): Likewise.
16183 (estimate_edge_size_and_time): Likewise.
16184 (estimate_calls_size_and_time): Likewise.
16185 (estimate_node_size_and_time): Likewise.
16186 (inline_update_callee_summaries): Likewise.
16187 (remap_edge_change_prob): Likewise.
16188 (remap_edge_summaries): Likewise.
16189 (ipa_merge_fn_summary_after_inlining): Likewise.
16190 (ipa_update_overall_fn_summary): Likewise.
16191 (read_ipa_call_summary): Likewise.
16192 (inline_read_section): Likewise.
16193 (write_ipa_call_summary): Likewise.
16194 (ipa_fn_summary_write): Likewise.
16195 (ipa_free_fn_summary): Likewise.
16196 * ipa-hsa.c (process_hsa_functions): Likewise.
16197 (ipa_hsa_write_summary): Likewise.
16198 (ipa_hsa_read_section): Likewise.
16199 * ipa-icf.c (sem_function::merge): Likewise.
16200 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
16201 (do_estimate_edge_time): Likewise.
16202 (estimate_size_after_inlining): Likewise.
16203 (estimate_growth): Likewise.
16204 (growth_likely_positive): Likewise.
16205 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
16206 (inline_call): Likewise.
16207 * ipa-inline.c (caller_growth_limits): Likewise.
16208 (can_inline_edge_p): Likewise.
16209 (can_inline_edge_by_limits_p): Likewise.
16210 (compute_uninlined_call_time): Likewise.
16211 (compute_inlined_call_time): Likewise.
16212 (want_inline_small_function_p): Likewise.
16213 (edge_badness): Likewise.
16214 (update_caller_keys): Likewise.
16215 (update_callee_keys): Likewise.
16216 (recursive_inlining): Likewise.
16217 (inline_small_functions): Likewise.
16218 (inline_to_all_callers_1): Likewise.
16219 (dump_overall_stats): Likewise.
16220 (early_inline_small_functions): Likewise.
16221 (early_inliner): Likewise.
16222 * ipa-inline.h (estimate_edge_growth): Likewise.
16223 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16224 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16225 * ipa-prop.h (IPA_NODE_REF): Likewise.
16226 (IPA_EDGE_REF): Likewise.
16227 * ipa-pure-const.c (malloc_candidate_p): Likewise.
16228 (propagate_malloc): Likewise.
16229 * ipa-split.c (execute_split_functions): Likewise.
16230 * symbol-summary.h: Rename get to get_create.
16232 (get_create): Likewise.
16233 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
16235 2018-06-08 Martin Liska <mliska@suse.cz>
16237 * symbol-summary.h (release): Move definition out of class
16239 (symtab_removal): Likewise.
16240 (symtab_duplication): Likewise.
16242 2018-06-08 Martin Liska <mliska@suse.cz>
16244 * symbol-summary.h (function_summary): Move constructor
16245 implementation out of class declaration.
16246 (release): Likewise.
16247 (symtab_insertion): Likewise.
16248 (symtab_removal): Likewise.
16249 (symtab_duplication): Likewise.
16252 2018-06-08 Martin Liska <mliska@suse.cz>
16254 * Makefile.in: Remove support for MPX (macros, related functions,
16255 fields in cgraph_node, ...).
16256 * builtin-types.def (BT_BND): Likewise.
16257 (BT_FN_BND_CONST_PTR): Likewise.
16258 (BT_FN_CONST_PTR_BND): Likewise.
16259 (BT_FN_VOID_PTR_BND): Likewise.
16260 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
16261 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
16262 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
16263 (expand_builtin_mempcpy_with_bounds): Likewise.
16264 (expand_builtin_memset_with_bounds): Likewise.
16265 (expand_builtin_memset_args): Likewise.
16266 (std_expand_builtin_va_start): Likewise.
16267 (expand_builtin): Likewise.
16268 (expand_builtin_with_bounds): Likewise.
16269 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
16270 (DEF_LIB_BUILTIN_CHKP): Likewise.
16271 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
16272 (DEF_CHKP_BUILTIN): Likewise.
16273 (BUILT_IN_MEMCPY): Likewise.
16274 (BUILT_IN_MEMMOVE): Likewise.
16275 (BUILT_IN_MEMPCPY): Likewise.
16276 (BUILT_IN_MEMSET): Likewise.
16277 (BUILT_IN_STPCPY): Likewise.
16278 (BUILT_IN_STRCAT): Likewise.
16279 (BUILT_IN_STRCHR): Likewise.
16280 (BUILT_IN_STRCPY): Likewise.
16281 (BUILT_IN_STRLEN): Likewise.
16282 (BUILT_IN_MEMCPY_CHK): Likewise.
16283 (BUILT_IN_MEMMOVE_CHK): Likewise.
16284 (BUILT_IN_MEMPCPY_CHK): Likewise.
16285 (BUILT_IN_MEMSET_CHK): Likewise.
16286 (BUILT_IN_STPCPY_CHK): Likewise.
16287 (BUILT_IN_STRCAT_CHK): Likewise.
16288 (BUILT_IN_STRCPY_CHK): Likewise.
16289 * calls.c (store_bounds): Likewise.
16290 (emit_call_1): Likewise.
16291 (special_function_p): Likewise.
16292 (maybe_warn_nonstring_arg): Likewise.
16293 (initialize_argument_information): Likewise.
16294 (finalize_must_preallocate): Likewise.
16295 (compute_argument_addresses): Likewise.
16296 (expand_call): Likewise.
16297 * cfgexpand.c (expand_call_stmt): Likewise.
16298 (expand_return): Likewise.
16299 (expand_gimple_stmt_1): Likewise.
16300 (pass_expand::execute): Likewise.
16301 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16302 (cgraph_node::remove): Likewise.
16303 (cgraph_node::dump): Likewise.
16304 (cgraph_node::verify_node): Likewise.
16305 * cgraph.h (chkp_function_instrumented_p): Likewise.
16306 (symtab_node::get_alias_target): Likewise.
16307 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
16308 (cgraph_local_p): Likewise.
16309 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
16310 (cgraph_edge::rebuild_references): Likewise.
16311 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
16312 (walk_polymorphic_call_targets): Likewise.
16313 (cgraph_node::expand_thunk): Likewise.
16314 (symbol_table::output_weakrefs): Likewise.
16315 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
16316 (ix86_handle_option): Likewise.
16317 * config/i386/constraints.md: Likewise.
16318 * config/i386/i386-builtin-types.def (BND): Likewise.
16322 * config/i386/i386-builtin.def (BDESC_END): Likewise.
16323 (BDESC_FIRST): Likewise.
16325 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
16326 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
16327 * config/i386/i386.c (enum reg_class): Likewise.
16328 (ix86_target_string): Likewise.
16329 (ix86_option_override_internal): Likewise.
16330 (ix86_conditional_register_usage): Likewise.
16331 (ix86_valid_target_attribute_inner_p): Likewise.
16332 (ix86_set_indirect_branch_type): Likewise.
16333 (ix86_set_current_function): Likewise.
16334 (ix86_function_arg_regno_p): Likewise.
16335 (init_cumulative_args): Likewise.
16336 (ix86_function_arg_advance): Likewise.
16337 (ix86_function_arg): Likewise.
16338 (ix86_pass_by_reference): Likewise.
16339 (ix86_function_value_regno_p): Likewise.
16340 (ix86_function_value_1): Likewise.
16341 (ix86_function_value_bounds): Likewise.
16342 (ix86_return_in_memory): Likewise.
16343 (ix86_setup_incoming_vararg_bounds): Likewise.
16344 (ix86_va_start): Likewise.
16345 (indirect_thunk_need_prefix): Likewise.
16346 (print_reg): Likewise.
16347 (ix86_print_operand): Likewise.
16348 (ix86_expand_call): Likewise.
16349 (ix86_output_function_return): Likewise.
16350 (reg_encoded_number): Likewise.
16351 (BDESC_VERIFYS): Likewise.
16352 (ix86_init_mpx_builtins): Likewise.
16353 (ix86_init_builtins): Likewise.
16354 (ix86_emit_cmove): Likewise.
16355 (ix86_emit_move_max): Likewise.
16356 (ix86_expand_builtin): Likewise.
16357 (ix86_builtin_mpx_function): Likewise.
16358 (ix86_get_arg_address_for_bt): Likewise.
16359 (ix86_load_bounds): Likewise.
16360 (ix86_store_bounds): Likewise.
16361 (ix86_load_returned_bounds): Likewise.
16362 (ix86_store_returned_bounds): Likewise.
16363 (ix86_class_likely_spilled_p): Likewise.
16364 (ix86_hard_regno_mode_ok): Likewise.
16365 (x86_order_regs_for_local_alloc): Likewise.
16366 (ix86_mitigate_rop): Likewise.
16367 (ix86_bnd_prefixed_insn_p): Likewise.
16368 (ix86_mpx_bound_mode): Likewise.
16369 (ix86_make_bounds_constant): Likewise.
16370 (ix86_initialize_bounds): Likewise.
16371 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
16372 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
16373 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
16374 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
16375 (TARGET_CHKP_BOUND_MODE): Likewise.
16376 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
16377 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
16378 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
16379 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
16380 * config/i386/i386.h (TARGET_MPX): Likewise.
16381 (TARGET_MPX_P): Likewise.
16382 (VALID_BND_REG_MODE): Likewise.
16383 (FIRST_BND_REG): Likewise.
16384 (LAST_BND_REG): Likewise.
16385 (enum reg_class): Likewise.
16386 (BND_REG_P): Likewise.
16387 (BND_REGNO_P): Likewise.
16388 (BNDmode): Likewise.
16389 (ADJUST_INSN_LENGTH): Likewise.
16390 * config/i386/i386.md: Likewise.
16391 * config/i386/i386.opt: Likewise.
16392 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
16393 (defined): Likewise.
16394 (LINK_MPX): Likewise.
16395 (MPX_SPEC): Likewise.
16396 (LIBMPX_SPEC): Likewise.
16397 (LIBMPXWRAPPERS_SPEC): Likewise.
16398 (CHKP_SPEC): Likewise.
16399 * config/i386/predicates.md: Likewise.
16400 * dbxout.c (dbxout_type): Likewise.
16401 * doc/extend.texi: Likewise.
16402 * doc/invoke.texi: Likewise.
16403 * doc/md.texi: Likewise.
16404 * doc/tm.texi: Likewise.
16405 * doc/tm.texi.in: Likewise.
16406 * dwarf2out.c (is_base_type): Likewise.
16407 (gen_formal_types_die): Likewise.
16408 (gen_subprogram_die): Likewise.
16409 (gen_type_die_with_usage): Likewise.
16410 (gen_decl_die): Likewise.
16411 (dwarf2out_late_global_decl): Likewise.
16412 * expr.c (expand_assignment): Likewise.
16413 (emit_storent_insn): Likewise.
16414 (store_expr_with_bounds): Likewise.
16415 (store_expr): Likewise.
16416 (expand_expr_real_1): Likewise.
16417 * expr.h (store_expr_with_bounds): Likewise.
16418 * function.c (use_register_for_decl): Likewise.
16419 (struct bounds_parm_data): Likewise.
16420 (assign_parms_augmented_arg_list): Likewise.
16421 (assign_parm_find_entry_rtl): Likewise.
16422 (assign_parm_is_stack_parm): Likewise.
16423 (assign_parm_load_bounds): Likewise.
16424 (assign_bounds): Likewise.
16425 (assign_parms): Likewise.
16426 (expand_function_start): Likewise.
16427 * gcc.c (CHKP_SPEC): Likewise.
16428 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16429 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
16430 (wrestrict_dom_walker::check_call): Likewise.
16431 * gimple.c (gimple_build_call_from_tree): Likewise.
16432 * gimple.h (enum gf_mask): Likewise.
16433 (gimple_call_with_bounds_p): Likewise.
16434 (gimple_call_set_with_bounds): Likewise.
16435 * gimplify.c (gimplify_init_constructor): Likewise.
16436 * ipa-cp.c (initialize_node_lattices): Likewise.
16437 (propagate_constants_across_call): Likewise.
16438 (find_more_scalar_values_for_callers_subset): Likewise.
16439 * ipa-hsa.c (process_hsa_functions): Likewise.
16440 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
16441 * ipa-icf.c (sem_function::merge): Likewise.
16442 * ipa-inline.c (early_inliner): Likewise.
16443 * ipa-pure-const.c (warn_function_noreturn): Likewise.
16444 (warn_function_cold): Likewise.
16445 (propagate_pure_const): Likewise.
16446 * ipa-ref.h (enum GTY): Likewise.
16447 * ipa-split.c (find_retbnd): Likewise.
16448 (consider_split): Likewise.
16449 (split_function): Likewise.
16450 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
16451 * ipa.c (walk_polymorphic_call_targets): Likewise.
16452 (symbol_table::remove_unreachable_nodes): Likewise.
16453 (process_references): Likewise.
16454 (cgraph_build_static_cdtor_1): Likewise.
16455 * lto-cgraph.c (lto_output_node): Likewise.
16456 (output_refs): Likewise.
16457 (compute_ltrans_boundary): Likewise.
16458 (input_overwrite_node): Likewise.
16459 (input_node): Likewise.
16460 (input_cgraph_1): Likewise.
16461 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
16462 * passes.c (pass_manager::execute_early_local_passes): Likewise.
16463 (class pass_chkp_instrumentation_passes): Likewise.
16464 (make_pass_chkp_instrumentation_passes): Likewise.
16465 * passes.def: Likewise.
16466 * rtl.h (struct GTY): Likewise.
16467 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
16468 * stor-layout.c (layout_type): Likewise.
16469 * symtab.c: Likewise.
16470 * target.def: Likewise.
16471 * targhooks.c (default_chkp_bound_type): Likewise.
16472 (default_chkp_bound_mode): Likewise.
16473 (default_builtin_chkp_function): Likewise.
16474 (default_chkp_function_value_bounds): Likewise.
16475 (default_chkp_make_bounds_constant): Likewise.
16476 (default_chkp_initialize_bounds): Likewise.
16477 * targhooks.h (default_chkp_bound_type): Likewise.
16478 (default_chkp_bound_mode): Likewise.
16479 (default_builtin_chkp_function): Likewise.
16480 (default_chkp_function_value_bounds): Likewise.
16481 (default_chkp_make_bounds_constant): Likewise.
16482 (default_chkp_initialize_bounds): Likewise.
16483 * toplev.c (compile_file): Likewise.
16484 (process_options): Likewise.
16485 * tree-core.h (DEF_BUILTIN): Likewise.
16486 (DEF_BUILTIN_CHKP): Likewise.
16487 * tree-inline.c (declare_return_variable): Likewise.
16488 (remap_gimple_stmt): Likewise.
16489 (copy_bb): Likewise.
16490 (initialize_inlined_parameters): Likewise.
16491 (expand_call_inline): Likewise.
16492 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
16493 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
16494 (make_pass_ipa_chkp_produce_thunks): Likewise.
16495 (make_pass_chkp): Likewise.
16496 (make_pass_chkp_opt): Likewise.
16497 (make_pass_chkp_instrumentation_passes): Likewise.
16498 * tree-pretty-print.c (dump_generic_node): Likewise.
16499 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
16500 * tree-ssa-dce.c (propagate_necessity): Likewise.
16501 (eliminate_unnecessary_stmts): Likewise.
16502 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
16503 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
16504 * tree-ssa-sccvn.h: Likewise.
16505 * tree-ssa-strlen.c (get_string_length): Likewise.
16506 (valid_builtin_call): Likewise.
16507 (adjust_last_stmt): Likewise.
16508 (handle_builtin_strchr): Likewise.
16509 (handle_builtin_strcpy): Likewise.
16510 (handle_builtin_stxncpy): Likewise.
16511 (handle_builtin_memcpy): Likewise.
16512 (handle_builtin_strcat): Likewise.
16513 (strlen_check_and_optimize_stmt): Likewise.
16514 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
16515 * tree-streamer-in.c: Likewise.
16516 * tree-streamer.c (record_common_node): Likewise.
16517 * tree.c (tree_code_size): Likewise.
16518 (wide_int_to_tree_1): Likewise.
16519 (type_contains_placeholder_1): Likewise.
16520 (build_common_tree_nodes): Likewise.
16521 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
16522 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
16523 (POINTER_BOUNDS_P): Likewise.
16524 (BOUNDED_TYPE_P): Likewise.
16525 (BOUNDED_P): Likewise.
16526 (CALL_WITH_BOUNDS_P): Likewise.
16527 (pointer_bounds_type_node): Likewise.
16528 * value-prof.c (gimple_ic): Likewise.
16529 * var-tracking.c (vt_add_function_parameters): Likewise.
16530 * varasm.c (make_decl_rtl): Likewise.
16531 (assemble_start_function): Likewise.
16532 (output_constant): Likewise.
16533 (maybe_assemble_visibility): Likewise.
16534 * varpool.c (ctor_for_folding): Likewise.
16535 * chkp-builtins.def: Remove.
16536 * ipa-chkp.c: Remove.
16537 * ipa-chkp.h: Remove.
16538 * rtl-chkp.c: Remove.
16539 * rtl-chkp.h: Remove.
16540 * tree-chkp-opt.c: Remove.
16541 * tree-chkp.c: Remove.
16542 * tree-chkp.h: Remove.
16544 2018-06-07 Carl Love <cel@us.ibm.com>
16546 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16547 vextract_fp_from_shortl): Add BE support.
16549 2018-06-07 Paul Koning <ni1d@arrl.net>
16551 * compare-elim.c (try_merge_compare): Don't merge compare if
16552 address contains a side effect.
16553 (try_eliminate_compare): Likewise.
16555 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
16557 * config.gcc: Support "tremont".
16558 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
16559 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16561 * config/i386/i386.c (m_TREMONT): Define.
16562 (processor_target_table): Add "tremont".
16563 (PTA_TREMONT): Define.
16564 (ix86_lea_outperforms): Add TARGET_TREMONT.
16565 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
16566 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
16567 and M_INTEL_GOLDMONT_PLUS.
16568 (fold_builtin_cpu): Add "tremont".
16569 (ix86_add_stmt_cost): Add TARGET_TREMONT.
16570 (ix86_option_override_internal): Add "tremont".
16571 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
16572 (processor_type): Add PROCESSOR_TREMONT.
16573 * config/i386/x86-tune.def: Add m_TREMONT.
16574 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
16576 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16578 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
16579 symbol defined for msp430i* devices to be lower case.
16581 2018-06-07 Richard Biener <rguenther@suse.de>
16583 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
16584 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
16585 Properly wrap signed arithmetic if overflow wraps.
16587 2018-06-07 Jakub Jelinek <jakub@redhat.com>
16589 PR tree-optimization/69615
16590 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
16591 of a cast from a same precision integral SSA_NAME in a bb dominated
16592 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
16593 cast to utype if rhs2 has already a compatible type.
16595 2018-06-07 Richard Biener <rguenther@suse.de>
16597 PR tree-optimization/85935
16598 * graphite-scop-detection.c (find_params_in_bb): Analyze
16599 condition operands with respect to the correct loop. Assert
16600 the analysis doesn't fail.
16602 2018-06-04 Carl Love <cel@us.ibm.com>
16604 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
16605 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
16606 as it is slightly cheaper.
16607 (first_match_or_eos_index_<mode>):
16608 Calculate index using natural element order.
16609 (first_match_index_<mode>):
16610 Calculate index using natural element order.
16611 (first_match_or_eos_index_<mode>):
16612 Calculate index using natural order.
16613 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
16614 for BE and LE modes.
16615 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
16616 P9V_BUILTIN_VCLZLSBB_V16QI.
16617 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
16620 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
16622 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16623 indentation and line wrap for many prototypes. Add missing
16624 @smallexample directives around block of prototypes for vec_xl and
16627 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
16629 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
16630 track if we pass or return IEEE 128-bit floating point.
16631 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
16632 C++ mangling that is compatible with GCC 8.1.
16633 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
16634 (init_cumulative_args): Note if we pass or return IEEE 128-bit
16635 floating point types.
16636 (rs6000_function_arg_advance_1): Likewise.
16637 (rs6000_mangle_type): Optionally generate mangled names that match
16638 what GCC 8.1 generated for IEEE 128-bit floating point types.
16639 (rs6000_globalize_decl_name): If we have an external function that
16640 passes or returns IEEE 128-bit types, generate a weak reference
16641 from the mangled name used in GCC 8.1 to the current mangled
16643 (rs6000_init_builtins): Make __ibm128 use the long double type if
16644 long double is IBM extended double. Make __float128 use the long
16645 double type if long double is IEEE 128-bit.
16648 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
16649 macro for __ibm128 built-in functions.
16650 (PACK_IF): Add __ibm128 pack/unpack functions.
16651 (UNPACK_IF): Likewise.
16652 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
16653 enable long double built-in functions if long double is IEEE
16654 128-bit floating point.
16655 (rs6000_invalid_builtin): Update long double built-in function
16657 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
16658 functions, adjust the built-in function to use the long double
16659 built-in function if __ibm128 and long double are the same type.
16660 * doc/extend.texi (PowerPC builtins): Update documention for
16661 __builtin_{,un}pack_longdouble. Add documentation for
16662 __builtin_{,un}pack_ibm128.
16664 2018-06-06 Jim Wilson <jimw@sifive.com>
16666 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
16667 (struct machine_function): New field interrupt_mode.
16668 (riscv_handle_type_attribute): New function. Add forward declaration.
16669 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
16670 (riscv_expand_epilogue): Check interrupt_mode field.
16671 (riscv_set_current_function): Check interrupt attribute args and
16672 set interrupt_mode field.
16673 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
16674 (riscv_sret, riscv_uret): New.
16675 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
16676 new arguments to interrupt attribute.
16678 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
16681 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
16682 Don't handle -mcpu=power8 if -mpower9-vector is also used.
16684 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
16686 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
16687 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
16688 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
16689 several redundant entries.
16691 2018-06-06 David Malcolm <dmalcolm@redhat.com>
16693 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
16694 type from "rtx" to "rtx_insn *".
16695 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
16696 for local "call_insn", removing cast.
16697 (ix86_expand_call): Likewise, introducing a "call_insn" local.
16699 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
16701 PR tree-optimization/86066
16702 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
16703 for BIT_INSERT_EXPR stores.
16705 2018-06-06 Richard Biener <rguenther@suse.de>
16707 PR tree-optimization/86062
16708 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
16709 component refs ontop
16710 of to be offsetted base.
16712 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16714 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
16715 to be static and remove check on interrupt attribute name.
16717 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
16719 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
16720 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
16722 2018-06-05 Steve Ellcey <sellcey@cavium.com>
16725 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
16727 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
16728 Remove second argument, change how error is called.
16729 (aarch64_layout_arg): Remove second argument from
16730 aarch64_err_no_fpadvsimd call.
16731 (aarch64_init_cumulative_args): Ditto.
16732 (aarch64_gimplify_va_arg_expr): Ditto.
16733 * config/aarch64/aarch64.md (mov<mode>): Ditto.
16735 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
16737 * config/i386/i386.md (simple_return_indirect_internal): New expander.
16738 (*simple_return_indirect_internal<mode>): Rename from
16739 simple_return_indirect_internal. Use W mode iterator.
16740 (rstorssp): New expander.
16741 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
16742 (clrssbsy): New expander.
16743 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
16745 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16747 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
16749 (cmse_check_pointed_object): Likewise.
16751 2018-06-05 Martin Liska <mliska@suse.cz>
16753 PR gcov-profile/47618
16754 * doc/invoke.texi: Document how -fprofile-dir format
16757 2018-06-05 Richard Biener <rguenther@suse.de>
16759 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
16760 removal pretend DOM info isn't available so we do not update
16761 it and only remove edges, not dominated blocks. Actually free
16762 DOM info in case we removed something. Remove unreachable blocks.
16763 (mfb_keep_latches): Work with either DOM info or marked backedges.
16764 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
16765 first. Mark backedges if DOM info isn't available.
16766 (Re-)compute DOM info after cleanup_control_flow_pre.
16768 2018-06-05 Richard Biener <rguenther@suse.de>
16770 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
16771 (locus_discrim_hasher::hash): Adjust.
16772 (locus_discrim_hasher::equal): Likewise.
16773 (next_discriminator_for_locus): Work on line directly.
16774 (same_line_p): Pass in expanded locus1 as well.
16775 (assign_discriminators): Avoid redundant location expansions.
16777 2018-06-05 Richard Biener <rguenther@suse.de>
16779 PR tree-optimization/86046
16780 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
16781 if required after clearing TREE_ADDRESSABLE.
16783 2018-06-05 Richard Biener <rguenther@suse.de>
16785 PR tree-optimization/86047
16786 * tree-ssa-loop.c (for_each_index): Glob handling of all
16787 decls and constants and really handle all of them.
16789 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16792 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
16793 qualifier_void_pointer and qualifier_const_void_pointer.
16794 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
16795 (arm_init_builtins): Handle the above.
16796 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
16797 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
16798 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
16801 2018-06-05 Martin Liska <mliska@suse.cz>
16803 * auto-profile.c (read_autofdo_file): Do not use
16804 gcov_ctr_summary struct.
16805 (afdo_callsite_hot_enough_for_early_inline): Likewise.
16806 * coverage.c (struct counts_entry): Likewise.
16807 (read_counts_file): Read just single summary entry.
16808 (get_coverage_counts): Use gcov_summary struct.
16809 * coverage.h (get_coverage_counts): Likewise.
16810 * gcov-dump.c (dump_working_sets): Likewise.
16811 (tag_summary): Dump just single summary.
16812 * gcov-io.c (gcov_write_summary): Write just histogram
16814 (gcov_read_summary): Read just single summary.
16815 (compute_working_sets): Use gcov_summary struct.
16816 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
16817 of GCOV_COUNTERS_SUMMABLE.
16818 (GCOV_COUNTERS_SUMMABLE): Remove.
16819 (GCOV_FIRST_VALUE_COUNTER): Replace with
16820 GCOV_COUNTER_V_INTERVAL.
16821 (struct gcov_ctr_summary): Remove.
16822 (struct gcov_summary): Directly use fields of former
16824 (compute_working_sets): Use gcov_summary struct.
16825 * gcov.c (read_count_file): Do not use ctrs fields.
16826 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
16828 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
16830 * profile.c: Likewise.
16831 * profile.h: Likewise.
16833 2018-06-05 Martin Liska <mliska@suse.cz>
16835 PR gcov-profile/84846
16836 * gcov.c (output_lines): Print working directory only
16837 in intermediate format.
16839 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
16841 * config/s390/s390-builtin-types.def: Add void function type.
16842 * config/s390/s390-builtins.def: Use the function type for the
16845 2018-06-04 Jim Wilson <jimw@sifive.com>
16847 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
16849 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
16850 and maybe_eh_return. Change regno to unsigned int. Use new args to
16851 handle EH_RETURN_DATA_REGNO registers properly.
16852 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
16853 (riscv_expand_epilogue): Update comment. Change argument name and
16854 type. Update code to use new name and type. Pass new args to
16855 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
16857 * config/riscv/riscv.md (NORMAL_RETURN): New.
16858 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
16859 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
16860 (eh_return): Call gen_eh_return_internal and emit barrier.
16861 (eh_return_internal): Call riscv_expand_epilogue.
16863 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
16865 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
16866 bit_insertion field and declare can_be_merged_into method.
16867 (merged_store_group::can_be_merged_into): New method.
16868 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
16869 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
16870 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
16872 2018-06-04 Richard Biener <rguenther@suse.de>
16874 PR tree-optimization/85955
16875 * builtins.c (fold_builtin_sincos): Convert pointers to
16876 destination to appropriate type before dereferencing.
16878 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
16880 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
16882 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
16884 * expr.c (expand_expr_real_1): Force the operand into memory if
16885 its TYPE_MODE is BLKmode and if there is no integer mode for
16886 the number of bits being extracted.
16888 2018-06-04 Jakub Jelinek <jakub@redhat.com>
16892 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
16893 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
16895 2018-06-04 Richard Biener <rguenther@suse.de>
16897 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
16898 (cleanup_tree_cfg_noloop): ... single caller. Do
16899 start_recording_case_labels later.
16901 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
16903 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
16904 to _IMMINTRIN_H_INCLUDED.
16905 * config/i386/pconfigintrin.h: Ditto.
16906 * config/i386/waitpkgintrin.h: Ditto.
16907 * config/i386/immintrin.h: Add includes for sgxintrin.h,
16908 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
16909 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
16910 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
16911 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
16912 waitpkgintrin.h and cldemoteintrin.h.
16914 2018-06-04 Richard Biener <rguenther@suse.de>
16916 PR tree-optimization/86038
16917 * tracer.c (find_best_successor): Check probability for
16918 being initialized, bail out if not.
16920 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
16923 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
16924 of bits to ignore when comparing architectures.
16926 2018-06-04 Jakub Jelinek <jakub@redhat.com>
16928 PR tree-optimization/69615
16929 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
16930 maximum or minimum of the type, try to merge it also as if
16931 range1 is + [-, x - 1] or + [x + 1, -].
16934 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
16936 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
16938 PR tree-optimization/86034
16939 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
16940 the unsigned bitfield type in a bit insertion sequence if it does not
16941 have a larger precision than the bitfield size.
16942 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
16944 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
16946 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
16948 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
16950 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
16951 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
16952 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
16953 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
16955 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
16957 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
16958 Disable -fdelete-null-pointer-checks for ELF toolchain.
16960 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
16961 Kito Cheng <kito.cheng@gmail.com>
16963 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
16964 (nds32le-*-*, nds32be-*-*): Integrate checking process.
16965 (nds32*-*-*): Add glibc and uclibc conditions.
16966 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
16967 (TARGET_EXCEPT_UNWIND_INFO): Define.
16968 * config/nds32/elf.h: New file.
16969 * config/nds32/linux.h: New file.
16970 * config/nds32/nds32-elf.opt: New file.
16971 * config/nds32/nds32-linux.opt: New file.
16972 * config/nds32/nds32-fp-as-gp.c
16973 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
16974 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
16976 (nds32_asm_file_end): Ditto.
16977 (nds32_print_operand): Ditto.
16978 (nds32_insert_attributes): Ditto.
16979 (nds32_init_libfuncs): New function.
16980 (TARGET_HAVE_TLS): Define.
16981 (TARGET_INIT_LIBFUNCS): Define.
16982 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
16984 (TARGET_ELF): Apply different mcmodel setting.
16985 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
16986 been migrated into elf.h and linux.h files.
16987 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
16988 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
16989 (mcmodel): The content has been migrated into nds32-elf.opt and
16990 nds32-linux.opt files.
16991 * config/nds32/t-elf: New file.
16992 * config/nds32/t-linux: New file.
16994 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
16995 Shiva Chen <shiva0217@gmail.com>
16997 * config/nds32/constants.md (unspec_volatile_element): Add
16998 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
16999 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
17001 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
17002 (make_pass_nds32_fp_as_gp): Declare.
17003 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
17005 (nds32_asm_function_end_prologue): Remove unused asm output.
17006 (nds32_asm_function_begin_epilogue): Remove unused asm output.
17007 (nds32_asm_file_start): Output necessary fp_as_gp information.
17008 (nds32_option_override): Adjust register usage.
17009 (nds32_expand_prologue): Consider fp_as_gp situation.
17010 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
17011 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
17014 (simple_return): Ditto.
17015 (omit_fp_begin): Output special directive for fp_as_gp.
17016 (omit_fp_end): Output special directive for fp_as_gp.
17017 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
17018 mforbid-fp-as-gp): New options.
17020 2018-06-01 Mark Wielaard <mark@klomp.org>
17022 * dwarf2out.c (dwarf2out_finish): Remove generation of
17023 DW_AT_loclists_base.
17025 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
17027 * gimple-ssa-store-merging.c: Include gimple-fold.h.
17028 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
17029 (struct merged_store_group): Add bit_insertion field.
17030 (dump_char_array): Use standard hexadecimal format.
17031 (merged_store_group::merged_store_group): Set bit_insertion to false.
17032 (merged_store_group::apply_stores): Use optimal buffer size. Deal
17033 with BIT_INSERT_EXPR stores. Move up code updating the mask and
17034 also print the mask in the dump file.
17035 (pass_store_merging::gate): Minor tweak.
17036 (imm_store_chain_info::coalesce_immediate): Fix wrong association
17037 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
17038 stores with INTEGER_CST stores.
17039 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
17040 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
17041 and use it throughout. Generate bit insertion sequences if need be.
17042 (pass_store_merging::process_store): Remove redundant condition.
17043 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
17045 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
17047 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
17048 the 128-bit floating point types. Fix function comment.
17050 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17052 * config/aarch64/aarch64-simd.md
17053 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
17055 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
17058 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
17060 PR tree-optimization/85989
17061 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
17063 (backprop::intersect_uses): Check it when deciding whether this
17064 is a backedge reference.
17065 (backprop::process_block): Add each phi to m_visited_phis
17066 after visiting it, then clear it at the end.
17068 2018-06-01 Richard Biener <rguenther@suse.de>
17070 * tree-vectorizer.h (vect_dr_stmt): New function.
17071 (vect_get_load_cost): Adjust.
17072 (vect_get_store_cost): Likewise.
17073 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
17074 Use vect_dr_stmt instead of DR_SMTT.
17075 (vect_record_base_alignments): Likewise.
17076 (vect_calculate_target_alignment): Likewise.
17077 (vect_compute_data_ref_alignment): Likewise and make static.
17078 (vect_update_misalignment_for_peel): Likewise.
17079 (vect_verify_datarefs_alignment): Likewise.
17080 (vector_alignment_reachable_p): Likewise.
17081 (vect_get_data_access_cost): Likewise. Pass down
17082 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
17083 (vect_get_peeling_costs_all_drs): Likewise.
17084 (vect_peeling_hash_get_lowest_cost): Likewise.
17085 (vect_enhance_data_refs_alignment): Likewise.
17086 (vect_find_same_alignment_drs): Likewise.
17087 (vect_analyze_data_refs_alignment): Likewise.
17088 (vect_analyze_group_access_1): Likewise.
17089 (vect_analyze_group_access): Likewise.
17090 (vect_analyze_data_ref_access): Likewise.
17091 (vect_analyze_data_ref_accesses): Likewise.
17092 (vect_vfa_segment_size): Likewise.
17093 (vect_small_gap_p): Likewise.
17094 (vectorizable_with_step_bound_p): Likewise.
17095 (vect_prune_runtime_alias_test_list): Likewise.
17096 (vect_analyze_data_refs): Likewise.
17097 (vect_supportable_dr_alignment): Likewise.
17098 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
17099 (vect_gen_prolog_loop_niters): Likewise.
17100 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17101 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
17103 (vect_recog_mask_conversion_pattern): Likewise.
17104 (vect_try_gather_scatter_pattern): Likewise.
17105 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
17106 to vect_get_store_cost.
17107 (vect_get_store_cost): Get stmt_info instead of DR.
17108 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
17109 (vect_get_load_cost): Get stmt_info instead of DR.
17111 2018-06-01 Richard Biener <rguenther@suse.de>
17113 PR middle-end/86017
17114 * gimple-fold.c (var_decl_component_p): Also allow offsetted
17115 vars wrapped in MEM_REFs.
17117 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
17119 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17120 Fix subreg tests so that we only return a choice between
17121 GENERAL_REGS and FP_REGS if the original classes included both.
17123 2018-06-01 Richard Biener <rguenther@suse.de>
17126 * tree-ssa-structalias.c (get_function_part_constraint):
17127 Handle NULL fi->decl.
17128 (find_func_aliases_for_call): Properly handle indirect
17129 fi from direct call.
17130 (find_func_clobbers): Likewise.
17131 (ipa_pta_execute): Likewise.
17132 (create_variable_info_for): For functions that are ifunc_resolver
17133 resolve to a varinfo that contains the result of the resolver call.
17134 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
17137 2018-05-31 Michael Collison <michael.collison@arm.com>
17139 * config/aarch64/aarch64.md:
17140 (*fix_to_zero_extenddfdi2): New pattern.
17141 * gcc.target/aarch64/fix_extend1.c: New testcase.
17143 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
17145 PR middle-end/78809
17146 PR middle-end/83026
17147 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
17148 and BUILT_IN_STRNCMP_EQ.
17149 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
17150 BUILT_IN_STRNCMP_EQ.
17151 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
17152 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
17153 (gimple_fold_builtin): Likewise.
17154 * tree-ssa-strlen.c (compute_string_length): New function.
17155 (determine_min_obsize): New function.
17156 (handle_builtin_string_cmp): New function to handle calls to
17157 string compare functions.
17158 (strlen_optimize_stmt): Add handling to builtin string compare
17160 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17161 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
17162 * tree.c (build_common_builtin_nodes): Add new defines of
17163 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
17165 2018-05-31 Jakub Jelinek <jakub@redhat.com>
17168 * bb-reorder.c (pass_partition_blocks::gate): Return false for
17169 functions with naked attribute.
17171 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
17173 * config/i386/sse.md (avx_vec_concat<mode>):
17174 Substitute concat_tg_mode mode attribute with xtg_mode.
17175 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
17176 (concat_tg_mode): Remove mode attribute.
17178 2018-05-31 Martin Sebor <msebor@redhat.com>
17181 * calls.c (alloc_max_size): Correct a logic error/typo.
17182 Treat excessive arguments as infinite. Warn for invalid arguments.
17183 * doc/invoke.texi (-Walloc-size-larger-than): Update.
17185 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
17188 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
17189 and movx for Haswell.
17191 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
17192 Cesar Philippidis <cesar@codesourcery.com>
17194 PR middle-end/85879
17195 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
17196 when emitting error on private/firstprivate reductions.
17197 * omp-low.c (lower_omp_target): Avoid reference-type processing
17198 on pointers for firstprivate clause.
17200 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
17202 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
17205 * config/aarch64/aarch64-simd.md
17206 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
17207 (aarch64_ld1_x3_<mode>): Likewise
17208 (aarch64_st1x2<VALLDIF:mode>): Likewise
17209 (aarch64_st1_x2_<mode>): Likewise
17210 (aarch64_st1x3<VALLDIF:mode>): Likewise
17211 (aarch64_st1_x3_<mode>): Likewise
17212 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
17213 (vld1_s8_x3): Likewise.
17214 (vld1_u16_x3): Likewise.
17215 (vld1_s16_x3): Likewise.
17216 (vld1_u32_x3): Likewise.
17217 (vld1_s32_x3): Likewise.
17218 (vld1_u64_x3): Likewise.
17219 (vld1_s64_x3): Likewise.
17220 (vld1_f16_x3): Likewise.
17221 (vld1_f32_x3): Likewise.
17222 (vld1_f64_x3): Likewise.
17223 (vld1_p8_x3): Likewise.
17224 (vld1_p16_x3): Likewise.
17225 (vld1_p64_x3): Likewise.
17226 (vld1q_u8_x3): Likewise.
17227 (vld1q_s8_x3): Likewise.
17228 (vld1q_u16_x3): Likewise.
17229 (vld1q_s16_x3): Likewise.
17230 (vld1q_u32_x3): Likewise.
17231 (vld1q_s32_x3): Likewise.
17232 (vld1q_u64_x3): Likewise.
17233 (vld1q_s64_x3): Likewise.
17234 (vld1q_f16_x3): Likewise.
17235 (vld1q_f32_x3): Likewise.
17236 (vld1q_f64_x3): Likewise.
17237 (vld1q_p8_x3): Likewise.
17238 (vld1q_p16_x3): Likewise.
17239 (vld1q_p64_x3): Likewise.
17240 (vst1_s64_x2): Likewise.
17241 (vst1_u64_x2): Likewise.
17242 (vst1_f64_x2): Likewise.
17243 (vst1_s8_x2): Likewise.
17244 (vst1_p8_x2): Likewise.
17245 (vst1_s16_x2): Likewise.
17246 (vst1_p16_x2): Likewise.
17247 (vst1_s32_x2): Likewise.
17248 (vst1_u8_x2): Likewise.
17249 (vst1_u16_x2): Likewise.
17250 (vst1_u32_x2): Likewise.
17251 (vst1_f16_x2): Likewise.
17252 (vst1_f32_x2): Likewise.
17253 (vst1_p64_x2): Likewise.
17254 (vst1q_s8_x2): Likewise.
17255 (vst1q_p8_x2): Likewise.
17256 (vst1q_s16_x2): Likewise.
17257 (vst1q_p16_x2): Likewise.
17258 (vst1q_s32_x2): Likewise.
17259 (vst1q_s64_x2): Likewise.
17260 (vst1q_u8_x2): Likewise.
17261 (vst1q_u16_x2): Likewise.
17262 (vst1q_u32_x2): Likewise.
17263 (vst1q_u64_x2): Likewise.
17264 (vst1q_f16_x2): Likewise.
17265 (vst1q_f32_x2): Likewise.
17266 (vst1q_f64_x2): Likewise.
17267 (vst1q_p64_x2): Likewise.
17268 (vst1_s64_x3): Likewise.
17269 (vst1_u64_x3): Likewise.
17270 (vst1_f64_x3): Likewise.
17271 (vst1_s8_x3): Likewise.
17272 (vst1_p8_x3): Likewise.
17273 (vst1_s16_x3): Likewise.
17274 (vst1_p16_x3): Likewise.
17275 (vst1_s32_x3): Likewise.
17276 (vst1_u8_x3): Likewise.
17277 (vst1_u16_x3): Likewise.
17278 (vst1_u32_x3): Likewise.
17279 (vst1_f16_x3): Likewise.
17280 (vst1_f32_x3): Likewise.
17281 (vst1_p64_x3): Likewise.
17282 (vst1q_s8_x3): Likewise.
17283 (vst1q_p8_x3): Likewise.
17284 (vst1q_s16_x3): Likewise.
17285 (vst1q_p16_x3): Likewise.
17286 (vst1q_s32_x3): Likewise.
17287 (vst1q_s64_x3): Likewise.
17288 (vst1q_u8_x3): Likewise.
17289 (vst1q_u16_x3): Likewise.
17290 (vst1q_u32_x3): Likewise.
17291 (vst1q_u64_x3): Likewise.
17292 (vst1q_f16_x3): Likewise.
17293 (vst1q_f32_x3): Likewise.
17294 (vst1q_f64_x3): Likewise.
17295 (vst1q_p64_x3): Likewise.
17297 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17299 * config/msp430/msp430.c (msp430_output_labelref): Prepend
17300 user_label_prefix to name.
17302 * tree-core.h: Update comment about the format of NAME string
17303 passed to handler in attribute_spec.
17305 * config/msp430/msp430.md: Remove erroneous subreg expression from
17306 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
17307 zero_extend{q,h}isi2.
17309 2018-05-30 Borislav Petkov <bp@suse.de>
17311 * doc/extend.texi: Document some architecture specific
17312 constraints and sort entries.
17314 2018-05-30 Martin Sebor <msebor@redhat.com>
17316 PR middle-end/85369
17317 * builtins.c (expand_builtin_stpcpy_1): New function.
17318 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
17319 only if the former succeeds.
17321 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
17323 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
17326 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17328 * doc/invoke.texi (-flinker-output): Document
17330 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17332 * passes.c (ipa_write_summaries): Only modify statements if body
17334 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
17335 incrementally linking.
17336 (ipa_passes): Likewise.
17337 * lto-cgraph.c (lto_output_node): When incrementally linking do not
17338 pass down resolution info.
17339 * common.opt (flag_incremental_link): Update info.
17340 * gcc.c (plugin specs): Turn flinker-output=* to
17341 -plugin-opt=-linker-output-known
17342 * toplev.c (compile_file): Also cut compilation when doing incremental
17344 * flag-types. (enum lto_partition_model): Add
17345 LTO_LINKER_OUTPUT_NOLTOREL.
17346 (invoke.texi): Add -flinker-output docs.
17347 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
17348 link same way as WPA; do not stream in dead initializers.
17350 * dwarf2out.c (dwarf2out_die_ref_for_decl,
17351 darf2out_register_external_decl): Support incremental link.
17353 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17355 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
17357 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17359 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
17360 it down to simple_object_copy_lto_debug_sections.
17361 (run_gcc): Determine incremental LTO link time and configure
17362 lto1 into non-wpa mode, disable renaming of debug sections.
17364 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
17366 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
17367 descriptions of various incorrectly documented functions.
17369 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
17372 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
17373 address check not strict.
17375 2018-05-30 Richard Biener <rguenther@suse.de>
17377 PR tree-optimization/85964
17378 * tracer.c (better_p): Drop initialized count check, we only
17379 call the function with initialized counts now.
17380 (find_best_successor): Do find a best edge if one
17381 has uninitialized count.
17382 (find_best_predecessor): Likewise. Do BB frequency check only
17383 if count is initialized.
17385 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
17387 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
17388 (aarch64_ldrstr_offset_compare): New.
17389 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
17390 load/store orderings.
17391 (aarch64_gen_adjusted_ldpstp): Likewise.
17393 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17395 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17396 Check for subset of GENERAL_REGS and FP_REGS.
17397 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
17400 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
17402 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
17403 and wi::to_poly_offset. Add the current offset and then check
17404 whether the sum fits, rather than using an unchecked addition of
17405 a checked term. Check for a shwi rather than a uhwi.
17406 * expr.c (get_bit_range): Use tree_to_poly_uint64.
17407 (store_constructor): Use poly_int_tree_p.
17408 (expand_expr_real_1): Likewise.
17409 * function.c (assign_temp): Likewise.
17410 * fold-const.c (const_binop): Use poly_int_tree_p and
17411 wi::to_poly_offset.
17412 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
17414 * ipa-icf-gimple.c (func_checker::compare_operand): Use
17415 to_poly_offset for MEM offsets.
17416 * ipa-icf.c (sem_variable::equals): Likewise.
17417 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
17418 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
17419 wi::to_poly_offset for BIT_FIELD_REF offsets.
17420 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
17421 wi::to_poly_offset.
17422 * var-tracking.c (emit_note_insn_var_location): Use
17423 tree_to_poly_uint64.
17425 2018-05-29 Jim Wilson <jimw@sifive.com>
17427 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
17429 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
17432 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
17433 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
17435 (sse4_1_round<mode>2): Use nonimmediate_operand
17436 for operand 1 predicate.
17438 2018-05-29 Martin Sebor <msebor@redhat.com>
17439 Richard Biener <rguenther@suse.de>
17442 * calls.c (get_size_range): Call determine_value_range instead
17443 of get_value_range..
17444 * tree-vrp.h (determine_value_range): Declared new function.
17445 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
17447 2018-05-29 Richard Biener <rguenther@suse.de>
17449 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
17450 sure to use non-pattern stmts for get_earlier_stmt arguments.
17451 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
17452 called on pattern stmts.
17453 (get_later_stmt): Likewise.
17455 2018-05-29 Martin Liska <mliska@suse.cz>
17457 PR gcov-profile/85759
17458 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
17461 2018-05-29 Jakub Jelinek <jakub@redhat.com>
17463 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
17465 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
17469 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
17470 VEC_PACK_FLOAT_EXPR): New tree codes.
17471 * tree-pretty-print.c (op_code_prio): Handle
17472 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
17473 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
17474 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
17475 * tree-inline.c (estimate_operator_cost): Likewise.
17476 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
17477 * fold-const.c (const_binop): Likewise.
17478 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
17479 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
17480 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17481 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
17482 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
17483 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
17484 * expr.c (expand_expr_real_2): Likewise.
17485 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
17486 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
17487 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
17489 * optabs.c (expand_widen_pattern_expr): For
17490 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
17491 sign from result type rather than operand's type.
17492 (expand_binop_directly): For vec_packu_float_optab and
17493 vec_packs_float_optab allow result type to be different from operand's
17495 * optabs-tree.c (optab_for_tree_code): Handle
17496 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
17497 VEC_PACK_FLOAT_EXPR. Formatting fixes.
17498 * tree-vect-generic.c (expand_vector_operations_1): Handle
17499 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
17500 VEC_PACK_FLOAT_EXPR.
17501 * tree-vect-stmts.c (supportable_widening_operation): Handle
17503 (supportable_narrowing_operation): Handle FLOAT_EXPR.
17504 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
17505 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
17506 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
17507 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
17509 (vec_pack<floatprefix>_float_<mode>): New expander.
17510 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
17512 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
17513 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
17514 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
17515 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
17516 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
17518 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
17519 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
17520 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
17521 VEC_PACK_FLOAT_EXPR): Document.
17523 2018-05-29 Richard Biener <rguenther@suse.de>
17525 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
17527 (stmt_vec_info_vec): Make pointer.
17528 (init_stmt_vec_info_vec): Remove.
17529 (free_stmt_vec_info_vec): Likewise.
17530 (set_stmt_vec_info_vec): New function.
17531 (free_stmt_vec_infos): Likewise.
17532 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
17533 (set_vinfo_for_stmt): Likewise.
17534 (get_earlier_stmt): Likewise.
17535 (get_later_stmt): Likewise.
17536 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
17537 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
17538 (vec_info::~vec_info): Free stmt_vec_infos.
17539 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
17540 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
17541 (pass_slp_vectorize::execute): Likewise.
17542 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
17543 (free_stmt_vec_info_vec): Likewise.
17544 (set_stmt_vec_info_vec): New function.
17545 (free_stmt_vec_infos): Likewise.
17546 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
17547 the global stmt_vec_info_vec.
17548 * tree-parloops.c (gather_scalar_reductions): Use
17549 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
17552 2018-05-29 Richard Biener <rguenther@suse.de>
17554 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
17556 2018-05-29 Martin Liska <mliska@suse.cz>
17557 David Malcolm <dmalcolm@redhat.com>
17559 * vec.c (test_reverse): New.
17560 (vec_c_tests): Add new test.
17561 * vec.h (vl_ptr>::reverse): New function.
17563 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
17565 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
17567 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
17570 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17572 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
17574 2018-05-28 Richard Biener <rguenther@suse.de>
17576 PR tree-optimization/85933
17577 * tree-vect-data-refs.c (vect_record_base_alignments): Only
17578 look at stmts marked as vectorizable.
17580 2018-05-28 Richard Biener <rguenther@suse.de>
17582 PR tree-optimization/85934
17583 * tree-vect-generic.c (expand_vector_operations_1): Hoist
17584 vector boolean check before scalar optimization.
17586 2018-05-28 Jakub Jelinek <jakub@redhat.com>
17588 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
17591 2018-05-28 Mark Wielaard <mark@klomp.org>
17593 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
17594 if it is an expression containing a minus sign.
17596 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
17598 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
17600 2018-05-27 Paul Koning <ni1d@arrl.net>
17602 * config/pdp11/pdp11.md (truncsihi2): Remove.
17604 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
17605 Chung-Ju Wu <jasonwucj@gmail.com>
17607 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
17609 (unaligned_store_dw): Ditto.
17610 * config/nds32/nds32-memory-manipulation.c
17611 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
17612 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
17613 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
17614 (emit_setmem_word_loop): Rename to ...
17615 (emit_setmem_doubleword_loop): ... this.
17616 (nds32_gen_dup_4_byte_to_word_value): New function.
17617 (nds32_gen_dup_8_byte_to_double_word_value): New function.
17618 (nds32_expand_setmem_loop): Refine implementation.
17619 (nds32_expand_setmem_loop_v3m): Ditto.
17620 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
17623 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
17625 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
17627 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
17629 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
17630 (nds32_init_machine_status): Initialize machine->attr_naked_p and
17631 machine->attr_no_prologue_p.
17632 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
17633 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
17634 (nds32_expand_epilogue): Consider attr_naked_p.
17635 (nds32_expand_epilogue_v3pop): Likewise.
17636 (nds32_can_use_return_insn): Likewise.
17637 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
17638 attr_no_prologue_p fields.
17639 * config/nds32/nds32.opt (mret-in-naked-func): New option.
17641 2018-05-27 Jakub Jelinek <jakub@redhat.com>
17644 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
17646 * config/i386/sse.md
17647 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
17649 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
17651 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
17653 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
17655 (*<floatsuffix>floatv2div2sf2): Rename to ...
17656 (*float<floatunssuffix>v2div2sf2): ... this.
17657 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
17658 (float<floatunssuffix>v2div2sf2_mask): ... this.
17659 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
17660 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
17661 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
17663 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
17665 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
17667 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
17669 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
17671 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
17673 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
17674 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
17675 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
17676 gen_ufix_truncv8dfv8si2.
17677 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
17678 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
17679 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
17680 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
17681 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
17682 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
17683 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
17684 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
17686 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
17690 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
17692 2018-05-25 Jim Wilson <jimw@sifive.com>
17694 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
17695 * config/riscv/riscv.c (struct machine_function): Add
17696 interrupt_handler_p and attribute_checked_p fields.
17697 (riscv_attribute_table): Add interrupt.
17698 (riscv_interrupt_type_p): New.
17699 (riscv_save_reg_p): Save extra regs for interrupt handler.
17700 (riscv_use_save_libcall): Return false for interrupt handler.
17701 (riscv_first_stack_step): Add forward declaration.
17702 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
17703 for interrupt handler with large frame. Use it for saved reg list.
17704 (riscv_expand_prologue): Move flag_stack_usage_info support to
17705 eliminate duplication.
17706 (riscv_expand_epilogue): Generate mret for interrupt handler.
17707 (riscv_epilogue_uses): New.
17708 (riscv_can_use_return_insn): Return false for interrupt handler.
17709 (riscv_function_ok_for_sibcall): Likewise.
17710 (riscv_set_current_function): Add interrupt handler support.
17711 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
17712 * config/riscv/riscv.md (UNSPECV_MRET): New.
17714 (riscv_frflags, riscv_fsflags): Use tab after opcode.
17716 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
17718 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
17720 PR tree-optimization/85712
17721 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
17722 this candidate has already been replaced in-situ by a copy.
17724 2018-05-25 Jason Merrill <jason@redhat.com>
17726 PR c++/80485 - inline function non-zero address.
17727 * symtab.c (nonzero_address): Check DECL_COMDAT.
17729 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
17732 * config/alpha/alpha.md (ashlsi3): New insn pattern.
17733 (*ashlsi_se): Rename from *ashldi_se. Define as sign
17734 extension of SImode operation. Use const123_operand predicate.
17735 (*saddsi_1): Remove.
17736 (*saddl_se_1): Ditto.
17737 (*ssubsi_1): Ditto.
17738 (*ssubl_se_1): Ditto.
17739 * config/alpha/predicates.md (const123_operand): New predicate.
17740 * config/alpha/constraints.md (P): Use IN_RANGE.
17742 2018-05-25 Richard Biener <rguenther@suse.de>
17744 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
17746 (ref_maybe_used_by_stmt_p): Likewise.
17747 (stmt_may_clobber_ref_p): Likewise.
17748 (stmt_may_clobber_ref_p_1): Likewise.
17749 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
17751 (ref_maybe_used_by_stmt_p): Likewise.
17752 (stmt_may_clobber_ref_p): Likewise.
17753 (stmt_may_clobber_ref_p_1): Likewise.
17754 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
17755 the alias oracle to disambiguate DRs with stmts DR analysis
17757 (vect_analyze_data_refs): Do not give up on not analyzable
17758 DRs for BB vectorization. Remove code truncating the dataref
17761 2018-05-25 Jakub Jelinek <jakub@redhat.com>
17764 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
17765 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
17766 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
17768 2018-05-25 Richard Biener <rguenther@suse.de>
17770 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
17771 function, combining stmt data ref gathering and fatal analysis
17773 (vect_analyze_data_refs): Remove now redudnant code and simplify.
17774 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
17775 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
17776 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
17777 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
17779 2018-05-25 Bin Cheng <bin.cheng@arm.com>
17781 PR tree-optimization/85720
17782 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
17783 SCC if all partitions are builtins.
17784 (version_loop_by_alias_check): New parameter. Generate cancelable
17785 runtime alias check if all partitions are builtins.
17786 (distribute_loop): Update call to above function.
17788 2018-05-25 Bin Cheng <bin.cheng@arm.com>
17790 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
17791 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
17792 (parm_default_def_partition_arg): Ditto.
17793 (set_parm_default_def_partition): Ditto.
17794 (get_parm_default_def_partitions): Ditto and make it static.
17795 (get_undefined_value_partitions): Ditto and make it static.
17796 (remove_ssa_form): Refactor call to init_var_map here.
17797 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
17798 computation for loop region.
17799 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
17800 (register_default_def): Delete.
17801 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
17802 (parm_default_def_partition_arg): Ditto.
17803 (set_parm_default_def_partition): Ditto.
17804 (get_parm_default_def_partitions): Ditto and make it static.
17805 (get_undefined_value_partitions): Ditto and make it static.
17806 (coalesce_with_default, coalesce_with_default): Update comment.
17807 (create_coalesce_list_for_region): New func factored out from
17808 create_outofssa_var_map.
17809 (populate_coalesce_list_for_outofssa): New func factored out from
17810 create_outofssa_var_map and coalesce_ssa_name.
17811 (create_outofssa_var_map): Delete.
17812 (coalesce_ssa_name): Refactor to support live range computation.
17813 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
17814 (get_parm_default_def_partitions): Delete.
17815 (get_undefined_value_partitions): Ditto.
17816 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
17817 computation for loop region.
17818 (new_tree_live_info, loe_visit_block): Ditto.
17819 (live_worklist, set_var_live_on_entry): Ditto.
17820 (calculate_live_on_exit, verify_live_on_entry): Ditto.
17821 * tree-ssa-live.h (struct _var_map): New fields.
17822 (init_var_map): Change decl.
17823 (region_contains_p): New.
17825 2018-05-25 Bin Cheng <bin.cheng@arm.com>
17827 * tree-ssa-live.h (live_merge_and_clear): Delete.
17829 2018-05-25 Richard Biener <rguenther@suse.de>
17832 * tree-dump.c (dequeue_and_dump): Remove access to removed
17833 operand 2 of a SWITCH_EXPR.
17835 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17837 * doc/sourcebuild.texi (vect_double_cond_arith): Include
17838 multiplication and division.
17839 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
17840 (cond_udiv@var{m}, cond_umod@var{m}): Document.
17841 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
17842 (cond_udiv_optab, cond_umod_optab): New optabs.
17843 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
17844 (IFN_COND_RDIV): New internal functions.
17845 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
17846 TRUNC_MOD_EXPR and RDIV_EXPR.
17847 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
17848 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
17850 (SVE_INT_BINARY): Include mult.
17851 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
17852 (optab, sve_int_op): Handle mult.
17853 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
17855 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
17856 for SVE_INT_BINARY_SD.
17858 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17860 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
17861 (optab, sve_int_op): Handle div and udiv.
17862 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
17863 for SVE_INT_BINARY_SD.
17864 (*<optab><mode>3): New insn for the same.
17866 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17868 * tree-vect-patterns.c: Include predict.h.
17869 (vect_recog_divmod_pattern): Restrict check for division support
17870 to when optimizing for size.
17872 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17874 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
17875 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
17876 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
17877 (gimple_match_op::set_op): Likewise.
17878 (gimple_resimplify4): Declare.
17879 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
17880 (expr::gen_transform): Likewise.
17881 (decision_tree::gen): Generate a simplification routine for 4 operands.
17882 * gimple-match-head.c (gimple_simplify): Add an overload for
17883 4 operands. In the top-level function, handle up to 4 call
17884 arguments and call gimple_resimplify4.
17885 (gimple_resimplify4): New function.
17886 (build_call_internal): Pass a fourth operand.
17887 (maybe_push_to_seq): Likewise.
17888 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
17889 Fold VEC_COND_EXPRs of an operation and a default value into
17890 an IFN_COND_* function if possible.
17891 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
17893 (SVE_COND_FP_BINARY): Include them.
17894 (optab, sve_fp_op): Handle them.
17895 (SVE_INT_BINARY_REV): New code iterator.
17896 (SVE_COND_FP_BINARY_REV): New int iterator.
17897 (commutative): New int attribute.
17898 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
17900 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
17902 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
17903 (*cond_<optab><mode>): New patterns for reversed operands.
17905 2018-05-25 Richard Biener <rguenther@suse.de>
17907 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
17908 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
17909 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
17910 (STMT_VINFO_GROUPED_ACCESS): Adjust.
17911 * tree-vect-data-refs.c (everywhere): Adjust users.
17912 * tree-vect-loop.c (everywhere): Likewise.
17913 * tree-vect-slp.c (everywhere): Likewise.
17914 * tree-vect-stmts.c (everywhere): Likewise.
17915 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
17917 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17919 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
17921 (gcc_cv_as_section_exclude): ... this.
17922 Try Solaris as #exclude syntax.
17923 * configure: Regenerate.
17924 * config.in: Regenerate.
17925 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
17927 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
17928 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
17930 * varasm.c (default_elf_asm_named_section): Don't check if
17931 HAVE_GAS_SECTION_EXCLUDE is defined.
17933 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17935 * doc/md.texi: Update the documentation of the cond_* optabs
17936 to mention the new final operand. Fix GET_MODE_NUNITS call.
17937 Describe the scalar case too.
17938 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
17939 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
17941 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
17942 (get_conditional_internal_fn): Update comment.
17943 * tree-vect-loop.c (vectorizable_reduction): Pass the original
17944 accumulator value as a final argument to conditional functions.
17945 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
17946 a define_expand and add an "else" operand. Assert for now that
17947 the else operand is equal to operand 2. Use SVE_INT_BINARY and
17948 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
17949 (*cond_<optab><mode>): New patterns.
17950 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
17951 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
17952 (UNSPEC_COND_EOR): Delete.
17953 (optab): Remove associated mappings.
17954 (SVE_INT_BINARY): New code iterator.
17955 (sve_int_op): Remove int attribute and add "minus" to the code
17957 (SVE_COND_INT_OP): Delete.
17958 (SVE_COND_FP_OP): Rename to...
17959 (SVE_COND_FP_BINARY): ...this.
17961 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17963 * optabs.c (can_reuse_operands_p): New function.
17964 (maybe_legitimize_operands): Try to reuse the results for
17967 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
17969 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
17970 Add {q} suffix to insn mnemonic.
17972 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17974 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
17975 (msp430_warn_func_return): New.
17977 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
17979 * fold-const.c (tree_nonzero_bits): New function.
17980 * fold-const.h (tree_nonzero_bits): Likewise.
17981 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
17982 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
17984 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
17988 * varasm.c (assemble_alias): Check ifunc_resolver only on
17991 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
17994 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
17995 when memory input operand is handled.
17997 2018-05-24 Luis Machado <luis.machado@linaro.org>
17999 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
18001 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
18003 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18005 * match.pd: Delay FMA folds until after vectorization.
18007 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
18010 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
18011 address check not strict.
18013 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18015 * gimple-match.h (gimple_match_op): New class.
18016 (mprts_hook): Replace parameters with a gimple_match_op *.
18017 (maybe_build_generic_op): Likewise.
18018 (gimple_simplified_result_is_gimple_val): Replace parameters with
18019 a const gimple_match_op *.
18020 (gimple_simplify): Replace code_helper * and tree * parameters with
18021 a gimple_match_op * parameter.
18022 (gimple_resimplify1): Replace code_helper *, tree and tree *
18023 parameters with a gimple_match_op * parameter.
18024 (gimple_resimplify2): Likewise.
18025 (gimple_resimplify3): Likewise.
18026 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
18027 parameters with a gimple_match_op * parameter.
18028 * gimple-match-head.c (gimple_simplify): Change prototypes of
18029 auto-generated functions to take a gimple_match_op * instead of
18030 separate code_helper * and tree * parameters. Make the same
18031 change in the top-level overload and update calls to the
18032 gimple_resimplify routines. Update calls to the auto-generated
18033 functions and to maybe_push_res_to_seq in the publicly-facing
18034 operation-specific gimple_simplify overloads.
18035 (gimple_match_op::MAX_NUM_OPS): Define.
18036 (gimple_resimplify1): Replace rcode and ops with a single res_op
18037 parameter. Update call to gimple_simplify.
18038 (gimple_resimplify2): Likewise.
18039 (gimple_resimplify3): Likewise.
18040 (mprts_hook): Replace parameters with a gimple_match_op *.
18041 (maybe_build_generic_op): Likewise.
18042 (build_call_internal): Replace type, nargs and ops with
18043 a gimple_match_op *.
18044 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
18045 with a single gimple_match_op *. Update calls to mprts_hook,
18046 build_call_internal and gimple_simplified_result_is_gimple_val.
18047 Factor out code that is common to the tree_code and combined_fn cases.
18048 * genmatch.c (expr::gen_transform): Replace tem_code and
18049 tem_ops with a gimple_match_op called tem_op. Update calls
18050 to the gimple_resimplify functions and maybe_push_res_to_seq.
18051 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
18052 res_ops. Update call to the gimple_resimplify functions.
18053 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
18054 (decision_tree::gen): Make the functions take a gimple_match_op *
18055 called res_op instead of separate res_code and res_ops parameters.
18056 Update call accordingly.
18057 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
18058 and ops with a single res_op parameter. Update calls to
18059 maybe_build_generic_op and maybe_push_res_to_seq.
18060 (fold_stmt_1): Update calls to gimple_simplify and
18061 replace_stmt_with_simplification.
18062 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
18063 and gimple_simplified_result_is_gimple_val.
18064 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
18066 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
18067 with a gimple_match_op *.
18068 (vn_nary_build_or_lookup): Likewise. Update call to
18069 vn_nary_build_or_lookup_1.
18070 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
18071 gimple_match_op *. Update calls to the gimple_resimplify routines
18072 and to gimple_simplified_result_is_gimple_val.
18073 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
18074 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
18075 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
18076 (visit_nary_op): Likewise.
18077 (visit_reference_op_load): Likewise.
18079 2018-05-23 Luis Machado <luis.machado@linaro.org>
18081 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
18082 modifier for printing the step amount.
18084 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18087 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
18090 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
18092 * doc/sourcebuild.texi (Endianness): New subsubsection.
18094 2018-05-23 Luis Machado <luis.machado@linaro.org>
18096 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18097 <prefetch_dynamic_strides>: New const bool field.
18098 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18099 prefetch_dynamic_strides.
18100 (exynosm1_prefetch_tune): Likewise.
18101 (thunderxt88_prefetch_tune): Likewise.
18102 (thunderx_prefetch_tune): Likewise.
18103 (thunderx2t99_prefetch_tune): Likewise.
18104 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
18106 (aarch64_override_options_internal): Update to set
18107 PARAM_PREFETCH_DYNAMIC_STRIDES.
18108 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18109 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18110 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18111 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18112 prefetch-dynamic-strides setting.
18114 2018-05-23 Luis Machado <luis.machado@linaro.org>
18116 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18117 <minimum_stride>: New const int field.
18118 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18119 minimum_stride field defaulting to -1.
18120 (exynosm1_prefetch_tune): Likewise.
18121 (thunderxt88_prefetch_tune): Likewise.
18122 (thunderx_prefetch_tune): Likewise.
18123 (thunderx2t99_prefetch_tune): Likewise.
18124 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
18125 <default_opt_level>: Set to 3.
18126 (aarch64_override_options_internal): Update to set
18127 PARAM_PREFETCH_MINIMUM_STRIDE.
18128 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18129 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18130 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18131 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18132 stride is constant and is below the minimum stride threshold.
18134 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18136 * config/arm/arm-cpus.in (mode26): Delete.
18137 (armv4): Delete mode26 reference.
18138 * config/arm/arm.c (arm_configure_build_target): Delete use of
18141 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
18143 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
18145 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
18146 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
18148 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
18149 (floatunsdidf2): Ditto.
18151 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
18153 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
18154 (fixuns_trunc<mode>si2_avx512f): Ditto.
18155 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
18156 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
18157 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
18159 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
18161 PR rtl-optimization/79985
18162 * df-scan.c (df_insn_refs_collect): Remove special case for
18163 global registers and asm statements.
18165 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
18167 * extend.texi (Global Register Variables): Rewrite the bullet list.
18168 Note that the register is available for allocation. Note that access
18169 via inline asm must use constraints. Add note about async-signal
18170 handlers. Remove paragraph about automagic register selection.
18172 2018-05-23 Richard Biener <rguenther@suse.de>
18174 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
18175 of fixed offset from memset VN.
18177 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
18179 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
18180 first_interp field.
18181 (alloc_cand_and_find_basis): Initialize first_interp field.
18182 (slsr_process_mul): Modify first_interp field.
18183 (slsr_process_add): Likewise.
18184 (slsr_process_cast): Modify first_interp field for each new
18186 (slsr_process_copy): Likewise.
18187 (dump_candidate): Dump first_interp field.
18188 (replace_mult_candidate): Process all interpretations, not just
18190 (replace_rhs_if_not_dup): Likewise.
18191 (replace_one_candidate): Likewise.
18193 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
18195 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
18197 (aarch64_needs_frame_chain): New function.
18198 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
18200 2018-05-23 Sudakshina Das <sudi.das@arm.com>
18203 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
18204 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
18205 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
18206 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
18207 as true for strict-align.
18208 (aarch64_can_inline_p): Perform checks even when callee has no
18209 attributes to check for strict alignment.
18210 * doc/extend.texi (AArch64 Function Attributes): Document
18212 * doc/invoke.texi: (AArch64 Options): Likewise.
18214 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
18216 PR tree-optimization/85853
18217 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
18218 the handling of the root of the node to...
18219 (vect_slp_analyze_node_operations_1): ...this new function,
18220 and run the whole thing with the child nodes' def types
18221 set according to their SLP node's def type.
18223 2018-05-23 Richard Biener <rguenther@suse.de>
18225 PR middle-end/85874
18226 * tree-data-ref.c (create_runtime_alias_checks): Defer
18227 and ignore overflow warnings.
18229 2018-05-23 Yury Gribov <tetra2005@gmail.com>
18231 PR tree-optimization/85822
18232 * tree-vrp.c (is_masked_range_test): Fix handling of negative
18235 2018-05-23 Richard Biener <rguenther@suse.de>
18237 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
18238 memset constants via native_interpret_expr.
18240 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
18243 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
18245 (cgraph_node::create_alias): Likewise.
18246 (cgraph_node::get_availability): Check ifunc_resolver instead
18247 of looking up ifunc attribute.
18248 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
18249 * varasm.c (do_assemble_alias): Likewise.
18250 (assemble_alias): Likewise.
18251 (default_binds_local_p_3): Likewise.
18252 * cgraph.h (cgraph_node): Add ifunc_resolver.
18253 (cgraph_node::only_called_directly_or_aliased_p): Return false
18254 for IFUNC resolver.
18255 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
18257 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
18258 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
18259 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
18260 instead of looking up ifunc attribute.
18262 2018-05-22 Luis Machado <luis.machado@linaro.org>
18264 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
18266 2018-05-22 Martin Sebor <msebor@redhat.com>
18268 PR middle-end/85359
18269 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
18270 only when expasion succeeds.
18271 (expand_builtin_strcmp): Same.
18272 (expand_builtin_strncmp): Same.
18274 2018-05-22 Martin Sebor <msebor@redhat.com>
18276 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
18278 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
18279 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18281 * config/aarch64/aarch64-ldpstp.md: Replace uses of
18282 aarch64_mem_pair_operand with memory_operand and delete operand swapping
18284 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
18285 Add check for legitimate_address.
18286 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
18287 (aarch64_swap_ldrstr_operands): New.
18288 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
18291 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
18292 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18294 * config/aarch64/aarch64.md: New patterns to generate stp
18296 (store_pair_sw, store_pair_dw): New patterns to generate stp for
18297 single words and double words.
18298 (load_pair_sw, load_pair_dw): Likewise.
18299 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
18301 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
18303 * config/aarch64/aarch64-ldpstp.md: Modify peephole
18304 for different mode ldpstp and add peephole for merged zero stores.
18305 Likewise for loads.
18306 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
18308 (aarch64_gen_store_pair): Rename calls to match new patterns.
18309 (aarch64_gen_load_pair): Rename calls to match new patterns.
18310 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
18311 (load_pair<DREG:mode><DREG2:mode>): ... This.
18312 (store_pair<mode>): Rename to...
18313 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
18314 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
18315 New mode iterators.
18316 (V_INT_EQUIV): Handle SImode.
18317 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
18320 2018-05-22 Martin Sebor <msebor@redhat.com>
18323 * calls.c (maybe_warn_nonstring_arg): Use string length to set
18324 or ajust the presumed bound on an operation to avoid unnecessary
18327 2018-05-22 Martin Sebor <msebor@redhat.com>
18329 PR tree-optimization/85826
18330 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
18331 assuming that a DECL necesarily has a constant size.
18333 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
18335 PR middle-end/85862
18336 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
18338 2018-05-22 Richard Biener <rguenther@suse.de>
18340 PR tree-optimization/85834
18341 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
18342 non-constant and non-zero memset arguments.
18344 2018-05-22 Martin Liska <mliska@suse.cz>
18347 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
18349 2018-05-22 Richard Biener <rguenther@suse.de>
18351 PR tree-optimization/85863
18352 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
18353 comparisons when vectype is specified.
18354 (vectorizable_condition): Do not specify vectype for
18355 vect_is_simple_cond when SLP vectorizing.
18357 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
18360 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
18361 define __ibm128 as long double.
18362 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
18363 as a distinct type when IEEE 128-bit support is enabled.
18364 (init_float128_ieee): Fix up conversions between IFmode and IEEE
18365 128-bit types to use the correct functions.
18366 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
18367 convert between 128-bit floating point types that have different
18368 modes but the same representation, instead of using gen_lowpart to
18370 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
18372 (IFKF_reg): New attributes to give the register constraints for
18374 (extend<mode>tf2_internal): New insns to mark an explicit
18375 conversion between 128-bit floating point types that have a
18376 different mode but share the same representation.
18378 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
18380 PR tree-optimization/85814
18381 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
18382 a null return from get_strinfo when unsharing the next
18383 strinfo in the chain.
18385 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
18388 * varasm.c (weak_finish): Clean up weak_decls.
18390 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18392 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
18393 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
18394 UNSPEC_UADALP values.
18395 * config/aarch64/iterators.md (ABAL): New int iterator.
18398 (sur): Add mappings for the above.
18399 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
18401 (aarch64_<sur>abal<mode>_4): Likewise.
18402 (aarch64_<sur>adalp<mode>_3): Likewise.
18403 (<sur>sadv16qi): New define_expand.
18405 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
18407 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
18408 (*movdf_internal): Ditto.
18409 (*rcpsf2_sse): Ditto.
18410 (*rsqrtsf2_sse): Ditto.
18411 (*sqrt<mode>2_sse): Ditto.
18413 2018-05-21 Tamar Christina <tamar.christina@arm.com>
18415 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
18417 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
18418 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
18419 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
18420 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
18422 * config/aarch64/arm_neon.h: Likewise.
18423 * config/aarch64/iterators.md (VQ_I): New.
18425 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
18427 * config.gcc: Add arc/t-multilib-linux to tmake_file for
18429 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
18432 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
18434 * config/nds32/constraints.md (S): New constraint.
18435 * config/nds32/nds32.md (call_internal): Use constraint S.
18436 (call_value_internal): Likewise.
18437 (sibcall_internal): Likewise.
18438 (sibcall_value_internal): Likewise.
18440 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
18441 Chung-Ju Wu <jasonwucj@gmail.com>
18443 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
18444 into consideration.
18446 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
18447 Chung-Ju Wu <jasonwucj@gmail.com>
18449 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
18450 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
18451 (nds32_rtx_costs_impl): Simplify.
18452 (nds32_address_cost_impl): Simplify.
18453 (nds32_init_rtx_costs): New function.
18454 (nds32_rtx_costs_speed_prefer): Likewise.
18455 (nds32_rtx_costs_size_prefer): Likewise.
18456 (nds32_address_cost_speed_prefer): Likewise.
18457 (nds32_address_cost_speed_fwprop): Likewise.
18458 (nds32_address_cost_size_prefer): Likewise.
18459 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
18460 * config/nds32/nds32.c (nds32_option_override): Use
18461 nds32_init_rtx_costs function.
18463 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
18465 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
18466 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
18467 (TARGET_PIPELINE_N8): Likewise.
18468 (TARGET_PIPELINE_N10): Likewise.
18469 (TARGET_PIPELINE_N13): Likewise.
18470 (TARGET_PIPELINE_GRAYWOLF): Likewise.
18472 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
18474 * config/nds32/nds32-fpu.md: Update copyright year.
18476 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18478 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
18480 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18482 * config/nds32/nds32.c
18483 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
18484 * config/nds32/nds32.opt (minline-asm-r15): New option.
18486 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18488 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
18490 * config/nds32/nds32.md (abssi2): New pattern.
18492 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
18494 * config/i386/i386.md (rex64namesuffix): New mode attribute.
18495 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
18496 Merge insn pattern from sse_cvtsi2ss<round_name> and
18497 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
18498 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
18499 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
18500 using SWI48 mode iterator.
18501 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
18502 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
18503 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
18504 pattern from sse_cvttss2si<round_saeonly_name>
18505 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
18506 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
18507 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
18508 using SWI48 mode iterator.
18509 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
18510 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
18511 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
18512 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
18513 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
18514 using SWI48 mode iterator.
18515 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
18516 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
18517 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
18518 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
18519 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
18520 SWI48 mode iterator.
18521 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
18522 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
18523 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
18524 pattern from sse_cvttsd2si<round_saeonly_name>
18525 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
18527 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18529 * config/nds32/nds32-md-auxiliary.c
18530 (nds32_valid_smw_lwm_base_p): Refine.
18531 (nds32_output_smw_single_word): Refine.
18532 (nds32_output_smw_double_word): New.
18533 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
18535 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18537 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
18538 (nds32_output_stack_pop): Refine.
18539 (nds32_expand_unaligned_load): Refine.
18540 (nds32_expand_unaligned_store): Refine.
18542 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18543 Chung-Ju Wu <jasonwucj@gmail.com>
18545 * config/nds32/constants.md: Add TP_REGNUM constant.
18546 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
18547 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
18549 * config/nds32/nds32-doubleword.md: Consider flag_pic.
18550 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
18551 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
18552 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
18553 and PIC code generation.
18554 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
18556 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
18558 * config/nds32/nds32.md: Support TLS and PIC.
18559 * config/nds32/nds32.c: Support TLS and PIC.
18560 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
18561 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
18564 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18566 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
18567 mode with E_ prefix.
18569 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18570 Chung-Ju Wu <jasonwucj@gmail.com>
18572 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
18573 * config/nds32/nds32-md-auxiliary.c
18574 (symbolic_reference_mentioned_p): New.
18575 (nds32_legitimize_ict_address): New.
18576 (nds32_expand_ict_move): New.
18577 (nds32_indirect_call_referenced_p): New.
18578 (nds32_symbol_binds_local_p): Delete.
18579 (nds32_long_call_p): Modify.
18580 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
18581 * config/nds32/nds32-protos.h
18582 (symbolic_reference_mentioned_p): Declare.
18583 (nds32_legitimize_ict_address): Declare.
18584 (nds32_expand_ict_move): Declare.
18585 (nds32_indirect_call_referenced_p): Declare.
18586 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
18587 (nds32_relax_group): Use nds32_ict_const_p as condition.
18588 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
18589 (nds32_asm_file_start): Output ict_model directive in asm code.
18590 (nds32_legitimate_address_p): Consider indirect call.
18591 (nds32_print_operand): Consider indirect call.
18592 (nds32_print_operand_address): Consider indirect call.
18593 (nds32_insert_attributes): Handle "indirect_call" attribute.
18594 (TARGET_LEGITIMATE_ADDRESS_P): Define.
18595 (TARGET_LEGITIMATE_CONSTANT_P): Define.
18596 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18597 (TARGET_DELEGITIMIZE_ADDRESS): Define.
18598 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
18599 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
18600 (TARGET_ICT_MODEL_SMALL): Define.
18601 (TARGET_ICT_MODEL_LARGE): Define.
18602 * config/nds32/nds32.md (movsi): Consider ict model.
18603 (call, call_value): Consider ict model.
18604 (sibcall, sibcall_value): Consider ict model.
18605 * config/nds32/nds32.opt (mict-model): New option.
18606 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
18609 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
18610 Monk Chiang <sh.chiang04@gmail.com>
18611 Jim Wilson <jimw@sifive.com>
18613 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
18614 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
18615 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
18616 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
18617 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
18618 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
18619 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
18620 compute save_libcall_adjustment properly.
18621 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
18622 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
18623 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
18624 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
18625 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
18626 (ABI_SPEC): Handle mabi=ilp32e.
18627 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
18628 (RVE): Add RVE mask.
18629 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
18630 <-march>: Add rv32e as an example.
18632 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
18635 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
18636 (intra_create_variable_infos): Handle C++ constructors.
18638 2018-05-18 Martin Liska <mliska@suse.cz>
18640 * passes.def: Remove a redundant pass.
18642 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
18645 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
18647 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18649 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
18651 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
18653 (armv2, armv2a, armv3, armv3m): Delete architectures.
18654 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
18655 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
18656 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
18658 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
18659 (*mulsidi3adddi): Likewise.
18660 (mulsidi3): Likewise.
18661 (*mulsidi3_nov6): Likewise.
18662 (umulsidi3): Likewise.
18663 (umulsidi3_nov6): Likewise.
18664 (umaddsidi4): Likewise.
18665 (*umulsidi3adddi): Likewise.
18666 (smulsi3_highpart): Likewise.
18667 (*smulsi3_highpart_nov6): Likewise.
18668 (umulsi3_highpart): Likewise.
18669 (*umulsi3_highpart_nov6): Likewise.
18670 * config/arm/arm.h (arm_arch3m): Delete.
18671 * config/arm/arm.c (arm_arch3m): Delete.
18672 (arm_option_override_internal): Update armv3-related comment.
18673 (arm_configure_build_target): Delete use of isa_bit_mode32.
18674 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
18675 (arm_rtx_costs_internal): Delete check of arm_arch3m.
18676 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
18677 (mulsa3): Likewise.
18678 (mulusa3): Likewise.
18679 * config/arm/arm-protos.h (arm_arch3m): Delete.
18680 * config/arm/arm-tables.opt: Regenerate.
18681 * config/arm/arm-tune.md: Likewise.
18682 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
18683 deleted architectures.
18685 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18687 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
18688 (armv5t, armv5te): New features.
18689 (ARMv5, ARMv5e): Delete fgroups.
18690 (ARMv5t, ARMv5te): Adjust for above changes.
18691 (ARMv6m): Likewise.
18692 (armv5, armv5e): Delete arches.
18693 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
18695 (*call_reg_arm): Likewise.
18696 (*call_value_reg_armv5): Likewise.
18697 (*call_value_reg_arm): Likewise.
18698 (*call_symbol): Likewise.
18699 (*call_value_symbol): Likewise.
18700 (*sibcall_insn): Likewise.
18701 (*sibcall_value_insn): Likewise.
18702 (clzsi2): Likewise.
18703 (prefetch): Likewise.
18704 (define_split and define_peephole2 dependent on arm_arch5):
18706 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
18708 (TARGET_ARM_QBIT): Likewise.
18709 (TARGET_DSP_MULTIPLY): Likewise.
18710 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
18711 (arm_arch5, arm_arch5e): Delete.
18712 (arm_arch5t, arm_arch5te): Declare.
18713 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
18714 (arm_arch5t): Declare.
18715 (arm_option_reconfigure_globals): Update for the above.
18716 (arm_options_perform_arch_sanity_checks): Update comment, replace
18717 use of arm_arch5 with arm_arch5t.
18718 (use_return_insn): Likewise.
18719 (arm_emit_call_insn): Likewise.
18720 (output_return_instruction): Likewise.
18721 (arm_final_prescan_insn): Likewise.
18722 (arm_coproc_builtin_available): Likewise.
18723 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
18724 arm_arch5e with arm_arch5t and arm_arch5te.
18725 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
18726 (arm_arch5t, arm_arch5te): Declare.
18727 * config/arm/arm-tables.opt: Regenerate.
18728 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
18729 * config/arm/t-multilib: Likewise.
18730 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
18731 instead of arm_arch5.
18732 (*call_reg_thumb1): Likewise.
18733 (*call_value_reg_thumb1_v5): Likewise.
18734 (*call_value_reg_thumb1): Likewise.
18735 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
18737 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
18739 2018-05-18 Martin Liska <mliska@suse.cz>
18741 PR gcov-profile/84846
18742 * doc/gcov.texi: Document -t option of gcov tool.
18744 2018-05-18 Martin Liska <mliska@suse.cz>
18746 PR gcov-profile/84846
18747 * gcov.c (print_usage): Add new -t option.
18748 (process_args): Handle the option.
18749 (generate_results): Use stdout as output when requested by
18752 2018-05-18 Martin Liska <mliska@suse.cz>
18754 PR gcov-profile/84846
18755 * coverage.c (coverage_init): Write PWD to .gcno file.
18756 * doc/gcov.texi: Document how working directory is printed.
18757 * gcov-dump.c (dump_gcov_file): Print PWD.
18758 * gcov.c (output_intermediate_file): Likewise.
18759 (read_graph_file): Read PWD string.
18760 (output_lines): Print PWD.
18762 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18764 PR middle-end/85817
18765 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
18766 for retval and return false if all args to phi are zero.
18768 2018-05-18 Richard Biener <rguenther@suse.de>
18770 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
18772 (evrp_dom_walker::before_dom_children): Call it.
18774 2018-05-18 Richard Biener <rguenther@suse.de>
18776 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
18777 results when processing array refs with variable index.
18779 2018-05-18 Toon Moene <toon@moene.org>
18781 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
18782 directly after that of -floop-interchange. Indicate that both
18783 options are enabled by default when specifying -O3.
18785 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18787 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
18788 iterator. Delete separate integer-mode vec_set<mode> expander.
18789 (aarch64_simd_vec_setv2di): Delete.
18790 (vec_setv2di): Delete.
18791 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
18792 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
18793 the "w, r" alternative.
18795 2018-05-18 Martin Liska <mliska@suse.cz>
18797 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
18798 * tree-pass.h (make_pass_lower_switch_O0): New function.
18799 * tree-switch-conversion.c (node_has_low_bound): Remove.
18800 (node_has_high_bound): Likewise.
18801 (node_is_bounded): Likewise.
18802 (class pass_lower_switch): Make it a template type and create
18804 (pass_lower_switch::execute): Add template argument.
18805 (make_pass_lower_switch): New function.
18806 (make_pass_lower_switch_O0): New function.
18807 (do_jump_if_equal): Remove.
18808 (emit_case_nodes): Simplify to just handle all 3 cases and leave
18809 all the hard work to tree optimization passes.
18811 2018-05-18 Martin Liska <mliska@suse.cz>
18813 * dbgcnt.c (limit_low): Renamed from limit.
18814 (limit_high): New variable.
18815 (dbg_cnt_is_enabled): Check for upper limit.
18816 (dbg_cnt): Adjust dumping.
18817 (dbg_cnt_set_limit_by_index): Add new argument for high
18819 (dbg_cnt_set_limit_by_name): Likewise.
18820 (dbg_cnt_process_single_pair): Parse new format.
18821 (dbg_cnt_process_opt): Use strtok.
18822 (dbg_cnt_list_all_counters): Remove 'value' and add
18824 * doc/invoke.texi: Document changes.
18826 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
18828 * doc/sourcebuild.texi (scalar_all_fma): Document.
18829 * tree.def (FMA_EXPR): Delete.
18830 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
18831 * internal-fn.c (ternary_direct): New macro.
18832 (expand_ternary_optab_fn): Likewise.
18833 (direct_ternary_optab_supported_p): Likewise.
18834 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
18835 * builtins.c (fold_builtin_fma): Delete.
18836 (fold_builtin_3): Don't call it.
18837 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
18838 * expr.c (expand_expr_real_2): Likewise.
18839 * fold-const.c (operand_equal_p): Likewise.
18840 (fold_ternary_loc): Likewise.
18841 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
18842 * gimple.c (DEFTREECODE): Likewise.
18843 * gimplify.c (gimplify_expr): Likewise.
18844 * optabs-tree.c (optab_for_tree_code): Likewise.
18845 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18846 * tree-eh.c (operation_could_trap_p): Likewise.
18847 (stmt_could_throw_1_p): Likewise.
18848 * tree-inline.c (estimate_operator_cost): Likewise.
18849 * tree-pretty-print.c (dump_generic_node): Likewise.
18850 (op_code_prio): Likewise.
18851 * tree-ssa-loop-im.c (stmt_cost): Likewise.
18852 * tree-ssa-operands.c (get_expr_operands): Likewise.
18853 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
18854 * fold-const-call.h (fold_fma): Delete.
18855 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
18856 CFN_FNMA and CFN_FNMS.
18857 (fold_fma): Delete.
18858 * genmatch.c (combined_fn): New enum.
18859 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
18860 (commutative_op): New function.
18861 (commutate): Use it. Handle more than 2 operands.
18862 (dt_operand::gen_gimple_expr): Use commutative_op.
18863 (parser::parse_expr): Allow :c to be used with non-binary
18864 operators if the commutative operand is known.
18865 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
18866 CFN_FMS, CFN_FNMA and CFN_FNMS.
18867 (backprop::process_assign_use): Remove FMA_EXPR handling.
18868 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
18869 (gen_hsa_fma): New function.
18870 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
18871 IFN_FNMA and IFN_FNMS.
18872 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
18873 * gimple-fold.h (follow_all_ssa_edges): Declare.
18874 * gimple-fold.c (follow_all_ssa_edges): New function.
18875 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
18876 gimple_build interface and use follow_all_ssa_edges to fold the result.
18877 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
18878 instead of checking for optabs directly.
18879 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
18880 rather than FMA_EXPRs.
18881 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
18882 call to IFN_FMA instead of an FMA_EXPR.
18884 2018-05-17 Jim Wilson <jimw@sifive.com>
18886 * expr.c (do_tablejump): When converting index to Pmode, if we have a
18887 sign extended promoted subreg, and the range does not have the sign bit
18888 set, then do a sign extend.
18890 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
18891 test, check for sign extended subreg and/or constant operands, and
18892 do a sign extend in that case.
18894 2018-05-17 Steve Ellcey <sellcey@cavium.com>
18896 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
18897 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
18899 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
18900 Change logics_shift_reg to logics_shift_imm.
18901 (thunderx2t99_fp_loadpair_basic): Delete.
18902 (thunderx2t99_fp_storepair_basic): Delete.
18903 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
18904 (thunderx2t99_asimd_polynomial): Delete.
18905 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
18906 and neon_fp_mul_d_scalar_q.
18907 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
18908 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
18909 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
18910 (thunderx2t99_asimd_lut): Add missing tbl types.
18911 (thunderx2t99_asimd_ext): Delete.
18912 (thunderx2t99_asimd_load1_1_mult): Delete.
18913 (thunderx2t99_asimd_load1_2_mult): Delete.
18914 (thunderx2t99_asimd_load1_ldp): New.
18915 (thunderx2t99_asimd_load1): New.
18916 (thunderx2t99_asimd_load2): Add missing *load2* types.
18917 (thunderx2t99_asimd_load3): New.
18918 (thunderx2t99_asimd_load4): New.
18919 (thunderx2t99_asimd_store1_1_mult): Delete.
18920 (thunderx2t99_asimd_store1_2_mult): Delete.
18921 (thunderx2t99_asimd_store2_mult): Delete.
18922 (thunderx2t99_asimd_store2_onelane): Delete.
18923 (thunderx2t99_asimd_store_stp): New.
18924 (thunderx2t99_asimd_store1): New.
18925 (thunderx2t99_asimd_store2): New.
18926 (thunderx2t99_asimd_store3): New.
18927 (thunderx2t99_asimd_store4): New.
18929 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
18931 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
18932 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
18934 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
18935 Segher Boessenkool <segher@kernel.crashing.org>
18938 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
18941 2018-05-17 Richard Biener <rguenther@suse.de>
18943 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
18944 for pruning loop and prune defs feeding only already visited PHIs.
18946 2018-05-17 Richard Biener <rguenther@suse.de>
18948 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
18950 2018-05-17 Bin Cheng <bin.cheng@arm.com>
18951 Richard Biener <rguenther@suse.de>
18953 PR tree-optimization/85793
18954 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
18955 for VMAT_ELEMENTWISE.
18957 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
18959 * internal-fn.h (lookup_internal_fn): Declare
18960 * internal-fn.c (lookup_internal_fn): New function.
18961 * gimple.c (gimple_build_call_from_tree): Handle calls to
18962 internal functions.
18963 * gimple-pretty-print.c (dump_gimple_call): Print "." before
18964 internal function names.
18965 * tree-pretty-print.c (dump_generic_node): Likewise.
18966 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
18968 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
18970 * gimple-fold.h (gimple_build): Make the function forms take
18971 combined_fn rather than built_in_function.
18972 (gimple_simplify): Likewise.
18973 * gimple-match-head.c (gimple_simplify): Likewise.
18974 * gimple-fold.c (gimple_build): Likewise.
18975 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
18976 rather than gimple_build_call_internal.
18977 (get_initial_defs_for_reduction): Likewise.
18978 (vect_create_epilog_for_reduction): Likewise.
18979 (vectorizable_live_operation): Likewise.
18981 2018-05-17 Martin Liska <mliska@suse.cz>
18983 * gimple-ssa-sprintf.c (format_directive): Do not use
18984 space in between 'G_' and '('.
18986 2018-05-17 Jakub Jelinek <jakub@redhat.com>
18989 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
18990 even if the mask is not all ones.
18993 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
18995 (ix86_gimple_fold_builtin): Likewise.
18998 * config/i386/i386.c: Include tree-vector-builder.h.
18999 (ix86_vector_shift_count): New function.
19000 (ix86_fold_builtin): Fold shift builtins by scalar count.
19001 (ix86_gimple_fold_builtin): Likewise.
19003 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
19004 _mm512_setzero): New intrinsics.
19006 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
19007 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19009 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
19010 code generation for cases where splatting a value is not useful.
19011 * simplify-rtx.c (simplify_ternary_operation): Simplify
19012 vec_merge across a vec_duplicate and a paradoxical subreg forming
19013 a vector mode to a vec_concat.
19015 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
19017 * config.gcc: Support "goldmont-plus".
19018 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19020 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19021 PROCESSOR_GOLDMONT_PLUS.
19022 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
19023 (processor_target_table): Add "goldmont-plus".
19024 (PTA_GOLDMONT_PLUS): Define.
19025 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
19026 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
19027 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
19028 (fold_builtin_cpu): Add "goldmont-plus".
19029 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
19030 (ix86_option_override_internal): Add "goldmont-plus".
19031 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
19032 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
19033 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
19034 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
19036 2018-05-17 Richard Biener <rguenther@suse.de>
19038 PR tree-optimization/85757
19039 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
19040 remove defs that only feed that PHI from further processing.
19042 2018-05-16 Jim Wilson <jimw@sifive.com>
19044 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
19046 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
19048 2018-05-16 Mark Wielaard <mark@klomp.org>
19050 * dwarf2out.c (count_index_strings): New function.
19051 (output_indirect_strings): Call count_index_strings and generate
19052 header for dwarf_version >= 5.
19054 2018-05-16 Mark Wielaard <mark@klomp.org>
19056 * dwarf2out.c (dwarf_FORM): New function.
19057 (set_indirect_string): Use dwarf_FORM.
19058 (reset_indirect_string): Likewise.
19059 (size_of_die): Likewise.
19060 (value_format): Likewise.
19061 (output_die): Likewise.
19062 (add_skeleton_AT_string): Likewise.
19063 (output_macinfo_op): Likewise.
19064 (index_string): Likewise.
19065 (output_index_string_offset): Likewise.
19066 (output_index_string): Likewise.
19067 (count_index_strings): Likewise.
19069 2018-05-16 Carl Love <cel@us.ibm.com>
19071 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
19072 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
19074 2018-05-16 Martin Jambor <mjambor@suse.cz>
19076 * ipa-prop.c (ipa_free_all_edge_args): Remove.
19077 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
19079 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
19081 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
19082 (fnma<mode>4): Likewise.
19083 (fms<mode>4): Likewise.
19084 (fnms<mode>4): Likewise.
19085 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
19086 (aarch64_fnma<mode>4): Likewise.
19087 (aarch64_fms<mode>4): Likewise.
19088 (aarch64_fnms<mode>4): Likewise.
19089 (aarch64_fnmadd<mode>4): Likewise.
19091 2018-05-16 Jason Merrill <jason@redhat.com>
19093 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
19095 2018-05-16 Richard Biener <rguenther@suse.de>
19097 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
19098 (dump_stmt_cost): Declare.
19099 (add_stmt_cost): Dump cost we add.
19100 (add_stmt_costs): New function.
19101 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
19102 No longer exported.
19103 (vect_analyze_stmt): Adjust prototype.
19104 (vectorizable_condition): Likewise.
19105 (vectorizable_live_operation): Likewise.
19106 (vectorizable_reduction): Likewise.
19107 (vectorizable_induction): Likewise.
19108 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
19109 cost vector to pass to vectorizable_ and record afterwards.
19110 (vect_model_reduction_cost): Take cost vector argument and adjust.
19111 (vect_model_induction_cost): Likewise.
19112 (vectorizable_reduction): Likewise.
19113 (vectorizable_induction): Likewise.
19114 (vectorizable_live_operation): Likewise.
19115 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
19116 SLP_TREE_NUMBER_OF_VEC_STMTS.
19117 (vect_analyze_slp_cost_1): Remove.
19118 (vect_analyze_slp_cost): Likewise.
19119 (vect_slp_analyze_node_operations): Take visited args and
19120 a target cost vector. Avoid processing already visited stmt sets.
19121 (vect_slp_analyze_operations): Use a local cost vector to gather
19122 costs and register those of non-discarded instances.
19123 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
19124 (vect_schedule_slp_instance): Remove copying of
19125 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
19127 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
19128 adding cost. Record cost entry location.
19129 (vect_prologue_cost_for_slp_op): Function to compute cost of
19130 a constant or invariant generated for SLP vect in the prologue,
19131 split out from vect_analyze_slp_cost_1.
19132 (vect_model_simple_cost): Make static. Adjust for SLP costing.
19133 (vect_model_promotion_demotion_cost): Likewise.
19134 (vect_model_store_cost): Likewise, make static.
19135 (vect_model_load_cost): Likewise.
19136 (vectorizable_bswap): Add cost vector arg and adjust.
19137 (vectorizable_call): Likewise.
19138 (vectorizable_simd_clone_call): Likewise.
19139 (vectorizable_conversion): Likewise.
19140 (vectorizable_assignment): Likewise.
19141 (vectorizable_shift): Likewise.
19142 (vectorizable_operation): Likewise.
19143 (vectorizable_store): Likewise.
19144 (vectorizable_load): Likewise.
19145 (vectorizable_condition): Likewise.
19146 (vectorizable_comparison): Likewise.
19147 (can_vectorize_live_stmts): Likewise.
19148 (vect_analyze_stmt): Likewise.
19149 (vect_transform_stmt): Adjust calls to vectorizable_*.
19150 * tree-vectorizer.c: Include gimple-pretty-print.h.
19151 (dump_stmt_cost): New function.
19153 2018-05-16 Richard Biener <rguenther@suse.de>
19155 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
19156 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
19157 * tree-ssa-dse.c: Include tree-ssa-loop.h.
19158 (check_name): New callback.
19159 (dse_classify_store): Track cycles via a visited bitmap of PHI
19160 defs and simplify handling of in-loop and across loop dead stores
19161 and properly fail for loop-variant refs. Handle byte-tracking with
19162 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
19165 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
19167 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
19168 (vect_get_mask_type_for_stmt): Likewise.
19169 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
19171 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
19172 to determine the statement's vector type and the vector type that
19173 should be used for calculating nunits. Deal with cases in which
19174 the type has to be deferred.
19175 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
19176 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
19177 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
19178 (vect_determine_vf_for_stmt): New functions, split out from...
19179 (vect_determine_vectorization_factor): ...here.
19180 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
19181 (vect_get_mask_type_for_stmt): New functions, split out from
19182 vect_determine_vectorization_factor.
19184 2018-05-16 Richard Biener <rguenther@suse.de>
19186 * tree-cfg.c (verify_gimple_assign_ternary): Properly
19187 verify the [VEC_]COND_EXPR embedded comparison.
19189 2018-05-15 Martin Sebor <msebor@redhat.com>
19191 PR tree-optimization/85753
19192 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
19193 RECORD_TYPE in addition to ARRAY_TYPE.
19195 2018-05-15 Martin Sebor <msebor@redhat.com>
19197 PR middle-end/85643
19198 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
19200 2018-05-15 Richard Biener <rguenther@suse.de>
19202 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
19203 add by_clobber_p one. Change algorithm to collect all defs
19204 representing uses we need to walk and try reducing them to
19205 a single one before failing.
19206 (dse_dom_walker::dse_optimize_stmt): Adjust.
19208 2018-05-13 Mark Wielaard <mark@klomp.org>
19210 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
19211 (size_of_loc_descr): Likewise.
19212 (output_loc_operands): Likewise.
19213 (output_loc_operands_raw): Likewise.
19214 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
19215 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
19216 (hash_loc_operands): Likewise.
19217 (compare_loc_operands): Likewise.
19219 2018-05-14 Mark Wielaard <mark@klomp.org>
19221 * dwarf2out.c (count_index_addrs): New function.
19222 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
19224 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19226 PR tree-optimization/83648
19227 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
19228 return value as malloc candidate.
19230 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19233 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
19234 param as true in call to suggest_attribute.
19236 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
19238 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
19239 -mreadonly-in-sdata.
19241 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19243 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
19245 (aarch64_crypto_aesd_fused): Likewise.
19247 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
19249 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
19250 (movsi_aarch64): Likewise.
19251 (load_pairsi): Likewise.
19252 (load_pairdi): Likewise.
19253 (store_pairsi): Likewise.
19254 (store_pairdi): Likewise.
19255 (load_pairsf): Likewise.
19256 (load_pairdf): Likewise.
19257 (store_pairsf): Likewise.
19258 (store_pairdf): Likewise.
19259 (zero_extend): Likewise.
19260 (trunc): Swap alternatives.
19261 (fcvt_target): Add '?' to prefer w over r.
19263 2018-05-14 Jakub Jelinek <jakub@redhat.com>
19266 * config/i386/i386.md: Disallow non-commutative arithmetics in
19267 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
19268 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
19269 in the peephole2 before it.
19271 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
19273 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
19274 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
19275 (ix86_handle_option): Handle -mcldemote.
19276 * config.gcc: New header.
19277 * config/i386/cldemoteintrin.h: New file.
19278 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
19279 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19281 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19282 OPTION_MASK_ISA_CLDEMOTE.
19283 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
19284 (ix86_valid_target_attribute_inner_p): Ditto.
19285 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
19286 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
19287 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
19288 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
19289 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
19291 * config/i386/i386.opt: Add -mcldemote.
19292 * config/i386/x86intrin.h: New header.
19293 * doc/invoke.texi: Add -mcldemote.
19295 2018-05-14 Richard Biener <rguenther@suse.de>
19297 * doc/match-and-simplify.texi: Adjust :s documentation.
19299 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
19301 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
19302 intended memcpy size.
19303 (REORDER_45): Likewise.
19305 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
19307 * sort.cc: New file.
19308 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
19309 * vec.c (qsort_chk): Use gcc_qsort.
19310 * Makefile.in (OBJS-libcommon): Add sort.o.
19311 (build/sort.o): New target. Use it...
19312 (BUILD_RTL): ... here, and...
19313 (build/gencfn-macros): ... here, and...
19314 (build/genmatch): ... here.
19316 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19317 Chung-Ju Wu <jasonwucj@gmail.com>
19319 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
19320 * config/nds32/nds32-graywolf.md: New file.
19321 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
19322 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
19324 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
19325 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
19326 * config/nds32/nds32.md (pipeline_model): Add graywolf.
19327 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
19328 * config/nds32/pipelines.md: Include n15 settings.
19330 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19331 Chung-Ju Wu <jasonwucj@gmail.com>
19333 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
19334 * config/nds32/nds32-n13.md: New file.
19335 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
19336 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
19338 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
19339 * config/nds32/nds32.md (pipeline_model): Add n13.
19340 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
19341 * config/nds32/pipelines.md: Include n13 settings.
19343 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19344 Chung-Ju Wu <jasonwucj@gmail.com>
19346 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
19347 * config/nds32/nds32-n10.md: New file.
19348 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
19349 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
19351 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
19352 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
19353 * config/nds32/nds32.md (pipeline_model): Add n10.
19354 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
19355 * config/nds32/pipelines.md: Include n10 settings.
19357 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
19358 Kito Cheng <kito.cheng@gmail.com>
19359 Chung-Ju Wu <jasonwucj@gmail.com>
19361 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
19362 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19363 Add enum values for DSP extension instructions.
19364 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
19366 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
19367 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
19368 New code iterators.
19369 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
19370 * config/nds32/nds32-dspext.md: New file for DSP implementation.
19371 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
19372 * config/nds32/nds32-intrinsic.md: Likewise.
19373 * config/nds32/nds32_intrinsic.h: Likewise.
19374 * config/nds32/nds32-md-auxiliary.c: Likewise.
19375 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
19376 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
19377 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
19378 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
19379 * config/nds32/nds32-protos.h: New declarations for DSP extension.
19380 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
19381 TYPE_DMAC in switch statement.
19382 * config/nds32/nds32.c: New checking and implementation for DSP
19383 extension instructions.
19384 * config/nds32/nds32.h: Likewise.
19385 * config/nds32/nds32.md: Likewise.
19386 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
19387 * config/nds32/predicates.md: Implement new predicates for DSP
19390 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19392 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
19393 Reformat alternatives and attributes so it is easier to identify
19394 which constraints/attributes go with which instruction.
19395 (mov<mode>_hardfloat32, FMOVE64): Likewise.
19396 (mov<mode>_softfloat32, FMOVE64): Likewise.
19397 (mov<mode>_hardfloat64, FMOVE64): Likewise.
19398 (mov<mode>_softfloat64, FMOVE64): Likewise.
19400 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19402 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19404 (Basic PowerPC Built-in Functions): The new name of the
19405 subsection previously known as "PowerPC Built-in Functions".
19406 (Basic PowerPC Built-in Functions Available on all Configurations):
19408 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
19409 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
19410 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
19411 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
19413 2018-05-11 Martin Jambor <mjambor@suse.cz>
19416 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
19419 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
19422 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
19424 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
19426 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
19427 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
19428 (ix86_handle_option): Handle -mwaitpkg.
19429 * config.gcc: New header.
19430 * config/i386/cpuid.h (bit_WAITPKG): New bit.
19431 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
19432 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
19434 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19435 OPTION_MASK_ISA_WAITPKG.
19436 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
19437 (ix86_option_override_internal): Add PTA_WAITPKG.
19438 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
19439 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
19440 IX86_BUILTIN_TPAUSE.
19441 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
19442 __builtin_ia32_umwait and __builtin_ia32_tpause.
19443 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
19444 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
19445 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
19446 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
19447 UNSPECV_TPAUSE): New.
19448 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
19449 * config/i386/i386.opt: Add -mwaitpkg.
19450 * config/i386/waitpkgintrin.h: New file.
19451 * config/i386/x86intrin.h: New header.
19452 * doc/invoke.texi: Add -mwaitpkg.
19454 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
19457 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
19459 (cortex-m0): Use armv6s-m isa.
19460 (cortex-m0plus): Likewise.
19461 (cortex-m1): Likewise.
19462 (cortex-m0.small-multiply): Likewise.
19463 (cortex-m0plus.small-multiply): Likewise.
19464 (cortex-m1.small-multiply): Likewise.
19466 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
19467 Jakub Jelinek <jakub@redhat.com>
19469 PR tree-optimization/85692
19470 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
19471 source permute as well.
19473 2018-05-11 Martin Liska <mliska@suse.cz>
19476 * doc/extend.texi: Document LLVM style format for no_sanitize
19479 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
19481 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
19482 mode_supports_vsx_dform_quad to mode_supports_dq_form.
19483 (mode_supports_vsx_dform_quad): Likewise.
19484 (mode_supports_vmx_dform): Move these functions to be next to the
19485 other mode_supports functions.
19486 (mode_supports_dq_form): Likewise.
19487 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
19488 mode_supports_dq_form.
19489 (reg_offset_addressing_ok_p): Likewise.
19490 (offsettable_ok_by_alignment): Likewise.
19491 (rs6000_legitimate_offset_address_p): Likewise.
19492 (legitimate_lo_sum_address_p): Likewise.
19493 (rs6000_legitimize_address): Likewise.
19494 (rs6000_legitimize_reload_address): Likewise.
19495 (rs6000_secondary_reload_inner): Likewise.
19496 (rs6000_preferred_reload_class): Likewise.
19497 (rs6000_output_move_128bit): Likewise.
19499 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
19501 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
19502 Generate SImode target register for null target.
19503 <case IX86_BUILTIN_XGETBV>: Ditto.
19504 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
19505 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
19507 2018-05-10 Carl Love <cel@us.ibm.com>
19509 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
19510 dcbtt and dcbtstt if operands[2] is 0.
19512 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
19515 * config/i386/sse.md (usadv64qi): New expander.
19517 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
19519 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
19520 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
19521 -maltivec=be support.
19522 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
19523 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
19524 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19525 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
19526 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
19527 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
19528 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
19529 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
19530 altivec_vsumsws): Adjust.
19531 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
19532 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
19533 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
19535 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
19536 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
19537 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
19538 (altivec_lve<VI_char>x): Delete expand.
19539 (*altivec_lve<VI_char>x_internal): Rename to...
19540 (altivec_lve<VI_char>x): ... this.
19541 (altivec_lvxl_<mode>): Delete expand.
19542 (*altivec_lvxl_<mode>_internal): Rename to ...
19543 (altivec_lvxl_<mode>): ... this.
19544 (altivec_stvxl_<mode>): Delete expand.
19545 (*altivec_stvxl_<mode>_internal): Rename to ...
19546 (altivec_stvxl_<mode>): ... this.
19547 (altivec_stve<VI_char>x): Delete expand.
19548 (*altivec_stve<VI_char>x_internal): Rename to ...
19549 (altivec_stve<VI_char>x): ... this.
19550 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
19551 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
19552 reduc_plus_scal_<mode>): Adjust.
19553 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
19555 (rs6000_cpu_cpp_builtins): Adjust.
19556 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
19557 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
19558 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
19559 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
19560 -maltivec=be support.
19561 (rs6000_split_vec_extract_var): Adjust.
19562 (rs6000_split_v4si_init): Adjust.
19563 (swap_selector_for_mode): Delete.
19564 (altivec_expand_lvx_be, altivec_expand_stvx_be,
19565 altivec_expand_stvex_be): Delete.
19566 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
19567 -maltivec=be support.
19568 (rs6000_gimple_fold_builtin): Ditto.
19569 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
19571 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
19572 (TARGET_DIRECT_MOVE_64BIT): Adjust.
19573 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
19574 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
19575 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
19576 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
19577 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
19578 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
19579 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
19580 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
19581 anonymous split): Adjust.
19582 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
19583 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
19585 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
19587 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
19588 when --with-gxx-include-dir is also specified.
19589 * configure: Regenerate.
19591 2018-05-09 Jim Wilson <jimw@sifive.com>
19594 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
19595 * config/riscv/t-withmultilib: New.
19596 * config/riscv/withmultilib.h: New.
19597 * doc/install.texi: Document RISC-V --with-multilib-list support.
19599 2018-05-09 Richard Biener <rguenther@suse.de>
19601 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
19603 (vect_bb_vectorization_profitable_p): Adjust. Compute
19604 actual scalar cost using the cost vector and the add_stmt_cost
19607 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19609 PR rtl-optimization/85645
19610 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
19611 in the REG_CFA_REGISTER note for LR, don't leave it empty.
19613 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19615 PR rtl-optimization/85645
19616 * shrink-wrap.c (spread_components): Return a boolean saying if
19617 anything was changed.
19618 (try_shrink_wrapping_separate): Iterate spread_components until
19619 nothing changes anymore.
19621 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19623 PR rtl-optimization/85645
19624 * regrename.c (build_def_use): Also kill the chains that include the
19625 destination of a REG_CFA_REGISTER note.
19627 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19629 PR rtl-optimization/85645
19630 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
19631 insn that has a REG_CFA_REGISTER note.
19633 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
19635 * cfgexpand.c (expand_clobber): New function.
19636 (expand_gimple_stmt_1): Use it.
19637 * tree-vect-stmts.c (vect_clobber_variable): New function,
19639 (vectorizable_simd_clone_call): ...here.
19640 (vectorizable_store): Emit a clobber either side of an
19641 IFN_STORE_LANES sequence.
19642 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
19644 2018-05-09 Tom de Vries <tom@codesourcery.com>
19647 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
19648 (define_insn "trap_if_false"): Add exit after trap.
19650 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
19652 PR rtl-optimization/85638
19653 * bb-reorder.c: Include common/common-target.h.
19654 (create_forwarder_block): New function extracted from...
19655 (fix_up_crossing_landing_pad): ...here. Rename into...
19656 (dw2_fix_up_crossing_landing_pad): ...this.
19657 (sjlj_fix_up_crossing_landing_pad): New function.
19658 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
19659 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
19660 from both partitions and exit the loop after one iteration.
19662 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19665 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19667 (Basic PowerPC Built-in Functions): The new name of the
19668 subsection previously known as "PowerPC Built-in Functions".
19669 (Basic PowerPC Built-in Functions Available on all Configurations):
19671 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
19673 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
19675 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
19677 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
19680 2018-05-08 Jim Wilson <jimw@sifive.com>
19682 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
19683 (LD_EMUL_SUFFIX): New.
19684 (LINK_SPEC): Use it.
19686 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19688 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19690 (Basic PowerPC Built-in Functions): The new name of the
19691 subsection previously known as "PowerPC Built-in Functions".
19692 (Basic PowerPC Built-in Functions Available on all Configurations):
19694 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
19696 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
19698 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
19700 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
19703 2018-05-08 Jakub Jelinek <jakub@redhat.com>
19706 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
19707 after cmpelim optimization.
19709 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
19711 * config.gcc: Support "goldmont".
19712 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
19713 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19714 PROCESSOR_GOLDMONT.
19715 * config/i386/i386.c (m_GOLDMONT): Define.
19716 (processor_target_table): Add "goldmont".
19717 (PTA_GOLDMONT): Define.
19718 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
19719 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
19720 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
19721 (fold_builtin_cpu): Add "goldmont".
19722 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
19723 (ix86_option_override_internal): Add "goldmont".
19724 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
19725 (processor_type): Add PROCESSOR_GOLDMONT.
19726 * config/i386/i386.md: Add CPU "glm".
19727 * config/i386/glm.md: New file.
19728 * config/i386/x86-tune.def: Add m_GOLDMONT.
19729 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
19731 2018-05-08 Jakub Jelinek <jakub@redhat.com>
19734 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
19736 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
19737 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
19738 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
19741 * config/i386/i386.c (ix86_fold_builtin): Handle
19742 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
19745 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
19746 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
19748 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
19751 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
19752 (check_arch): Likewise.
19753 (check_fpu): Return the result rather than printing it.
19754 (end arch): Fix operator precedence.
19755 (end cpu): Likewise.
19756 (END): Print the result from check_fpu.
19758 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
19759 Alan Hayward <alan.hayward@arm.com>
19760 David Sherwood <david.sherwood@arm.com>
19762 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
19763 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
19764 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
19765 (*fcmuo<mode>_and): New patterns.
19767 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
19769 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
19770 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
19771 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
19772 (cmp_op, sve_imm_con): New code attributes.
19773 (SVE_COND_INT_CMP, imm_con): Delete.
19774 (cmp_op): Remove above unspecs from int attribute.
19775 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
19777 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
19778 comparison-specific unspecs.
19779 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
19780 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
19781 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
19782 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
19783 (*vec_fcm<cmp_op><mode>): Rename to...
19784 (*fcm<cmp_op><mode>): ...this and adjust likewise.
19785 (*vec_fcmuo<mode>): Rename to...
19786 (*fcmuo<mode>): ...this and adjust likewise.
19787 (*pred_fcm<cmp_op><mode>): New pattern.
19788 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
19789 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
19791 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
19793 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
19794 (aarch64_emit_sve_predicated_cond): New function.
19795 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
19796 (aarch64_emit_unspec_cond_or): Replace with...
19797 (aarch64_emit_sve_or_conds): ...this new function. Use
19798 aarch64_emit_sve_ptrue_op for the individual comparisons and
19799 aarch64_emit_binop to OR them together.
19800 (aarch64_emit_inverted_unspec_cond): Replace with...
19801 (aarch64_emit_sve_inverted_cond): ...this new function. Use
19802 aarch64_emit_sve_ptrue_op for the comparison and
19803 aarch64_emit_unop to invert the result.
19804 (aarch64_expand_sve_vec_cmp_float): Update after the above
19805 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
19807 2018-05-07 Nathan Sidwell <nathan@acm.org>
19809 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
19810 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
19811 (Backwards Compatibility): Likewise.
19813 2018-05-07 Luis Machado <luis.machado@linaro.org>
19817 2018-05-07 Luis Machado <luis.machado@linaro.org>
19819 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19820 <prefetch_dynamic_strides>: New const bool field.
19821 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19822 prefetch_dynamic_strides.
19823 (exynosm1_prefetch_tune): Likewise.
19824 (thunderxt88_prefetch_tune): Likewise.
19825 (thunderx_prefetch_tune): Likewise.
19826 (thunderx2t99_prefetch_tune): Likewise.
19827 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
19829 (aarch64_override_options_internal): Update to set
19830 PARAM_PREFETCH_DYNAMIC_STRIDES.
19831 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19832 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19833 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19834 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19835 prefetch-dynamic-strides setting.
19837 2018-05-07 Luis Machado <luis.machado@linaro.org>
19839 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19840 <minimum_stride>: New const int field.
19841 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19842 minimum_stride field.
19843 (exynosm1_prefetch_tune): Likewise.
19844 (thunderxt88_prefetch_tune): Likewise.
19845 (thunderx_prefetch_tune): Likewise.
19846 (thunderx2t99_prefetch_tune): Likewise.
19847 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
19848 (aarch64_override_options_internal): Update to set
19849 PARAM_PREFETCH_MINIMUM_STRIDE.
19850 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
19851 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
19852 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
19853 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
19854 stride is constant and is below the minimum stride threshold.
19856 2018-05-07 Luis Machado <luis.machado@linaro.org>
19858 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
19861 2018-05-07 Luis Machado <luis.machado@linaro.org>
19863 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19864 <prefetch_dynamic_strides>: New const bool field.
19865 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19866 prefetch_dynamic_strides.
19867 (exynosm1_prefetch_tune): Likewise.
19868 (thunderxt88_prefetch_tune): Likewise.
19869 (thunderx_prefetch_tune): Likewise.
19870 (thunderx2t99_prefetch_tune): Likewise.
19871 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
19873 (aarch64_override_options_internal): Update to set
19874 PARAM_PREFETCH_DYNAMIC_STRIDES.
19875 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19876 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19877 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19878 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19879 prefetch-dynamic-strides setting.
19881 2018-05-07 Luis Machado <luis.machado@linaro.org>
19883 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19884 <minimum_stride>: New const int field.
19885 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19886 minimum_stride field.
19887 (exynosm1_prefetch_tune): Likewise.
19888 (thunderxt88_prefetch_tune): Likewise.
19889 (thunderx_prefetch_tune): Likewise.
19890 (thunderx2t99_prefetch_tune): Likewise.
19891 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
19892 (aarch64_override_options_internal): Update to set
19893 PARAM_PREFETCH_MINIMUM_STRIDE.
19894 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
19895 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
19896 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
19897 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
19898 stride is constant and is below the minimum stride threshold.
19900 2018-05-06 Jakub Jelinek <jakub@redhat.com>
19903 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
19904 the type is addressable. Don't force op into register if it has
19907 2018-05-05 Roland McGrath <mcgrathr@google.com>
19910 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
19911 any section for which we don't know a specific type it should have,
19912 regardless of name. Previously this was done only for the exact
19913 names ".init_array", ".fini_array", and ".preinit_array".
19914 (default_elf_asm_named_section): Add comment about
19915 relationship with default_section_type_flags and SECTION_NOTYPE.
19916 (get_section): Don't consider it a type conflict if one side has
19917 SECTION_NOTYPE and the other doesn't, as long as neither has the
19918 SECTION_BSS et al used in the default_section_type_flags logic.
19920 2018-05-05 Tom de Vries <tom@codesourcery.com>
19923 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
19924 (workaround_barsyncs): New function.
19925 (nvptx_reorg): Use workaround_barsyncs.
19926 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19927 (define_expand "nvptx_membar_cta"): New define_expand.
19928 (define_insn "*nvptx_membar_cta"): New insn.
19930 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19932 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
19933 To improve optimization opportunities.
19934 * builtin-types.def: The new needed builtin types for the above.
19936 2018-05-04 Richard Biener <rguenther@suse.de>
19938 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
19939 * gimple-ssa-store-merging.c
19940 (imm_store_chain_info::output_merged_store): Remove redundant create,
19941 release split_store vector contents on failure.
19942 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
19943 scalar stmt vector on cache hit.
19945 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
19947 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19949 * config.gcc (powerpc-xilinx-eabi*): Remove.
19950 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
19952 (fusion_addis_mem_combo_load): Ditto.
19953 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
19955 (rs6000_cpu_cpp_builtins): Ditto.
19956 * config/rs6000/rs6000-linux.c
19957 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
19958 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
19959 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
19961 (rs6000_setup_reg_addr_masks): Ditto.
19962 (rs6000_init_hard_regno_mode_ok): Ditto.
19963 (rs6000_option_override_internal): Ditto.
19964 (legitimate_lo_sum_address_p): Ditto.
19965 (rs6000_legitimize_address): Ditto.
19966 (rs6000_legitimize_reload_address): Ditto.
19967 (rs6000_legitimate_address_p): Ditto.
19968 (abi_v4_pass_in_fpr): Ditto.
19969 (setup_incoming_varargs): Ditto.
19970 (rs6000_gimplify_va_arg): Ditto.
19971 (rs6000_split_multireg_move): Ditto.
19972 (rs6000_savres_strategy): Ditto.
19973 (rs6000_emit_prologue_components): Ditto.
19974 (rs6000_emit_epilogue_components): Ditto.
19975 (rs6000_emit_prologue): Ditto.
19976 (rs6000_emit_epilogue): Ditto.
19977 (rs6000_elf_file_end): Ditto.
19978 (rs6000_function_value): Ditto.
19979 (rs6000_libcall_value): Ditto.
19980 * config/rs6000/rs6000.h: Ditto.
19981 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
19982 (TARGET_MINMAX): ... this. New.
19983 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
19984 * config/rs6000/rs6000.md: Remove Xilinx FP support.
19985 (*movsi_internal1_single): Delete.
19986 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
19987 mfpu=, mxilinx-fpu): Delete.
19988 * config/rs6000/singlefp.h: Delete.
19989 * config/rs6000/sysv4.h: Remove Xilinx FP support.
19990 * config/rs6000/t-rs6000: Ditto.
19991 * config/rs6000/t-xilinx: Delete.
19992 * config/rs6000/titan.md: Adjust for fp_type removal.
19993 * config/rs6000/vsx.md: Remove Xilinx FP support.
19994 (VStype_simple): Delete.
19995 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
19996 * config/rs6000/xfpu.h: Delete.
19997 * config/rs6000/xfpu.md: Delete.
19998 * config/rs6000/xilinx.h: Delete.
19999 * config/rs6000/xilinx.opt: Delete.
20000 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
20001 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
20003 2018-05-04 Tom de Vries <tom@codesourcery.com>
20006 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
20009 2018-05-04 Richard Biener <rguenther@suse.de>
20011 PR middle-end/85627
20012 * tree-complex.c (update_complex_assignment): We are always in SSA form.
20013 (expand_complex_div_wide): Likewise.
20014 (expand_complex_operations_1): Likewise.
20015 (expand_complex_libcall): Preserve EH info of the original stmt.
20016 (tree_lower_complex): Handle removed blocks.
20017 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
20018 on complex multiplication and division libcall builtins.
20020 2018-05-04 Richard Biener <rguenther@suse.de>
20022 PR middle-end/85574
20023 * fold-const.c (negate_expr_p): Restrict negation of operand
20024 zero of a division to when we know that can happen without
20026 (fold_negate_expr_1): Likewise.
20028 2018-05-04 Jakub Jelinek <jakub@redhat.com>
20031 * real.h (real_nextafter): Declare.
20032 * real.c (real_nextafter): New function.
20033 * fold-const-call.c (fold_const_nextafter): New function.
20034 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
20035 CASE_CFN_NEXTTOWARD.
20036 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
20037 even when arg1_mode is different from arg0_mode.
20039 2018-05-03 Nathan Sidwell <nathan@acm.org>
20041 * doc/extend.texi (Deprecated Features): Remove
20042 -ffriend-injection.
20043 (Backwards Compatibility): Likewise.
20044 * doc/invoke.texi (C++ Language Options): Likewise.
20045 (C++ Dialect Options): Likewise.
20047 2018-05-03 Jakub Jelinek <jakub@redhat.com>
20050 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
20051 _mm512_mask_mullox_epi64): New intrinsics.
20053 2018-05-03 Tom de Vries <tom@codesourcery.com>
20056 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20057 dump files): Add offload-tree.
20059 2018-05-03 Richard Biener <rguenther@suse.de>
20061 PR tree-optimization/85615
20062 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
20063 to loops not nested in BBs loop father to avoid creating multi-entry
20066 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20068 PR tree-optimization/70291
20069 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
20070 arguments. Change return type to tree. Emit libcall as a new
20071 statement rather than replacing existing one when inplace_p is true.
20072 (expand_complex_multiplication_components): New function.
20073 (expand_complex_multiplication): Expand floating-point complex
20074 multiplication using the above.
20075 (expand_complex_division): Rename inner_type parameter to type.
20076 Update expand_complex_libcall call-site.
20077 (expand_complex_operations_1): Update expand_complex_multiplication
20078 and expand_complex_division call-sites.
20080 2018-05-02 Jakub Jelinek <jakub@redhat.com>
20083 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
20084 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
20085 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
20086 the highest significant bit of the shift count mask is clear. In
20087 check whether and[sq]i3 is needed verify that all significant bits
20088 of the shift count other than the highest are set.
20090 2018-05-02 Tom de Vries <tom@codesourcery.com>
20093 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
20094 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
20095 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
20096 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
20097 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
20098 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
20099 __builtin_goacc_parlevel_size.
20101 2018-05-02 Richard Biener <rguenther@suse.de>
20103 PR tree-optimization/85597
20104 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
20105 do not use split vect_get_vec_defs call but call vect_get_slp_defs
20108 2018-05-02 Tom de Vries <tom@codesourcery.com>
20111 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20112 dump files): Add ltrans-tree.
20114 2018-05-02 Tom de Vries <tom@codesourcery.com>
20117 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20118 dump files): Add wpa-ipa.
20120 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
20122 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
20123 powerpc*-*-linux*paired* target.
20124 * config/rs6000/750cl.h: Delete.
20125 * config/rs6000/paired.h: Delete.
20126 * config/rs6000/paired.md: Delete.
20127 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
20129 * config/rs6000/rs6000-builtin.def: Remove paired float support.
20130 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20131 comment. Remove paired float support.
20132 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
20133 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
20135 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
20136 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
20138 * config/rs6000/rs6000.c: Remove paired float support.
20139 (paired_expand_vector_init, paired_expand_vector_move,
20140 paired_emit_vector_compare, paired_emit_vector_cond_expr,
20141 (paired_expand_lv_builtin, paired_expand_stv_builtin,
20142 paired_expand_builtin, paired_expand_predicate_builtin,
20143 paired_init_builtins): Delete.
20144 * config/rs6000/rs6000.h: Remove paired float support.
20145 * config/rs6000/rs6000.md: Remove paired float support.
20146 (move_from_CR_ov_bit): Delete.
20147 * config/rs6000/rs6000.opt (mpaired): Delete.
20148 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
20149 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
20151 2018-05-02 Richard Biener <rguenther@suse.de>
20153 PR middle-end/85567
20154 * gimplify.c (gimplify_save_expr): When in SSA form allow
20155 SAVE_EXPRs to compute to SSA vars.
20157 2018-05-02 Jakub Jelinek <jakub@redhat.com>
20160 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
20161 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
20162 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
20163 clobber operands[2], instead use a new pseudo. Formatting fixes.
20165 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
20167 PR tree-optimization/85586
20168 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
20169 exit early for statements in the same group if the accesses are
20172 2018-05-02 Tom de Vries <tom@codesourcery.com>
20175 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
20178 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
20180 PR tree-optimization/85143
20181 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
20183 2018-05-01 Tom de Vries <tom@codesourcery.com>
20186 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
20187 not found" error message.
20189 2018-05-01 Tom de Vries <tom@codesourcery.com>
20192 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
20193 * vec.c (test_ordered_remove_if): New function.
20194 (vec_c_tests): Call test_ordered_remove_if.
20195 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
20196 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
20197 * tree-vect-patterns.c (vect_pattern_recog_1): Use
20198 VEC_ORDERED_REMOVE_IF.
20200 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20202 PR tree-optimization/82665
20203 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
20204 pointer subtraction where arguments come from a memchr call.
20206 2018-05-01 Jakub Jelinek <jakub@redhat.com>
20208 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
20209 --push-state --as-needed and --pop-state instead of --as-needed and
20210 --no-as-needed if ld supports it.
20211 * configure: Regenerated.
20214 * doc/install.texi2html: Replace _002d with - and _002a with * in
20215 generated html files using sed.
20217 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20220 * gcc-rich-location.c (blank_line_before_p): New function.
20221 (use_new_line): New function.
20222 (gcc_rich_location::add_fixit_insert_formatted): New function.
20223 * gcc-rich-location.h
20224 (gcc_rich_location::add_fixit_insert_formatted): New function.
20226 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20228 * selftest.c (assert_streq): Rename "expected" and "actual" to
20229 "val1" and "val2". Extend NULL-handling to cover both inputs
20230 symmetrically, while still requiring both to be non-NULL for a pass.
20231 * selftest.h (assert_streq): Rename "expected" and "actual" to
20233 (ASSERT_EQ): Likewise.
20234 (ASSERT_EQ_AT): Likewise.
20235 (ASSERT_KNOWN_EQ): Likewise.
20236 (ASSERT_KNOWN_EQ_AT): Likewise.
20237 (ASSERT_NE): Likewise.
20238 (ASSERT_MAYBE_NE): Likewise.
20239 (ASSERT_MAYBE_NE_AT): Likewise.
20240 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
20241 the assertion to pass.
20242 (ASSERT_STREQ_AT): Likewise.
20244 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
20246 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
20247 interaction with -pie.
20249 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20251 * selftest.h: Fix alphabetization of per-source-file selftest
20254 2018-04-30 Jason Merrill <jason@redhat.com>
20256 PR c++/61982 - dead stores to destroyed objects.
20257 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
20260 2018-04-30 Jason Merrill <jason@redhat.com>
20262 * tree.c (build_clobber): New.
20263 * tree.h: Declare it.
20264 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
20266 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20268 * diagnostic-show-locus.c (layout::layout): Update for
20269 location_get_source_line returning a char_span.
20270 (struct char_span): Move to input.h.
20271 (struct correction): Update for fields in char_span becoming
20273 (struct source_line): Update for location_get_source_line
20274 returning a char_span.
20275 (layout::print_line): Likewise.
20276 * edit-context.c (edited_file::print_content): Likewise.
20277 (edited_file::print_diff_hunk): Likewise.
20278 (edited_file::print_run_of_changed_lines): Likewise.
20279 (edited_file::get_num_lines): Likewise.
20280 (edited_line::edited_line): Likewise.
20281 * final.c (asm_show_source): Likewise.
20282 * input.c (location_get_source_line): Convert return type
20283 from const char * to char_span, losing the final "line_len"
20285 (dump_location_info): Update for the above.
20286 (get_substring_ranges_for_loc): Likewise. Use a char_span
20287 when handling the literal within the line.
20288 (test_reading_source_line): Update for location_get_source_line
20289 returning a char_span.
20290 * input.h (class char_span): Move here from
20291 diagnostic-show-locus.c, converting from a struct to a class.
20292 Make data members private.
20293 (char_span::operator bool): New.
20294 (char_span::length): New.
20295 (char_span::get_buffer): New.
20296 (char_span::operator[]): New.
20297 (char_span::subspan): Make const.
20298 (char_span::xstrdup): New.
20299 (location_get_source_line): Convert return type from const char *
20300 to char_span, losing the final "line_size" param.
20302 2018-04-30 Jan Hubicka <jh@suse.cz>
20304 * lto-wrapper.c (ltrans_priorities): New static var.
20305 (cmp_priority): New.
20306 (run_gcc): Read priorities and if doing parallel build order
20307 the Makefile by them.
20309 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20311 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
20313 2018-04-30 Richard Biener <rguenther@suse.de>
20315 * tree-cfg.c (verify_address): Remove base argument, add
20316 flag whether to check TREE_ADDRESSABLE and do that.
20317 (verify_expr): Remove.
20318 (verify_types_in_gimple_reference): Add pieces from verify_expr.
20319 (verify_gimple_assign_single): Likewise.
20320 (verify_gimple_switch): Likewise.
20321 (verify_expr_location_1): Dereference tp once. Add (disabled)
20322 piece from verify_expr.
20323 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
20325 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20327 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
20329 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20331 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
20332 (small_data_pattern): Likewise.
20333 (arc_rewrite_small_data): Likewise.
20334 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
20335 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
20336 (get_symbol_alignment): New function.
20337 (legitimate_small_data_address_p): Likewise.
20338 (legitimate_scaled_address): Update, call
20339 legitimate_small_data_address_p.
20340 (output_sdata): New static variable.
20341 (arc_print_operand): Update how we handle small data operands.
20342 (arc_print_operand_address): Likewise.
20343 (arc_legitimate_address_p): Update, use
20344 legitimate_small_data_address_p.
20345 (arc_rewrite_small_data_p): Remove.
20346 (arc_rewrite_small_data_1): Likewise.
20347 (arc_rewrite_small_data): Likewise.
20348 (small_data_pattern): Likewise.
20349 (compact_sda_memory_operand): Update to use
20350 legitimate_small_data_address_p and get_symbol_alignment.
20351 (prepare_move_operands): Don't rewite sdata pattern.
20352 (prepare_extend_operands): Remove.
20353 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
20355 (zero_extendqisi2): Likewise.
20356 (zero_extendhisi2): Likewise.
20357 (extendqihi2): Likewise.
20358 (extendqisi2): Likewise.
20359 (extendhisi2): Likewise.
20360 (addsi3): Likewise.
20361 (subsi3): Likewise.
20362 (andsi3): Likewise.
20363 * config/arc/constraints.md (Usd): Change it to memory constraint.
20365 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20367 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
20368 as source of std instructions.
20369 * config/arc/arc.md (movsi_insn): Update pattern predicate to
20370 allow 6-bit constants as source for store instructions.
20371 (movdi_insn): Update instruction pattern to allow 6-bit constants
20372 as source for store instructions.
20374 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
20376 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
20378 2018-04-30 Nathan Sidwell <nathan@acm.org>
20379 Sandra Loosemore <sandra@codesourcery.com>
20381 * dumpfile.c (dump_open): Allow '-' for stdout.
20382 * doc/invoke.texi (Developer Options): Document dump filename
20383 determination early. Document stdin/stdout selection.
20385 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
20387 Microblaze Target: PIC data text relative
20389 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
20390 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
20392 * config/microblaze/microblaze.h (microblaze_constant_address_p):
20393 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
20394 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
20395 New addressing mode for data-text relative position indepenedent code.
20396 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
20397 'ADDRESS_SYMBOLIC_TXT_REL'.
20398 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
20399 (microblaze_legitimate_pic_operand): Exclude function calls from
20400 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
20401 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
20403 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
20404 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
20405 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
20406 for 'address + offset'.
20407 (microblaze_expand_prologue): Add new function prologue call for
20409 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
20410 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
20411 table in case of TARGET_PIC_DATA_TEXT_REL.
20412 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
20413 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
20414 Add new macros 'UNSPEC_TEXT',
20415 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
20416 + exclude function calls from 'UNSPEC_PLT' in case of data text
20418 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
20419 new target hook for generating address diff vector tables in case of
20421 * doc/tm.texi : Regenerate.
20422 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
20423 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
20424 of addr diff vector generation.
20425 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
20426 target hook definition.
20427 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
20428 Add default function for generate_pic_addr_diff_vec -> flag_pic.
20429 * doc/invoke.texi (Add new pic option): Add new microblaze pic
20430 option for data text relative.
20432 2018-04-30 Richard Biener <rguenther@suse.de>
20434 * tree-chrec.h (evolution_function_is_constant_p): Remove
20436 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
20438 2018-04-30 Richard Biener <rguenther@suse.de>
20441 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
20443 2018-04-30 Richard Biener <rguenther@suse.de>
20445 PR tree-optimization/28364
20446 PR tree-optimization/85275
20447 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
20448 copying first exit test.
20450 2018-04-28 Mark Wielaard <mark@klomp.org>
20452 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
20453 dwarf_version >= 5.
20454 (dwarf_AT): Handle DW_AT_addr_base.
20455 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
20457 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
20460 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
20461 (*ashl<dwi>3_doubleword_mask_1): Ditto.
20462 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
20463 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
20465 2018-04-28 Richard Biener <rguenther@suse.de>
20467 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
20468 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
20469 to reflect use. Only add interesting stmts.
20471 2018-04-27 Martin Jambor <mjambor@suse.cz>
20474 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
20475 the jump function allows for passing through aggregate values.
20477 2018-04-27 David Malcolm <dmalcolm@redhat.com>
20479 * input.h (in_system_header_at): Convert from macro to inline
20481 (from_macro_expansion_at): Likewise.
20482 (from_macro_definition_at): Likewise.
20484 2018-04-27 Jeff Law <law@redhat.com>
20486 * config.gcc: Mark tile* targets as deprecated/obsolete.
20488 2018-04-27 Richard Biener <rguenther@suse.de>
20490 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
20493 2018-04-27 Richard Biener <rguenther@suse.de>
20495 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
20497 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
20499 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
20500 with Yd constraint. Set "preferred_for_speed" attribute from
20501 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
20502 with Yd constraint.
20503 (*movdi_internal): Ditto.
20504 (movti_interunit splitters): Remove
20505 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
20506 (movdi_interunit splitters): Ditto.
20507 * config/i386/constraints.md (Ye): Remove.
20508 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
20510 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20513 * config/aarch64/constraints.md (Usg): Limit to 31.
20514 (Usj): Limit to 63.
20516 2018-04-27 Jakub Jelinek <jakub@redhat.com>
20518 PR tree-optimization/85529
20519 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
20520 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
20521 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
20522 zero extension or masking of the MSB bit.
20523 (optimize_range_tests): Add FIRST_BB argument, pass it through
20524 to optimize_range_tests_var_bound.
20525 (maybe_optimize_range_tests, reassociate_bb): Adjust
20526 optimize_range_tests callers.
20528 2018-04-26 Richard Biener <rguenther@suse.de>
20529 Jakub Jelinek <jakub@redhat.com>
20531 * cgraph.h (symbol_table): Just declare debug method here.
20532 * symtab.c (symbol_table::debug): Define.
20534 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
20536 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
20538 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
20540 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
20541 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
20542 (*movdi_internal): Substitute Yi and Yj constraint with x
20543 and Ym and Yn constraint with y constraint. Update "isa"
20544 attribute and set "preferred_for_speed" attribute from
20545 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
20546 (*movsi_internal): Ditto.
20547 (*movdf_internal): Ditto.
20548 (*movsf_internal): Ditto.
20549 (*zero_extendsidi2): Ditto.
20550 * config/i386/sse.md (vec_set<mode>_0): Ditto.
20551 (sse2_loadld): Ditto.
20552 (*vec_extract<ssevecmodelower>_0): Ditto.
20553 (*vec_extractv4si_0_zext_sse4): Ditto.
20554 (vec_concatv2di): Ditto.
20555 (*vec_dup<mode>): Ditto.
20556 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
20557 * config/i386/constraints.md (Yi): Remove.
20562 2018-04-26 Nathan Sidwell <nathan@acm.org>
20564 * dumpfile.c (dump_open): New.
20565 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
20566 (dump_finish): Detect stdio/stderr by value not name.
20568 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
20570 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
20572 2018-04-26 Tom de Vries <tom@codesourcery.com>
20575 * config/nvptx/nvptx.c (verify_neutering_jumps)
20576 (verify_neutering_labels): New function
20577 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
20579 2018-04-26 Tom de Vries <tom@codesourcery.com>
20582 * config/nvptx/nvptx.c (needs_neutering_p): New function.
20583 (nvptx_single): Use needs_neutering_p to skip over insns that do not
20586 2018-04-26 Richard Biener <rguenther@suse.de>
20587 Tom de Vries <tom@codesourcery.com>
20590 * lto-streamer-out.c (output_function): Fixup loops if required to match
20591 discovery done in the reader.
20593 2018-04-26 Richard Biener <rguenther@suse.de>
20595 PR tree-optimization/85116
20596 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
20597 have a loop exit from the single latch predecessor. Remove
20598 case of header with just condition.
20599 (ch_base::copy_headers): Exclude infinite loops from any
20601 (pass_ch::execute): Record exits.
20603 2018-04-26 Richard Biener <rguenther@suse.de>
20605 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
20606 prologue cost vector and pass it to vect_get_load_cost.
20607 (vect_get_peeling_costs_all_drs): Likewise.
20608 (vect_peeling_hash_get_lowest_cost): Likewise.
20609 (vect_enhance_data_refs_alignment): Likewise.
20611 2018-04-26 Richard Biener <rguenther@suse.de>
20613 PR middle-end/85450
20614 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
20615 checking of integer<->pointer conversions.
20616 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
20617 sign-/zero-extending pointer types.
20618 (expand_omp_for_static_chunk): Likewise.
20620 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
20621 Jean Lee <xiaoyur347@gmail.com>
20623 * config/mips/mips.c (mips_asan_shadow_offset): New function.
20624 (TARGET_ASAN_SHADOW_OFFSET): Define.
20625 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
20626 true for -fsanitize=address.
20628 2018-04-25 Mark Wielaard <mark@klomp.org>
20630 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
20633 2018-04-25 Jakub Jelinek <jakub@redhat.com>
20635 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
20636 than "alu", remove explicit "memory" and "imm_disp" attributes.
20637 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
20639 PR middle-end/85414
20640 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
20641 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
20642 gen_lowpart_no_emit.
20644 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
20647 * config/i386/i386.c (ix86_expand_builtin): Change memory
20648 operand to XI, extend p0 to Pmode.
20649 * config/i386/i386.md: Change unspec volatile and operand
20650 1 mode to XI, change operand 0 mode to P.
20652 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20654 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
20655 GET_MODE_MASK before any checking.
20656 (nds32_can_use_bset_p): Likewise.
20657 (nds32_can_use_btgl_p): Likewise.
20659 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20661 * config/nds32/nds32-doubleword.md: New define_split pattern for
20662 illegal register number.
20664 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20666 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
20668 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20670 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
20672 2018-04-25 Richard Biener <rguenther@suse.de>
20674 * lto-streamer.h (LTO_major_version): Bump to 8.
20676 2018-04-25 Jakub Jelinek <jakub@redhat.com>
20678 * BASE-VER: Set to 9.0.0.
20680 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
20682 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
20683 in __abskf2 and __powikf2.
20685 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20688 * config/aarch64/constraints.md (Usg, Usj): New constraints.
20689 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
20690 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
20691 Use the above on operand 2. Reindent.
20692 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
20694 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
20697 * common/config/i386/i386-common.c (ix86_handle_option): Don't
20699 * config/i386/i386.opt (mcet): Removed.
20700 * doc/install.texi: Remove -mcet documentation.
20701 * doc/invoke.texi: Likewise.
20703 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
20706 * doc/install.texi: Remove -mcet from bootstrap-cet.
20708 2018-04-24 Jakub Jelinek <jakub@redhat.com>
20711 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
20712 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
20716 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
20717 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
20718 containing a CONST_VECTOR.
20720 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
20722 * doc/install.texi: Update newlib dependency for nvptx.
20724 2018-04-24 Jakub Jelinek <jakub@redhat.com>
20727 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
20728 instead of INTVAL when shifting x left.
20730 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
20732 PR tree-optimization/85478
20733 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
20734 vect_grouped_store_supported for single element vectors.
20736 2018-04-24 Richard Biener <rguenther@suse.de>
20739 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
20740 load cost increase to the case of non-constant step.
20742 2018-04-24 Jakub Jelinek <jakub@redhat.com>
20745 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
20746 destination if any_malformed_asm.
20748 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
20750 PR middle-end/85496
20751 * expr.c (store_field): In the bitfield case, if the value comes from
20752 a function call and is returned in registers by means of a PARALLEL,
20753 do not change the mode of the temporary unless BLKmode and VOIDmode.
20755 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
20757 PR rtl-optimization/85423
20758 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
20759 dependencies to debug insns when the previous insn is non-debug.
20761 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
20763 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
20764 enums into a single definition.
20765 (fls): Fix predicates and printing.
20768 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
20770 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
20771 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
20772 and short u6 immediate.
20773 (check_if_valid_sleep_operand): Remove.
20774 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
20776 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20778 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
20779 flag_always_save_lp condition.
20780 * config/nds32/nds32.opt (malways-save-lp): New option.
20782 2018-04-22 Shiva Chen <shiva0217@gmail.com>
20784 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
20785 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
20786 * config/nds32/nds32.h
20787 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
20788 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
20790 2018-04-22 Shiva Chen <shiva0217@gmail.com>
20792 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
20793 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
20795 2018-04-22 Shiva Chen <shiva0217@gmail.com>
20796 Chung-Ju Wu <jasonwucj@gmail.com>
20798 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
20800 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
20801 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
20803 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20805 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
20807 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20809 * config/nds32/nds32-protos.h (nds32_data_alignment,
20810 nds32_local_alignment): Declare.
20811 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
20812 nds32_local_alignment): New functions.
20813 (TARGET_CONSTANT_ALIGNMENT): Define.
20814 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
20816 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20818 * config/nds32/nds32.c
20819 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
20820 (TARGET_MODES_TIEABLE_P): Likewise.
20822 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20824 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
20825 level Ofast and Og.
20827 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
20828 Chung-Ju Wu <jasonwucj@gmail.com>
20830 * config/nds32/constants.md (unspec_volatile_element): Add enum values
20831 for unaligned access.
20832 * config/nds32/nds32-intrinsic.c: Implementation of expanding
20834 * config/nds32/nds32-intrinsic.md: Likewise.
20835 * config/nds32/nds32_intrinsic.h: Likewise.
20836 * config/nds32/nds32.h (nds32_builtins): Likewise.
20837 * config/nds32/nds32.opt (munaligned-access): New option.
20838 * config/nds32/nds32.c (nds32_asm_file_start): Display
20839 flag_unaligned_access status.
20841 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
20843 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
20844 -mno-relax is present.
20845 * config/riscv/linux.h (LINK_SPEC): Ditto.
20847 2018-04-20 Martin Sebor <msebor@redhat.com>
20850 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
20852 (gimple_fold_builtin_stxcpy_chk): Same.
20853 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
20855 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
20858 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
20859 __powikf2 when long double is IEEE 128-bit.
20861 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
20863 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
20864 step to make sure stack always aligned.
20866 2018-04-20 Carl Love <cel@us.ibm.com>
20869 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
20870 size check for arg0.
20872 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
20873 Tom de Vries <tom@codesourcery.com>
20876 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
20877 Emit insns for calls too.
20878 (nvptx_find_par): Always look for worker-level predecessor insn.
20879 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
20881 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
20882 (nvptx_process_pars): Propagate frames for calls.
20884 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
20887 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
20889 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
20890 (ix86_handle_option): Don't handle OPT_mibt.
20891 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
20893 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
20895 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
20896 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
20897 (ix86_target_macros): Define __CET__ with flag_cf_protection
20898 for -fcf-protection.
20899 * config/i386/i386.c (isa2_opts): Remove -mibt.
20900 * config/i386/i386.h (TARGET_IBT): Removed.
20901 (TARGET_IBT_P): Likewise.
20902 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
20903 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
20904 * config/i386/i386.opt (mcet): Update help message.
20905 (mshstk): Likewise.
20907 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
20908 -mcet as an alias for -mshstk.
20910 2018-04-20 Richard Biener <rguenther@suse.de>
20912 PR middle-end/85475
20913 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
20914 complexity by forcing a single use of the multiply operand.
20916 2018-04-20 Martin Jambor <mjambor@suse.cz>
20919 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
20920 recursion dependency to only apply to non-clones.
20922 2018-04-20 Martin Jambor <mjambor@suse.cz>
20925 * ipa-cp.c (create_specialized_node): Check that clones of
20926 self-recursive edges exist during IPA-CP.
20928 2018-04-19 Toon Moene <toon@moene.org>
20930 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
20933 2018-04-19 Jakub Jelinek <jakub@redhat.com>
20935 PR tree-optimization/85467
20936 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
20937 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
20938 VECTOR_CST element to type.
20940 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
20943 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
20944 * config/i386/i386.md (builtin_setjmp_setup): Removed.
20945 (builtin_longjmp): Likewise.
20946 (save_stack_nonlocal): New pattern.
20947 (restore_stack_nonlocal): Likewise.
20949 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
20952 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
20953 Replace ASM_OUTPUT_LABEL with fprintf.
20955 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
20958 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
20959 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
20960 * config/i386/i386-c.c (ix86_target_macros_internal): Also
20961 define __IBT__ and __SHSTK__ for -fcf-protection.
20962 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
20964 (ix86_trampoline_init): Likewise.
20965 (x86_output_mi_thunk): Likewise.
20966 (ix86_notrack_prefixed_insn_p): Likewise.
20967 (ix86_option_override_internal): Don't disallow -fcf-protection.
20968 * config/i386/i386.md (rdssp<mode>): Also enable for
20970 (incssp<mode>): Likewise.
20971 (nop_endbr): Likewise.
20972 * config/i386/i386.opt (mcet): Change help message to built-in
20975 (mshstk): Likewise.
20976 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
20977 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
20978 enable CET built-in functions.
20980 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
20982 * common/config/i386/i386-common.c
20983 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
20984 OPTION_MASK_ISA_MOVDIRI_UNSET,
20985 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
20986 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
20987 * config.gcc (movdirintrin.h): New header.
20988 * config/i386/cpuid.h (bit_MOVDIRI,
20989 bit_MOVDIR64B): New bits.
20990 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
20992 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
20993 (VOID, PVOID, PCVOID)): New function types.
20994 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
20995 __builtin_ia32_directstoreu_u64,
20996 __builtin_ia32_movdir64b): New builtins.
20997 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
20998 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
21000 (ix86_valid_target_attribute_inner_p): Ditto.
21001 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
21002 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
21003 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
21004 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
21005 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
21006 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
21007 (movdiri<mode>, movdir64b_<mode>): New.
21008 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
21009 * config/i386/immintrin.h: Include movdirintrin.h.
21010 * config/i386/movdirintrin.h: New file.
21011 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
21013 2018-04-19 Richard Biener <rguenther@suse.de>
21015 PR middle-end/85455
21016 * cfg.c (clear_bb_flags): When loop state says we have
21017 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
21019 2018-04-19 Richard Biener <rguenther@suse.de>
21021 PR tree-optimization/84737
21022 * tree-vect-data-refs.c (vect_copy_ref_info): New function
21023 copying restrict info.
21024 (vect_setup_realignment): Use it.
21025 * tree-vectorizer.h (vect_copy_ref_info): Declare.
21026 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
21027 the first DR to all generated stores.
21028 (vectorizable_load): Likewise for loads.
21030 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21032 PR tree-optimization/85446
21033 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
21034 the integral and pointer types to have the same precision.
21036 * doc/install.texi: Document --disable-cet being the default and
21039 2018-04-18 Martin Liska <mliska@suse.cz>
21041 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
21044 2018-04-18 Martin Liska <mliska@suse.cz>
21047 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21050 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21051 arguments if they are comparable.
21053 2018-04-18 Martin Liska <mliska@suse.cz>
21056 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21059 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21062 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
21065 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
21066 ENDBR after calling __morestack.
21068 2018-04-18 David Malcolm <dmalcolm@redhat.com>
21071 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
21072 by using gcc_base_ver to generate a gcc_driver_version, and use
21073 it when generating GCC_DRIVER_NAME.
21074 * configure: Regenerate.
21076 2018-04-18 Jakub Jelinek <jakub@redhat.com>
21079 * config.gcc: Obsolete powerpc*-*-*spe*.
21081 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21084 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
21085 (stabstr_D): Change type of unum from unsigned int to
21086 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
21089 2018-04-17 Jim Wilson <jimw@sifive.com>
21092 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
21093 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
21094 Set arg_pointer_offset after using pretend_args_size.
21096 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21098 PR rtl-optimization/85431
21099 * dse.c (record_store): Ignore zero width stores.
21102 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
21103 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
21104 __builtin_stack_restore rather than after it.
21105 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
21106 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
21107 argument instead of virtual_dynamic_stack_rtx.
21109 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
21111 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
21113 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21114 Add note to error message to explain internal mapping of overloaded
21115 built-in function name to non-overloaded built-in function name.
21116 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
21119 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21122 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
21123 where the inputs overlap with the output.
21125 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21127 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
21128 (=v, v) alternative and explicit "memory" attribute.
21129 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
21130 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21132 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
21133 "sselog1" type instead of "sselog".
21134 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
21135 "sselog". Remove explicit "memory" attribute.
21136 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
21137 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21139 (vec_extract_hi_v32hi): Merge all alternatives into one, use
21140 "sselog1" type instead of "sselog". Remove explicit "memory"
21142 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
21143 use "sselog1" type instead of "sselog". Remove explicit "memory"
21145 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
21146 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21148 (vec_extract_hi_v64qi): Merge all alternatives into one, use
21149 "sselog1" type instead of "sselog". Remove explicit "memory"
21151 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
21152 use "sselog1" type instead of "sselog". Remove explicit "memory"
21156 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
21158 PR middle-end/85414
21159 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
21162 2018-04-17 Martin Jambor <mjambor@suse.cz>
21165 * ipa-cp.c (create_specialized_node): Call
21166 expand_all_artificial_thunks if necessary.
21168 2018-04-17 Martin Liska <mliska@suse.cz>
21171 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
21172 in message, remote space in between '_G' and '('.
21174 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21177 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
21178 avx512f_vmcmp<mode>3<round_saeonly_name>,
21179 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
21180 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
21181 avx512f_rndscale<mode><round_saeonly_name>,
21182 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
21183 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
21184 Use %<iptr>2 instead of %2 for -masm=intel.
21185 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
21186 avx512f_vcvttss2usi<round_saeonly_name>,
21187 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
21189 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
21190 avx512f_vcvttsd2usi<round_saeonly_name>,
21191 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
21192 Use %q1 instead of %1 for -masm=intel.
21193 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
21194 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
21195 of %3 for -masm=intel.
21196 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
21198 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
21200 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
21202 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
21203 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
21205 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
21207 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
21208 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
21209 %g1 and one with %0 and %1.
21210 (avx512er_vmrcp28<mode><round_saeonly_name>,
21211 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
21212 %1 for -masm=intel.
21213 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
21214 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
21215 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
21216 of %0 and %{%4%} for -masm=intel.
21217 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
21218 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
21219 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
21220 order of %0 and %{%5%}%{z%} for -masm=intel.
21222 2018-04-17 Jan Hubicka <jh@suse.cz>
21225 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
21227 2018-04-17 Martin Liska <mliska@suse.cz>
21230 * multiple_target.c (create_dispatcher_calls): Set apostrophes
21231 for target_clone error message. Make default implementation
21232 clone to be a local declaration.
21233 (separate_attrs): Add new argument and check for an empty
21235 (expand_target_clones): Handle it.
21236 (ipa_target_clone): Make redirection just for target_clones
21239 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
21240 Tom de Vries <tom@codesourcery.com>
21242 PR middle-end/84955
21243 * omp-expand.c (expand_oacc_for): Add dummy false branch for
21244 tiled basic blocks without omp continue statements.
21246 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
21249 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
21250 vec_extract expression as having side effects to make sure it gets
21253 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
21256 * config/i386/i386.c (get_builtin_code_for_version): Check
21259 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
21262 * config.gcc: Support "skylake".
21263 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21265 * config/i386/i386.c (m_SKYLAKE): Define.
21266 (processor_target_table): Add "skylake".
21267 (ix86_option_override_internal): Add "skylake".
21268 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
21269 PROCESSOR_CANNONLAKE.
21270 (get_builtin_code_for_version): Fix priority for
21271 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
21272 PROCESSOR_SKYLAKE-AVX512.
21273 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
21274 (processor_type): Add PROCESSOR_SKYLAKE.
21276 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
21277 Jason Merrill <jason@redhat.com>
21280 * convert.c (convert_to_integer_1): Use direct recursion for
21281 enumeral types and types with a precision less than the number
21282 of bits in their mode.
21284 2018-04-16 Julia Koval <julia.koval@intel.com>
21287 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
21288 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
21290 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
21293 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
21294 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
21295 and -mno-direct-move.
21297 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
21300 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
21301 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
21302 Ensure negative shifts result in {0}.
21304 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
21306 PR rtl-optimization/79916
21307 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
21308 regs (if any) to define how to gnerate SD moves when LRA is in
21311 2018-04-13 Jakub Jelinek <jakub@redhat.com>
21313 PR rtl-optimization/85393
21314 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
21315 * except.c (expand_dw2_landing_pad_for_region): Make static.
21316 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
21317 a label and unconditional jump to old_bb, rather than
21318 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
21321 PR rtl-optimization/85376
21322 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
21323 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
21324 instead of a specific value.
21326 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
21327 Bin Cheng <bin.cheng@arm.com>
21329 PR tree-optimization/82965
21330 PR tree-optimization/83991
21331 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
21332 by_profile_only parameter.
21333 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
21334 information if the loop was predicted to iterate too many times.
21335 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
21337 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
21340 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
21343 2018-04-13 Martin Liska <mliska@suse.cz>
21344 Jakub Jelinek <jakub@redhat.com>
21346 PR middle-end/81657
21347 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
21348 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
21349 * builtins.c (expand_builtin_memory_copy_args): Use
21350 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
21351 handle dest_addr == pc_rtx.
21353 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21356 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
21357 asked to not generate direct moves.
21358 (fix_trunc<mode>si2_stfiwx): Similar.
21359 (fix_trunc<mode>si2_internal): Similar.
21361 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21364 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
21365 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
21366 lookup if dest in some wider mode is known to be const0_rtx and
21367 if so, record permanent equivalence for it to be ZERO_EXTEND of
21368 the narrower mode destination.
21370 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
21372 * lto-streamer-out.c (output_function): Revert 259346.
21373 * omp-expand.c (expand_oacc_for): Likewise.
21375 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
21377 PR rtl-optimization/85354
21378 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
21379 * sel-sched.c (sel_global_init): ... here.
21381 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
21384 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
21385 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
21386 mode for PE-COFF targets.
21387 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
21388 (i386_pe_asm_lto_end): Likewise.
21389 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
21390 (TARGET_ASM_LTO_END): Likewise.
21391 * config/i386/winnt.c (saved_debug_info_level): New static variable.
21392 (i386_pe_asm_lto_start): New function.
21393 (i386_pe_asm_lto_end): Likewise.
21395 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
21396 Richard Biener <rguenther@suse.de>
21398 PR middle-end/84955
21399 * lto-streamer-out.c (output_function): Fix CFG loop state before
21401 * omp-expand.c (expand_oacc_for): Handle calls to internal
21402 functions like regular functions.
21404 2018-04-12 Richard Biener <rguenther@suse.de>
21407 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
21408 for the early LTO debug to properly generate references to it
21409 during DIE emission. Do not re-use that for the skeleton for
21411 (dwarf2out_early_finish): Likewise.
21413 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21416 * config/i386/sse.md
21417 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
21418 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
21419 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
21420 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
21421 and output is a reg, avoid creating invalid lowpart subreg, but
21422 instead split into a 512-bit move. Don't split if not AVX512VL,
21423 input is xmm16+ reg and output is a mem.
21424 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
21425 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
21426 xmm16+ reg and output is a mem.
21428 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21430 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
21431 also for flag_dwarf2_cfi_asm.
21433 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21435 PR rtl-optimization/85342
21436 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
21437 a bool scalar var inside of the loop instead. Don't try to update
21438 recog_data.operand after failed apply_change_group.
21440 2018-04-12 Tom de Vries <tom@codesourcery.com>
21443 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
21444 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
21445 array with flexible array member as array without given dimension.
21446 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
21447 argument for undefined param to true.
21449 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
21452 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
21453 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
21454 from PowerPC section.
21455 * config/rs6000/sysv4.opt (mcall-): Improve help text.
21456 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
21457 help text that is too long.
21458 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
21459 help text that is too long.
21460 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
21461 help text that is too long.
21463 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
21465 * config/alpha/alpha.md (stack_probe_internal): Rename
21466 from "probe_stack". Update all callers.
21468 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21470 PR rtl-optimization/84566
21471 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
21472 sched_macro_fuse_insns.
21474 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21477 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
21478 (compute_block_dependences): ... from here.
21480 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21482 PR tree-optimization/85331
21483 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
21484 from int to HOST_WIDE_INT.
21486 2018-04-11 Martin Jambor <mjambor@suse.cz>
21489 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
21490 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
21491 not the same as the source val.
21492 (cgraph_edge_brings_value_p): New parameter.
21493 (gather_edges_for_value): Pass destination value to
21494 cgraph_edge_brings_value_p.
21495 (perhaps_add_new_callers): Likewise.
21496 (get_info_about_necessary_edges): Likewise and exclude values brought
21497 only by self-recursive edges.
21498 (create_specialized_node): Redirect only clones of self-calling edges.
21499 (+self_recursive_pass_through_p): New function.
21500 (find_more_scalar_values_for_callers_subset): Use it.
21501 (find_aggregate_values_for_callers_subset): Likewise.
21502 (known_aggs_to_agg_replacement_list): Removed.
21503 (decide_whether_version_node): Re-calculate known constants for all
21504 remaining context clones.
21506 2018-04-11 Richard Biener <rguenther@suse.de>
21509 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
21510 from early DWARF output.
21511 (dwarf2out_early_finish): Output line info unconditionally into
21512 early DWARF and add reference to it.
21514 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21517 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
21518 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
21519 other than V2DFmode using iptr mode attribute.
21520 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
21522 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21524 PR rtl-optimization/84659
21525 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
21527 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21530 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
21532 (output_loc_list): Pass address of a dummy size variable even in the
21533 locview handling loop.
21534 (index_location_lists): Add comment on why skip_loc_list_entry can't
21537 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
21540 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
21543 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
21546 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
21547 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
21548 and -mstring-compare-inline-limit.
21550 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21553 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
21554 for stack clash protection in a register whenever we need it to be in
21557 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21559 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
21560 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
21562 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21565 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
21567 (mlong-double-): Ditto.
21568 * config/rs6000/sysv4.opt (msdata=): Ditto.
21569 (mtls-size=): Ditto.
21571 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21573 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
21574 erroneous entries for
21575 "vector int vec_ldl (int, long int *)", and
21576 "vector unsigned int vec_ldl (int, unsigned long int *)".
21577 Add comments and entries for
21578 "vector bool char vec_ldl (int, bool char *)",
21579 "vector bool short vec_ldl (int, bool short *)",
21580 "vector bool int vec_ldl (int, bool int *)",
21581 "vector bool long long vec_ldl (int, bool long long *)",
21582 "vector pixel vec_ldl (int, pixel *)",
21583 "vector long long vec_ldl (int, long long *)",
21584 "vector unsigned long long vec_ldl (int, unsigned long long *)".
21585 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
21586 type tree bool_long_long_type_node and correct definition of
21587 bool_V2DI_type_node to make reference to this new type tree.
21588 (rs6000_mangle_type): Replace erroneous reference to
21589 bool_long_type_node with bool_long_long_type_node.
21590 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
21591 comments to emphasize sign distinctions for char and int types and
21592 replace RS6000_BTI_bool_long constant with
21593 RS6000_BTI_bool_long_long constant. Also add comment to restrict
21594 use of RS6000_BTI_pixel.
21595 (bool_long_type_node): Remove this macro definition.
21596 (bool_long_long_type_node): New macro definition
21598 2018-04-10 Jakub Jelinek <jakub@redhat.com>
21600 PR rtl-optimization/85300
21601 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
21602 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
21603 simplify_unary_operation fails.
21605 2018-04-10 Martin Liska <mliska@suse.cz>
21607 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
21608 cgraph_edge and ipa_ref.
21610 2018-04-10 Jakub Jelinek <jakub@redhat.com>
21614 * config/i386/sse.md
21615 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
21616 computation of the VEC_MERGE selector from mask.
21617 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
21618 Fix decoding of the VEC_MERGE selector into mask.
21620 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
21622 PR tree-optimization/85286
21623 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
21625 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
21627 * final.c (final_1): Set insn_last_address as well as
21628 insn_current_address.
21630 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21633 * explow.c (emit_stack_probe): Call validize_mem on memory location
21634 before passing it to gen_probe_stack. Create address operand and
21635 legitimize it for the probe_stack_address case.
21637 2018-04-09 Jan Hubicka <jh@suse.cz>
21640 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
21641 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
21642 * tree.c (free_lang_data_in_type): Fix handling of binfos;
21644 (free_lang_data): Rebuild type inheritance graph.
21646 2018-04-09 Martin Sebor <msebor@redhat.com>
21648 * invoke.texi (-finline-small-functions): Mention other optimization
21650 (-findirect-inlining, -fpartial-inlining): Same.
21651 (-finline-functions-called-once): Same.
21652 (-freorder-blocks-and-partition): Same.
21654 2018-04-09 Jan Hubicka <jh@suse.cz>
21657 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
21658 jumps; choose last target that matches the criteria (i.e.
21659 no partition changes for non-crossing jumps).
21660 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
21661 support for redirecting crossing jumps to non-crossing.
21663 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
21665 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
21666 also for naked functions.
21668 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
21670 * config/arc/arc.md (add_shift): New pattern.
21671 (add_shift2): Likewise.
21672 (sub_shift): Likewise.
21673 (sub_shift_cmp0_noout): Likewise.
21674 (compare_si_ashiftsi): Likewise.
21675 (xbfu_cmp0_noout): New combine pattern.
21676 (xbfu_cmp0"): Likewise.
21677 (movsi_set_cc_insn): Place the predicable variant first.
21678 (commutative_binary_cmp0_noout): Remove clobber.
21679 (commutative_binary_cmp0): New pattern.
21680 (noncommutative_binary_cmp0): Likewise.
21681 (noncommutative_binary_cmp0_noout): Likewise.
21682 (noncommutative_binary_comparison_result_used): Removed.
21683 (rsub_cmp0): New pattern.
21684 (rsub_cmp0_noout): Likewise.
21685 (extzvsi): Changed, keep only meaningful variants.
21686 (SQH, SEZ): New iterators.
21687 (SQH_postfix): New mode attribute.
21688 (SEZ_prefix): New code attribute.
21689 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
21690 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
21691 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
21692 of numerical value.
21693 (noncommutative_operator): Check the availability of barrel
21696 2018-04-09 Richard Biener <rguenther@suse.de>
21698 PR tree-optimization/85284
21699 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
21700 Only use the niter constraining form of simple_iv when the exit
21701 is always executed.
21703 2018-04-09 Tom de Vries <tom@codesourcery.com>
21706 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
21707 (define_expand "*memory_barrier"): New define_expand.
21708 (define_insn "memory_barrier"): New insn.
21710 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21712 PR rtl-optimization/80463
21713 PR rtl-optimization/83972
21714 PR rtl-optimization/83480
21716 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
21717 correct producer for the insn.
21718 (tidy_control_flow): Fixup seqnos in case of debug insns.
21720 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21722 PR rtl-optimization/83913
21724 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
21725 different sched-times when merging exprs.
21727 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21729 PR rtl-optimization/83962
21731 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
21732 tidy_fallthru_edge and tidy_control_flow.
21734 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21736 PR rtl-optimization/83530
21738 * sel-sched.c (force_next_insn): New global variable.
21739 (remove_insn_for_debug): When force_next_insn is true, also leave only
21740 next insn in the ready list.
21741 (sel_sched_region): When the region wasn't scheduled, make another pass
21742 over it with force_next_insn set to 1.
21744 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
21746 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
21748 * config/nds32/constants.md (unspec_volatile_element): Add enum values
21749 for interrupt control.
21750 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
21751 functions for interrupt control.
21752 * config/nds32/nds32-intrinsic.md: Likewise.
21753 * config/nds32/nds32_intrinsic.h: Likewise.
21754 * config/nds32/nds32.h (nds32_builtins): Likewise.
21756 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
21758 * config/nds32/nds32.c (nds32_init_machine_status,
21759 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
21760 strict_aligned_p field.
21761 (nds32_expand_to_rtl_hook): New function.
21762 (TARGET_EXPAND_TO_RTL_HOOK): Define.
21763 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
21765 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21766 Chung-Ju Wu <jasonwucj@gmail.com>
21768 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
21769 * config/nds32/nds32-n7.md: New file.
21770 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
21771 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
21773 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
21774 * config/nds32/nds32.md (pipeline_model): Add n7.
21775 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
21776 * config/nds32/pipelines.md: Include n7 settings.
21778 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21779 Chung-Ju Wu <jasonwucj@gmail.com>
21781 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
21782 * config/nds32/nds32-e8.md: New file.
21783 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
21784 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
21786 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
21787 * config/nds32/nds32.md (pipeline_model): Add e8.
21788 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
21789 * config/nds32/pipelines.md: Include e8 settings.
21791 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21792 Chung-Ju Wu <jasonwucj@gmail.com>
21794 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
21795 * config/nds32/nds32-n8.md: New file.
21796 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
21797 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
21799 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
21800 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
21801 * config/nds32/nds32.md (pipeline_model): Add n8.
21802 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
21803 * config/nds32/pipelines.md: Include n8 settings.
21805 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21806 Chung-Ju Wu <jasonwucj@gmail.com>
21808 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
21809 * config/nds32/nds32-n9-2r1w.md: New file.
21810 * config/nds32/nds32-n9-3r2w.md: New file.
21811 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
21812 nds32_register_ports): New or modify for cpu n9.
21813 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
21815 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
21816 * config/nds32/nds32-utils.c: New file.
21817 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
21818 TARGET_MUL_SLOW): Define.
21819 * config/nds32/nds32.md (pipeline_model): New attribute.
21820 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
21821 New options that support cpu n9.
21822 * config/nds32/pipelines.md: Include n9 settings.
21823 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
21825 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
21827 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
21828 information if necessary.
21829 (output_cond_branch_compare_zero): Likewise.
21830 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
21831 (nds32_target_alignment): Refine for alignment.
21832 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
21833 (FUNCTION_BOUNDARY): Modify.
21834 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
21836 * config/nds32/nds32.opt (malways-align, malign-functions): New.
21838 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
21840 * config/nds32/constants.md (unspec_volatile_element): Add values for
21841 TLB operation and data prefetch.
21842 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
21843 functions for TLB operation and data prefetch.
21844 * config/nds32/nds32-intrinsic.md: Likewise.
21845 * config/nds32/nds32_intrinsic.h: Likewise.
21846 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
21847 (nds32_print_operand): Likewise.
21848 * config/nds32/nds32.h (nds32_builtins): Likewise.
21850 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21851 Andrew Pinski <pinsika@gcc.gnu.org>
21853 PR middle-end/82976
21854 * match.pd: Use constant_boolean_node of correct type instead of
21855 boolean_true_node or boolean_false_node for simplifying
21856 pointer comparisons to zero.
21858 2018-04-07 Jakub Jelinek <jakub@redhat.com>
21860 PR tree-optimization/80021
21861 * tree.c (verify_type_variant): Make error call in verify_variant_match
21862 translatable and remove final full stop.
21864 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21866 * config/nds32/constants.md (unspec_volatile_element): Add
21867 UNSPEC_VOLATILE_EH_RETURN.
21868 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
21869 nds32_output_stack_pop): Support dwarf exception handling process.
21870 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
21871 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
21872 exception handling process.
21873 (nds32_compute_stack_frame): Likewise.
21874 (nds32_return_addr_rtx): Likewise.
21875 (nds32_initial_elimination_offset): Likewise.
21876 (nds32_expand_prologue): Likewise.
21877 (nds32_expand_epilogue): Likewise.
21878 (nds32_dynamic_chain_address): New function.
21879 * config/nds32/nds32.h (machine_function): Add fields for dwarf
21880 exception handling.
21881 (DYNAMIC_CHAIN_ADDRESS): Define.
21882 (EH_RETURN_DATA_REGNO): Define.
21883 (EH_RETURN_STACKADJ_RTX): Define.
21884 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
21885 patterns for dwarf exception handling.
21887 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21889 * config/nds32/nds32.h: Clean up obsolete macros.
21891 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21893 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
21894 Add enum values for particular instructions.
21895 * config/nds32/nds32-intrinsic.c: Implementation of expanding
21896 particular intrinsic functions.
21897 * config/nds32/nds32-intrinsic.md: Likewise.
21898 * config/nds32/nds32_intrinsic.h: Likewise.
21899 * config/nds32/nds32.h (nds32_builtins): Likewise.
21900 * config/nds32/nds32.md (type): Add pbsad and pbsada.
21901 (btst, ave): New patterns for particular instructions.
21903 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21905 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
21906 Add enum values for atomic load/store and memory sync.
21907 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
21909 * config/nds32/nds32-intrinsic.md: Likewise.
21910 * config/nds32/nds32_intrinsic.h: Likewise.
21911 * config/nds32/nds32.h (nds32_builtins): Likewise.
21913 2018-04-07 Jakub Jelinek <jakub@redhat.com>
21915 PR tree-optimization/85257
21916 * fold-const.c (native_encode_vector): If not all elts could fit
21917 and off is -1, return 0 rather than offset.
21918 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
21919 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
21920 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
21921 adjust buffer in native_interpret_expr call.
21923 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21925 * config/nds32/constants.md (unspec_volatile_element): Add cache
21926 control enum values.
21927 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
21928 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
21929 * config/nds32/nds32.c (nds32_cctl_names): New.
21930 (nds32_print_operand): Handle cache control register names.
21931 * config/nds32/nds32.h (nds32_builtins): New enum values.
21932 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
21934 * config/nds32/nds32.md (type): Add mmu.
21935 * config/nds32/pipelines.md (simple_insn): Add mmu.
21937 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21939 * config/nds32/nds32.md (type): Remove call.
21940 * config/nds32/pipelines.md (simple_insn): Likewise.
21942 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21944 * config/nds32/constants.md (unspec_volatile_element): Add
21945 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
21946 UNSPEC_VOLATILE_FMFCFG.
21947 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
21948 description for fmfcfg and fmfcsr.
21949 (bdesc_1arg): Add fmtcsr.
21950 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
21951 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
21952 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
21953 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
21954 unspec_fmfcfg): New patterns.
21955 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
21956 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
21957 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
21958 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
21959 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
21960 __nds32__fmfcfg): Define.
21962 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21964 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
21965 intrinsic register names.
21966 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
21967 intrinsic register enum values and macros.
21969 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21971 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
21972 for load/store addressing form.
21973 (nds32_print_operand_address): Likewise.
21975 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
21978 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
21979 based on LABEL_REF. Remove useless assertion.
21980 (pic_address_needs_scratch): Fix formatting.
21981 (sparc_legitimize_pic_address): Minor tweaks.
21982 (sparc_delegitimize_address): Adjust assertion accordingly.
21983 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
21984 into symbolic_operand.
21985 (movsi_high_pic_label_ref): Likewise.
21986 (movsi_lo_sum_pic_label_ref): Likewise.
21987 (movdi_pic_label_ref): Likewise.
21988 (movdi_high_pic_label_ref): Likewise.
21989 (movdi_lo_sum_pic_label_ref): Likewise.
21991 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
21993 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
21994 custom LIB_SPEC setup.
21996 2018-04-06 Ruslan Bukin <br@bsdpad.com>
21997 Kito Cheng <kito.cheng@gmail.com>
21999 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
22000 * config/riscv/freebsd.h: New.
22002 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22004 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
22005 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
22008 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22009 Kito Cheng <kito.cheng@gmail.com>
22011 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
22012 nds32_output_call, nds32_symbol_binds_local_p): New functions.
22013 * config/nds32/nds32-protos.h (nds32_output_call,
22014 nds32_output_return): Declare.
22015 * config/nds32/nds32.md: Refine all the call and return patterns.
22017 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22020 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
22021 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
22023 PR rtl-optimization/84872
22024 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
22025 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
22026 EDGE_CROSSING edge.
22028 2018-04-06 Tamar Christina <tamar.christina@arm.com>
22030 * expr.c (copy_blkmode_to_reg): Revert 254862.
22031 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
22033 2018-04-06 Richard Biener <rguenther@suse.de>
22035 PR middle-end/85244
22036 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
22037 after seeing a component reference with an adjacent field. Treat
22038 refs to arrays at struct end of external decls similar to
22039 refs to unconstrained commons.
22041 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22044 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
22045 look through SAVE_EXPRs with non-side-effects argument. Adjust
22047 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
22050 2018-04-06 Richard Biener <rguenther@suse.de>
22052 PR middle-end/85180
22053 * alias.c (find_base_term): New wrapper around find_base_term
22054 unwinding CSELIB_VAL_PTR changes.
22055 (find_base_term): Do not restore CSELIB_VAL_PTR during the
22058 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22060 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
22062 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
22063 constant definitions.
22064 ("nop"): lr 0,0 -> nopr r0
22065 ("nop_lr0", "nop_lr1"): New insn definitions.
22067 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22069 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
22070 NDS32_V3PUSH_AVAILABLE_P macro.
22072 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
22073 Chung-Ju Wu <jasonwucj@gmail.com>
22075 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
22076 (nds32*-*-*): Add float and fpu_config into supported_defaults.
22077 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
22078 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
22079 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
22080 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
22081 * config/nds32/constraints.md: New constraints and checking for hard
22082 float configuration.
22083 * config/nds32/iterators.md: New mode iterator and attribute for hard
22084 float configuration.
22085 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
22087 * config/nds32/nds32-fpu.md: New file.
22088 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
22089 deal with hard float code generation.
22090 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
22092 (abi_type, float_reg_number): New enum type.
22093 * config/nds32/nds32-predicates.c: New predicates for hard float.
22094 * config/nds32/nds32-protos.h: Declare functions for hard float.
22095 * config/nds32/nds32.c: Implementation for hard float configuration.
22096 * config/nds32/nds32.h: Definitions for hard float configuration.
22097 * config/nds32/nds32.md: Include hard float machine description and
22098 modify patterns for hard float configuration.
22099 * config/nds32/nds32.opt: New options for hard float configuration.
22100 * config/nds32/predicates.md: New predicates for hard float
22103 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22105 * common/config/nds32/nds32-common.c
22106 (nds32_option_optimization_table): Enable -mreleax-hint by default.
22108 2018-04-05 Jakub Jelinek <jakub@redhat.com>
22110 PR middle-end/85195
22111 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
22112 CONSTRUCTOR_ELT (ctor, ...)->value.
22114 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
22117 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
22119 2018-04-05 Tom de Vries <tom@codesourcery.com>
22122 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
22125 2018-04-05 Shiva Chen <shiva0217@gmail.com>
22126 Kito Cheng <kito.cheng@gmail.com>
22128 * config/nds32/constraints.md (U33): Fine-tune checking condition.
22129 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
22130 * config/nds32/nds32.h (nds32_16bit_address_type): Add
22131 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
22133 2018-04-05 Shiva Chen <shiva0217@gmail.com>
22134 Kito Cheng <kito.cheng@gmail.com>
22136 * config/nds32/constraints.md (Ufe): New memory constraint.
22137 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
22138 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
22139 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
22141 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
22142 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
22144 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22146 * config/nds32/nds32.md: Use optimize_size in the condition for
22147 alu-shift instructions.
22149 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22151 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
22153 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22155 * config/nds32/nds32.md (negsi2): Refine pattern.
22157 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
22158 Chung-Ju Wu <jasonwucj@gmail.com>
22160 * config/nds32/iterators.md (shift_rotate): New code iterator.
22161 (shift): New code attribute.
22162 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
22163 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
22164 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
22165 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
22166 bit-wise operations.
22167 (andsi3, *andsi3): Ditto.
22168 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
22169 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
22170 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
22171 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
22172 nds32_ior_operand, nds32_xor_operand): New predicates.
22174 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22176 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
22177 (addsi3, subsi3): ... this.
22179 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22181 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
22183 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22185 * config/nds32/nds32.md: Adjust indention.
22187 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
22189 * config/nds32/nds32.md (feature): New attribute.
22191 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22193 * config/nds32/nds32.md (subtype): New attribute.
22195 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22198 * config/arm/arm-builtins.c (arm_expand_builtin): Change
22199 expansion to perform a bitwise AND of the argument followed by a
22200 boolean negation of the result.
22202 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
22204 PR rtl-optimization/84878
22205 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
22206 the basic block. Assert the use reference is not artificial and that
22207 it has an associated insn.
22209 2018-04-04 Michael Matz <matz@suse.de>
22211 * builtins.c (compute_objsize): Pass correct operand
22212 to array_at_struct_end_p.
22214 2018-04-04 Richard Biener <rguenther@suse.de>
22217 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
22218 from contexts for DINFO_LEVEL_TERSE and below.
22220 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22222 * config/nds32/nds32-doubleword.md (move_<mode>): Require
22223 resiter_operand condition.
22224 * config/nds32/nds32.md (*move<mode>): Ditto.
22226 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22227 Monk Chiang <sh.chiang04@gmail.com>
22229 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
22231 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22233 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
22235 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22236 Kito Cheng <kito.cheng@gmail.com>
22238 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
22239 nds32_cond_code_str, output_cond_branch,
22240 output_cond_branch_compare_zero, nds32_expand_cbranch,
22241 nds32_expand_cstore, nds32_expand_movcc,
22242 nds32_output_cbranchsi4_equality_zero,
22243 nds32_output_cbranchsi4_equality_reg,
22244 nds32_output_cbranchsi4_equality_reg_or_const_int,
22245 nds32_output_cbranchsi4_greater_less_zero: New functions.
22246 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
22247 nds32_expand_cstore, nds32_expand_movcc,
22248 nds32_output_cbranchsi4_equality_zero,
22249 nds32_output_cbranchsi4_equality_reg,
22250 nds32_output_cbranchsi4_equality_reg_or_const_int,
22251 nds32_output_cbranchsi4_greater_less_zero): Declare.
22252 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
22253 nds32_rimm11s_operand): New predicates.
22254 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
22255 * config/nds32/nds32.md: Rewrite all the branch and conditional move
22258 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22260 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
22261 * config/nds32/nds32.md: Ditto.
22262 * config/nds32/pipelines.md: Ditto.
22264 2018-04-04 Richard Biener <rguenther@suse.de>
22266 PR tree-optimization/85168
22267 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
22268 propagating abnormals.
22270 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22272 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
22274 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22275 Kito Cheng <kito.cheng@gmail.com>
22277 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
22278 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
22279 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
22280 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
22281 * config/nds32/nds32.md (sibcall_internal): New.
22282 (sibcall_register): Remove.
22283 (sibcall_immediate): Remove.
22284 (sibcall_value_internal): New.
22285 (sibcall_value_register): Remove.
22286 (sibcall_value_immediate): Remove.
22287 * config/nds32/predicates.md (nds32_general_register_operand): New.
22288 (nds32_call_address_operand): New.
22290 2018-04-03 Jakub Jelinek <jakub@redhat.com>
22292 PR rtl-optimization/85167
22293 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
22294 bb_defs if *split_p, instead preinitialize it to NULL.
22296 PR tree-optimization/85156
22297 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
22298 evaluating the argument multiple times.
22300 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
22302 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
22304 (_mm_cvtpd_ps): Likewise.
22305 (_mm_cvttpd_epi32): Likewise.
22306 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
22307 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
22308 vector, pixel, and bool following altivec.h include.
22310 2018-04-03 Martin Sebor <msebor@redhat.com>
22312 * doc/extend.texi (Common Function Attributes): Clarify.
22313 (const attribute): Likewise.
22314 (pure attribute): Likewise.
22316 2018-04-03 Jakub Jelinek <jakub@redhat.com>
22319 * config/i386/i386.c (ix86_expand_vector_set): Use
22320 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
22322 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
22324 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
22325 instructions when changing rounding bits to preserve precision bits
22326 in the x87 control word.
22328 2018-04-03 Martin Liska <mliska@suse.cz>
22330 PR tree-optimization/82491
22331 * rtl.h (strip_offset_and_add): Replace += suboffset with
22332 poly_uint64 () + suboffset.
22334 2018-03-29 Martin Liska <mliska@suse.cz>
22335 Martin Jambor <mjambor@suse.cz>
22338 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
22339 param_type is not an integral or pointer type.
22341 2018-04-03 Richard Biener <rguenther@suse.de>
22343 * sese.h (recompute_all_dominators): Remove.
22345 2018-04-02 Martin Sebor <msebor@redhat.com>
22347 * doc/invoke.texi (-Wrestrict): Fix typos.
22349 2018-04-02 Jim Wilson <jimw@sifive.com>
22351 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
22352 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
22353 (<optab>di3, <optab>si3_extend): Likewise.
22354 (<optab>si3_mask, <optab>si3_mask_1): New.
22355 (<optab>di3_mask, <optab>di3_mask_1): New.
22356 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
22357 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
22358 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
22360 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
22362 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
22365 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
22367 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
22368 (nds32_canonicalize_comparison): New function.
22370 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22371 Kito Cheng <kito.cheng@gmail.com>
22372 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22374 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
22375 * config/nds32/constants.md (unspec_volatile_element): Add
22376 UNSPEC_VOLATILE_RELAX_GROUP.
22377 * config/nds32/nds32-relax-opt.c: New file.
22378 * config/nds32/nds32-predicates.c
22379 (nds32_symbol_load_store_p): New function.
22380 * config/nds32/nds32-protos.h
22381 (nds32_symbol_load_store_p): Declare function.
22382 (make_pass_nds32_relax_opt): Declare new rtl pass function.
22383 * config/nds32/nds32.c
22384 (nds32_register_pass): New function to register pass.
22385 (nds32_register_passes): New function to register passes.
22386 * config/nds32/nds32.md (relax_group): New pattern.
22387 * config/nds32/nds32.opt (mrelax-hint): New option.
22388 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
22390 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
22392 * config/nds32/t-nds32: Modify files dependency.
22394 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22396 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
22397 (PROFILE_HOOK): Define its implementation.
22399 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22401 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
22402 type and 32-bit size.
22404 2018-04-01 Jakub Jelinek <jakub@redhat.com>
22406 PR middle-end/85090
22407 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
22408 (V_128_256): New mode iterator.
22409 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
22410 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
22411 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
22413 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
22414 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
22416 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22419 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
22420 NaN inputs correctly.
22422 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
22425 * config/rs6000/vsx.md (??r): New mode attribute.
22426 (*vsx_mov<mode>_64bit): Use it.
22427 (*vsx_mov<mode>_32bit): Likewise.
22429 2018-03-30 Martin Sebor <msebor@redhat.com>
22431 PR tree-optimization/84818
22432 * builtins.c (check_access): Use warning_n.
22434 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22437 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
22439 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
22442 2018-03-30 Julia Koval <julia.koval@intel.com>
22445 * x86-tune.def (movx, partial_reg_dependency): Enable for
22448 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
22450 PR inline-asm/84985
22451 * lra-constraints.c (process_alt_operands): Move setting
22452 this_alternative_matches below.
22454 2018-03-29 Martin Liska <mliska@suse.cz>
22457 * doc/invoke.texi: Document how LTO works with debug info.
22458 Describe auto-load support of binutils. Mention 'x86-64'
22459 as valid option value of -march option.
22461 2018-03-29 Jakub Jelinek <jakub@redhat.com>
22463 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
22466 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
22467 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
22468 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
22471 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
22474 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
22475 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
22476 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
22477 for RS6000_BTM_POWERPC64.
22478 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
22479 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
22480 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
22483 (DIVDEU): Likewise.
22485 2018-03-28 Carl Love <cel@us.ibm.com>
22488 2017-09-27 Carl Love <cel@us.ibm.com>
22490 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
22491 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
22492 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
22495 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22497 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
22498 instead of __vector bool.
22499 (_mm_max_pu8): Likewise.
22500 (_mm_min_pi16): Likewise.
22502 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
22505 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
22506 (DIVWEUO): Likewise.
22507 (DIVDEO): Likewise.
22508 (DIVDEUO): Likewise.
22509 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
22510 DIVWEUO and DIVDEUO.
22511 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
22512 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
22513 (div_extend): Likewise.
22514 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
22516 (__builtin_divweuo): Likewise.
22517 (__builtin_divdeo): Likewise.
22518 (__builtin_divdeuo): Likewise.
22520 2018-03-28 Jakub Jelinek <jakub@redhat.com>
22523 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
22524 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
22526 PR tree-optimization/82004
22527 * gimple-match-head.c (optimize_pow_to_exp): New function.
22528 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
22529 Don't fold to exp if optimize_pow_to_exp is false.
22531 2018-03-28 Martin Liska <mliska@suse.cz>
22534 * calls.c (initialize_argument_information): Fix trailing space.
22535 * common.opt: Fix typo and provide better explanation for
22536 -fsanitize-coverage option.
22537 * config/i386/i386.opt: Fix typo.
22539 2018-03-28 Jakub Jelinek <jakub@redhat.com>
22540 Martin Liska <mliska@suse.cz>
22543 * gimplify.c (asan_poison_variable): Don't do the check for
22544 gimplify_omp_ctxp here.
22545 (gimplify_decl_expr): Do it here.
22546 (gimplify_target_expr): Likewise.
22548 2018-03-28 Martin Liska <mliska@suse.cz>
22551 * config/i386/i386.c (ix86_function_arg_advance): Do not call
22552 chkp_type_bounds_count if MPX is not enabled.
22554 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
22556 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
22558 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22561 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
22562 function to create the function decl for complex long double
22563 multiply and divide for -mabi=ieeelongdouble.
22564 (init_float128_ieee): Call it.
22566 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
22569 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
22570 -fcf-protection=branch -mibt.
22571 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
22573 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22576 * config/arm/arm.c (arm_valid_symbolic_address): Handle
22577 arm_word_relocations.
22579 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
22582 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
22583 extern array declarations.
22585 2018-03-27 Richard Biener <rguenther@suse.de>
22587 PR middle-end/84067
22588 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
22589 explicit single_use checks.
22591 2018-03-27 Richard Biener <rguenther@suse.de>
22593 PR tree-optimization/85082
22594 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
22597 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22599 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
22600 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
22601 Turn on fasynchronous-unwind-tables and funwind-tables.
22603 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
22606 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
22607 (*bmi_blsr_<mode>_ccz): Ditto.
22609 2018-03-26 Tom de Vries <tom@codesourcery.com>
22611 PR tree-optimization/85063
22612 * omp-general.c (offloading_function_p): New function. Factor out
22614 * omp-offload.c (pass_omp_target_link::gate): ... here.
22615 * omp-general.h (offloading_function_p): Declare.
22616 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
22617 with attribute omp declare target for offloading functions.
22619 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
22621 PR tree-optimization/84005
22622 * tree-data-ref.h (get_base_for_alignment): Declare.
22623 * tree-data-ref.c (get_base_for_alignment_1): New function.
22624 (get_base_for_alignment): Likewise.
22625 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22626 get_base_for_alignment to find a suitable base object, instead
22627 of always using drb->base_address.
22629 2018-03-23 Jakub Jelinek <jakub@redhat.com>
22631 PR inline-asm/85022
22632 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
22633 known size by default.
22635 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
22637 PR inline-asm/85030
22638 * lra-constraints.c (process_alt_operands): Don't match BLKmode
22639 and non BLKmode operands.
22641 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22644 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
22645 Clean up attributes.
22647 2018-03-23 Richard Biener <rguenther@suse.de>
22650 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
22651 we are going to emit early debug for LTO.
22653 2018-03-23 Jakub Jelinek <jakub@redhat.com>
22655 PR inline-asm/85034
22656 * function.c (match_asm_constraints_1): Don't optimize if input
22657 doesn't satisfy general_operand predicate for output's mode.
22659 PR inline-asm/85022
22660 * alias.c (write_dependence_p): Don't require for x_canonicalized
22661 non-VOIDmode if x has VOIDmode.
22664 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
22665 just don't try to optimize it rather than assert it never happens.
22667 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
22669 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
22670 macro expansions for definition of ST_INTERNAL_<mode> and
22671 LD_INTERNAL_<mode> builtins.
22672 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
22674 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
22676 (altivec_expand_st_builtin): Likewise.
22677 (altivec_expand_builtin): Remove calls to deleted functions.
22678 (rs6000_address_for_altivec): Delete this function.
22679 * config/rs6000/vector.md: Remove expands for
22680 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
22682 2018-03-22 Sudakshina Das <sudi.das@arm.com>
22685 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
22686 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
22687 re-computing once computed.
22688 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
22689 (arm_init_machine_status): Initialize
22690 machine->static_chain_stack_bytes.
22692 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
22695 * doc/extend.texi: Add four new prototypes for vec_ld.
22696 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
22697 definitions for more logical presentation.
22698 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
22699 entries for V1TI variants of __builtin_altivec_ld builtin.
22700 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
22701 handling of V1TI variant of LVX icode pattern.
22702 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
22703 (rs6000_gimple_fold_builtin): Likewise.
22704 (altivec_init_builtins): Add code to define
22705 __builtin_altivec_lvx_v1ti function.
22707 2018-03-22 Jakub Jelinek <jakub@redhat.com>
22709 PR inline-asm/84941
22710 * function.c (match_asm_constraints_1): Don't do the optimization
22711 if input isn't a REG, SUBREG, MEM or constant.
22713 2018-03-22 Tom de Vries <tom@codesourcery.com>
22715 PR tree-optimization/84956
22716 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
22717 bb_has_abnormal_pred.
22719 2018-03-22 Jakub Jelinek <jakub@redhat.com>
22722 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
22723 DECL_INITIAL (decl) to decl at the end.
22724 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
22725 adjust the comment.
22727 2018-03-21 Joseph Myers <joseph@codesourcery.com>
22729 * doc/extend.texi (__builtin_tgmath): Document when complex
22730 integer types are treated as _Complex _Float64.
22732 2018-03-21 Tom de Vries <tom@codesourcery.com>
22734 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
22736 2018-03-21 Jakub Jelinek <jakub@redhat.com>
22738 PR tree-optimization/84960
22739 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
22740 if it is ENTRY block, move them into single succ of ENTRY in that case.
22742 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
22744 PR tree-optimization/84811
22745 * poly-int.h (poly_span_traits): Remove the T3 parameter and
22746 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
22747 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
22748 (known_subrange_p): Update accordingly. Cast each value involved
22749 in the size comparison, rather than casting the result of the
22752 2018-03-21 Jakub Jelinek <jakub@redhat.com>
22754 PR tree-optimization/84982
22755 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
22756 by flipping the least significant bit rather than all bits from
22757 bitpos to bitpos + bitsize - 1.
22759 2018-03-21 Nathan Sidwell <nathan@acm.org>
22761 * doc/extend.texi (Deprecated Features): Remove mention of
22762 long-deleted deprecations.
22764 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22767 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
22768 * configure: Regenerate.
22770 2018-03-21 Tom de Vries <tom@codesourcery.com>
22772 PR tree-optimization/83126
22773 * tree-parloops.c (num_phis): New function.
22774 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
22776 2018-03-21 Nathan Sidwell <nathan@acm.org>
22778 * doc/extend.texi (Deprecated Features): Update deprecated flags,
22779 mention anon-struct/union members and trailing attributes.
22781 2018-03-21 Bin Cheng <bin.cheng@arm.com>
22783 PR tree-optimization/84969
22784 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
22785 builtin memset partitions if they set different rhs values.
22787 2018-03-21 Jakub Jelinek <jakub@redhat.com>
22789 PR rtl-optimization/84989
22790 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
22791 VEC_DUPLICATE with scalar result mode.
22793 2018-03-21 Martin Liska <mliska@suse.cz>
22796 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
22797 not intended return statement.
22799 2018-03-21 Martin Liska <mliska@suse.cz>
22802 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
22803 (chkp_find_bound_slots_1): Limit number of iterations.
22805 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
22808 * Minor grammar fixes for x86 options.
22810 2018-03-20 Jakub Jelinek <jakub@redhat.com>
22813 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
22814 holding REG_CFA_RESTORE notes, instead turn them into a USE.
22816 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
22819 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
22820 (altivec_lvx_<mode>_1op): Likewise.
22821 (altivec_stvx_<mode>_2op): Likewise.
22822 (altivec_stvx_<mode>_1op): Likewise.
22823 (altivec_lvx_<VM2:mode>): New define_expand.
22824 (altivec_stvx_<VM2:mode>): Likewise.
22825 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
22826 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
22827 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
22828 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
22829 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
22830 (rs6000_gen_lvx): Likewise.
22831 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
22832 (altivec_expand_stv_builtin): Likewise.
22833 (altivec_expand_builtin): Likewise.
22834 * config/rs6000/vector.md: Likewise.
22836 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22839 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
22842 2018-03-20 Richard Biener <rguenther@suse.de>
22845 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
22846 sign-conversions as zero, fall back to standard scalar_stmt
22849 2018-03-20 Martin Liska <mliska@suse.cz>
22852 * predict.c (rebuild_frequencies): Handle case when we have
22853 PROFILE_ABSENT, but flag_guess_branch_prob is false.
22855 2018-03-20 Jakub Jelinek <jakub@redhat.com>
22858 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
22859 flag_section_anchors.
22860 * varasm.c (use_blocks_for_decl_p): Remove hack for
22861 dw2_force_const_mem.
22864 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
22866 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
22867 be created, use lowpart_subreg of operands[0] rather than operands[0]
22869 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
22870 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
22871 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
22872 and n constraint instead of aarch64_shift_imm_di and Usd.
22873 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
22874 (*aarch64_<optab>_reg_minus<mode>3): ... this.
22876 2018-03-20 Sudakshina Das <sudi.das@arm.com>
22879 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
22880 to favor GPR over NEON registers.
22881 (<shift>di3_neon): Likewise.
22883 2018-03-20 Tom de Vries <tom@codesourcery.com>
22886 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
22887 (nvptx_process_pars): Emit bar.sync asap and alap.
22889 2018-03-20 Tom de Vries <tom@codesourcery.com>
22892 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
22893 seen_label if seen_label is already set.
22895 2018-03-20 Jakub Jelinek <jakub@redhat.com>
22898 * config/i386/i386.c (fold_builtin_cpu): For features above 31
22899 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
22900 Use 1U instead of 1. Formatting fixes.
22903 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
22904 instead of TREE_TYPE (s1) for the return value.
22906 2018-03-19 Jakub Jelinek <jakub@redhat.com>
22908 PR tree-optimization/84946
22909 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
22910 bitsize + bitsize in poly_uint64 rather than poly_int64.
22913 * dwarf2asm.c: Include fold-const.c.
22914 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
22915 of decl rather than decl itself.
22917 PR rtl-optimization/84643
22918 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
22920 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
22923 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
22924 calling assemble_variable.
22926 2018-03-19 Sudakshina Das <sudi.das@arm.com>
22929 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
22930 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
22932 2018-03-19 Jim Wilson <jimw@sifive.com>
22935 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
22936 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
22937 (riscv_first_stack_step): Likewise.
22938 (riscv_option_override): Use STACK_BOUNDARY instead of
22939 MIN_STACK_BOUNDARY.
22940 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
22941 MIN_STACK_BOUNDARY.
22942 (BIGGEST_ALIGNMENT): Set to 128.
22943 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
22944 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
22947 2018-03-19 Richard Biener <rguenther@suse.de>
22949 PR tree-optimization/84933
22950 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
22951 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
22953 2018-03-19 Richard Biener <rguenther@suse.de>
22955 PR tree-optimization/84859
22956 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
22957 (cond_if_else_store_replacement): Perform sinking operation on
22958 single-store BBs regardless of MAX_STORES_TO_SINK setting.
22959 Generalize what a BB with a single eligible store is.
22961 2018-03-19 Richard Biener <rguenther@suse.de>
22963 PR tree-optimization/84929
22964 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
22965 chrec_is_positive against non-chrec arg.
22967 2018-03-19 Tamar Christina <tamar.christina@arm.com>
22970 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
22972 2018-03-18 Martin Liska <mliska@suse.cz>
22974 PR rtl-optimization/84635
22975 * regrename.c (build_def_use): Use matches_mode only when
22978 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
22980 PR tree-optimization/84913
22981 * tree-vect-loop.c (vectorizable_reduction): Don't try to
22982 vectorize chains of COND_EXPRs.
22984 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
22986 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
22988 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
22990 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
22992 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
22994 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
22996 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
22997 Kito Cheng <kito.cheng@gmail.com>
22999 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
23000 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
23001 (nds32_adjust_reg_alloc_order): New function.
23002 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
23004 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
23006 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
23007 nds32_print_operand, nds32_print_operand_address): Use
23008 HOST_WIDE_INT_PRINT_DEC instead.
23010 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23012 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
23014 2018-03-17 Jakub Jelinek <jakub@redhat.com>
23017 * config/i386/i386.c (initial_ix86_tune_features,
23018 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
23019 unsigned long long.
23020 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
23021 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
23022 rather than 1u << ix86_tune. Formatting fix.
23023 (ix86_option_override_internal): Change ix86_arch_mask from
23024 unsigned int to unsigned HOST_WIDE_INT, initialize to
23025 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
23026 (ix86_function_specific_restore): Likewise.
23028 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23031 * postreload.c (reload_combine_recognize_pattern): Perform
23032 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
23033 truncate_int_for_mode the result for the destination's mode.
23036 * hsa-gen.c (mem_type_for_type): Fix comment typo.
23037 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
23039 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23042 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
23045 * lra-assigns.c (lra_split_hard_reg_for): Don't use
23046 regno_allocno_class_array and sorted_pseudos.
23047 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
23048 insns where regno is used.
23050 2018-03-16 Martin Liska <mliska@suse.cz>
23053 * multiple_target.c (create_dispatcher_calls): Redirect
23054 reference in the symbol table.
23056 2018-03-16 Martin Liska <mliska@suse.cz>
23059 * multiple_target.c (create_dispatcher_calls): Redirect also
23062 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23066 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
23067 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
23068 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
23070 2018-03-16 Julia Koval <julia.koval@intel.com>
23072 * doc/invoke.texi (Skylake Server): Add CLWB.
23073 Cannonlake): Remove CLWB.
23075 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23077 PR tree-optimization/84841
23078 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
23080 (FLOAT_ONE_CONST_TYPE): Define.
23081 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
23082 (sort_by_operand_rank): Put entries with higher constant_type last
23083 rather than first to match comments.
23085 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
23087 * config/nios2/nios2.md (movsi_internal): Fix thinko in
23090 2018-03-15 Jakub Jelinek <jakub@redhat.com>
23093 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
23094 check and use address of target always.
23096 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
23099 * config/i386/i386.c (indirect_thunk_needed): Update comments.
23100 (indirect_thunk_bnd_needed): Likewise.
23101 (indirect_thunks_used): Likewise.
23102 (indirect_thunks_bnd_used): Likewise.
23103 (indirect_return_needed): New.
23104 (indirect_return_bnd_needed): Likewise.
23105 (output_indirect_thunk_function): Add a bool argument for
23107 (output_indirect_thunk_function): Don't generate alias for
23108 function return thunk.
23109 (ix86_code_end): Call output_indirect_thunk_function to generate
23110 function return thunks.
23111 (ix86_output_function_return): Set indirect_return_bnd_needed
23112 and indirect_return_needed instead of indirect_thunk_bnd_needed
23113 and indirect_thunk_needed.
23115 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
23117 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
23118 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
23119 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
23121 2018-03-15 David Malcolm <dmalcolm@redhat.com>
23122 Paul Hua <paul.hua.gm@gmail.com>
23125 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
23127 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
23129 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
23130 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
23131 resp. SFmode cases.
23133 2018-03-15 Tamar Christina <tamar.christina@arm.com>
23136 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
23137 instead of GET_MODE_SIZE when comparing Units.
23139 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
23142 * varasm.c (hash_section): Return an unchangeble hash value
23143 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
23144 Return !aarch64_can_use_per_function_literal_pools_p ().
23146 2018-03-15 Jakub Jelinek <jakub@redhat.com>
23149 * optabs.c (emit_conditional_move): Pass address of cmode's copy
23150 rather than address of cmode as last argument to prepare_cmp_insn.
23152 2018-03-15 Julia Koval <julia.koval@intel.com>
23154 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
23155 F_AVX512VNNI, F_AVX512BITALG): New.
23157 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
23160 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
23161 insn for floating-point loads and stores.
23163 2018-03-14 Carl Love <cel@us.ibm.com>
23165 * config/rs6000/rs6000-c.c: Add macro definitions for
23166 ALTIVEC_BUILTIN_VEC_PERMXOR.
23167 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
23168 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
23169 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
23170 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
23172 * config/doc/extend.texi: Add prototypes for vec_permxor.
23174 2018-03-14 David Malcolm <dmalcolm@redhat.com>
23177 * diagnostic-show-locus.c (class layout_point): Convert m_line
23178 from int to linenum_type.
23179 (line_span::comparator): Use linenum "compare" function when
23180 comparing line numbers.
23181 (test_line_span): New function.
23182 (layout_range::contains_point): Convert param "row" from int to
23184 (layout_range::intersects_line_p): Likewise.
23185 (layout::will_show_line_p): Likewise.
23186 (layout::print_source_line): Likewise.
23187 (layout::should_print_annotation_line_p): Likewise.
23188 (layout::print_annotation_line): Likewise.
23189 (layout::print_leading_fixits): Likewise.
23190 (layout::annotation_line_showed_range_p): Likewise.
23191 (struct line_corrections): Likewise for field m_row.
23192 (line_corrections::line_corrections): Likewise for param "row".
23193 (layout::print_trailing_fixits): Likewise.
23194 (layout::get_state_at_point): Likewise.
23195 (layout::get_x_bound_for_row): Likewise.
23196 (layout::print_line): Likewise.
23197 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
23198 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
23199 * input.c (selftest::test_linenum_comparisons): New function.
23200 (selftest::input_c_tests): Call it.
23201 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
23202 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
23203 * selftest.h (ASSERT_GT): New macro.
23204 (ASSERT_GT_AT): New macro.
23205 (ASSERT_LT): New macro.
23206 (ASSERT_LT_AT): New macro.
23208 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
23210 PR rtl-optimization/84780
23211 * combine.c (distribute_links): Don't make a link based on pc_rtx.
23213 2018-03-14 Martin Liska <mliska@suse.cz>
23215 * tree.c (record_node_allocation_statistics): Use
23216 get_stats_node_kind.
23217 (get_stats_node_kind): New function extracted from
23218 record_node_allocation_statistics.
23219 (free_node): Use get_stats_node_kind.
23221 2018-03-14 Richard Biener <rguenther@suse.de>
23223 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
23224 that the value-set of ANTIC_IN doesn't grow.
23227 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
23229 (BB_VISITED_WITH_VISITED_SUCCS): New define.
23230 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
23232 2018-03-14 Julia Koval <julia.koval@intel.com>
23234 * config.gcc (icelake-client, icelake-server): New.
23236 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
23237 (initial_ix86_arch_features): Ditto.
23238 (PTA_SKYLAKE): Add SGX.
23239 (PTA_ICELAKE): Remove.
23240 (PTA_ICELAKE_CLIENT): New.
23241 (PTA_ICELAKE_SERVER): New.
23242 (ix86_option_override_internal): Split up icelake on icelake client and
23244 (get_builtin_code_for_version): Ditto.
23245 (fold_builtin_cpu): Ditto.
23246 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
23247 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
23248 * config/i386/i386.h (processor_type): Ditto.
23249 * doc/invoke.texi: Ditto.
23251 2018-03-14 Jakub Jelinek <jakub@redhat.com>
23254 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
23255 INTEGER_CST offset, add it together with bitpos / 8 and
23256 sign extend based on POINTER_SIZE.
23260 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23263 * config/i386/constraints.md (Yc): New register constraint.
23264 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23265 Use Yc constraint for alternative 2 of operand 0. Remove
23266 preferred_for_speed attribute.
23268 2018-03-14 Richard Biener <rguenther@suse.de>
23270 PR tree-optimization/84830
23271 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
23272 with the old one to avoid oscillations.
23274 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
23277 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
23279 (assign_by_spills): Return a flag of reload assignment failure.
23280 Do not process the reload assignment failures. Do not spill other
23281 reload pseudos if they has the same reg class. Update n if
23283 (lra_assign): Add a return arg. Set up from the result of
23284 assign_by_spills call.
23285 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23286 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23287 usage_insns if it is not NULL.
23288 (spill_hard_reg_in_range): New function.
23289 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23290 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23291 function prototypes.
23292 (lra_assign): Change prototype.
23293 * lra.c (lra): Add code to deal with fails by splitting hard reg
23296 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
23298 * config/riscv/riscv.opt (mrelax): New option.
23299 * config/riscv/riscv.c (riscv_file_start): Emit ".option
23300 "norelax" when riscv_mrelax is disabled.
23301 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
23303 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23306 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
23307 reassociation for int modes.
23309 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23311 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
23312 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
23314 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
23315 * config/aarch64/aarch64-sve.md
23316 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
23317 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
23318 (*extend<mode><Vwide>2): Rename to...
23319 (aarch64_sve_extend<mode><Vwide>2): ...this.
23320 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
23321 renaming the old pattern to...
23322 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
23324 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
23325 define_expand, renaming the old pattern to...
23326 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
23327 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
23328 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
23329 account when deciding which SVE instruction the optab should use.
23330 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
23332 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23334 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
23335 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
23336 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
23337 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
23338 (tlsdesc_small_<mode>): Turn a define_expand and use
23339 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
23340 (tlsdesc_small_advsimd_<mode>): ...this.
23341 (tlsdesc_small_sve_<mode>): New pattern.
23343 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23345 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
23346 (UNSPEC_UMUL_HIGHPART): New constants.
23347 (MUL_HIGHPART): New int iteraor.
23348 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
23349 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
23351 (*<su>mul<mode>3_highpart): New define_insn.
23353 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23356 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23359 2018-03-13 Martin Liska <mliska@suse.cz>
23362 * (sem_item_optimizer::sem_item_optimizer): Initialize new
23364 (sem_item_optimizer::~sem_item_optimizer): Release it.
23365 (sem_item_optimizer::merge_classes): Register variable aliases.
23366 (sem_item_optimizer::fixup_pt_set): New function.
23367 (sem_item_optimizer::fixup_points_to_sets): Likewise.
23368 * ipa-icf.h: Declare new variables and functions.
23370 2018-03-13 Jakub Jelinek <jakub@redhat.com>
23372 PR middle-end/84834
23373 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
23374 integer_pow2p@2 and test integer_pow2p in condition.
23375 (A < 0 ? C : 0): Similarly for @1.
23377 PR middle-end/84831
23378 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
23379 characters starting at p contain '\0' character, don't look beyond
23383 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
23384 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
23387 * reg-stack.c (change_stack): Change update_end var from int to
23388 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
23389 also call set_block_for_insn on the newly added insns and rescan.
23392 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
23393 on the last operand.
23396 * tree.c (stabilize_reference_1): Return save_expr (e) for
23397 STATEMENT_LIST even if it doesn't have side-effects.
23399 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
23401 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
23403 2018-03-12 Renlin Li <renlin.li@arm.com>
23405 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
23406 aarch64_output_scalar_simd_mov_immediate.
23408 2018-03-12 Martin Sebor <msebor@redhat.com>
23410 PR tree-optimization/83456
23411 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
23412 for perfectly overlapping calls to memcpy.
23413 (gimple_fold_builtin_memory_chk): Same.
23414 (gimple_fold_builtin_strcpy): Handle no-warning.
23415 (gimple_fold_builtin_stxcpy_chk): Same.
23416 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
23418 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
23420 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
23421 parameter. Use it for SFmode.
23422 (rs6000_function_arg_advance_1): Adjust.
23423 (rs6000_function_arg): Adjust.
23424 (rs6000_gimplify_va_arg): Pass false for that new parameter.
23426 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
23428 PR rtl-optimization/84169
23429 PR rtl-optimization/84780
23430 * combine.c (can_combine_p): Check for a 2-insn combination whether
23431 the destination register is used between the two insns, too.
23433 2018-03-12 Richard Biener <rguenther@suse.de>
23435 PR tree-optimization/84803
23436 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
23437 for refs DR analysis didn't process.
23439 2018-03-12 Richard Biener <rguenther@suse.de>
23441 PR tree-optimization/84777
23442 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
23443 force-vectorize loops ignore whether we are optimizing for size.
23445 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
23447 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
23448 (TARGET_MD_ASM_ADJUST): Define.
23450 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
23451 Kito Cheng <kito.cheng@gmail.com>
23452 Chung-Ju Wu <jasonwucj@gmail.com>
23454 * config/nds32/nds32.c (nds32_compute_stack_frame,
23455 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
23456 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
23457 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
23458 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
23459 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
23460 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
23461 * config/nds32/nds32.md (prologue, epilogue): Use macro
23462 NDS32_V3PUSH_AVAILABLE_P to do checking.
23464 2018-03-11 Jakub Jelinek <jakub@redhat.com>
23467 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
23468 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
23469 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
23470 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
23471 addition of most attributes on !orig_type_die or the attribute not
23472 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
23474 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23475 Chung-Ju Wu <jasonwucj@gmail.com>
23477 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
23478 __NDS32_VH__ macro.
23479 * config/nds32/nds32.opt (mvh): New option.
23481 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23482 Chung-Ju Wu <jasonwucj@gmail.com>
23484 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
23486 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
23487 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
23490 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23491 Chung-Ju Wu <jasonwucj@gmail.com>
23493 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
23495 * config/nds32/nds32-multiple.md (strlensi): New pattern.
23496 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
23498 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
23499 Kito Cheng <kito.cheng@gmail.com>
23500 Chung-Ju Wu <jasonwucj@gmail.com>
23502 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
23503 UNSPEC_FFMISM and UNSPEC_FLMISM.
23504 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
23505 for ffb, ffmism and flmism.
23506 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
23507 (unspec_ffmism): Ditto.
23508 (unspec_flmism): Ditto.
23509 (nds32_expand_builtin_impl): Check if string extension is available.
23510 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
23511 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
23513 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
23516 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
23519 * lra-assigns.c (assign_by_spills): Return a flag of reload
23520 assignment failure. Do not process the reload assignment
23521 failures. Do not spill other reload pseudos if they has the same
23523 (lra_assign): Add a return arg. Set up from the result of
23524 assign_by_spills call.
23525 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23526 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23527 usage_insns if it is not NULL.
23528 (spill_hard_reg_in_range): New function.
23529 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23530 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23531 function prototypes.
23532 (lra_assign): Change prototype.
23533 * lra.c (lra): Add code to deal with fails by splitting hard reg
23536 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
23539 * config/i386/i386.opt: Replace Enforcment with Enforcement.
23541 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
23544 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
23545 (dw_val_node): Add val_symbolic_view.
23546 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
23547 (symview_upper_bound): New.
23548 (new_line_info_table): Initialize symviews_since_reset.
23549 (dwarf2out_source_line): Count symviews_since_reset and set
23550 symview_upper_bound.
23551 (dw_val_equal_p): Handle symview.
23552 (add_AT_symview): New.
23553 (print_dw_val): Handle symview.
23554 (attr_checksum, attr_checksum_ordered): Likewise.
23555 (same_dw_val_p, size_of_die): Likewise.
23556 (value_format, output_die): Likewise.
23557 (add_high_low_attributes): Use add_AT_symview for entry_view.
23558 (dwarf2out_finish): Reset symview_upper_bound, clear
23561 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
23564 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
23565 Add strict argument and use it.
23566 (rs6000_split_multireg_move): Update for new strict argument.
23567 (mem_operand_gpr): Disallow all non-offsettable addresses.
23568 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
23570 2018-03-09 Jakub Jelinek <jakub@redhat.com>
23573 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
23574 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
23575 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
23578 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
23579 decl, use remap_type if we want to use the type.
23581 2018-03-09 Martin Sebor <msebor@redhat.com>
23583 PR tree-optimization/84526
23584 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23586 (builtin_access::generic_overlap): Be prepared to handle non-array
23589 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
23591 PR rtl-optimization/84682
23592 * lra-constraints.c (process_address_1): Check is_address flag
23593 for address constraints.
23594 (process_alt_operands): Likewise.
23595 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
23596 preprocess_constraints.
23597 * recog.h (preprocess_constraints): Add oploc parameter.
23599 * recog.c (preprocess_constraints): Test address_operand for
23600 CT_ADDRESS constraints.
23602 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
23605 * lra-assigns.c (assign_by_spills): Return a flag of reload
23606 assignment failure. Do not process the reload assignment
23607 failures. Do not spill other reload pseudos if they has the same
23609 (lra_assign): Add a return arg. Set up from the result of
23610 assign_by_spills call.
23611 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23612 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23613 usage_insns if it is not NULL.
23614 (spill_hard_reg_in_range): New function.
23615 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23616 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23617 function prototypes.
23618 (lra_assign): Change prototype.
23619 * lra.c (lra): Add code to deal with fails by splitting hard reg
23622 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23625 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
23626 Accept complain bool parameter. Only emit errors if it is true.
23627 (arm_parse_cpu_option_name): Likewise.
23628 (arm_target_thumb_only): Adjust callers of the above.
23629 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
23630 prototype to take a default true bool parameter.
23631 (arm_parse_arch_option_name): Likewise.
23633 2018-03-09 David Malcolm <dmalcolm@redhat.com>
23634 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
23638 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
23639 * configure: Regenerate.
23640 * configure.ac ("linker --version-script option"): New.
23641 ("linker soname option"): New.
23643 2018-03-09 Richard Biener <rguenther@suse.de>
23645 PR tree-optimization/84775
23646 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
23647 immediate uses of predicate stmts and mark them modified.
23650 PR tree-optimization/84178
23651 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
23653 (version_loop_for_if_conversion): Delay update_ssa call.
23654 (tree_if_conversion): Delay update_ssa until after predicate
23657 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
23660 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
23661 when the function accesses prior frames.
23663 2018-03-08 Jakub Jelinek <jakub@redhat.com>
23666 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
23667 gen_llsym, otherwise call maybe_gen_llsym.
23669 PR inline-asm/84742
23670 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
23671 has ',' character inside of it.
23673 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23676 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
23677 as clobbering CC_REGNUM.
23679 2018-03-08 Richard Biener <rguenther@suse.de>
23681 PR middle-end/84552
23682 * tree-scalar-evolution.c: Include tree-into-ssa.h.
23683 (follow_copies_to_constant): Do not follow SSA names registered
23686 2018-03-08 Richard Biener <rguenther@suse.de>
23688 PR tree-optimization/84178
23689 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
23691 (version_loop_for_if_conversion): Delay update_ssa call.
23692 (tree_if_conversion): Delay update_ssa until after predicate
23695 2018-03-08 David Malcolm <dmalcolm@redhat.com>
23697 PR tree-optimization/84178
23698 * tree-if-conv.c (release_bb_predicate): Remove the
23699 the assertion that the stmts have NULL use_ops.
23700 Discard the statements, asserting that they haven't
23701 yet been added to a BB.
23703 2018-03-08 Richard Biener <rguenther@suse.de>
23705 PR tree-optimization/84746
23706 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
23707 (phi_translate): Pass in destination ANTIC_OUT set.
23708 (phi_translate_1): Likewise. For a simplified result lookup
23709 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
23710 (phi_translate_set): Adjust.
23711 (do_pre_regular_insertion): Likewise.
23712 (do_pre_partial_partial_insertion): Likewise.
23714 2018-03-08 Martin Liska <mliska@suse.cz>
23716 PR gcov-profile/84735
23717 * doc/gcov.texi: Document usage of profile files.
23718 * gcov-io.h: Document changes in the format.
23720 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
23724 * dwarf2out.c (struct dw_line_info_table): Update comments for
23726 (FORCE_RESET_NEXT_VIEW): New.
23727 (FORCE_RESETTING_VIEW_P): New.
23728 (RESETTING_VIEW_P): Check for -1 too.
23729 (ZERO_VIEW_P): Likewise.
23730 (new_line_info_table): Force-reset next view.
23731 (dwarf2out_begin_function): Likewise.
23732 (dwarf2out_source_line): Simplify zero_view_p initialization.
23733 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
23734 view directly. Omit view when omitting .loc at line 0.
23736 2018-03-08 Jakub Jelinek <jakub@redhat.com>
23738 PR tree-optimization/84740
23739 * tree-switch-conversion.c (process_switch): Call build_constructors
23740 only if info.phi_count is non-zero.
23742 PR tree-optimization/84739
23743 * tree-tailcall.c (find_tail_calls): Check call arguments against
23744 DECL_ARGUMENTS (current_function_decl) rather than
23745 DECL_ARGUMENTS (func) when checking for tail recursion.
23747 2018-03-07 Jakub Jelinek <jakub@redhat.com>
23749 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
23750 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
23751 Volker Reichelt's entry and add entries for people that perform
23752 GCC fuzzy testing and report numerous bugs.
23754 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23757 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
23758 readonly data in sdata, if that is disabled.
23759 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
23760 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
23761 -mreadonly-in-sdata option.
23763 2018-03-07 Martin Sebor <msebor@redhat.com>
23765 PR tree-optimization/84468
23766 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
23767 basic block when looking for nul assignment.
23769 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
23772 * except.h (output_function_exception_table): Adjust prototype.
23773 * except.c (output_function_exception_table): Remove FNNAME parameter
23774 and add SECTION parameter. Ouput one part of the table at a time.
23775 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
23776 the first part of the exception table and emit unwind directives.
23777 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
23778 (i386_pe_seh_cold_init): Likewise.
23779 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
23780 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
23781 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
23782 (ix86_output_call_insn): Emit a nop in one more case for SEH.
23783 * config/i386/winnt.c: Include except.h.
23784 (struct seh_frame_state): Add reg_offset, after_prologue and
23785 in_cold_section fields.
23786 (i386_pe_seh_end_prologue): Set seh->after_prologue.
23787 (i386_pe_seh_cold_init): New function.
23788 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
23789 to seh->in_cold_section.
23790 (seh_emit_push): Record the offset of the push.
23791 (seh_emit_save): Record the offet of the save.
23792 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
23793 Test seh->after_prologue to disregard the epilogue.
23794 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
23795 (i386_pe_end_cold_function): New function.
23797 2018-03-07 Jakub Jelinek <jakub@redhat.com>
23800 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
23801 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
23804 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
23806 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
23807 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
23809 PR middle-end/84723
23810 * multiple_target.c: Include tree-inline.h and intl.h.
23811 (expand_target_clones): Diagnose and fail if node->definition and
23812 !tree_versionable_function_p (node->decl).
23814 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
23816 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
23818 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
23819 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
23820 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
23822 2018-03-06 Jakub Jelinek <jakub@redhat.com>
23825 * combine.c (try_combine): Use reg_or_subregno instead of handling
23826 just paradoxical SUBREGs and REGs.
23828 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
23830 * config/arc/arc.c (arc_finalize_pic): Remove function.
23831 (arc_must_save_register): We use single base PIC register, remove
23832 checks to save/restore the PIC register.
23833 (arc_expand_prologue): Likewise.
23834 * config/arc/arc-protos.h (arc_set_default_type_attributes):
23836 (arc_verify_short): Likewise.
23837 (arc_attr_type): Likewise.
23838 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
23839 (walk_stores): Likewise.
23840 (arc_address_cost): Make it static.
23841 (arc_verify_short): Likewise.
23842 (branch_dest): Likewise.
23843 (arc_attr_type): Likewise.
23844 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
23845 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
23846 (arc_final_prescan_insn): Remove inserting the nops due to
23847 hardware hazards. It is done in reorg step.
23848 (insn_length_variant_t): Remove.
23849 (insn_length_parameters_t): Likewise.
23850 (arc_insn_length_parameters): Likewise.
23851 (arc_get_insn_variants): Likewise.
23852 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
23854 2018-03-06 Jakub Jelinek <jakub@redhat.com>
23856 PR inline-asm/84683
23857 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
23860 PR tree-optimization/84687
23861 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
23863 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
23865 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23867 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
23868 Rename to ppc_speculation_barrier.
23869 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
23870 __builtin_ppc_speculation_barrier.
23872 2018-03-05 Jakub Jelinek <jakub@redhat.com>
23875 * combine.c (combine_simplify_rtx): Don't try to simplify if
23876 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
23879 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
23881 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
23882 to 32 bytes when compiling for POWER9.
23884 2018-03-05 Jakub Jelinek <jakub@redhat.com>
23887 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
23888 regparm >= 3 with no arg reg available also for calls with
23889 flag_force_indirect_call. Pass decl to ix86_function_regparm.
23892 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
23894 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
23896 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
23899 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
23901 2018-03-05 Richard Biener <rguenther@suse.de>
23903 PR tree-optimization/84486
23904 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
23905 When inserting a __builtin_assume_aligned call set the LHS
23906 SSA name alignment info accordingly.
23908 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
23910 PR tree-optimization/84114
23911 * config/aarch64/aarch64.c (aarch64_reassociation_width)
23912 Avoid reassociation of FLOAT_MODE addition.
23914 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
23916 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
23917 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
23918 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
23919 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
23920 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
23921 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
23922 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
23924 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
23925 __builtin_ia32_wbinvd): New builtins.
23926 (SPECIAL_ARGS2): New.
23927 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
23928 (SPECIAL_ARGS2): New.
23929 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
23930 (ix86_valid_target_attribute_inner_p): Ditto.
23931 (ix86_init_mmx_sse_builtins): Add special_args2.
23932 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
23933 TARGET_WBNOINVD_P): New.
23934 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
23935 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
23936 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
23937 * config/i386/immintrin.h (_wbinvd): New intrinsic.
23938 * config/i386/pconfigintrin.h: New file.
23939 * config/i386/wbnoinvdintrin.h: Ditto.
23940 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
23942 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
23944 2018-03-05 Richard Biener <rguenther@suse.de>
23946 PR tree-optimization/84670
23947 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
23949 (BB_VISITED_WITH_VISITED_SUCCS): New define.
23950 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
23951 (compute_antic_aux): Only assert the number of values in ANTIC_IN
23952 doesn't grow if all successors (recursively) were visited at least
23955 2018-03-05 Richard Biener <rguenther@suse.de>
23957 PR tree-optimization/84650
23958 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
23959 if executed in the loop pipeline.
23961 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
23963 * doc/configfiles.texi (Configuration Files): Move info about
23964 conditionalizing $target-protos.h to...
23965 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
23966 differs from $target-protos.h.
23968 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
23969 Chung-Ju Wu <jasonwucj@gmail.com>
23971 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
23972 * config/nds32/nds32-multiple.md (setmemsi): Define.
23973 * config/nds32/nds32-memory-manipulation.c
23974 (nds32_gen_dup_4_byte_to_word_value): New.
23975 (emit_setmem_word_loop): New.
23976 (emit_setmem_byte_loop): New.
23977 (nds32_expand_setmem_loop): New.
23978 (nds32_expand_setmem_loop_v3m): New.
23979 (nds32_expand_setmem_unroll): New.
23980 (nds32_expand_setmem): New.
23982 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
23983 Chung-Ju Wu <jasonwucj@gmail.com>
23985 * config/nds32/nds32-memory-manipulation.c
23986 (nds32_emit_load_store): New.
23987 (nds32_emit_post_inc_load_store): New.
23988 (nds32_emit_mem_move): New.
23989 (nds32_emit_mem_move_block): New.
23990 (nds32_expand_movmemsi_loop_unknown_size): New.
23991 (nds32_expand_movmemsi_loop_known_size): New.
23992 (nds32_expand_movmemsi_loop): New.
23993 (nds32_expand_movmemsi_unroll): New.
23994 (nds32_expand_movmemqi): Rename ...
23995 (nds32_expand_movmemsi): ... to this.
23996 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
23997 (movmemsi): ... to this.
23998 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
23999 (nds32_expand_movmemsi): ... to this.
24001 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24002 Monk Chiang <sh.chiang04@gmail.com>
24003 Chung-Ju Wu <jasonwucj@gmail.com>
24005 * config/nds32/nds32-protos.h
24006 (nds32_expand_load_multiple): New arguments.
24007 (nds32_expand_store_multiple): Ditto.
24008 (nds32_valid_multiple_load_store): Rename ...
24009 (nds32_valid_multiple_load_store_p): ... to this.
24010 * config/nds32/nds32-memory-manipulation.c
24011 (nds32_expand_load_multiple): Refine implementation.
24012 (nds32_expand_store_multiple): Ditto.
24013 * config/nds32/nds32-multiple.md
24014 (load_multiple): Update nds32_expand_load_multiple interface.
24015 (store_multiple): Update nds32_expand_store_multiple interface.
24016 * config/nds32/nds32-predicates.c
24017 (nds32_valid_multiple_load_store): Rename ...
24018 (nds32_valid_multiple_load_store_p): ... to this and refine
24020 * config/nds32/predicates.md
24021 (nds32_load_multiple_and_update_address_operation): New predicate.
24022 (nds32_store_multiple_and_update_address_operation): New predicate.
24024 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24025 Chung-Ju Wu <jasonwucj@gmail.com>
24027 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
24028 (combo): New attribute.
24029 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
24031 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
24033 * config/nds32/nds32.opt: Change -mcmodel= default value.
24035 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
24036 Monk Chiang <sh.chiang04@gmail.com>
24037 Chung-Ju Wu <jasonwucj@gmail.com>
24039 * config/nds32/constants.md (unspec_element): New enum.
24040 * config/nds32/constraints.md (Umw): New constraint.
24041 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
24042 * config/nds32/nds32-intrinsic.md: Likewise.
24043 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
24044 (nds32_valid_smw_lwm_base_p): New.
24045 (nds32_output_smw_single_word): New.
24046 (nds32_output_lmw_single_word): New.
24047 (nds32_expand_unaligned_load): New.
24048 (nds32_expand_unaligned_store): New.
24049 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
24050 (nds32_output_smw_single_word): Declare.
24051 (nds32_output_lmw_single_word): Declare.
24052 (nds32_expand_unaligned_load): Declare.
24053 (nds32_expand_unaligned_store): Declare.
24054 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
24055 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
24056 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
24057 NDS32_BUILTIN_UASTORE_DW.
24058 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
24061 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
24062 Kito Cheng <kito.cheng@gmail.com>
24063 Chung-Ju Wu <jasonwucj@gmail.com>
24065 * config/nds32/nds32-intrinsic.c
24066 (nds32_expand_builtin_null_ftype_reg): Delete.
24067 (nds32_expand_builtin_reg_ftype_imm): Ditto.
24068 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
24069 (nds32_read_argument): New.
24070 (nds32_legitimize_target): Ditto.
24071 (nds32_legitimize_argument): Ditto.
24072 (nds32_check_constant_argument): Ditto.
24073 (nds32_expand_unop_builtin): Ditto.
24074 (nds32_expand_unopimm_builtin): Ditto.
24075 (nds32_expand_binop_builtin): Ditto.
24076 (nds32_builtin_decl_impl): Ditto.
24077 (builtin_description): Ditto.
24078 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
24079 (nds32_init_builtins_impl): Ditto.
24080 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
24081 (nds32_builtin_decl): New.
24082 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
24083 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
24085 2018-03-02 Jeff Law <law@redhat.com>
24087 * reorg.c (stop_search_p): Handle DEBUG_INSN.
24088 (redundant_insn, fill_simple_delay_slots): Likewise.
24089 (fill_slots_from_thread): Likewise.
24090 * resource.c (mark_referenced_resources): Likewise.
24091 (mark_set_resources, find_dead_or_set_registers): Likewise.
24093 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24095 * substring-locations.h (format_warning_va): Formatting fix for
24096 ATTRIBUTE_GCC_DIAG.
24097 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
24099 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
24100 * substring-locations.c: Include intl.h.
24101 (format_warning_va): Turned into small wrapper around
24102 format_warning_n_va, renamed to ...
24103 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
24104 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
24106 (format_warning_at_substring_n): New function.
24107 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
24108 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
24109 format_warning_at_substring with just a shorter name instead of
24110 const function pointer.
24111 (fmtwarn_n): New function.
24112 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
24113 appropriate, get rid of all the fmtstr temporaries, move conditionals
24114 with G_() wrapped string literals directly into fmtwarn arguments,
24115 cast dir.len to (int), formatting fixes.
24117 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
24119 * doc/invoke.texi: Remove "Cilk Plus" references.
24121 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24122 Richard Biener <rguenther@suse.de>
24125 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
24126 for error or warning attributes if CALL_FROM_THUNK_P is set.
24129 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24132 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
24133 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
24136 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
24137 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
24139 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
24140 instead of -1U in last predictors element's probability member.
24142 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
24145 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
24146 arguments if they are comparable.
24148 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
24150 PR tree-optimization/84634
24151 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
24152 masks and masked_loop_p with a single loop_masks, making sure it's
24153 null for bb vectorization.
24155 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
24157 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
24158 (vect_analyze_data_ref_access): Use loop->safe_len rather than
24159 loop->force_vectorize to check whether there is no alias.
24161 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24164 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
24166 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
24168 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
24169 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
24170 instead of a loop around prev_real_insn.
24171 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
24174 PR inline-asm/84625
24175 * config/i386/i386.c (ix86_print_operand): Use conditional
24176 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
24179 2018-03-02 Richard Biener <rguenther@suse.de>
24181 PR tree-optimization/84427
24182 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
24183 (bitmap_set_subtract_values): Rewrite to handle multiple
24186 (prune_clobbered_mems): Likewise.
24187 (phi_translate): Take edge instead of pred/phiblock.
24188 (phi_translate_1): Likewise.
24189 (phi_translate_set): Likewise. Insert all translated
24190 exprs for a value into the set, keeping possibly multiple
24191 expressions per value.
24192 (compute_antic_aux): Adjust for phi_translate changes.
24193 When intersecting union the expressions and prune those
24194 not in the final value set, keeping possibly multiple
24195 expressions per value. Do not use value-insertion
24196 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
24197 all expressions. Add verification that the value-sets
24198 only shrink during iteration.
24199 (compute_partial_antic_aux): Adjust for the phi_translate changes.
24200 (do_pre_regular_insertion): Likewise.
24201 (do_pre_partial_partial_insertion): Likewise.
24203 2018-03-02 Richard Biener <rguenther@suse.de>
24206 * config/darwin.c (saved_debug_info_level): New static global.
24207 (darwin_asm_lto_start): Disable debug info generation for LTO out.
24208 (darwin_asm_lto_end): Restore debug info generation settings.
24210 2018-03-01 Martin Liska <mliska@suse.cz>
24213 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
24214 volatile arguments.
24216 2018-03-01 Richard Biener <rguenther@suse.de>
24219 * dwarf2out.c (gen_variable_die): Properly handle late VLA
24220 type annotation with LTO when debug was disabled at compile-time.
24222 2018-03-01 Matthew Fortune <mfortune@gmail.com>
24224 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
24226 (mips_final_postscan_insn): Likewise.
24228 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
24230 PR rtl-optimization/84528
24231 * alias.c (init_alias_target): Add commentary.
24232 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
24233 a unique base value if the frame pointer is not eliminated
24234 to the stack pointer.
24236 2018-03-01 Tom de Vries <tom@codesourcery.com>
24238 PR rtl-optimization/83327
24239 * lra-int.h (hard_regs_spilled_into): Declare.
24240 * lra.c (hard_regs_spilled_into): Define.
24241 (init_reg_info): Init hard_regs_spilled_into.
24242 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
24243 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
24244 (process_bb_lives): Handle hard_regs_spilled_into.
24245 (lra_create_live_ranges_1): Before doing liveness propagation, clear
24246 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
24248 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
24250 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
24251 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
24252 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
24253 * config/rs6000/aix72.h: New file.
24255 2018-02-28 Jakub Jelinek <jakub@redhat.com>
24257 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
24258 instead of warning_at with conditional singular and plural messages
24262 * stor-layout.c (update_alignment_for_field): For
24263 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
24264 && !DECL_PACKED (field), do the alignment update, just use
24265 only desired_align instead of MAX (type_align, desired_align)
24267 (place_field): Don't do known_align < desired_align handling
24268 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
24269 is non-NULL, instead do it after rli->prev_field handling and
24270 only if not within a bitfield word. For DECL_PACKED (field)
24271 use type_align of BITS_PER_UNIT.
24273 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
24275 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
24276 superfluous parentheses and trailing spaces.
24278 2018-02-28 Richard Biener <rguenther@suse.de>
24280 PR tree-optimization/84584
24281 * graphite-scop-detection.c (scop_detection::add_scop): Discard
24282 SCoPs with fake exit edge.
24284 2018-02-28 Martin Liska <mliska@suse.cz>
24287 * timevar.c (timer::print): Fix format to properly print 100%
24290 2018-02-28 Richard Biener <rguenther@suse.de>
24292 PR middle-end/84607
24293 * genmatch.c (capture_info::walk_match): Do not mark
24294 captured expressions without operands as expr_p given
24295 they act more like predicates and should be subject to
24296 "lost tail" side-effect preserving.
24298 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
24300 PR rtl-optimization/81611
24301 * auto-inc-dec.c (attempt_change): Move dead note from
24302 mem_insn if it's the next use of regno
24303 (find_address): Take address use of reg holding
24304 non-incremented value. Add parm to limit search to the named
24306 (merge_in_block): Attempt to use a mem insn that is the next
24307 use of the original regno.
24309 2018-02-27 Martin Sebor <msebor@redhat.com>
24312 * doc/invoke.texi (-Wmissing-attributes): New option.
24313 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
24315 2018-02-27 Martin Sebor <msebor@redhat.com>
24317 PR translation/84207
24318 * diagnostic-core.h (warning_n, error_n, inform_n): Change
24319 n argument to unsigned HOST_WIDE_INT.
24320 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
24321 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
24322 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
24323 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
24325 2018-02-27 Richard Biener <rguenther@suse.de>
24327 PR tree-optimization/84512
24328 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
24329 Do not use the estimate returned from record_stmt_cost for
24330 the scalar iteration cost but sum properly using add_stmt_cost.
24332 2018-02-27 Richard Biener <rguenther@suse.de>
24334 PR tree-optimization/84466
24335 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
24336 Adjust last change to less strictly validate use operands.
24338 2018-02-27 Martin Liska <mliska@suse.cz>
24340 PR gcov-profile/84548
24341 * gcov.c (process_file): Allow partial overlap and consider it
24342 also as group functions.
24343 (output_lines): Properly calculate range of lines for a group.
24345 2018-02-27 Martin Liska <mliska@suse.cz>
24347 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
24348 'ggc' suffixes. Change first column width.
24349 (timer::print): Fix formatting of the column.
24351 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
24353 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
24354 preserve inline entry blocks for the sake of debug inline
24355 entry point markers alone.
24356 (remove_unused_locals): Suggest in comments a better place to
24357 force the preservation of inline entry blocks that are
24358 otherwise unused, but do not preserve them.
24360 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24362 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
24364 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24367 * config/i386/constraints.md (Bs): Replace
24368 ix86_indirect_branch_register with
24369 TARGET_INDIRECT_BRANCH_REGISTER.
24371 * config/i386/i386.md (indirect_jump): Likewise.
24372 (tablejump): Likewise.
24373 (*sibcall_memory): Likewise.
24374 (*sibcall_value_memory): Likewise.
24375 Peepholes of indirect call and jump via memory: Likewise.
24376 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
24377 (*sibcall_value_GOT_32): Likewise.
24378 * config/i386/predicates.md (indirect_branch_operand): Likewise.
24379 (GOT_memory_operand): Likewise.
24380 (call_insn_operand): Likewise.
24381 (sibcall_insn_operand): Likewise.
24382 (GOT32_symbol_operand): Likewise.
24383 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
24385 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
24387 PR rtl-optimization/83496
24388 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
24389 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
24390 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
24391 redundant insn, if any.
24392 (relax_delay_slots): Likewise.
24393 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
24395 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
24397 PR tree-optimization/83965
24398 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
24399 that grouped statements are part of a reduction chain. Return
24400 true if the statement is not marked as a reduction itself but
24401 is part of a group.
24402 (vect_recog_dot_prod_pattern): Don't check whether the statement
24403 is part of a group here.
24404 (vect_recog_sad_pattern): Likewise.
24405 (vect_recog_widen_sum_pattern): Likewise.
24407 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
24410 * final.c (rest_of_clean_state): Also look for calls inside sequences.
24412 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24415 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
24417 (ix86_output_indirect_function_return): New prototype.
24418 (ix86_split_simple_return_pop_internal): Likewise.
24419 * config/i386/i386.c (indirect_return_via_cx): New.
24420 (indirect_return_via_cx_bnd): Likewise.
24421 (indirect_thunk_name): Handle return va CX_REG.
24422 (output_indirect_thunk_function): Create alias for
24423 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
24424 (ix86_output_indirect_jmp): Remove the bool argument.
24425 (ix86_output_indirect_function_return): New function.
24426 (ix86_split_simple_return_pop_internal): Likewise.
24427 * config/i386/i386.md (*indirect_jump): Don't pass false
24428 to ix86_output_indirect_jmp.
24429 (*tablejump_1): Likewise.
24430 (simple_return_pop_internal): Change it to define_insn_and_split.
24431 Call ix86_split_simple_return_pop_internal to split it for
24432 -mfunction-return=.
24433 (simple_return_indirect_internal): Call
24434 ix86_output_indirect_function_return instead of
24435 ix86_output_indirect_jmp.
24437 2018-02-26 Jakub Jelinek <jakub@redhat.com>
24440 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
24441 memset and value initialization afterwards.
24443 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
24445 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
24447 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24450 * common/config/aarch64/aarch64-common.c
24451 (aarch_option_optimization_table[]): Switch
24452 off fomit-frame-pointer
24454 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
24455 Chung-Ju Wu <jasonwucj@gmail.com>
24457 * config/nds32/nds32-multiple.md (load_multiple): Disallow
24459 (store_multiple): Ditto.
24461 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
24463 * config.gcc: Add --with-cpu support for nds32 target.
24464 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
24465 * config/nds32/nds32.opt: Add -mcpu= option.
24467 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
24469 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
24470 isel=yes): Warn for these deprecated options.
24472 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
24474 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
24475 ISA_2_5_MASKS_EMBEDDED.
24477 2018-02-23 Jakub Jelinek <jakub@redhat.com>
24479 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
24480 p->max as pointers rather than using iterative_hash_expr.
24482 2018-02-23 Carl Love <cel@us.ibm.com>
24484 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
24485 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
24487 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
24488 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
24489 P8V_BUILTIN_VEC_VUNSIGNED2.
24491 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
24494 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
24495 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
24497 * lra-constraints.c (curr_insn_transform): Set up
24498 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
24500 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
24501 * lra-eliminations.c (spill_pseudos): Ditto.
24502 (process_insn_for_elimination): Ditto.
24503 * lra-lives.c (reg_early_clobber_p): Use the new macros.
24504 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
24505 LRA_NON_CLOBBERED_ALT.
24507 2018-02-22 Martin Sebor <msebor@redhat.com>
24509 PR tree-optimization/84480
24510 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
24511 to maybe_diag_stxncpy_trunc. Call it.
24512 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
24513 from gimple_fold_builtin_strcpy. Print inlining stack.
24514 (handle_builtin_stxncpy): Print inlining stack.
24515 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
24517 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
24520 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
24521 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
24522 and -fcheck-pointer-bounds are used together.
24523 (indirect_thunk_prefix): New enum.
24524 (indirect_thunk_need_prefix): New function.
24525 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
24526 "_nt" instead of "_bnd" for NOTRACK prefix.
24527 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
24528 (output_indirect_thunk_function): Likewise.
24530 (ix86_code_end): Update output_indirect_thunk_function calls.
24531 (ix86_output_indirect_branch_via_reg): Replace
24532 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
24533 (ix86_output_indirect_branch_via_push): Likewise.
24534 (ix86_output_function_return): Likewise.
24535 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
24536 incompatible with -fcf-protection=branch and
24537 -fcheck-pointer-bounds.
24539 2018-02-22 Steve Ellcey <sellcey@cavium.com>
24542 * config/aarch64/aarch64.c (aarch64_print_address_internal):
24543 Change gcc_assert call to output_operand_lossage.
24545 2018-02-22 Steve Ellcey <sellcey@cavium.com>
24547 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
24549 2018-02-22 DJ Delorie <dj@redhat.com>
24550 Sebastian Perta <sebastian.perta@renesas.com>
24551 Oleg Endo <olegendo@gcc.gnu.org>
24553 * config/rx/rx.c (rx_rtx_costs): New function.
24554 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
24556 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24558 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
24560 2018-02-22 Martin Liska <mliska@suse.cz>
24563 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
24564 Add "native" as a possible value.
24566 2018-02-22 Martin Liska <mliska@suse.cz>
24569 * config/i386/i386.c (ix86_option_override_internal):
24570 Add "native" as a possible value for -march and -mtune.
24572 2018-02-22 Jakub Jelinek <jakub@redhat.com>
24575 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
24576 to all type variants.
24578 PR tree-optimization/84503
24579 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
24580 width as info->bitpos + info->bitsize - start.
24581 (merged_store_group::merge_overlapping): Simplify width computation.
24582 (check_no_overlap): New function.
24583 (imm_store_chain_info::try_coalesce_bswap): Compute expected
24584 start + width and last_order of the group, fail if check_no_overlap
24586 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
24587 to group if check_no_overlap fails.
24589 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
24591 * config/rs6000/altivec.md: Delete contraint arguments to
24592 define_expand, define_split, and define_peephole2, and in
24593 define_insn_and_split if always unused.
24594 * config/rs6000/darwin.md: Ditto.
24595 * config/rs6000/dfp.md: Ditto.
24596 * config/rs6000/rs6000.md: Ditto.
24597 * config/rs6000/sync.md: Ditto.
24598 * config/rs6000/vector.md: Ditto.
24599 * config/rs6000/vsx.md: Ditto.
24601 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
24603 * config/rs6000/altivec.md: Write output control strings as braced
24604 blocks instead of double-quoted strings.
24605 * config/rs6000/darwin.md: Ditto.
24606 * config/rs6000/rs6000.md: Ditto.
24607 * config/rs6000/vector.md: Ditto.
24608 * config/rs6000/vsx.md: Ditto.
24610 2018-02-21 Jason Merrill <jason@redhat.com>
24612 PR c++/84314 - ICE with templates and fastcall attribute.
24613 * attribs.c (build_type_attribute_qual_variant): Remove assert.
24615 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
24617 * ipa-cp.c (determine_versionability): Fix comment typos.
24619 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
24622 * ipa-cp.c (determine_versionability): Do not version functions caling
24625 2018-02-21 Martin Liska <mliska@suse.cz>
24628 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
24629 Add "native" as a possible value.
24630 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
24631 the macro when native cpu detection is available.
24633 2018-02-21 Martin Liska <mliska@suse.cz>
24636 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
24637 Add "native" as a possible value.
24638 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
24639 when native cpu detection is available.
24641 2018-02-21 Jakub Jelinek <jakub@redhat.com>
24642 Martin Sebor <msebor@redhat.com>
24644 PR tree-optimization/84478
24645 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
24647 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
24648 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
24649 support which is conservatively correct, for 2 only stay conservative
24650 for maxlen. Formatting and comment capitalization fixes. Add STRICT
24651 argument to the 2 argument get_range_strlen, adjust 6 arg
24652 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
24654 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
24655 (gimple_fold_builtin_strlen): Pass true as last argument to
24658 2018-02-20 Martin Sebor <msebor@redhat.com>
24660 PR middle-end/84095
24661 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
24662 (builtin_memref::set_base_and_offset): Same. Handle inner references.
24663 (builtin_memref::builtin_memref): Factor out parts into
24664 set_base_and_offset and call it.
24666 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
24668 PR middle-end/84406
24669 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
24670 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
24671 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
24672 search at the associated MODE_INT.
24674 2018-02-20 Jeff Law <law@redhat.com>
24676 PR middle-end/82123
24677 PR tree-optimization/81592
24678 PR middle-end/79257
24679 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
24680 for range data rather than using global data.
24681 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
24682 range data rather than using global data.
24683 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
24684 pass it to children as needed.
24685 (struct directive::fmtresult): Similarly.
24686 (struct directive::set_width): Similarly.
24687 (struct directive::set_precision): Similarly.
24688 (format_integer, format_directive, parse_directive): Similarly.
24689 (format_none): Accept unnamed vr_values parameter.
24690 (format_percent, format_floating, format_character): Similarly.
24691 (format_string, format_plain): Similarly.
24692 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
24693 the EVRP range analyzer for range data rather than using global data.
24694 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
24695 gimple-ssa-evrp-analyze.h
24696 (class sprintf_dom_walker): Add after_dom_children member function.
24697 Add evrp_range_analyzer member.
24698 (sprintf_dom_walker::before_dom_children): Call into the EVRP
24699 range analyzer as needed.
24700 (sprintf_dom_walker::after_dom_children): New member function.
24701 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
24703 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
24704 (evrp_range_analyzer::pop_to_marker): Likewise.
24706 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
24708 PR tree-optimization/84419
24709 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
24710 with the required type if its current type is compatible but
24713 2018-02-20 Jakub Jelinek <jakub@redhat.com>
24715 PR middle-end/82004
24716 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
24717 after vectorization.
24719 2018-02-20 Martin Liska <mliska@suse.cz>
24722 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
24723 possible values if we don't have a hint.
24725 2018-02-20 Martin Liska <mliska@suse.cz>
24729 * final.c (shorten_branches): Build align_tab array with one
24731 * opts.c (finish_options): Add alignment option limit check.
24732 (MAX_CODE_ALIGN): Likewise.
24733 (MAX_CODE_ALIGN_VALUE): Likewise.
24734 * doc/invoke.texi: Document maximum allowed option value for
24735 all -falign-* options.
24737 2018-02-19 Jakub Jelinek <jakub@redhat.com>
24740 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
24741 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
24742 * var-tracking.c (emit_note_insn_var_location): Remove all references
24743 to NOTE_INSN_CALL_ARG_LOCATION.
24744 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
24745 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
24746 Use copy_rtx_if_shared.
24747 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
24748 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
24749 (dwarf2out_var_location): Remove handling of
24750 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
24752 * final.c (final_scan_insn): Remove all references to
24753 NOTE_INSN_CALL_ARG_LOCATION.
24754 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
24755 before dumping final insns.
24756 * except.c (emit_note_eh_region_end): Remove all references to
24757 NOTE_INSN_CALL_ARG_LOCATION.
24758 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
24759 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
24760 * config/arc/arc.c (hwloop_optimize): Likewise.
24761 * config/arm/arm.c (create_fix_barrier): Likewise.
24762 * config/s390/s390.c (s390_chunkify_start): Likewise.
24763 * config/sh/sh.c (find_barrier): Likewise.
24764 * config/i386/i386.c (rest_of_insert_endbranch,
24765 ix86_seh_fixup_eh_fallthru): Likewise.
24766 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
24767 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
24768 * config/frv/frv.c (frv_function_prologue): Likewise.
24769 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
24771 (note_outside_basic_block_p): Remove all references to
24772 NOTE_INSN_CALL_ARG_LOCATION.
24773 * gengtype.c (adjust_field_rtx_def): Likewise.
24774 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
24776 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
24777 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
24780 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
24783 PR tree-optimization/84452
24784 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
24785 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
24788 2018-02-19 Martin Liska <mliska@suse.cz>
24791 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
24793 2018-02-19 Martin Liska <mliska@suse.cz>
24794 Richard Sandiford <richard.sandiford@linaro.org>
24796 PR tree-optimization/82491
24797 * gimple-fold.c (get_base_constructor): Make earlier bail out
24800 2018-02-19 Carl Love <cel@us.ibm.com>
24802 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
24803 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
24805 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
24806 P8V_BUILTIN_VEC_NEG.
24808 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
24810 * config/rl78/rl78.md (movdf): New define expand.
24812 2018-02-19 Martin Liska <mliska@suse.cz>
24815 * doc/invoke.texi: Fix typo.
24816 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
24818 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
24820 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
24821 handle rs6000_single_float and rs6000_double_float specially for
24824 2018-02-16 Jeff Law <law@redhat.com>
24826 * config/rx/rx.c (add_pop_cfi_notes): New function.;
24827 (pop_regs): Use it.
24829 2018-02-16 Jakub Jelinek <jakub@redhat.com>
24832 * ipa-inline.c (inline_small_functions): Fix a typo.
24834 2018-02-16 Nathan Sidwell <nathan@acm.org>
24836 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
24838 2018-02-16 Carl Love <cel@us.ibm.com>
24840 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
24841 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
24842 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
24843 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
24844 expansion to P8V_BUILTIN_VEC_FLOAT2.
24846 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
24848 PR rtl-optimization/70023
24849 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
24850 src_regno into account.
24852 2018-02-16 Carl Love <cel@us.ibm.com>
24854 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
24855 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
24856 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
24857 * config/rs6000/rs6000.c: Remove case statements for
24858 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
24859 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
24860 and P9V_BUILTIN_VEC_VINSERT4B.
24861 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
24862 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
24863 * config/rs6000/vsx.md:
24864 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
24867 2018-02-16 Carl Love <cel@us.ibm.com>
24869 * config/rs6000/altivec.h: Add builtin names vec_extract4b
24871 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
24873 * config/rs6000/rs6000-c.c: Add the definitions for
24874 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
24875 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
24876 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
24877 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
24878 definition for insert4b and define insn *insert3b_internal.
24879 * doc/extend.texi: Add documentation for vec_extract4b.
24881 2018-02-16 Nathan Sidwell <nathan@acm.org>
24883 * doc/extend.texi (Backwards Compatibility): Mention friend
24884 injection. Note for-scope is deprecated.
24885 * doc/invoke.texi (-ffriend-injection): Deprecate.
24887 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
24889 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
24890 that moved to I2, also allow destinations that are a paradoxical
24891 subreg (instead of a normal reg).
24893 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
24896 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
24899 2018-02-16 Richard Biener <rguenther@suse.de>
24901 PR tree-optimization/84037
24902 PR tree-optimization/84016
24904 * config/i386/i386.c (ix86_builtin_vectorization_cost):
24905 Adjust vec_construct for the fact we need additional higher latency
24906 128bit inserts for AVX256 and AVX512 vector builds.
24907 (ix86_add_stmt_cost): Scale vector construction cost for
24910 2018-02-16 Richard Biener <rguenther@suse.de>
24912 PR tree-optimization/84417
24913 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
24914 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
24915 (non_rewritable_lvalue_p): Likewise, use poly-ints.
24917 2018-02-16 Martin Liska <mliska@suse.cz>
24920 * internal-fn.def (ASAN_CHECK): Set proper flags.
24921 (ASAN_MARK): Likewise.
24923 2018-02-16 Julia Koval <julia.koval@intel.com>
24925 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
24926 from PTA_CANNONLAKE.
24928 2018-02-16 Jakub Jelinek <jakub@redhat.com>
24931 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
24932 Use ++iter rather than iter++ for std::list iterators.
24933 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
24934 defer deleting them until all nodes in the forest are processed. Do
24935 free even leaf nodes. Change to_process into auto_vec.
24938 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
24939 * vec.h (vec_default_construct): Use memset instead of placement new
24940 if BROKEN_VALUE_INITIALIZATION is defined.
24941 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
24942 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
24945 PR rtl-optimization/83723
24946 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
24947 * lra.c (lra_substitute_pseudo): Likewise. If true, use
24948 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
24950 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
24952 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
24954 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
24956 PR rtl-optimization/81443
24957 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
24958 from inner REGs to paradoxical SUBREGs.
24960 2018-02-16 Richard Biener <rguenther@suse.de>
24962 PR tree-optimization/84399
24963 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
24964 For operands we can analyze at their definition make sure we can
24965 analyze them at each use as well.
24967 2018-02-16 Richard Biener <rguenther@suse.de>
24969 PR tree-optimization/84190
24970 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
24971 volatile accesses if the decl isn't volatile.
24973 2018-02-15 Jason Merrill <jason@redhat.com>
24975 PR c++/84314 - ICE with templates and fastcall attribute.
24976 * attribs.c (build_type_attribute_qual_variant): Don't clobber
24977 TYPE_CANONICAL on an existing type.
24979 2018-02-15 Jakub Jelinek <jakub@redhat.com>
24981 PR tree-optimization/84383
24982 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
24983 dstoff nor call operand_equal_p if dstbase is NULL.
24985 PR tree-optimization/84334
24986 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
24987 also a CONSTANT_CLASS_P, punt.
24989 2018-02-14 Jim Wilson <jimw@sifive.com>
24991 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
24992 first SMALL_OPERAND check. New local min_second_step. Move assert
24993 to where locals are set. Add TARGET_RVC support.
24994 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
24996 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
24998 * doc/invoke.texi: Correct -Wformat-overflow code sample.
25000 2018-02-14 Martin Sebor <msebor@redhat.com>
25002 PR tree-optimization/83698
25003 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
25004 arrays constrain the offset range to their bounds.
25005 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
25006 (builtin_access::overlap): Avoid setting the size of overlap if it's
25008 (maybe_diag_overlap): Also consider arrays when deciding what values
25009 of offsets to include in diagnostics.
25011 2018-02-14 Martin Sebor <msebor@redhat.com>
25014 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
25015 that correspond to the kind of a declaration.
25017 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
25020 * config/pa/pa.md: Load address of PIC label using the linkage table
25021 if the label is nonlocal.
25023 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
25025 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
25026 warning message if user requests -maltivec=be.
25027 * doc/invoke.texi: Document deprecation of -maltivec=be.
25029 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
25032 * config/rs6000/rs6000-c.c: Update definitions for
25033 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
25034 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
25036 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25039 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
25040 add _get_ssp intrinsics. Remove argument from
25041 __builtin_ia32_rdssp[d|q].
25042 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
25043 * config/i386/i386-builtin.def: Remove argument from
25044 __builtin_ia32_rdssp[d|q].
25045 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
25046 ix86_expand_special_args_builtin for _rdssp[d|q].
25047 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
25048 Clear register before usage.
25049 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
25050 Add documentation for new _get_ssp and _inc_ssp intrinsics.
25052 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
25054 PR tree-optimization/84357
25055 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
25056 operand 1 of an ARRAY_REF too.
25058 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
25061 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
25062 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
25064 (set_of_reg): New struct.
25065 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
25066 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
25067 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
25069 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
25070 Split into bitclr, bitset, bitinvert patterns if appropriate.
25071 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
25072 use rx_fuse_in_memory_bitop.
25073 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
25074 to named insn, correct maximum insn length.
25076 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
25079 * machmode.def: Define a complex mode for PARTIAL_INT.
25080 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
25082 * doc/rtl.texi: Document CSPImode.
25083 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
25085 (msp430_hard_regno_nregs_with_padding): Likewise.
25087 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
25090 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
25092 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25094 PR rtl-optimization/84169
25095 * combine.c (try_combine): New variable split_i2i3. Set it to true if
25096 we generated a parallel as new i3 and we split that to new i2 and i3
25097 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
25098 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
25099 those to i2, not i1. Partially rewrite this scan code.
25101 2018-02-13 Jakub Jelinek <jakub@redhat.com>
25104 * stor-layout.c (place_field): For variable length fields, adjust
25105 offset_align afterwards not just based on the field's alignment,
25106 but also on the size.
25108 PR middle-end/84309
25109 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
25110 of exps and logs in the use_exp2 case.
25112 2018-02-13 Jeff Law <law@redhat.com>
25114 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
25115 entry for "vector".
25117 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
25120 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
25124 * common.opt (gas-loc-support, gas-locview-support): New.
25125 (ginline-points, ginternal-reset-location-views): New.
25126 * doc/invoke.texi: Document them. Use @itemx where intended.
25127 (gvariable-location-views): Adjust.
25128 * target.def (reset_location_view): New.
25129 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
25130 (TARGET_RESET_LOCATION_VIEW): New.
25131 * doc/tm.texi: Rebuilt.
25132 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
25133 (dwarf2out_default_as_locview_support): New.
25134 (output_asm_line_debug_info): Use option variables.
25135 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
25136 (output_loc_list): Likewise.
25137 (add_high_low_attributes): Check option variables.
25138 Don't output entry view attribute in strict mode.
25139 (gen_inlined_subroutine_die): Check option variables.
25140 (dwarf2out_inline_entry): Likewise.
25141 (init_sections_and_labels): Likewise.
25142 (dwarf2out_early_finish): Likewise.
25143 (maybe_reset_location_view): New, from...
25144 (dwarf2out_var_location): ... here. Call it.
25145 * debug.h (dwarf2out_default_as_loc_support): Declare.
25146 (dwarf2out_default_as_locview_support): Declare.
25147 * hooks.c (hook_int_rtx_insn_0): New.
25148 * hooks.h (hook_int_rtx_insn_0): Declare.
25149 * toplev.c (process_options): Take -gas-loc-support and
25150 -gas-locview-support from dwarf2out. Enable
25151 -gvariable-location-views by default only with locview
25152 assembler support. Enable -ginternal-reset-location-views by
25153 default only if the target defines the corresponding hook.
25154 Enable -ginline-points by default if location views are
25155 enabled; force it disabled if statement frontiers are
25157 * tree-inline.c (expand_call_inline): Check option variables.
25158 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
25160 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
25162 PR tree-optimization/84321
25163 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
25164 handling. Also check whether the anti-range contains any values
25165 that satisfy the mask; switch to a VR_RANGE if not.
25167 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
25170 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
25172 2018-02-13 Martin Jambor <mjambor@suse.cz>
25175 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
25176 of call statements, also set location of a load to a temporary.
25178 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
25180 * config/rl78/rl78.c (add_vector_labels): New function.
25181 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
25182 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
25183 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
25184 which checks that no arguments are passed.
25185 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
25186 * doc/extend.texi: Documentation for the new attribute.
25188 2018-02-13 Andreas Schwab <schwab@suse.de>
25190 * config/riscv/linux.h (CPP_SPEC): Define.
25192 2018-02-13 Jakub Jelinek <jakub@redhat.com>
25195 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
25196 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
25197 OPTION_MASK_ISA_AES as first argument to def_builtin_const
25198 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
25199 instead of OPTION_MASK_ISA_PCLMUL as first argument to
25200 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
25201 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
25202 temporarily for AES and PCLMUL builtins.
25204 PR tree-optimization/84339
25205 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
25206 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
25209 PR middle-end/84309
25210 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
25211 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
25212 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
25214 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
25216 * omp-simd-clone.h: New file.
25217 * omp-simd-clone.c: Include omp-simd-clone.h.
25218 (expand_simd_clones): No longer static.
25219 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
25220 cgraph.h and omp-simd-clone.h.
25221 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
25222 (vect_recog_widen_shift_pattern): Formatting fix.
25223 (vect_pattern_recog_1): Don't check optab for calls.
25226 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
25227 operands[2] into a REG before using gen_lowpart on it.
25229 2018-02-12 Jeff Law <law@redhat.com>
25232 * config/sh/sh.c (find_barrier): Consider a sibling call
25235 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
25236 successfully back substituting a reg.
25238 2018-02-12 Richard Biener <rguenther@suse.de>
25240 PR tree-optimization/84037
25241 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
25242 parameter, move visited init to caller.
25243 (vect_slp_analyze_operations): Separate cost from validity
25244 check, initialize visited once for all instances.
25245 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
25247 * tree-vect-stmts.c (vect_model_simple_cost): Make early
25249 (vect_model_promotion_demotion_cost): Likewise.
25250 (vectorizable_bswap): Guard cost modeling with !slp_node
25251 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
25253 (vectorizable_call): Likewise.
25254 (vectorizable_conversion): Likewise.
25255 (vectorizable_assignment): Likewise.
25256 (vectorizable_shift): Likewise.
25257 (vectorizable_operation): Likewise.
25258 (vectorizable_store): Likewise.
25259 (vectorizable_load): Likewise.
25260 (vectorizable_condition): Likewise.
25261 (vectorizable_comparison): Likewise.
25263 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
25266 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
25267 (ASAN_MARK): Fix fnspec to account for return value, change pointer
25268 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
25270 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
25272 PR middle-end/83665
25273 * params.def (inline-min-speedup): Increase from 8 to 15.
25274 (max-inline-insns-auto): Decrease from 40 to 30.
25275 * ipa-split.c (consider_split): Add some buffer for function to
25276 be considered inlining candidate.
25277 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
25280 2018-02-12 Richard Biener <rguenther@suse.de>
25282 PR tree-optimization/84037
25283 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
25284 matched stmts if we cannot swap the non-matched ones.
25286 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
25288 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
25289 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
25290 _mm_maskz_scalef_round_ss): New intrinsics.
25291 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
25292 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
25293 __builtin_ia32_scalefss_round): Remove.
25294 (__builtin_ia32_scalefsd_mask_round,
25295 __builtin_ia32_scalefss_mask_round): New intrinsics.
25296 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
25297 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
25298 ((match_operand:VF_128 2 "<round_nimm_predicate>"
25299 "<round_constraint>")): Changed to ...
25300 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
25301 "<round_scalar_constraint>")): ... this.
25302 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
25303 %0, %1, %2<round_op3>}"): Changed to ...
25304 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
25305 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
25306 %2<round_scalar_mask_op3>}"): ... this.
25307 * config/i386/subst.md (round_scalar_nimm_predicate): New.
25309 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
25311 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
25312 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
25313 (_mm_maskz_sqrt_round_ss): New intrinsics.
25314 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
25315 (__builtin_ia32_sqrtsd_mask_round)
25316 (__builtin_ia32_sqrtss_mask_round): New builtins.
25317 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
25318 (__builtin_ia32_sqrtss_round): Remove.
25319 (__builtin_ia32_sqrtsd_mask_round)
25320 (__builtin_ia32_sqrtss_mask_round): New builtins.
25321 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
25322 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
25323 ((match_operand:VF_128 1 "vector_operand"
25324 "xBm,<round_constraint>")): Changed to ...
25325 ((match_operand:VF_128 1 "vector_operand"
25326 "xBm,<round_scalar_constraint>")): ... this.
25327 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
25328 %0, %2, %<iptr>1<round_op3>}): Changed to ...
25329 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
25330 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
25331 %<iptr>1<round_scalar_mask_op3>}): ... this.
25332 ((set_attr "prefix" "<round_prefix>")): Changed to ...
25333 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
25335 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
25338 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
25339 Cast vec_cmpeq result to correct type.
25340 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
25341 Cast vec_cmpgt result to correct type.
25343 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
25345 * final.c (final_scan_insn_1): Renamed from...
25346 (final_scan_insn): ... this. New wrapper, to recover
25347 seen from the outermost call in recursive ones.
25348 * config/sparc/sparc.c (output_return): Drop seen from call.
25349 (output_sibcall): Likewise.
25350 * config/visium/visium.c (output_branch): Likewise.
25352 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
25354 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
25357 2018-02-10 Alan Modra <amodra@gmail.com>
25360 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
25361 Specify LR as an input.
25363 2018-02-10 Jakub Jelinek <jakub@redhat.com>
25366 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
25367 remove_member_access_dummy_vars): New functions.
25368 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
25369 lower_omp_1, execute_lower_omp): Use them.
25371 PR rtl-optimization/84308
25372 * shrink-wrap.c (spread_components): Release todo vector.
25374 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
25376 PR rtl-optimization/57193
25377 * ira-color.c (struct allocno_color_data): Add member
25378 conflict_allocno_hard_prefs.
25379 (update_conflict_allocno_hard_prefs): New.
25380 (bucket_allocno_compare_func): Add a preference based on
25381 conflict_allocno_hard_prefs.
25382 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
25383 (color_allocnos): Remove a dead code. Initiate
25384 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
25386 2018-02-09 Jakub Jelinek <jakub@redhat.com>
25389 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
25390 constraint from =wa to wa. Avoid a subreg on the output operand,
25391 instead use a pseudo and subreg it in a move.
25392 (p9_xxbrd_<mode>): Changed to ...
25393 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
25394 (p9_xxbrd_v2df): New expander.
25395 (p9_xxbrw_<mode>): Changed to ...
25396 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
25397 (p9_xxbrw_v4sf): New expander.
25399 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
25401 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
25403 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
25406 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
25407 multiply in 32-bit mode.
25408 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
25409 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
25412 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
25414 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
25415 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
25416 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
25417 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
25419 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
25422 * dwarf2out.c (is_trivial_indirect_ref): New function.
25423 (dwarf2out_late_global_decl): Do not generate a location
25424 attribute for variables that have a non-trivial DECL_VALUE_EXPR
25425 and that are not defined in the current unit.
25427 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
25429 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
25430 instead of a libcall for UNORDERED.
25432 2018-02-09 Tamar Christina <tamar.christina@arm.com>
25435 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
25436 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
25438 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25441 * config/s390/s390.c (s390_set_current_function): Invoke
25442 s390_indirect_branch_settings also if fndecl didn't change.
25444 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
25446 * config/rs6000/rs6000.md (blockage): Set length to zero.
25448 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
25450 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
25452 2018-02-09 Jakub Jelinek <jakub@redhat.com>
25455 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
25456 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
25460 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
25461 PARALLEL incoming that failed vt_get_decl_and_offset check.
25463 PR middle-end/84237
25464 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
25465 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
25467 (get_variable_section): For decls in named .bss* sections pass true as
25468 second argument to bss_initializer_p.
25470 2018-02-09 Marek Polacek <polacek@redhat.com>
25471 Jakub Jelinek <jakub@redhat.com>
25474 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
25475 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
25476 Sync some changes from cxx_fold_indirect_ref.
25478 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
25480 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
25482 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
25483 (BLOCK_INLINE_ENTRY_LABEL): New.
25484 (dwarf2out_var_location): Disregard inline entry markers.
25485 (inline_entry_data): New struct.
25486 (inline_entry_data_hasher): New hashtable type.
25487 (inline_entry_data_hasher::hash): New.
25488 (inline_entry_data_hasher::equal): New.
25489 (inline_entry_data_table): New variable.
25490 (add_high_low_attributes): Add DW_AT_entry_pc and
25491 DW_AT_GNU_entry_view attributes if a pending entry is found
25492 in inline_entry_data_table. Add old entry_pc attribute only
25493 if debug nonbinding markers are disabled.
25494 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
25495 markers are enabled.
25496 (block_within_block_p, dwarf2out_inline_entry): New.
25497 (dwarf2out_finish): Check that no entries remained in
25498 inline_entry_data_table.
25499 * final.c (reemit_insn_block_notes): Handle inline entry notes.
25500 (final_scan_insn, notice_source_line): Likewise.
25501 (rest_of_clean_state): Skip inline entry markers.
25502 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
25504 * gimple.c (gimple_build_debug_inline_entry): New.
25505 * gimple.h (enum gimple_debug_subcode): Add
25506 GIMPLE_DEBUG_INLINE_ENTRY.
25507 (gimple_build_debug_inline_entry): Declare.
25508 (gimple_debug_inline_entry_p): New.
25509 (gimple_debug_nonbind_marker_p): Adjust.
25510 * insn-notes.def (INLINE_ENTRY): New.
25511 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
25512 inline entry marker notes.
25513 (print_insn): Likewise.
25514 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
25515 (INSN_DEBUG_MARKER_KIND): Likewise.
25516 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
25517 * tree-inline.c (expand_call_inline): Build and insert
25518 debug_inline_entry stmt.
25519 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
25520 inline entry blocks early, if nonbind markers are enabled.
25521 (dump_scope_block): Dump fragment info.
25522 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
25523 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
25524 (gimple_build_debug_inline_entry): New.
25525 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
25526 Enable/disable inline entry points too.
25527 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
25528 (DEBUG_INSN): Describe inline entry markers.
25530 * common.opt (gvariable-location-views): New.
25531 (gvariable-location-views=incompat5): New.
25532 * config.in: Rebuilt.
25533 * configure: Rebuilt.
25534 * configure.ac: Test assembler for view support.
25535 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
25536 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
25537 * dwarf2out.c (var_loc_view): New typedef.
25538 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
25539 (dwarf2out_locviews_in_attribute): New.
25540 (dwarf2out_locviews_in_loclist): New.
25541 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
25542 (enum dw_line_info_opcode): Add LI_adv_address.
25543 (struct dw_line_info_table): Add view.
25544 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
25545 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
25546 (zero_view_p): New variable.
25547 (ZERO_VIEW_P): New macro.
25548 (output_asm_line_debug_info): New.
25549 (struct var_loc_node): Add view.
25550 (add_AT_view_list, AT_loc_list): New.
25551 (add_var_loc_to_decl): Add view param. Test it against last.
25552 (new_loc_list): Add view params. Record them.
25553 (AT_loc_list_ptr): Handle loc and view lists.
25554 (view_list_to_loc_list_val_node): New.
25555 (print_dw_val): Handle dw_val_class_view_list.
25556 (size_of_die): Likewise.
25557 (value_format): Likewise.
25558 (loc_list_has_views): New.
25559 (gen_llsym): Set vl_symbol too.
25560 (maybe_gen_llsym, skip_loc_list_entry): New.
25561 (dwarf2out_maybe_output_loclist_view_pair): New.
25562 (output_loc_list): Output view list or entries too.
25563 (output_view_list_offset): New.
25564 (output_die): Handle dw_val_class_view_list.
25565 (output_dwarf_version): New.
25566 (output_compilation_unit_header): Use it.
25567 (output_skeleton_debug_sections): Likewise.
25568 (output_rnglists, output_line_info): Likewise.
25569 (output_pubnames, output_aranges): Update version comments.
25570 (output_one_line_info_table): Output view numbers in asm comments.
25571 (dw_loc_list): Determine current endview, pass it to new_loc_list.
25572 Call maybe_gen_llsym.
25573 (loc_list_from_tree_1): Adjust.
25574 (add_AT_location_description): Create view list attribute if
25575 needed, check it's absent otherwise.
25576 (convert_cfa_to_fb_loc_list): Adjust.
25577 (maybe_emit_file): Call output_asm_line_debug_info for test.
25578 (dwarf2out_var_location): Reset views as needed. Precompute
25579 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
25580 attribute. Set view.
25581 (new_line_info_table): Reset next view.
25582 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
25583 (dwarf2out_source_line): Likewise. Output view resets and labels to
25584 the assembler, or select appropriate line info opcodes.
25585 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
25586 (optimize_string_length): Catch it. Adjust.
25587 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
25588 dw_val_class_view_list, and remove it if no longer needed.
25589 (hash_loc_list): Hash view numbers.
25590 (loc_list_hasher::equal): Compare them.
25591 (optimize_location_lists): Check whether a view list symbol is
25592 needed, and whether the locview attribute is present, and
25593 whether they match. Remove the locview attribute if no longer
25595 (index_location_lists): Call skip_loc_list_entry for test.
25596 (dwarf2out_finish): Call output_asm_line_debug_info for test.
25597 Use output_dwarf_version.
25598 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
25599 (struct dw_val_node): Add val_view_list.
25600 * final.c (SEEN_NEXT_VIEW): New.
25601 (set_next_view_needed): New.
25602 (clear_next_view_needed): New.
25603 (maybe_output_next_view): New.
25604 (final_start_function): Rename to...
25605 (final_start_function_1): ... this. Take pointer to FIRST,
25606 add SEEN parameter. Emit param bindings in the initial view.
25607 (final_start_function): Reintroduce SEEN-less interface.
25608 (final): Rename to...
25609 (final_1): ... this. Take SEEN parameter. Output final pending
25610 next view at the end.
25611 (final): Reintroduce seen-less interface.
25612 (final_scan_insn): Output pending next view before switching
25613 sections or ending a block. Mark the next view as needed when
25614 outputting variable locations. Notify debug backend of section
25615 changes, and of location view changes.
25616 (rest_of_handle_final): Adjust.
25617 * toplev.c (process_options): Autodetect value for debug variable
25618 location views option. Warn on incompat5 without -gdwarf-5.
25619 * doc/invoke.texi (gvariable-location-views): New.
25620 (gvariable-location-views=incompat5): New.
25621 (gno-variable-location-views): New.
25623 2018-02-08 David Malcolm <dmalcolm@redhat.com>
25625 PR tree-optimization/84136
25626 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
25627 that the result of find_edge is non-NULL.
25629 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
25632 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
25633 storing integer register in SImode. Fix cost of 256 and 512
25634 byte aligned SSE register store.
25636 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
25638 * config/i386/i386.c (ix86_multiplication_cost): Fix
25639 multiplication cost for TARGET_AVX512DQ.
25641 2018-02-08 Marek Polacek <polacek@redhat.com>
25643 PR tree-optimization/84238
25644 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
25647 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25649 PR tree-optimization/84265
25650 * tree-vect-stmts.c (vectorizable_store): Don't treat
25651 VMAT_CONTIGUOUS accesses as grouped.
25652 (vectorizable_load): Likewise.
25654 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25656 PR tree-optimization/81635
25657 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
25658 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
25659 (test_round_for_mask): New functions.
25660 (wide_int_cc_tests): Call test_round_for_mask.
25661 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
25662 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
25663 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
25664 range returned by get_range_info.
25666 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
25669 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
25670 * symtab.c: Include builtins.h
25671 (symtab_node::output_to_lto_symbol_table_p): Move here
25672 from lto-streamer-out.c:output_symbol_p.
25673 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
25674 (output_symbol_p): Move all logic to symtab.c
25675 (produce_symtab): Update.
25677 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25679 * config/s390/s390-opts.h (enum indirect_branch): Define.
25680 * config/s390/s390-protos.h (s390_return_addr_from_memory)
25681 (s390_indirect_branch_via_thunk)
25682 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
25683 (enum s390_indirect_branch_type): Define.
25684 * config/s390/s390.c (struct s390_frame_layout, struct
25685 machine_function): Remove.
25686 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
25687 (indirect_branch_table_label_no, indirect_branch_table_name):
25689 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
25690 (enum s390_indirect_branch_option): Define.
25691 (s390_return_addr_from_memory): New function.
25692 (s390_handle_string_attribute): New function.
25693 (s390_attribute_table): Add new attribute handler.
25694 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
25695 (s390_indirect_branch_via_thunk): New function.
25696 (s390_indirect_branch_via_inline_thunk): New function.
25697 (s390_function_ok_for_sibcall): When jumping via thunk disallow
25698 sibling call optimization for non z10 compiles.
25699 (s390_emit_call): Force indirect branch target to be a single
25700 register. Add r1 clobber for non-z10 compiles.
25701 (s390_emit_epilogue): Emit return jump via return_use expander.
25702 (s390_reorg): Handle JUMP_INSNs as execute targets.
25703 (s390_option_override_internal): Perform validity checks for the
25704 new command line options.
25705 (s390_indirect_branch_attrvalue): New function.
25706 (s390_indirect_branch_settings): New function.
25707 (s390_set_current_function): Invoke s390_indirect_branch_settings.
25708 (s390_output_indirect_thunk_function): New function.
25709 (s390_code_end): Implement target hook.
25710 (s390_case_values_threshold): Implement target hook.
25711 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
25713 * config/s390/s390.h (struct s390_frame_layout)
25714 (struct machine_function): Move here from s390.c.
25715 (TARGET_INDIRECT_BRANCH_NOBP_RET)
25716 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
25717 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
25718 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
25719 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
25720 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
25721 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
25722 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
25723 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
25724 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
25725 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
25726 (mnemonic attribute): Add values which aren't recognized
25728 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
25729 pattern for branch conversion. Fix mnemonic attribute.
25730 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
25731 indirect branch via thunk if requested.
25732 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
25733 ("*indirect_jump"): Disable for branch conversion using out of
25735 ("indirect_jump_via_thunk<mode>_z10")
25736 ("indirect_jump_via_thunk<mode>")
25737 ("indirect_jump_via_inlinethunk<mode>_z10")
25738 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
25739 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
25740 ("casesi_jump_via_inlinethunk<mode>_z10")
25741 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
25742 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
25743 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
25744 ("*indirect2_jump"): Disable for branch conversion.
25745 ("casesi_jump"): Turn into expander and expand patterns for branch
25747 ("return_use"): New expander.
25748 ("*return"): Emit return via thunk and rename it to ...
25749 ("*return<mode>"): ... this one.
25750 * config/s390/s390.opt: Add new options and and enum for the
25753 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25755 * lra-constraints.c (match_reload): Unconditionally use
25756 gen_lowpart_SUBREG, rather than selecting between that
25757 and equivalent gen_rtx_SUBREG code.
25759 2018-02-08 Richard Biener <rguenther@suse.de>
25761 PR tree-optimization/84233
25762 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
25763 changed flag instead of boguously re-using phi_inserted.
25765 2018-02-08 Martin Jambor <mjambor@suse.cz>
25767 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
25768 static local variables.
25770 2018-02-08 Richard Biener <rguenther@suse.de>
25772 PR tree-optimization/84278
25773 * tree-vect-stmts.c (vectorizable_store): When looking for
25774 smaller vector types to perform grouped strided loads/stores
25775 make sure the mode is supported by the target.
25776 (vectorizable_load): Likewise.
25778 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
25780 * config/aarch64/aarch64.c (aarch64_components_for_bb):
25781 Increase LDP/STP opportunities by adding adjacent callee-saves.
25783 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
25785 PR rtl-optimization/84068
25786 PR rtl-optimization/83459
25787 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
25789 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
25791 PR tree-optimization/84224
25792 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
25793 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
25794 non-zero arguments.
25796 2018-02-07 Iain Sandoe <iain@codesourcery.com>
25799 * config/rs6000/altivec.md (*restore_world): Remove LR use.
25800 * config/rs6000/predicates.md (restore_world_operation): Adjust op
25801 count, remove one USE.
25803 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25805 * doc/install.texi (Configuration): Document the
25806 --with-long-double-format={ibm,ieee} PowerPC configuration
25810 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
25811 Convert from define_expand to be define_insn_and_split. Rework
25812 float/double/_Float128 conversions to QI/HI/SImode to work with
25813 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
25814 conversions to QI/HImode types did a store and then a load to
25815 truncate the value. For conversions to VSX registers, don't split
25816 the insn, instead emit the code directly. Use the code iterator
25817 any_fix to combine signed and unsigned conversions.
25818 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
25819 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
25820 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
25821 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
25822 (fix_<mode>di2_hw): Likewise.
25823 (fixuns_<mode>di2_hw): Likewise.
25824 (fix_<mode>si2_hw): Likewise.
25825 (fixuns_<mode>si2_hw): Likewise.
25826 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
25827 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
25828 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
25829 fix<uns>_trunc<SFDF:mode>si2_p8.
25830 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
25831 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
25832 (fix<uns>_<mode>_mem): Likewise.
25833 (fctiw<u>z_<mode>_mem): Likewise.
25834 (fix<uns>_<mode>_mem): Likewise.
25835 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
25836 the register allocator from doing a direct move to the GPRs to do
25837 a store, and instead use the ISA 3.0 store byte/half-word from
25838 vector register instruction. For IEEE 128-bit floating point,
25839 also optimize stores of 32-bit ints.
25840 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
25842 2018-02-07 Alan Hayward <alan.hayward@arm.com>
25844 * genextract.c (push_pathstr_operand): New function to support
25846 (walk_rtx): Call push_pathstr_operand.
25847 (print_path): Support [a-zA-Z].
25849 2018-02-07 Richard Biener <rguenther@suse.de>
25851 PR tree-optimization/84037
25852 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
25853 (cse_and_gimplify_to_preheader): Declare.
25854 (vect_get_place_in_interleaving_chain): Likewise.
25855 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25857 (_loop_vec_info::~_loop_vec_info): Delete it.
25858 (cse_and_gimplify_to_preheader): New function.
25859 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
25860 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
25861 (vectorizable_load): Likewise. For grouped stores always base
25862 the IV on the first element.
25863 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
25864 condition before gimplifying.
25866 2018-02-07 Jakub Jelinek <jakub@redhat.com>
25868 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
25869 *DIV_EXPR and *MOD_EXPR.
25871 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
25874 * config/i386/i386.c (ix86_option_override_internal): Mask out
25875 the CF_SET bit when checking -fcf-protection.
25877 2018-02-07 Tom de Vries <tom@codesourcery.com>
25880 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
25883 2018-02-07 Richard Biener <rguenther@suse.de>
25885 PR tree-optimization/84204
25886 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
25889 PR tree-optimization/84205
25890 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
25891 special-case isl_ast_op_zdiv_r.
25893 PR tree-optimization/84223
25894 * graphite-scop-detection.c (gather_bbs::before_dom_children):
25895 Only add conditions from within the region.
25896 (gather_bbs::after_dom_children): Adjust.
25898 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
25901 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
25902 * config/avr/avr.md: Only post-reload split REG-REG moves if
25903 either register is GENERAL_REG_P.
25905 2018-02-07 Jakub Jelinek <jakub@redhat.com>
25907 PR tree-optimization/84235
25908 * tree-ssa-scopedtables.c
25909 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
25910 if the subtraction is performed in floating point type where NaNs are
25911 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
25912 build 1. Formatting fix.
25914 2018-02-06 Jakub Jelinek <jakub@redhat.com>
25917 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
25918 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
25919 and skip it regardless of bb boundaries. Use CALL_P macro,
25920 don't test INSN_P (insn) together with CALL_P or JUMP_P check
25921 unnecessarily, formatting fix.
25923 2018-02-06 Michael Collison <michael.collison@arm.com>
25925 * config/arm/thumb2.md:
25926 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
25927 (*thumb_mov_notscc): Ditto.
25929 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25932 * config/rs6000/rs6000.md (su code attribute): Use "u" for
25933 unsigned_fix, not "s".
25935 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25937 * configure.ac (gcc_fn_eh_frame_ro): New function.
25938 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
25939 correct .eh_frame permissions.
25940 * configure: Regenerate.
25942 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
25944 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
25945 irrelevant options.
25947 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25949 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25950 Display warning message for -mno-speculate-indirect-jumps.
25952 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
25954 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
25956 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
25958 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
25960 PR tree-optimization/84225
25961 * tree-eh.c (find_trapping_overflow): Only call
25962 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
25964 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25967 * config/i386/i386.c: Reimplement the check of possible options
25968 -mibt/-mshstk conbination. Change error messages.
25969 * doc/invoke.texi: Fix a typo: remove extra '='.
25971 2018-02-06 Marek Polacek <polacek@redhat.com>
25973 PR tree-optimization/84228
25974 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
25976 2018-02-06 Tamar Christina <tamar.christina@arm.com>
25979 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
25980 emitted arch directives.
25981 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
25982 __ARM_FEATURE_COPROC before changing architectures.
25984 2018-02-06 Richard Biener <rguenther@suse.de>
25986 * config/i386/i386.c (print_reg): Fix typo.
25987 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
25989 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
25991 * configure: Regenerate.
25993 2018-02-05 Martin Sebor <msebor@redhat.com>
25995 PR tree-optimization/83369
25996 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
25999 2018-02-05 Martin Liska <mliska@suse.cz>
26001 * doc/invoke.texi: Cherry-pick upstream r323995.
26003 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
26005 * ira.c (ira_init_register_move_cost): Adjust comment.
26007 2018-02-05 Martin Liska <mliska@suse.cz>
26009 PR gcov-profile/84137
26010 * doc/gcov.texi: Fix typo in documentation.
26012 2018-02-05 Martin Liska <mliska@suse.cz>
26014 PR gcov-profile/83879
26015 * doc/gcov.texi: Document necessity of --dynamic-list-data when
26016 using dlopen functionality.
26018 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
26020 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
26021 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
26022 _mm_maskz_range_ss, _mm_mask_range_round_ss,
26023 _mm_maskz_range_round_ss): New intrinsics.
26024 (__builtin_ia32_rangesd128_round)
26025 (__builtin_ia32_rangess128_round): Remove.
26026 (__builtin_ia32_rangesd128_mask_round,
26027 __builtin_ia32_rangess128_mask_round): New builtins.
26028 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
26029 __builtin_ia32_rangess128_round): Remove.
26030 (__builtin_ia32_rangesd128_mask_round,
26031 __builtin_ia32_rangess128_mask_round): New builtins.
26032 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
26033 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
26034 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
26035 "<round_saeonly_constraint>")): Changed to ...
26036 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
26037 "<round_saeonly_scalar_constraint>")): ... this.
26038 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
26039 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
26040 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
26041 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
26042 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
26044 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
26046 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
26048 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
26049 Remove all values except native, 8540 and 8548.
26051 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
26053 * config/i386/i386.c (ix86_output_function_return): Pass
26054 INVALID_REGNUM, instead of -1, as invalid register number to
26055 indirect_thunk_name and output_indirect_thunk.
26057 2018-02-02 Julia Koval <julia.koval@intel.com>
26059 * config.gcc: Add -march=icelake.
26060 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
26061 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
26062 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
26063 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
26064 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
26065 (processor_target_table): Add icelake.
26066 (ix86_option_override_internal): Handle new PTAs.
26067 (get_builtin_code_for_version): Handle icelake.
26068 (M_INTEL_COREI7_ICELAKE): New.
26069 (fold_builtin_cpu): Handle icelake.
26070 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
26071 * doc/invoke.texi: Add -march=icelake.
26073 2018-02-02 Julia Koval <julia.koval@intel.com>
26075 * config/i386/i386.c (ix86_option_override_internal): Change flags type
26076 to wide_int_bitmask.
26077 * wide-int-bitmask.h: New.
26079 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26082 * config/i386/i386.md: Replace Pmode with word_mode in
26083 builtin_setjmp_setup and builtin_longjmp to support x32.
26085 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
26089 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
26091 (rs6000_supported_cpu_names): New static variable.
26092 (linux_cpu_translation_table): Likewise.
26093 (elf_platform) <cpu>: Define new static variable and use it.
26094 Translate kernel AT_PLATFORM name to canonical name if needed.
26095 Error if platform name is unknown.
26097 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
26100 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
26102 2018-02-01 Jeff Law <law@redhat.com>
26105 * config/i386/i386.c (release_scratch_register_on_entry): Add new
26106 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
26107 the scratch if RELEASE_VIA_POP is false.
26108 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
26109 If we have to save a temporary register, decrement SIZE appropriately.
26110 Pass new arguments to release_scratch_register_on_entry.
26111 (ix86_adjust_stack_and_probe): Likewise.
26112 (ix86_emit_probe_stack_range): Pass new arguments to
26113 release_scratch_register_on_entry.
26115 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
26117 PR rtl-optimization/84157
26118 * combine.c (change_zero_ext): Use REG_P predicate in
26119 front of HARD_REGISTER_P predicate.
26121 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
26123 * config/avr/avr.c (avr_option_override): Move disabling of
26124 -fdelete-null-pointer-checks to...
26125 * common/config/avr/avr-common.c (avr_option_optimization_table):
26128 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26130 PR tree-optimization/81635
26131 * tree-data-ref.c (split_constant_offset_1): For types that
26132 wrap on overflow, try to use range info to prove that wrapping
26135 2018-02-01 Renlin Li <renlin.li@arm.com>
26138 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
26139 TAILCALL_ADDR_REGS.
26140 (aarch64_register_move_cost): Likewise.
26141 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
26142 TAILCALL_ADDR_REGS.
26143 (REG_CLASS_NAMES): Likewise.
26144 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
26145 TAILCALL_ADDR_REGS. Remove IP registers.
26146 * config/aarch64/aarch64.md (Ucs): Update register constraint.
26148 2018-02-01 Richard Biener <rguenther@suse.de>
26150 * domwalk.h (dom_walker::dom_walker): Add additional constructor
26151 for specifying RPO order and allow NULL for that.
26152 * domwalk.c (dom_walker::dom_walker): Likewise.
26153 (dom_walker::walk): Handle NULL RPO order.
26154 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
26156 (rewrite_update_dom_walker): Likewise.
26157 (mark_def_dom_walker): Likewise.
26159 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26161 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
26162 (aarch64_maybe_expand_sve_subreg_move): Declare.
26163 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
26164 * config/aarch64/predicates.md (aarch64_any_register_operand): New
26166 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
26167 that are semantically a reverse operation.
26168 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
26169 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
26170 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
26172 (aarch64_can_change_mode_class): For big-endian, forbid changes
26173 between two SVE modes if they have different element sizes.
26175 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26177 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
26178 the TImode handling for big-endian targets.
26180 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26182 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
26183 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
26185 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
26186 Remove BSWAP handing for big-endian targets and use the form of
26187 LD1RQ appropariate for the mode.
26189 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26191 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
26192 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
26193 duplicated element.
26195 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26198 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
26199 check for operands that need to go through aarch64_sve_reload_be.
26201 2018-02-01 Jakub Jelinek <jakub@redhat.com>
26203 PR tree-optimization/81661
26204 PR tree-optimization/84117
26205 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
26206 * tree-eh.c: Include gimplify.h.
26207 (find_trapping_overflow, replace_trapping_overflow,
26208 rewrite_to_non_trapping_overflow): New functions.
26209 * tree-vect-loop.c: Include tree-eh.h.
26210 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
26211 * tree-data-ref.c: Include tree-eh.h.
26212 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
26214 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
26216 PR rtl-optimization/84123
26217 * combine.c (change_zero_ext): Check if hard register satisfies
26218 can_change_dest_mode before calling gen_lowpart_SUBREG.
26220 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
26223 * ira.c (ira_init_register_move_cost): Remove assert.
26225 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26227 PR rtl-optimization/84071
26228 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
26229 * doc/tm.texi: Regenerate.
26231 2018-01-31 Richard Biener <rguenther@suse.de>
26233 PR tree-optimization/84132
26234 * tree-data-ref.c (analyze_miv_subscript): Properly
26235 check whether evolution_function_is_affine_multivariate_p
26236 before calling gcd_of_steps_may_divide_p.
26238 2018-01-31 Julia Koval <julia.koval@intel.com>
26241 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26242 * config/i386/i386.md (rdpid_rex64) New.
26243 (rdpid): Make 32bit only.
26245 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
26248 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
26249 an IDENTIFIER_NODE for FUNCTION_TYPE's.
26251 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26254 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26256 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26258 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26260 PR rtl-optimization/84071
26261 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
26262 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
26264 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
26266 * config/arc/arc.c (arc_handle_aux_attribute): New function.
26267 (arc_attribute_table): Add 'aux' attribute.
26268 (arc_in_small_data_p): Consider aux like variables.
26269 (arc_is_aux_reg_p): New function.
26270 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
26271 (arc_get_aux_arg): New function.
26272 (prepare_move_operands): Handle aux-register access.
26273 (arc_handle_aux_attribute): New function.
26274 * doc/extend.texi (ARC Variable attributes): Add subsection.
26276 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
26278 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
26279 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
26280 (arc_attribute_table): Add 'uncached' attribute.
26281 (arc_print_operand): Print '.di' flag for uncached memory
26283 (arc_in_small_data_p): Do not consider for small data the uncached
26285 (arc_is_uncached_mem_p): New function.
26286 * config/arc/predicates.md (compact_store_memory_operand): Check
26287 for uncached memory accesses.
26288 (nonvol_nonimm_operand): Likewise.
26289 * doc/extend.texi (ARC Type Attribute): New subsection.
26291 2018-01-31 Jakub Jelinek <jakub@redhat.com>
26294 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
26295 falign-loops=): Add Optimization flag.
26297 2018-01-30 Jeff Law <law@redhat.com>
26300 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
26301 INT_REGISTERS_SAVED. Check it prior to calling
26302 get_scratch_register_on_entry.
26303 (ix86_adjust_stack_and_probe): Similarly.
26304 (ix86_emit_probe_stack_range): Similarly.
26305 (ix86_expand_prologue): Corresponding changes.
26307 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26310 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
26311 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
26313 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
26316 * lra-constraints.c (curr_insn_transform): Process AND in the
26319 2018-01-30 Jakub Jelinek <jakub@redhat.com>
26321 PR rtl-optimization/83986
26322 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
26323 dependence against last_pending_memory_flush in addition to
26324 pending_jump_insns.
26326 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
26328 PR tree-optimization/81611
26329 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
26332 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26335 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
26338 2018-01-30 Richard Biener <rguenther@suse.de>
26339 Jakub Jelinek <jakub@redhat.com>
26341 PR tree-optimization/84111
26342 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
26343 inner loops added during recursion, as they don't have up-to-date
26346 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
26349 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
26350 (can_inline_edge_by_limits_p): ... here.
26351 (can_early_inline_edge_p, check_callers,
26352 update_caller_keys, update_callee_keys, recursive_inlining,
26353 add_new_edges_to_heap, speculation_useful_p,
26354 inline_small_functions,
26355 inline_small_functions, flatten_function,
26356 inline_to_all_callers_1): Update.
26358 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
26360 * profile-count.c (profile_count::combine_with_ipa_count): Handle
26363 2018-01-30 Richard Biener <rguenther@suse.de>
26365 PR tree-optimization/83008
26366 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
26367 invariant and constant vector uses in stmts when they need
26368 more than one stmt.
26370 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26373 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
26374 * configure: Regenerate.
26376 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26378 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
26380 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
26381 Use gen_rtx_REG rather than gen_lowpart.
26383 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26385 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
26386 rather than 0 when creating partial subregs.
26388 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26390 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
26393 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
26396 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
26397 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
26398 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
26399 flags. This restores the settings used before the 2017-07-24.
26400 Turning off pre increment/decrement/modify allows IVOPTS to
26401 optimize DF/SF loops where the index is an int.
26403 2018-01-29 Richard Biener <rguenther@suse.de>
26404 Kelvin Nilsen <kelvin@gcc.gnu.org>
26407 * tree-vect-stmts.c (vectorizable_call): Don't call
26408 targetm.vectorize_builtin_md_vectorized_function if callee is
26411 2018-01-22 Carl Love <cel@us.ibm.com>
26413 * doc/extend.tex: Fix typo in second arg in
26414 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
26416 2018-01-29 Richard Biener <rguenther@suse.de>
26418 PR tree-optimization/84086
26419 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
26420 (flush_ssaname_freelist): When SSA names were released reset
26421 the SCEV hash table.
26423 2018-01-29 Richard Biener <rguenther@suse.de>
26425 PR tree-optimization/84057
26426 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
26427 removed paths when removing edges.
26429 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
26431 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
26432 -mfunction-return=@var{choice}.
26434 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
26436 PR diagnostic/84034
26437 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
26438 Handle CR like TAB.
26439 (layout::print_source_line): Likewise.
26440 (test_get_line_width_without_trailing_whitespace): Add test cases.
26442 2018-01-27 Jakub Jelinek <jakub@redhat.com>
26444 PR middle-end/84040
26445 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
26448 2018-01-26 Jim Wilson <jimw@sifive.com>
26450 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
26452 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
26455 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26457 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
26458 and CMP + SUB-immediate -> SUBS.
26460 2018-01-26 Martin Sebor <msebor@redhat.com>
26462 PR tree-optimization/83896
26463 * tree-ssa-strlen.c (get_string_len): Rename...
26464 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
26465 Avoid assuming length is constant.
26466 (handle_char_store): Use HOST_WIDE_INT for string length.
26468 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
26471 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
26472 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
26474 2018-01-26 Richard Biener <rguenther@suse.de>
26476 PR rtl-optimization/84003
26477 * dse.c (record_store): Only record redundant stores when
26478 the earlier store aliases at least all accesses the later one does.
26480 2018-01-26 Jakub Jelinek <jakub@redhat.com>
26482 PR rtl-optimization/83985
26483 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
26484 REG_CFA_RESTORE insns.
26485 (delete_unmarked_insns): Don't ignore separate shrink wrapping
26486 REG_CFA_RESTORE insns here.
26489 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
26490 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
26492 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26494 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
26495 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
26496 (arc_init): Likewise.
26497 (arc_override_options): Likewise.
26498 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
26500 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
26502 * config/arc/arc.h (TARGET_DBNZ): Define.
26503 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
26504 properly set the tune attribute.
26505 (dbnz): Use TARGET_DBNZ guard.
26506 * config/arc/arc.opt (mtune): Add core3 option.
26508 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26510 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
26511 recognize new pic like addresses.
26512 (arc_delegitimize_address): Clean up.
26514 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26516 * config/arc/arc-arches.def: Option mrf16 valid for all
26518 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
26519 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
26520 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
26521 * config/arc/arc-tables.opt: Regenerate.
26522 * config/arc/arc.c (arc_conditional_register_usage): Handle
26523 reduced register file case.
26524 (arc_file_start): Set must have build attributes.
26525 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
26526 mrf16 option value.
26527 * config/arc/arc.opt (mrf16): Add new option.
26528 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
26529 * config/arc/genmultilib.awk: Handle new mrf16 option.
26530 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
26531 * config/arc/t-multilib: Regenerate.
26532 * doc/invoke.texi (ARC Options): Document mrf16 option.
26534 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26536 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
26537 * config/arc/arc.c (arc_handle_secure_attribute): New function.
26538 (arc_attribute_table): Add 'secure_call' attribute.
26539 (arc_print_operand): Print secure call operand.
26540 (arc_function_ok_for_sibcall): Don't optimize tail calls when
26542 (arc_is_secure_call_p): New function. * config/arc/arc.md
26543 (call_i): Add support for sjli instruction.
26544 (call_value_i): Likewise.
26545 * config/arc/constraints.md (Csc): New constraint.
26547 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26548 John Eric Martin <John.Martin@emmicro-us.com>
26550 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
26551 * config/arc/arc.c (_arc_jli_section): New struct.
26552 (arc_jli_section): New type.
26553 (rc_jli_sections): New static variable.
26554 (arc_handle_jli_attribute): New function.
26555 (arc_attribute_table): Add jli_always and jli_fixed attribute.
26556 (arc_file_end): New function.
26557 (TARGET_ASM_FILE_END): Define.
26558 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
26559 (arc_add_jli_section): New function.
26560 (jli_call_scan): Likewise.
26561 (arc_reorg): Call jli_call_scan.
26562 (arc_output_addsi): Remove 'S' from printing asm operand.
26563 (arc_is_jli_call_p): New function.
26564 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
26566 (movhi_insn): Likewise.
26567 (movsi_insn): Likewise.
26568 (movsi_set_cc_insn): Likewise.
26569 (loadqi_update): Likewise.
26570 (load_zeroextendqisi_update): Likewise.
26571 (load_signextendqisi_update): Likewise.
26572 (loadhi_update): Likewise.
26573 (load_zeroextendhisi_update): Likewise.
26574 (load_signextendhisi_update): Likewise.
26575 (loadsi_update): Likewise.
26576 (loadsf_update): Likewise.
26577 (movsicc_insn): Likewise.
26578 (bset_insn): Likewise.
26579 (bxor_insn): Likewise.
26580 (bclr_insn): Likewise.
26581 (bmsk_insn): Likewise.
26582 (bicsi3_insn): Likewise.
26583 (cmpsi_cc_c_insn): Likewise.
26584 (movsi_ne): Likewise.
26585 (movsi_cond_exec): Likewise.
26586 (clrsbsi2): Likewise.
26587 (norm_f): Likewise.
26597 (call_i): Remove 'S' asm letter, add jli instruction.
26598 (call_value_i): Likewise.
26599 * config/arc/arc.op (mjli-always): New option.
26600 * config/arc/constraints.md (Cji): New constraint.
26601 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
26603 (subsf3_fpx): Likewise.
26604 (mulsf3_fpx): Likewise.
26605 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
26607 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
26608 function attrbutes.
26609 * doc/invoke.texi (ARC): Document mjli-always option.
26611 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
26613 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
26614 avoid addition with 0 and use incw and decw where possible.
26616 2018-01-26 Richard Biener <rguenther@suse.de>
26618 PR tree-optimization/81082
26619 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
26620 association if it requires casting to unsigned.
26621 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
26622 from fold_plusminus_mult_expr to catch important cases late when
26623 range info is available.
26625 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26627 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
26628 * configure.ac (hidden_linkonce): New test.
26629 * configure: Regenerate.
26630 * config.in: Regenerate.
26632 2018-01-26 Julia Koval <julia.koval@intel.com>
26634 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
26635 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
26636 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
26637 _mm_mask_bitshuffle_epi64_mask): Fix type.
26638 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
26639 USI_FTYPE_V4DI_V4DI_USI): Remove.
26640 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
26641 __builtin_ia32_vpshufbitqmb256_mask,
26642 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
26643 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
26644 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
26646 2018-01-26 Alan Modra <amodra@gmail.com>
26649 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
26650 UNSPEC_VBPERMQ. Sort other unspecs.
26652 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
26654 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
26656 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
26658 PR middle-end/83055
26659 * predict.c (drop_profile): Do not push/pop cfun; update also
26661 (handle_missing_profiles): Fix logic looking for zero profiles.
26663 2018-01-25 Jakub Jelinek <jakub@redhat.com>
26665 PR middle-end/83977
26666 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
26667 on functions with #pragma omp declare simd or functions with simd
26669 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
26670 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
26671 Remove trailing \n from warning_at calls.
26673 2018-01-25 Tom de Vries <tom@codesourcery.com>
26676 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
26677 for neutered workers.
26679 2018-01-24 Joseph Myers <joseph@codesourcery.com>
26682 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
26683 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
26685 2018-01-24 Jeff Law <law@redhat.com>
26688 * i386.c (get_probe_interval): Move to earlier point.
26689 (ix86_compute_frame_layout): If -fstack-clash-protection and
26690 the frame is larger than the probe interval, then use pushes
26691 to save registers rather than reg->mem moves.
26692 (ix86_expand_prologue): Remove conditional for int_registers_saved
26695 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
26698 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
26699 min/max for never referenced object.
26701 2018-01-24 Jakub Jelinek <jakub@redhat.com>
26703 PR middle-end/83977
26704 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
26706 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
26707 attributes from DECL_ATTRIBUTES (decl) without affecting
26708 DECL_ATTRIBUTES (current_function_decl).
26709 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
26710 functions with non-NULL DECL_ABSTRACT_ORIGIN.
26712 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
26714 PR tree-optimization/83979
26715 * fold-const.c (fold_comparison): Use constant_boolean_node
26716 instead of boolean_{true,false}_node.
26718 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
26720 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
26723 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26725 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
26726 Simplify the clause that sets the length attribute.
26727 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
26728 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
26729 clause that sets the length attribute.
26730 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
26732 2018-01-24 Tom de Vries <tom@codesourcery.com>
26735 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
26736 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
26737 Add strict parameter.
26738 (prevent_branch_around_nothing): Insert dummy insn between branch to
26739 label and label with no ptx insn inbetween.
26740 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
26742 2018-01-24 Tom de Vries <tom@codesourcery.com>
26745 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
26746 for neutered threads in warp.
26747 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
26749 2018-01-24 Richard Biener <rguenther@suse.de>
26751 PR tree-optimization/83176
26752 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
26755 2018-01-24 Richard Biener <rguenther@suse.de>
26757 PR tree-optimization/82819
26758 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
26759 code generating pluses that are no-ops in the target precision.
26761 2018-01-24 Richard Biener <rguenther@suse.de>
26763 PR middle-end/84000
26764 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
26766 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
26768 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
26769 to merge probabilities.
26770 * predict.c (probably_never_executed): Also mark as cold functions
26771 with global 0 profile and guessed local profile.
26772 * profile-count.c (profile_probability::combine_with_count): New
26774 * profile-count.h (profile_probability::operator*,
26775 profile_probability::operator*=, profile_probability::operator/,
26776 profile_probability::operator/=): Reduce precision to adjusted
26777 and set value to guessed on contradictory divisions.
26778 (profile_probability::combine_with_freq): Remove.
26779 (profile_probability::combine_wiht_count): Declare.
26780 (profile_count::force_nonzero):: Set to adjusted.
26781 (profile_count::probability_in):: Set quality to adjusted.
26782 * tree-ssa-tail-merge.c (replace_block_by): Use
26783 combine_with_count.
26785 2018-01-23 Andrew Waterman <andrew@sifive.com>
26786 Jim Wilson <jimw@sifive.com>
26788 * config/riscv/riscv.c (riscv_stack_boundary): New.
26789 (riscv_option_override): Set riscv_stack_boundary. Handle
26790 riscv_preferred_stack_boundary_arg.
26791 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
26792 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
26793 (STACK_BOUNDARY): Set to riscv_stack_boundary.
26794 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
26795 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
26796 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
26798 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
26801 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
26802 of struct ix86_frame.
26803 (ix86_expand_epilogue): Likewise. Add a local variable for
26804 the reg_save_offset field in struct ix86_frame.
26806 2018-01-23 Bin Cheng <bin.cheng@arm.com>
26808 PR tree-optimization/82604
26809 * tree-loop-distribution.c (enum partition_kind): New enum item
26810 PKIND_PARTIAL_MEMSET.
26811 (partition_builtin_p): Support above new enum item.
26812 (generate_code_for_partition): Ditto.
26813 (compute_access_range): Differentiate cases that equality can be
26814 proven at all loops, the innermost loops or no loops.
26815 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
26816 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
26817 (finalize_partitions, distribute_loop): Don't fuse partition of
26818 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
26819 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
26820 parloop is enabled.
26822 2018-01-23 Martin Liska <mliska@suse.cz>
26824 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
26825 order to ignore the predictor.
26826 (PRED_POLYMORPHIC_CALL): Likewise.
26827 (PRED_RECURSIVE_CALL): Likewise.
26829 2018-01-23 Martin Liska <mliska@suse.cz>
26831 * tree-profile.c (tree_profiling): Print function header to
26832 aware reader which function we are working on.
26833 * value-prof.c (gimple_find_values_to_profile): Do not print
26834 not interesting value histograms.
26836 2018-01-23 Martin Liska <mliska@suse.cz>
26838 * profile-count.h (enum profile_quality): Add
26839 profile_uninitialized as the first value. Do not number values
26840 as they are zero based.
26841 (profile_count::verify): Update sanity check.
26842 (profile_probability::verify): Likewise.
26844 2018-01-23 Nathan Sidwell <nathan@acm.org>
26846 * doc/invoke.texi (ffor-scope): Deprecate.
26848 2018-01-23 David Malcolm <dmalcolm@redhat.com>
26850 PR tree-optimization/83510
26851 * domwalk.c (set_all_edges_as_executable): New function.
26852 (dom_walker::dom_walker): Convert bool param
26853 "skip_unreachable_blocks" to enum reachability. Move setup of
26854 edge flags to set_all_edges_as_executable and only do it when
26855 reachability is REACHABLE_BLOCKS.
26856 * domwalk.h (enum dom_walker::reachability): New enum.
26857 (dom_walker::dom_walker): Convert bool param
26858 "skip_unreachable_blocks" to enum reachability.
26859 (set_all_edges_as_executable): New decl.
26860 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
26861 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
26863 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
26864 but converting true to REACHABLE_BLOCKS.
26865 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
26867 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
26868 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
26869 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
26871 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
26872 if check_all_array_refs will be called.
26874 2018-01-23 David Malcolm <dmalcolm@redhat.com>
26876 * tree.c (selftest::test_location_wrappers): Add more test
26879 2018-01-23 David Malcolm <dmalcolm@redhat.com>
26881 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
26882 (selftest::test_bit_in_range): Likewise.
26884 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
26887 * doc/sourcebuild.texi (vect_float): Say that the selector
26888 only describes the situation when -funsafe-math-optimizations is on.
26889 (vect_float_strict): Document.
26891 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
26893 PR tree-optimization/83965
26894 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
26895 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
26896 instead of checking only for a reduction.
26897 (vect_recog_widen_sum_pattern): Likewise.
26899 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
26901 * predict.c (probably_never_executed): Only use precise profile info.
26902 (compute_function_frequency): Skip after inlining hack since we now
26903 have quality checking.
26905 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
26907 * profile-count.h (profile_probability::very_unlikely,
26908 profile_probability::unlikely, profile_probability::even): Set
26909 precision to guessed.
26911 2018-01-23 Richard Biener <rguenther@suse.de>
26913 PR tree-optimization/83963
26914 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26915 Properly terminate dominator walk when crossing the exit edge not
26916 when visiting its source block.
26918 2018-01-23 Jakub Jelinek <jakub@redhat.com>
26921 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
26922 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
26924 2018-01-22 Jakub Jelinek <jakub@redhat.com>
26926 PR tree-optimization/83957
26927 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
26928 semicolon after for body surrounded by braces.
26930 PR tree-optimization/83081
26931 * profile-count.h (profile_probability::split): New method.
26932 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
26933 Use profile_probability::split.
26934 (do_compare_rtx_and_jump): Fix adjustment of probabilities
26935 when splitting a single conditional jump into 2.
26937 2018-01-22 David Malcolm <dmalcolm@redhat.com>
26939 PR tree-optimization/69452
26940 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
26943 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26945 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
26946 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
26947 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
26949 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26951 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
26953 * config/rl78/rl78.md (movdi): New define_expand.
26954 * config/rl78/rl78.c (rl78_split_movdi): New function.
26956 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26959 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
26961 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
26962 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
26963 128-bit to produce an UNSPEC move to get the double word with the
26964 signbit and then a shift directly to do signbit.
26965 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
26966 implementation with a new version that just does either a direct
26967 move or a regular move. Move memory interface to separate insns.
26968 Move insns so they are next to the expander.
26969 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
26970 with signbit move. Split big and little endian case.
26971 (signbit<mode>2_dm_mem_le): Likewise.
26972 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
26973 (signbit<mode>2_dm2): Likewise.
26975 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26977 * config/rl78/rl78.md (anddi3): New define_expand.
26979 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26981 * config/rl78/rl78.md (umindi3): New define_expand.
26983 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26985 * config/rl78/rl78.md (smindi3): New define_expand.
26987 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26989 * config/rl78/rl78.md (smaxdi3): New define_expand.
26991 2018-01-22 Carl Love <cel@us.ibm.com>
26993 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
26994 LVX_V1TI): Add macro expansion.
26995 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
26996 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
26997 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
26998 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
26999 Change check to determine if the instruction is a byte reversing
27000 entry. Fix typo in comment.
27001 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
27002 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
27003 Add def_builtin calls for new builtins.
27004 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
27005 Add define_insn expansion.
27007 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27009 * config/rl78/rl78.md (umaxdi3): New define_expand.
27011 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27013 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
27014 for non-QImode registers.
27016 2018-01-22 Richard Biener <rguenther@suse.de>
27018 PR tree-optimization/83963
27019 * graphite-scop-detection.c (scop_detection::get_sese): Delay
27020 including the loop exit block.
27021 (scop_detection::merge_sese): Likewise.
27022 (scop_detection::add_scop): Do it here instead.
27024 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27026 * doc/sourcebuild.texi (arm_softfloat): Document.
27028 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
27031 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
27032 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
27033 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
27035 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27036 David Edelsohn <dje.gcc@gmail.com>
27039 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27040 Change "crset eq" to "crset 2".
27041 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27042 (*call_indirect_aix<mode>_nospec): Likewise.
27043 (*call_value_indirect_aix<mode>_nospec): Likewise.
27044 (*call_indirect_elfv2<mode>_nospec): Likewise.
27045 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
27046 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
27047 change assembly output from . to $.
27048 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27049 (indirect_jump<mode>_nospec): Change assembly output from . to $.
27050 (*tablejump<mode>_internal1_nospec): Likewise.
27052 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
27055 * config/sh/sh_optimize_sett_clrt.cc:
27056 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
27058 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
27060 PR tree-optimization/83940
27061 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
27062 offset_dt to vect_constant_def rather than vect_unknown_def_type.
27063 (vect_check_load_store_mask): Add a mask_dt_out parameter and
27064 use it to pass back the definition type.
27065 (vect_check_store_rhs): Likewise rhs_dt_out.
27066 (vect_build_gather_load_calls): Add a mask_dt argument and use
27067 it instead of a call to vect_is_simple_use.
27068 (vectorizable_store): Update calls to vect_check_load_store_mask
27069 and vect_check_store_rhs. Use the dt returned by the latter instead
27070 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
27071 instead of calls to vect_is_simple_use. Pass the scalar rather
27072 than the vector operand to vect_is_simple_use when handling
27073 second and subsequent copies of an rhs value.
27074 (vectorizable_load): Update calls to vect_check_load_store_mask
27075 and vect_build_gather_load_calls. Use the cached mask_dt and
27076 gs_info.offset_dt instead of calls to vect_is_simple_use.
27078 2018-01-20 Jakub Jelinek <jakub@redhat.com>
27080 PR middle-end/83945
27081 * tree-emutls.c: Include gimplify.h.
27082 (lower_emutls_2): New function.
27083 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
27084 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
27085 it before further processing.
27088 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
27089 UINTVAL (trueop1) instead of INTVAL (op1).
27091 2018-01-19 Jakub Jelinek <jakub@redhat.com>
27095 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
27096 INCOMING_FRAME_SP_OFFSET if not defined.
27097 (scan_trace): Add ENTRY argument. If true and
27098 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
27099 emit a note to adjust the CFA offset.
27100 (create_cfi_notes): Adjust scan_trace callers.
27101 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
27102 INCOMING_FRAME_SP_OFFSET in the CIE.
27103 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
27104 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
27106 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
27107 * doc/tm.texi: Regenerated.
27109 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27111 PR rtl-optimization/83147
27112 * lra-constraints.c (remove_inheritance_pseudos): Use
27113 lra_substitute_pseudo_within_insn.
27115 2018-01-19 Tom de Vries <tom@codesourcery.com>
27116 Cesar Philippidis <cesar@codesourcery.com>
27119 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
27121 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
27124 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
27125 spaces for function labels.
27127 2018-01-19 Martin Liska <mliska@suse.cz>
27129 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
27130 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
27131 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
27132 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
27133 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
27134 (PRED_CONST_RETURN): Change from 69 to 65.
27135 (PRED_NULL_RETURN): Change from 91 to 71.
27136 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
27137 (PRED_LOOP_GUARD): Change from 66 to 73.
27139 2018-01-19 Martin Liska <mliska@suse.cz>
27141 * predict.c (predict_insn_def): Add new assert.
27142 (struct branch_predictor): Change type to signed integer.
27143 (test_prediction_value_range): Amend test to cover
27144 PROB_UNINITIALIZED.
27145 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
27146 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
27147 (PRED_LOOP_ITERATIONS_MAX): Likewise.
27148 (PRED_LOOP_IV_COMPARE): Likewise.
27149 * predict.h (PROB_UNINITIALIZED): Define new constant.
27151 2018-01-19 Martin Liska <mliska@suse.cz>
27153 * predict.c (dump_prediction): Add new format for
27154 analyze_brprob.py script which is enabled with -details
27156 * profile-count.h (precise_p): New function.
27158 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
27160 PR tree-optimization/83922
27161 * tree-vect-loop.c (vect_verify_full_masking): Return false if
27162 there are no statements that need masking.
27163 (vect_active_double_reduction_p): New function.
27164 (vect_analyze_loop_operations): Use it when handling phis that
27165 are not in the loop header.
27167 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
27169 PR tree-optimization/83914
27170 * tree-vect-loop.c (vectorizable_induction): Don't convert
27171 init_expr or apply the peeling adjustment for inductions
27172 that are nested within the vectorized loop.
27174 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27176 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
27179 2018-01-18 Jakub Jelinek <jakub@redhat.com>
27183 * function.h (gimplify_parameters): Add gimple_seq * argument.
27184 * function.c: Include gimple.h and options.h.
27185 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
27186 for the added local temporaries if needed.
27187 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
27188 if there are any parameter cleanups, wrap whole body into a
27189 try/finally with the cleanups.
27191 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27194 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27195 Use GET_MODE_CLASS for scalar floating point.
27197 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
27201 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
27202 Fix call of call_cgraph_insertion_hooks.
27204 2018-01-18 Martin Sebor <msebor@redhat.com>
27206 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
27208 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
27211 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
27214 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
27217 * common.opt: (-ffile-prefix-map): New option.
27218 * opts.c (common_handle_option): Defer it.
27219 * opts-global.c (handle_common_deferred_options): Handle it.
27220 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
27221 * file-prefix-map.h: New file.
27222 (remap_debug_filename, add_debug_prefix_map): ...here.
27223 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
27224 * final.c (debug_prefix_map, add_debug_prefix_map
27225 remap_debug_filename): Move to...
27226 * file-prefix-map.c: New file.
27227 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
27228 generalize, get rid of alloca(), use strrchr() instead of strchr().
27229 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
27230 Implement in terms of add_prefix_map().
27231 (remap_macro_filename, remap_debug_filename): Implement in term of
27233 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
27234 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
27235 * dbxout.c: Include file-prefix-map.h.
27236 * varasm.c: Likewise.
27237 * vmsdbgout.c: Likewise.
27238 * xcoffout.c: Likewise.
27239 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
27240 * doc/cppopts.texi (-fmacro-prefix-map): Document.
27241 * doc/invoke.texi (-ffile-prefix-map): Document.
27242 (-fdebug-prefix-map): Update description.
27244 2018-01-18 Martin Liska <mliska@suse.cz>
27246 * config/i386/i386.c (indirect_thunk_name): Document that also
27248 (output_indirect_thunk): Document why both instructions
27249 (pause and lfence) are generated.
27251 2018-01-18 Richard Biener <rguenther@suse.de>
27253 PR tree-optimization/83887
27254 * graphite-scop-detection.c
27255 (scop_detection::get_nearest_dom_with_single_entry): Remove.
27256 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
27257 (scop_detection::merge_sese): Re-implement with a flood-fill
27258 algorithm that properly finds a SESE region if it exists.
27260 2018-01-18 Jakub Jelinek <jakub@redhat.com>
27263 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
27264 pointer_diff optimizations use view_convert instead of convert.
27266 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27268 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27269 Generate different code for -mno-speculate-indirect-jumps.
27270 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27271 (*call_indirect_aix<mode>): Disable for
27272 -mno-speculate-indirect-jumps.
27273 (*call_indirect_aix<mode>_nospec): New define_insn.
27274 (*call_value_indirect_aix<mode>): Disable for
27275 -mno-speculate-indirect-jumps.
27276 (*call_value_indirect_aix<mode>_nospec): New define_insn.
27277 (*sibcall_nonlocal_sysv<mode>): Generate different code for
27278 -mno-speculate-indirect-jumps.
27279 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27281 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
27283 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
27284 long double type, set the flags for noting the default long double
27285 type, even if we don't pass or return a long double type.
27287 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
27290 * ipa-inline.c (flatten_function): Do not overwrite final inlining
27293 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
27295 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
27296 support for merge[hl].
27297 (fold_mergehl_helper): New helper function.
27298 (tree-vector-builder.h): New #include for tree_vector_builder usage.
27299 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
27300 (altivec_vmrglw_direct): Add xxmrglw insn.
27302 2018-01-17 Andrew Waterman <andrew@sifive.com>
27304 * config/riscv/riscv.c (riscv_conditional_register_usage): If
27305 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
27307 2018-01-17 David Malcolm <dmalcolm@redhat.com>
27310 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
27311 call the lto_location_cache before reading the
27312 DECL_SOURCE_LOCATION of the types.
27314 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27315 Richard Sandiford <richard.sandiford@linaro.org>
27317 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
27318 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
27319 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
27320 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
27321 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
27323 * config/aarch64/constraints.md (aarch64_movti_operand):
27325 * config/aarch64/predicates.md (Uti): Add new constraint.
27327 2018-01-17 Carl Love <cel@us.ibm.com>
27329 * config/rs6000/vsx.md (define_expand xl_len_r,
27330 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
27331 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
27333 (define_expand, define_insn): Move the shift left from the
27334 define_insn to the define_expand for lxvl and stxvl instructions.
27335 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
27336 and XL_LEN_R definitions to PURE.
27338 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
27340 * config/i386/i386.c (indirect_thunk_name): Declare regno
27341 as unsigned int. Compare regno with INVALID_REGNUM.
27342 (output_indirect_thunk): Ditto.
27343 (output_indirect_thunk_function): Ditto.
27344 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
27345 in the call to output_indirect_thunk_function.
27347 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
27349 PR middle-end/83884
27350 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
27351 rather than the size of inner_type to determine the stack slot size
27352 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
27354 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
27357 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
27360 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27362 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
27363 endian Linux systems to optionally enable multilibs for selecting
27364 the long double type if the user configured an explicit type.
27365 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
27366 have no long double multilibs if not defined.
27367 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
27368 warn if the user used -mabi={ieee,ibm}longdouble and we built
27369 multilibs for long double.
27370 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
27371 appropriate multilib option.
27372 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
27374 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
27375 for building long double multilibs.
27376 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
27378 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
27380 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
27383 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
27385 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
27388 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
27391 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
27394 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
27396 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
27397 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
27399 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
27401 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
27402 different rtl trees depending on TARGET_64BIT.
27403 (rs6000_gen_lvx): Likewise.
27405 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
27407 * config/visium/visium.md (nop): Tweak comment.
27408 (hazard_nop): Likewise.
27410 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27412 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
27413 -mspeculate-indirect-jumps.
27414 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
27415 for -mno-speculate-indirect-jumps.
27416 (*call_indirect_elfv2<mode>_nospec): New define_insn.
27417 (*call_value_indirect_elfv2<mode>): Disable for
27418 -mno-speculate-indirect-jumps.
27419 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
27420 (indirect_jump): Emit different RTL for
27421 -mno-speculate-indirect-jumps.
27422 (*indirect_jump<mode>): Disable for
27423 -mno-speculate-indirect-jumps.
27424 (*indirect_jump<mode>_nospec): New define_insn.
27425 (tablejump): Emit different RTL for
27426 -mno-speculate-indirect-jumps.
27427 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
27428 (tablejumpsi_nospec): New define_expand.
27429 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
27430 (tablejumpdi_nospec): New define_expand.
27431 (*tablejump<mode>_internal1): Disable for
27432 -mno-speculate-indirect-jumps.
27433 (*tablejump<mode>_internal1_nospec): New define_insn.
27434 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
27437 2018-01-16 Artyom Skrobov tyomitch@gmail.com
27439 * caller-save.c (insert_save): Drop unnecessary parameter. All
27442 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27443 Richard Biener <rguenth@suse.de>
27446 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
27447 return early, inline manually is_gimple_sizepos. Make sure if we
27448 call gimplify_expr we don't end up with a gimple constant.
27449 * tree.c (variably_modified_type_p): Don't return true for
27450 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
27451 * gimplify.h (is_gimple_sizepos): Remove.
27453 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27455 PR tree-optimization/83857
27456 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
27457 vectorizable_live_operation for pure SLP statements.
27458 (vectorizable_live_operation): Handle PHIs.
27460 2018-01-16 Richard Biener <rguenther@suse.de>
27462 PR tree-optimization/83867
27463 * tree-vect-stmts.c (vect_transform_stmt): Precompute
27464 nested_in_vect_loop_p since the scalar stmt may get invalidated.
27466 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27469 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
27470 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
27471 If off is not INTEGER_CST, issue a may not be aligned warning
27472 rather than isn't aligned. Use isn%'t rather than isn't.
27473 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
27475 <case MULT_EXPR>: Improve the case when bottom and one of the
27476 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
27477 operand, in that case check if the other operand is multiple of
27478 bottom divided by the INTEGER_CST operand.
27480 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27483 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
27484 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
27485 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
27486 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
27487 * config/pa/pa.c (pa_function_arg_advance): Likewise.
27488 (pa_function_arg, pa_arg_partial_bytes): Likewise.
27489 (pa_function_arg_size): New function.
27491 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27493 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
27494 in a separate statement.
27496 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27498 PR tree-optimization/83847
27499 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
27500 group gathers and scatters.
27502 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27504 PR rtl-optimization/86620
27505 * params.def (max-sched-ready-insns): Bump minimum value to 1.
27507 PR rtl-optimization/83213
27508 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
27509 to last if both are JUMP_INSNs.
27511 PR tree-optimization/83843
27512 * gimple-ssa-store-merging.c
27513 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
27514 store_immediate_info for bswap/nop orig_stores.
27516 2018-01-15 Andrew Waterman <andrew@sifive.com>
27518 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
27520 <UDIV>: Increase cost if !TARGET_DIV.
27522 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27524 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
27525 (define_attr "cr_logical_3op"): New.
27526 (cceq_ior_compare): Adjust.
27527 (cceq_ior_compare_complement): Adjust.
27528 (*cceq_rev_compare): Adjust.
27529 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
27530 (is_cracked_insn): Adjust.
27531 (insn_must_be_first_in_group): Adjust.
27532 * config/rs6000/40x.md: Adjust.
27533 * config/rs6000/440.md: Adjust.
27534 * config/rs6000/476.md: Adjust.
27535 * config/rs6000/601.md: Adjust.
27536 * config/rs6000/603.md: Adjust.
27537 * config/rs6000/6xx.md: Adjust.
27538 * config/rs6000/7450.md: Adjust.
27539 * config/rs6000/7xx.md: Adjust.
27540 * config/rs6000/8540.md: Adjust.
27541 * config/rs6000/cell.md: Adjust.
27542 * config/rs6000/e300c2c3.md: Adjust.
27543 * config/rs6000/e500mc.md: Adjust.
27544 * config/rs6000/e500mc64.md: Adjust.
27545 * config/rs6000/e5500.md: Adjust.
27546 * config/rs6000/e6500.md: Adjust.
27547 * config/rs6000/mpc.md: Adjust.
27548 * config/rs6000/power4.md: Adjust.
27549 * config/rs6000/power5.md: Adjust.
27550 * config/rs6000/power6.md: Adjust.
27551 * config/rs6000/power7.md: Adjust.
27552 * config/rs6000/power8.md: Adjust.
27553 * config/rs6000/power9.md: Adjust.
27554 * config/rs6000/rs64.md: Adjust.
27555 * config/rs6000/titan.md: Adjust.
27557 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27559 * config/i386/predicates.md (indirect_branch_operand): Rewrite
27560 ix86_indirect_branch_register logic.
27562 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27564 * config/i386/constraints.md (Bs): Update
27565 ix86_indirect_branch_register check. Don't check
27566 ix86_indirect_branch_register with GOT_memory_operand.
27568 * config/i386/predicates.md (GOT_memory_operand): Don't check
27569 ix86_indirect_branch_register here.
27570 (GOT32_symbol_operand): Likewise.
27572 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27574 * config/i386/predicates.md (constant_call_address_operand):
27575 Rewrite ix86_indirect_branch_register logic.
27576 (sibcall_insn_operand): Likewise.
27578 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27580 * config/i386/constraints.md (Bs): Replace
27581 ix86_indirect_branch_thunk_register with
27582 ix86_indirect_branch_register.
27584 * config/i386/i386.md (indirect_jump): Likewise.
27585 (tablejump): Likewise.
27586 (*sibcall_memory): Likewise.
27587 (*sibcall_value_memory): Likewise.
27588 Peepholes of indirect call and jump via memory: Likewise.
27589 * config/i386/i386.opt: Likewise.
27590 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27591 (GOT_memory_operand): Likewise.
27592 (call_insn_operand): Likewise.
27593 (sibcall_insn_operand): Likewise.
27594 (GOT32_symbol_operand): Likewise.
27596 2018-01-15 Jakub Jelinek <jakub@redhat.com>
27598 PR middle-end/83837
27599 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
27600 type rather than type addr's type points to.
27601 (expand_omp_atomic_mutex): Likewise.
27602 (expand_omp_atomic): Likewise.
27604 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27607 * config/i386/i386.c (output_indirect_thunk_function): Use
27608 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
27609 for __x86_return_thunk.
27611 2018-01-15 Richard Biener <rguenther@suse.de>
27613 PR middle-end/83850
27614 * expmed.c (extract_bit_field_1): Fix typo.
27616 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27619 * config/arm/iterators.md (VF): New mode iterator.
27620 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
27621 Remove integer-related logic from pattern.
27622 (neon_vabd<mode>_3): Likewise.
27624 2018-01-15 Jakub Jelinek <jakub@redhat.com>
27626 PR middle-end/82694
27627 * common.opt (fstrict-overflow): No longer an alias.
27628 (fwrapv-pointer): New option.
27629 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
27630 also for pointer types based on flag_wrapv_pointer.
27631 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
27632 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
27633 opts->x_flag_wrapv got set.
27634 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
27635 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
27636 POINTER_TYPE_OVERFLOW_UNDEFINED.
27637 * match.pd: Likewise in address comparison pattern.
27638 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
27640 2018-01-15 Richard Biener <rguenther@suse.de>
27643 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
27644 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
27645 Reset type names to their identifier if their TYPE_DECL doesn't
27646 have linkage (and thus is used for ODR and devirt).
27647 (save_debug_info_for_decl): Remove.
27648 (save_debug_info_for_type): Likewise.
27649 (add_tree_to_fld_list): Adjust.
27650 * tree-pretty-print.c (dump_generic_node): Make dumping of
27651 type names more robust.
27653 2018-01-15 Richard Biener <rguenther@suse.de>
27655 * BASE-VER: Bump to 8.0.1.
27657 2018-01-14 Martin Sebor <msebor@redhat.com>
27660 * builtins.c (check_access): Avoid warning when the no-warning bit
27663 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
27665 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
27666 * ira-color (allocno_hard_regs_compare): Likewise.
27668 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
27671 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
27672 Use .pushsection/.popsection.
27674 2018-01-14 Martin Sebor <msebor@redhat.com>
27677 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
27679 2018-01-14 Jakub Jelinek <jakub@redhat.com>
27681 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
27682 entry from extra_headers.
27683 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
27684 extra_headers, make the list bitwise identical to the i?86-*-* one.
27686 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27688 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
27689 -mcmodel=large with -mindirect-branch=thunk,
27690 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
27691 -mfunction-return=thunk-extern.
27692 * doc/invoke.texi: Document -mcmodel=large is incompatible with
27693 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
27694 -mfunction-return=thunk and -mfunction-return=thunk-extern.
27696 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27698 * config/i386/i386.c (print_reg): Print the name of the full
27699 integer register without '%'.
27700 (ix86_print_operand): Handle 'V'.
27701 * doc/extend.texi: Document 'V' modifier.
27703 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27705 * config/i386/constraints.md (Bs): Disallow memory operand for
27706 -mindirect-branch-register.
27708 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27709 (GOT_memory_operand): Likewise.
27710 (call_insn_operand): Likewise.
27711 (sibcall_insn_operand): Likewise.
27712 (GOT32_symbol_operand): Likewise.
27713 * config/i386/i386.md (indirect_jump): Call convert_memory_address
27714 for -mindirect-branch-register.
27715 (tablejump): Likewise.
27716 (*sibcall_memory): Likewise.
27717 (*sibcall_value_memory): Likewise.
27718 Disallow peepholes of indirect call and jump via memory for
27719 -mindirect-branch-register.
27720 (*call_pop): Replace m with Bw.
27721 (*call_value_pop): Likewise.
27722 (*sibcall_pop_memory): Replace m with Bs.
27723 * config/i386/i386.opt (mindirect-branch-register): New option.
27724 * doc/invoke.texi: Document -mindirect-branch-register option.
27726 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27728 * config/i386/i386-protos.h (ix86_output_function_return): New.
27729 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
27730 set function_return_type.
27731 (indirect_thunk_name): Add ret_p to indicate thunk for function
27733 (output_indirect_thunk_function): Pass false to
27734 indirect_thunk_name.
27735 (ix86_output_indirect_branch_via_reg): Likewise.
27736 (ix86_output_indirect_branch_via_push): Likewise.
27737 (output_indirect_thunk_function): Create alias for function
27738 return thunk if regno < 0.
27739 (ix86_output_function_return): New function.
27740 (ix86_handle_fndecl_attribute): Handle function_return.
27741 (ix86_attribute_table): Add function_return.
27742 * config/i386/i386.h (machine_function): Add
27743 function_return_type.
27744 * config/i386/i386.md (simple_return_internal): Use
27745 ix86_output_function_return.
27746 (simple_return_internal_long): Likewise.
27747 * config/i386/i386.opt (mfunction-return=): New option.
27748 (indirect_branch): Mention -mfunction-return=.
27749 * doc/extend.texi: Document function_return function attribute.
27750 * doc/invoke.texi: Document -mfunction-return= option.
27752 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27754 * config/i386/i386-opts.h (indirect_branch): New.
27755 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
27756 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
27757 with local indirect jump when converting indirect call and jump.
27758 (ix86_set_indirect_branch_type): New.
27759 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
27760 (indirectlabelno): New.
27761 (indirect_thunk_needed): Likewise.
27762 (indirect_thunk_bnd_needed): Likewise.
27763 (indirect_thunks_used): Likewise.
27764 (indirect_thunks_bnd_used): Likewise.
27765 (INDIRECT_LABEL): Likewise.
27766 (indirect_thunk_name): Likewise.
27767 (output_indirect_thunk): Likewise.
27768 (output_indirect_thunk_function): Likewise.
27769 (ix86_output_indirect_branch_via_reg): Likewise.
27770 (ix86_output_indirect_branch_via_push): Likewise.
27771 (ix86_output_indirect_branch): Likewise.
27772 (ix86_output_indirect_jmp): Likewise.
27773 (ix86_code_end): Call output_indirect_thunk_function if needed.
27774 (ix86_output_call_insn): Call ix86_output_indirect_branch if
27776 (ix86_handle_fndecl_attribute): Handle indirect_branch.
27777 (ix86_attribute_table): Add indirect_branch.
27778 * config/i386/i386.h (machine_function): Add indirect_branch_type
27779 and has_local_indirect_jump.
27780 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
27782 (tablejump): Likewise.
27783 (*indirect_jump): Use ix86_output_indirect_jmp.
27784 (*tablejump_1): Likewise.
27785 (simple_return_indirect_internal): Likewise.
27786 * config/i386/i386.opt (mindirect-branch=): New option.
27787 (indirect_branch): New.
27790 (thunk-inline): Likewise.
27791 (thunk-extern): Likewise.
27792 * doc/extend.texi: Document indirect_branch function attribute.
27793 * doc/invoke.texi: Document -mindirect-branch= option.
27795 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
27798 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
27800 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
27802 * ipa-inline.c (want_inline_small_function_p): Return false if
27803 inlining has already failed with CIF_FINAL_ERROR.
27804 (update_caller_keys): Call want_inline_small_function_p before
27806 (update_callee_keys): Likewise.
27808 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27810 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
27812 (rs6000_quadword_masked_address_p): Likewise.
27813 (quad_aligned_load_p): Likewise.
27814 (quad_aligned_store_p): Likewise.
27815 (const_load_sequence_p): Add comment to describe the outer-most loop.
27816 (mimic_memory_attributes_and_flags): New function.
27817 (rs6000_gen_stvx): Likewise.
27818 (replace_swapped_aligned_store): Likewise.
27819 (rs6000_gen_lvx): Likewise.
27820 (replace_swapped_aligned_load): Likewise.
27821 (replace_swapped_load_constant): Capitalize argument name in
27822 comment describing this function.
27823 (rs6000_analyze_swaps): Add a third pass to search for vector loads
27824 and stores that access quad-word aligned addresses and replace
27825 with stvx or lvx instructions when appropriate.
27826 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
27827 New function prototype.
27828 (rs6000_quadword_masked_address_p): Likewise.
27829 (rs6000_gen_lvx): Likewise.
27830 (rs6000_gen_stvx): Likewise.
27831 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
27832 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
27833 when memory address is aligned.
27834 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
27835 this split to select lvx instruction when memory address is aligned.
27836 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
27837 instruction when memory address is aligned.
27838 (*vsx_le_perm_load_v16qi): Likewise.
27839 (four unnamed splitters): Modify to select the stvx instruction
27840 when memory is aligned.
27842 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
27844 * predict.c (determine_unlikely_bbs): Handle correctly BBs
27845 which appears in the queue multiple times.
27847 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27848 Alan Hayward <alan.hayward@arm.com>
27849 David Sherwood <david.sherwood@arm.com>
27851 * tree-vectorizer.h (vec_lower_bound): New structure.
27852 (_loop_vec_info): Add check_nonzero and lower_bounds.
27853 (LOOP_VINFO_CHECK_NONZERO): New macro.
27854 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
27855 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
27856 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
27857 fields. Make seg_len the distance travelled, not including the
27859 (dr_direction_indicator): Declare.
27860 (dr_zero_step_indicator): Likewise.
27861 (dr_known_forward_stride_p): Likewise.
27862 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
27864 (runtime_alias_check_p): Allow runtime alias checks with
27866 (operator ==): Compare access_size and align.
27867 (prune_runtime_alias_test_list): Rework for new distinction between
27868 the access_size and seg_len.
27869 (create_intersect_range_checks_index): Likewise. Cope with polynomial
27871 (get_segment_min_max): New function.
27872 (create_intersect_range_checks): Use it.
27873 (dr_step_indicator): New function.
27874 (dr_direction_indicator): Likewise.
27875 (dr_zero_step_indicator): Likewise.
27876 (dr_known_forward_stride_p): Likewise.
27877 * tree-loop-distribution.c (data_ref_segment_size): Return
27878 DR_STEP * (niters - 1).
27879 (compute_alias_check_pairs): Update call to the dr_with_seg_len
27881 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
27882 (vect_preserves_scalar_order_p): New function, split out from...
27883 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
27884 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
27885 (vect_vfa_access_size): New function.
27886 (vect_vfa_align): Likewise.
27887 (vect_compile_time_alias): Take access_size_a and access_b arguments.
27888 (dump_lower_bound): New function.
27889 (vect_check_lower_bound): Likewise.
27890 (vect_small_gap_p): Likewise.
27891 (vectorizable_with_step_bound_p): Likewise.
27892 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
27893 depencies if the vectorization factor is 1. Convert the checks
27894 for nonzero steps into checks on the bounds of DR_STEP. Try using
27895 a bunds check for variable steps if the minimum required step is
27896 relatively small. Update calls to the dr_with_seg_len
27897 constructor and to vect_compile_time_alias.
27898 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
27900 (vect_loop_versioning): Call it.
27901 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
27903 (vect_estimate_min_profitable_iters): Account for any bounds checks.
27905 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27906 Alan Hayward <alan.hayward@arm.com>
27907 David Sherwood <david.sherwood@arm.com>
27909 * doc/sourcebuild.texi (vect_scatter_store): Document.
27910 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
27912 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
27914 * genopinit.c (main): Add supports_vec_scatter_store and
27915 supports_vec_scatter_store_cached to target_optabs.
27916 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
27917 IFN_MASK_SCATTER_STORE.
27918 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
27920 * internal-fn.h (internal_store_fn_p): Declare.
27921 (internal_fn_stored_value_index): Likewise.
27922 * internal-fn.c (scatter_store_direct): New macro.
27923 (expand_scatter_store_optab_fn): New function.
27924 (direct_scatter_store_optab_supported_p): New macro.
27925 (internal_store_fn_p): New function.
27926 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
27927 IFN_MASK_SCATTER_STORE.
27928 (internal_fn_mask_index): Likewise.
27929 (internal_fn_stored_value_index): New function.
27930 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
27931 for scatter stores.
27932 * optabs-query.h (supports_vec_scatter_store_p): Declare.
27933 * optabs-query.c (supports_vec_scatter_store_p): New function.
27934 * tree-vectorizer.h (vect_get_store_rhs): Declare.
27935 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
27936 true for scatter stores.
27937 (vect_gather_scatter_fn_p): Handle scatter stores too.
27938 (vect_check_gather_scatter): Consider using scatter stores if
27939 supports_vec_scatter_store_p.
27940 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
27941 scatter stores too.
27942 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
27943 internal_fn_stored_value_index.
27944 (check_load_store_masking): Handle scatter stores too.
27945 (vect_get_store_rhs): Make public.
27946 (vectorizable_call): Use internal_store_fn_p.
27947 (vectorizable_store): Handle scatter store internal functions.
27948 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
27949 when deciding whether the end of the group has been reached.
27950 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
27951 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
27952 (mask_scatter_store<mode>): New insns.
27954 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27955 Alan Hayward <alan.hayward@arm.com>
27956 David Sherwood <david.sherwood@arm.com>
27958 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
27959 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
27960 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
27962 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
27963 Use vect_truncate_gather_scatter_offset if we can't treat the
27964 operation as a normal gather load or scatter store.
27965 (get_group_load_store_type): Take the gather_scatter_info
27966 as argument. Try using a gather load or scatter store for
27967 single-element groups.
27968 (get_load_store_type): Update calls to get_group_load_store_type
27969 and vect_use_strided_gather_scatters_p.
27971 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27972 Alan Hayward <alan.hayward@arm.com>
27973 David Sherwood <david.sherwood@arm.com>
27975 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
27976 optional tree argument.
27977 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
27979 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
27980 but continue to use the current value as a fallback.
27981 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
27982 to compare the updates.
27983 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
27984 (get_load_store_type): Use it when handling a strided access.
27985 (vect_get_strided_load_store_ops): New function.
27986 (vect_get_data_ptr_increment): Likewise.
27987 (vectorizable_load): Handle strided gather loads. Always pass
27988 a step to vect_create_data_ref_ptr and bump_vector_ptr.
27990 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27991 Alan Hayward <alan.hayward@arm.com>
27992 David Sherwood <david.sherwood@arm.com>
27994 * doc/md.texi (gather_load@var{m}): Document.
27995 (mask_gather_load@var{m}): Likewise.
27996 * genopinit.c (main): Add supports_vec_gather_load and
27997 supports_vec_gather_load_cached to target_optabs.
27998 * optabs-tree.c (init_tree_optimization_optabs): Use
27999 ggc_cleared_alloc to allocate target_optabs.
28000 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
28001 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
28003 * internal-fn.h (internal_load_fn_p): Declare.
28004 (internal_gather_scatter_fn_p): Likewise.
28005 (internal_fn_mask_index): Likewise.
28006 (internal_gather_scatter_fn_supported_p): Likewise.
28007 * internal-fn.c (gather_load_direct): New macro.
28008 (expand_gather_load_optab_fn): New function.
28009 (direct_gather_load_optab_supported_p): New macro.
28010 (direct_internal_fn_optab): New function.
28011 (internal_load_fn_p): Likewise.
28012 (internal_gather_scatter_fn_p): Likewise.
28013 (internal_fn_mask_index): Likewise.
28014 (internal_gather_scatter_fn_supported_p): Likewise.
28015 * optabs-query.c (supports_at_least_one_mode_p): New function.
28016 (supports_vec_gather_load_p): Likewise.
28017 * optabs-query.h (supports_vec_gather_load_p): Declare.
28018 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
28019 and memory_type field.
28020 (NUM_PATTERNS): Bump to 15.
28021 * tree-vect-data-refs.c: Include internal-fn.h.
28022 (vect_gather_scatter_fn_p): New function.
28023 (vect_describe_gather_scatter_call): Likewise.
28024 (vect_check_gather_scatter): Try using internal functions for
28025 gather loads. Recognize existing calls to a gather load function.
28026 (vect_analyze_data_refs): Consider using gather loads if
28027 supports_vec_gather_load_p.
28028 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
28029 (vect_get_gather_scatter_offset_type): Likewise.
28030 (vect_convert_mask_for_vectype): Likewise.
28031 (vect_add_conversion_to_patterm): Likewise.
28032 (vect_try_gather_scatter_pattern): Likewise.
28033 (vect_recog_gather_scatter_pattern): New pattern recognizer.
28034 (vect_vect_recog_func_ptrs): Add it.
28035 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28036 internal_fn_mask_index and internal_gather_scatter_fn_p.
28037 (check_load_store_masking): Take the gather_scatter_info as an
28038 argument and handle gather loads.
28039 (vect_get_gather_scatter_ops): New function.
28040 (vectorizable_call): Check internal_load_fn_p.
28041 (vectorizable_load): Likewise. Handle gather load internal
28043 (vectorizable_store): Update call to check_load_store_masking.
28044 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
28045 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
28046 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
28047 (aarch64_gather_scale_operand_d): New predicates.
28048 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
28049 (mask_gather_load<mode>): New insns.
28051 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28052 Alan Hayward <alan.hayward@arm.com>
28053 David Sherwood <david.sherwood@arm.com>
28055 * optabs.def (fold_left_plus_optab): New optab.
28056 * doc/md.texi (fold_left_plus_@var{m}): Document.
28057 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
28058 * internal-fn.c (fold_left_direct): Define.
28059 (expand_fold_left_optab_fn): Likewise.
28060 (direct_fold_left_optab_supported_p): Likewise.
28061 * fold-const-call.c (fold_const_fold_left): New function.
28062 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
28063 * tree-parloops.c (valid_reduction_p): New function.
28064 (gather_scalar_reductions): Use it.
28065 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
28066 (vect_finish_replace_stmt): Declare.
28067 * tree-vect-loop.c (fold_left_reduction_fn): New function.
28068 (needs_fold_left_reduction_p): New function, split out from...
28069 (vect_is_simple_reduction): ...here. Accept reductions that
28070 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
28071 (vect_force_simple_reduction): Also store the reduction type in
28072 the assignment's STMT_VINFO_REDUC_TYPE.
28073 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
28074 (merge_with_identity): New function.
28075 (vect_expand_fold_left): Likewise.
28076 (vectorize_fold_left_reduction): Likewise.
28077 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
28078 scalar phi in place for it. Check for target support and reject
28079 cases that would reassociate the operation. Defer the transform
28080 phase to vectorize_fold_left_reduction.
28081 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
28082 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
28083 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
28085 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28087 * tree-if-conv.c (predicate_mem_writes): Remove redundant
28088 call to ifc_temp_var.
28090 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28091 Alan Hayward <alan.hayward@arm.com>
28092 David Sherwood <david.sherwood@arm.com>
28094 * target.def (legitimize_address_displacement): Take the original
28095 offset as a poly_int.
28096 * targhooks.h (default_legitimize_address_displacement): Update
28098 * targhooks.c (default_legitimize_address_displacement): Likewise.
28099 * doc/tm.texi: Regenerate.
28100 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
28101 as an argument, moving assert of ad->disp == ad->disp_term to...
28102 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
28103 Try calling targetm.legitimize_address_displacement before expanding
28104 the address rather than afterwards, and adjust for the new interface.
28105 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
28106 Match the new hook interface. Handle SVE addresses.
28107 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
28108 new hook interface.
28110 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28112 * Makefile.in (OBJS): Add early-remat.o.
28113 * target.def (select_early_remat_modes): New hook.
28114 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
28115 * doc/tm.texi: Regenerate.
28116 * targhooks.h (default_select_early_remat_modes): Declare.
28117 * targhooks.c (default_select_early_remat_modes): New function.
28118 * timevar.def (TV_EARLY_REMAT): New timevar.
28119 * passes.def (pass_early_remat): New pass.
28120 * tree-pass.h (make_pass_early_remat): Declare.
28121 * early-remat.c: New file.
28122 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
28124 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
28126 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28127 Alan Hayward <alan.hayward@arm.com>
28128 David Sherwood <david.sherwood@arm.com>
28130 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
28131 vfm1 with a bound_epilog parameter.
28132 (vect_do_peeling): Update calls accordingly, and move the prologue
28133 call earlier in the function. Treat the base bound_epilog as 0 for
28134 fully-masked loops and retain vf - 1 for other loops. Add 1 to
28135 this base when peeling for gaps.
28136 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
28137 with fully-masked loops.
28138 (vect_estimate_min_profitable_iters): Handle the single peeled
28139 iteration in that case.
28141 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28142 Alan Hayward <alan.hayward@arm.com>
28143 David Sherwood <david.sherwood@arm.com>
28145 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
28146 single-element interleaving even if the size is not a power of 2.
28147 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
28148 accesses for single-element interleaving if the group size is
28151 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28152 Alan Hayward <alan.hayward@arm.com>
28153 David Sherwood <david.sherwood@arm.com>
28155 * doc/md.texi (fold_extract_last_@var{m}): Document.
28156 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
28157 * optabs.def (fold_extract_last_optab): New optab.
28158 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
28159 * internal-fn.c (fold_extract_direct): New macro.
28160 (expand_fold_extract_optab_fn): Likewise.
28161 (direct_fold_extract_optab_supported_p): Likewise.
28162 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
28163 * tree-vect-loop.c (vect_model_reduction_cost): Handle
28164 EXTRACT_LAST_REDUCTION.
28165 (get_initial_def_for_reduction): Do not create an initial vector
28166 for EXTRACT_LAST_REDUCTION reductions.
28167 (vectorizable_reduction): Leave the scalar phi in place for
28168 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
28169 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
28170 epilogue code for EXTRACT_LAST_REDUCTION and defer the
28171 transform phase to vectorizable_condition.
28172 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
28174 (vect_finish_stmt_generation): ...here.
28175 (vect_finish_replace_stmt): New function.
28176 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
28177 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
28179 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
28181 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28182 Alan Hayward <alan.hayward@arm.com>
28183 David Sherwood <david.sherwood@arm.com>
28185 * doc/md.texi (extract_last_@var{m}): Document.
28186 * optabs.def (extract_last_optab): New optab.
28187 * internal-fn.def (EXTRACT_LAST): New internal function.
28188 * internal-fn.c (cond_unary_direct): New macro.
28189 (expand_cond_unary_optab_fn): Likewise.
28190 (direct_cond_unary_optab_supported_p): Likewise.
28191 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
28192 loops using EXTRACT_LAST.
28193 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
28194 (extract_last_<mode>): ...this optab.
28195 (vec_extract<mode><Vel>): Update accordingly.
28197 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28198 Alan Hayward <alan.hayward@arm.com>
28199 David Sherwood <david.sherwood@arm.com>
28201 * target.def (empty_mask_is_expensive): New hook.
28202 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
28203 * doc/tm.texi: Regenerate.
28204 * targhooks.h (default_empty_mask_is_expensive): Declare.
28205 * targhooks.c (default_empty_mask_is_expensive): New function.
28206 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
28207 if the target says that empty masks are expensive.
28208 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
28210 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
28212 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28213 Alan Hayward <alan.hayward@arm.com>
28214 David Sherwood <david.sherwood@arm.com>
28216 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
28217 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
28218 (vect_use_loop_mask_for_alignment_p): New function.
28219 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
28220 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
28221 niters_skip argument. Make sure that the first niters_skip elements
28222 of the first iteration are inactive.
28223 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
28224 Update call to vect_set_loop_masks_directly.
28225 (get_misalign_in_elems): New function, split out from...
28226 (vect_gen_prolog_loop_niters): ...here.
28227 (vect_update_init_of_dr): Take a code argument that specifies whether
28228 the adjustment should be added or subtracted.
28229 (vect_update_init_of_drs): Likewise.
28230 (vect_prepare_for_masked_peels): New function.
28231 (vect_do_peeling): Skip prologue peeling if we're using a mask
28232 instead. Update call to vect_update_inits_of_drs.
28233 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28235 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
28236 alignment. Do not include the number of peeled iterations in
28237 the minimum threshold in that case.
28238 (vectorizable_induction): Adjust the start value down by
28239 LOOP_VINFO_MASK_SKIP_NITERS iterations.
28240 (vect_transform_loop): Call vect_prepare_for_masked_peels.
28241 Take the number of skipped iterations into account when calculating
28243 * tree-vect-stmts.c (vect_gen_while_not): New function.
28245 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28246 Alan Hayward <alan.hayward@arm.com>
28247 David Sherwood <david.sherwood@arm.com>
28249 * doc/sourcebuild.texi (vect_fully_masked): Document.
28250 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
28251 default value to 0.
28252 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
28254 (vect_analyze_loop_2): ...here. Don't check the vectorization
28255 factor against the number of loop iterations if the loop is
28258 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28259 Alan Hayward <alan.hayward@arm.com>
28260 David Sherwood <david.sherwood@arm.com>
28262 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
28263 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
28264 (dump_groups): Update accordingly.
28265 (iv_use::mem_type): New member variable.
28266 (address_p): New function.
28267 (record_use): Add a mem_type argument and initialize the new
28269 (record_group_use): Add a mem_type argument. Use address_p.
28270 Remove obsolete null checks of base_object. Update call to record_use.
28271 (find_interesting_uses_op): Update call to record_group_use.
28272 (find_interesting_uses_cond): Likewise.
28273 (find_interesting_uses_address): Likewise.
28274 (get_mem_type_for_internal_fn): New function.
28275 (find_address_like_use): Likewise.
28276 (find_interesting_uses_stmt): Try find_address_like_use before
28277 calling find_interesting_uses_op.
28278 (addr_offset_valid_p): Use the iv mem_type field as the type
28279 of the addressed memory.
28280 (add_autoinc_candidates): Likewise.
28281 (get_address_cost): Likewise.
28282 (split_small_address_groups_p): Use address_p.
28283 (split_address_groups): Likewise.
28284 (add_iv_candidate_for_use): Likewise.
28285 (autoinc_possible_for_pair): Likewise.
28286 (rewrite_groups): Likewise.
28287 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
28288 (determine_group_iv_cost): Update after split of USE_ADDRESS.
28289 (get_alias_ptr_type_for_ptr_address): New function.
28290 (rewrite_use_address): Rewrite address uses in calls that were
28291 identified by find_address_like_use.
28293 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28294 Alan Hayward <alan.hayward@arm.com>
28295 David Sherwood <david.sherwood@arm.com>
28297 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
28299 * gimple-expr.h (is_gimple_addressable: Likewise.
28300 * gimple-expr.c (is_gimple_address): Likewise.
28301 * internal-fn.c (expand_call_mem_ref): New function.
28302 (expand_mask_load_optab_fn): Use it.
28303 (expand_mask_store_optab_fn): Likewise.
28305 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28306 Alan Hayward <alan.hayward@arm.com>
28307 David Sherwood <david.sherwood@arm.com>
28309 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
28310 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
28311 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
28312 (cond_umax@var{mode}): Document.
28313 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
28314 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
28315 (cond_umin_optab, cond_umax_optab): New optabs.
28316 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
28317 (COND_IOR, COND_XOR): New internal functions.
28318 * internal-fn.h (get_conditional_internal_fn): Declare.
28319 * internal-fn.c (cond_binary_direct): New macro.
28320 (expand_cond_binary_optab_fn): Likewise.
28321 (direct_cond_binary_optab_supported_p): Likewise.
28322 (get_conditional_internal_fn): New function.
28323 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
28324 Cope with reduction statements that are vectorized as calls rather
28326 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
28327 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
28328 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
28329 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
28330 (UNSPEC_COND_EOR): New unspecs.
28331 (optab): Add mappings for them.
28332 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
28333 (sve_int_op, sve_fp_op): New int attributes.
28335 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28336 Alan Hayward <alan.hayward@arm.com>
28337 David Sherwood <david.sherwood@arm.com>
28339 * optabs.def (while_ult_optab): New optab.
28340 * doc/md.texi (while_ult@var{m}@var{n}): Document.
28341 * internal-fn.def (WHILE_ULT): New internal function.
28342 * internal-fn.h (direct_internal_fn_supported_p): New override
28343 that takes two types as argument.
28344 * internal-fn.c (while_direct): New macro.
28345 (expand_while_optab_fn): New function.
28346 (convert_optab_supported_p): Likewise.
28347 (direct_while_optab_supported_p): New macro.
28348 * wide-int.h (wi::udiv_ceil): New function.
28349 * tree-vectorizer.h (rgroup_masks): New structure.
28350 (vec_loop_masks): New typedef.
28351 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
28352 and fully_masked_p.
28353 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
28354 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
28355 (vect_max_vf): New function.
28356 (slpeel_make_loop_iterate_ntimes): Delete.
28357 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
28358 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
28359 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
28360 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
28361 internal-fn.h, stor-layout.h and optabs-query.h.
28362 (vect_set_loop_mask): New function.
28363 (add_preheader_seq): Likewise.
28364 (add_header_seq): Likewise.
28365 (interleave_supported_p): Likewise.
28366 (vect_maybe_permute_loop_masks): Likewise.
28367 (vect_set_loop_masks_directly): Likewise.
28368 (vect_set_loop_condition_masked): Likewise.
28369 (vect_set_loop_condition_unmasked): New function, split out from
28370 slpeel_make_loop_iterate_ntimes.
28371 (slpeel_make_loop_iterate_ntimes): Rename to..
28372 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
28373 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
28374 (vect_do_peeling): Update call accordingly.
28375 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
28377 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28378 mask_compare_type, can_fully_mask_p and fully_masked_p.
28379 (release_vec_loop_masks): New function.
28380 (_loop_vec_info): Use it to free the loop masks.
28381 (can_produce_all_loop_masks_p): New function.
28382 (vect_get_max_nscalars_per_iter): Likewise.
28383 (vect_verify_full_masking): Likewise.
28384 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
28385 retries, and free the mask rgroups before retrying. Check loop-wide
28386 reasons for disallowing fully-masked loops. Make the final decision
28387 about whether use a fully-masked loop or not.
28388 (vect_estimate_min_profitable_iters): Do not assume that peeling
28389 for the number of iterations will be needed for fully-masked loops.
28390 (vectorizable_reduction): Disable fully-masked loops.
28391 (vectorizable_live_operation): Likewise.
28392 (vect_halve_mask_nunits): New function.
28393 (vect_double_mask_nunits): Likewise.
28394 (vect_record_loop_mask): Likewise.
28395 (vect_get_loop_mask): Likewise.
28396 (vect_transform_loop): Handle the case in which the final loop
28397 iteration might handle a partial vector. Call vect_set_loop_condition
28398 instead of slpeel_make_loop_iterate_ntimes.
28399 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
28400 (check_load_store_masking): New function.
28401 (prepare_load_store_mask): Likewise.
28402 (vectorizable_store): Handle fully-masked loops.
28403 (vectorizable_load): Likewise.
28404 (supportable_widening_operation): Use vect_halve_mask_nunits for
28406 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
28407 (vect_gen_while): New function.
28408 * config/aarch64/aarch64.md (umax<mode>3): New expander.
28409 (aarch64_uqdec<mode>): New insn.
28411 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28412 Alan Hayward <alan.hayward@arm.com>
28413 David Sherwood <david.sherwood@arm.com>
28415 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
28416 (reduc_xor_scal_optab): New optabs.
28417 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
28418 (reduc_xor_scal_@var{m}): Document.
28419 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
28420 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
28421 internal functions.
28422 * fold-const-call.c (fold_const_call): Handle them.
28423 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
28424 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
28425 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
28426 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
28427 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
28428 (UNSPEC_XORV): New unspecs.
28429 (optab): Add entries for them.
28430 (BITWISEV): New int iterator.
28431 (bit_reduc_op): New int attributes.
28433 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28434 Alan Hayward <alan.hayward@arm.com>
28435 David Sherwood <david.sherwood@arm.com>
28437 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
28438 * internal-fn.def (VEC_SHL_INSERT): New internal function.
28439 * optabs.def (vec_shl_insert_optab): New optab.
28440 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
28441 (duplicate_and_interleave): Likewise.
28442 * tree-vect-loop.c: Include internal-fn.h.
28443 (neutral_op_for_slp_reduction): New function, split out from
28444 get_initial_defs_for_reduction.
28445 (get_initial_def_for_reduction): Handle option 2 for variable-length
28446 vectors by loading the neutral value into a vector and then shifting
28447 the initial value into element 0.
28448 (get_initial_defs_for_reduction): Replace the code argument with
28449 the neutral value calculated by neutral_op_for_slp_reduction.
28450 Use gimple_build_vector for constant-length vectors.
28451 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
28452 but the first group_size elements have a neutral value.
28453 Use duplicate_and_interleave otherwise.
28454 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
28455 Update call to get_initial_defs_for_reduction. Handle SLP
28456 reductions for variable-length vectors by creating one vector
28457 result for each scalar result, with the elements associated
28458 with other scalar results stubbed out with the neutral value.
28459 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
28460 Require IFN_VEC_SHL_INSERT for double reductions on
28461 variable-length vectors, or SLP reductions that have
28462 a neutral value. Require can_duplicate_and_interleave_p
28463 support for variable-length unchained SLP reductions if there
28464 is no neutral value, such as for MIN/MAX reductions. Also require
28465 the number of vector elements to be a multiple of the number of
28466 SLP statements when doing variable-length unchained SLP reductions.
28467 Update call to vect_create_epilog_for_reduction.
28468 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
28469 and remove initial values.
28470 (duplicate_and_interleave): Make public.
28471 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
28472 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
28474 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28475 Alan Hayward <alan.hayward@arm.com>
28476 David Sherwood <david.sherwood@arm.com>
28478 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
28479 (can_duplicate_and_interleave_p): New function.
28480 (vect_get_and_check_slp_defs): Take the vector of statements
28481 rather than just the current one. Remove excess parentheses.
28482 Restriction rejectinon of vect_constant_def and vect_external_def
28483 for variable-length vectors to boolean types, or types for which
28484 can_duplicate_and_interleave_p is false.
28485 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
28486 (duplicate_and_interleave): New function.
28487 (vect_get_constant_vectors): Use gimple_build_vector for
28488 constant-length vectors and suitable variable-length constant
28489 vectors. Use duplicate_and_interleave for other variable-length
28490 vectors. Don't defer the update when inserting new statements.
28492 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28493 Alan Hayward <alan.hayward@arm.com>
28494 David Sherwood <david.sherwood@arm.com>
28496 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
28497 min_profitable_iters doesn't go negative.
28499 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28500 Alan Hayward <alan.hayward@arm.com>
28501 David Sherwood <david.sherwood@arm.com>
28503 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
28504 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
28505 * optabs.def (vec_mask_load_lanes_optab): New optab.
28506 (vec_mask_store_lanes_optab): Likewise.
28507 * internal-fn.def (MASK_LOAD_LANES): New internal function.
28508 (MASK_STORE_LANES): Likewise.
28509 * internal-fn.c (mask_load_lanes_direct): New macro.
28510 (mask_store_lanes_direct): Likewise.
28511 (expand_mask_load_optab_fn): Handle masked operations.
28512 (expand_mask_load_lanes_optab_fn): New macro.
28513 (expand_mask_store_optab_fn): Handle masked operations.
28514 (expand_mask_store_lanes_optab_fn): New macro.
28515 (direct_mask_load_lanes_optab_supported_p): Likewise.
28516 (direct_mask_store_lanes_optab_supported_p): Likewise.
28517 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
28519 (vect_load_lanes_supported): Likewise.
28520 * tree-vect-data-refs.c (strip_conversion): New function.
28521 (can_group_stmts_p): Likewise.
28522 (vect_analyze_data_ref_accesses): Use it instead of checking
28523 for a pair of assignments.
28524 (vect_store_lanes_supported): Take a masked_p parameter.
28525 (vect_load_lanes_supported): Likewise.
28526 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
28527 vect_store_lanes_supported and vect_load_lanes_supported.
28528 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
28529 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
28530 parameter. Don't allow gaps for masked accesses.
28531 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
28532 and vect_load_lanes_supported.
28533 (get_load_store_type): Take a masked_p parameter and update
28534 call to get_group_load_store_type.
28535 (vectorizable_store): Update call to get_load_store_type.
28536 Handle IFN_MASK_STORE_LANES.
28537 (vectorizable_load): Update call to get_load_store_type.
28538 Handle IFN_MASK_LOAD_LANES.
28540 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28541 Alan Hayward <alan.hayward@arm.com>
28542 David Sherwood <david.sherwood@arm.com>
28544 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
28546 * config/aarch64/aarch64-protos.h
28547 (aarch64_sve_struct_memory_operand_p): Declare.
28548 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
28549 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
28550 (VPRED, vpred): Handle SVE structure modes.
28551 * config/aarch64/constraints.md (Utx): New constraint.
28552 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
28553 (aarch64_sve_struct_nonimmediate_operand): New predicates.
28554 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
28555 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
28556 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
28557 structure modes. Split into pieces after RA.
28558 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
28559 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
28561 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
28562 SVE structure modes.
28563 (aarch64_classify_address): Likewise.
28564 (sizetochar): Move earlier in file.
28565 (aarch64_print_operand): Handle SVE register lists.
28566 (aarch64_array_mode): New function.
28567 (aarch64_sve_struct_memory_operand_p): Likewise.
28568 (TARGET_ARRAY_MODE): Redefine.
28570 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28571 Alan Hayward <alan.hayward@arm.com>
28572 David Sherwood <david.sherwood@arm.com>
28574 * target.def (array_mode): New target hook.
28575 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
28576 * doc/tm.texi: Regenerate.
28577 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
28578 * hooks.c (hook_optmode_mode_uhwi_none): New function.
28579 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
28580 targetm.array_mode.
28581 * stor-layout.c (mode_for_array): Likewise. Support polynomial
28584 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28585 Alan Hayward <alan.hayward@arm.com>
28586 David Sherwood <david.sherwood@arm.com>
28588 * fold-const.c (fold_binary_loc): Check the argument types
28589 rather than the result type when testing for a vector operation.
28591 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28593 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
28594 * doc/tm.texi: Regenerate.
28596 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28597 Alan Hayward <alan.hayward@arm.com>
28598 David Sherwood <david.sherwood@arm.com>
28600 * doc/invoke.texi (-msve-vector-bits=): Document new option.
28601 (sve): Document new AArch64 extension.
28602 * doc/md.texi (w): Extend the description of the AArch64
28603 constraint to include SVE vectors.
28604 (Upl, Upa): Document new AArch64 predicate constraints.
28605 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
28607 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
28608 (msve-vector-bits=): New option.
28609 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
28610 SVE when these are disabled.
28611 (sve): New extension.
28612 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
28613 modes. Adjust their number of units based on aarch64_sve_vg.
28614 (MAX_BITSIZE_MODE_ANY_MODE): Define.
28615 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
28616 aarch64_addr_query_type.
28617 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
28618 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
28619 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
28620 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
28621 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
28622 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
28623 (aarch64_simd_imm_zero_p): Delete.
28624 (aarch64_check_zero_based_sve_index_immediate): Declare.
28625 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
28626 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
28627 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
28628 (aarch64_sve_float_mul_immediate_p): Likewise.
28629 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
28630 rather than an rtx.
28631 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
28632 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
28633 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
28634 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
28635 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
28636 (aarch64_regmode_natural_size): Likewise.
28637 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
28638 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
28640 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
28641 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
28642 for VG and the SVE predicate registers.
28643 (V_ALIASES): Add a "z"-prefixed alias.
28644 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
28645 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
28646 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
28647 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
28648 (REG_CLASS_NAMES): Add entries for them.
28649 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
28650 and the predicate registers.
28651 (aarch64_sve_vg): Declare.
28652 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
28653 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
28654 (REGMODE_NATURAL_SIZE): Define.
28655 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
28657 * config/aarch64/aarch64.c: Include cfgrtl.h.
28658 (simd_immediate_info): Add a constructor for series vectors,
28659 and an associated step field.
28660 (aarch64_sve_vg): New variable.
28661 (aarch64_dbx_register_number): Handle VG and the predicate registers.
28662 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
28663 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
28664 (VEC_ANY_DATA, VEC_STRUCT): New constants.
28665 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
28666 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
28667 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
28668 (aarch64_get_mask_mode): New functions.
28669 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
28670 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
28671 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
28672 predicate modes and predicate registers. Explicitly restrict
28673 GPRs to modes of 16 bytes or smaller. Only allow FP registers
28674 to store a vector mode if it is recognized by
28675 aarch64_classify_vector_mode.
28676 (aarch64_regmode_natural_size): New function.
28677 (aarch64_hard_regno_caller_save_mode): Return the original mode
28679 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
28680 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
28681 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
28682 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
28684 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
28685 does not overlap dest if the function is frame-related. Handle
28687 (aarch64_split_add_offset): New function.
28688 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
28689 them aarch64_add_offset.
28690 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
28691 and update call to aarch64_sub_sp.
28692 (aarch64_add_cfa_expression): New function.
28693 (aarch64_expand_prologue): Pass extra temporary registers to the
28694 functions above. Handle the case in which we need to emit new
28695 DW_CFA_expressions for registers that were originally saved
28696 relative to the stack pointer, but now have to be expressed
28697 relative to the frame pointer.
28698 (aarch64_output_mi_thunk): Pass extra temporary registers to the
28700 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
28701 IP0 and IP1 values for SVE frames.
28702 (aarch64_expand_vec_series): New function.
28703 (aarch64_expand_sve_widened_duplicate): Likewise.
28704 (aarch64_expand_sve_const_vector): Likewise.
28705 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
28706 Handle SVE constants. Use emit_move_insn to move a force_const_mem
28707 into the register, rather than emitting a SET directly.
28708 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
28709 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
28710 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
28711 (offset_9bit_signed_scaled_p): New functions.
28712 (aarch64_replicate_bitmask_imm): New function.
28713 (aarch64_bitmask_imm): Use it.
28714 (aarch64_cannot_force_const_mem): Reject expressions involving
28715 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
28716 (aarch64_classify_index): Handle SVE indices, by requiring
28717 a plain register index with a scale that matches the element size.
28718 (aarch64_classify_address): Handle SVE addresses. Assert that
28719 the mode of the address is VOIDmode or an integer mode.
28720 Update call to aarch64_classify_symbol.
28721 (aarch64_classify_symbolic_expression): Update call to
28722 aarch64_classify_symbol.
28723 (aarch64_const_vec_all_in_range_p): New function.
28724 (aarch64_print_vector_float_operand): Likewise.
28725 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
28726 "vN" for FP registers with SVE modes. Handle (const ...) vectors
28727 and the FP immediates 1.0 and 0.5.
28728 (aarch64_print_address_internal): Handle SVE addresses.
28729 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
28730 (aarch64_regno_regclass): Handle predicate registers.
28731 (aarch64_secondary_reload): Handle big-endian reloads of SVE
28733 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
28734 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
28735 (aarch64_convert_sve_vector_bits): New function.
28736 (aarch64_override_options): Use it to handle -msve-vector-bits=.
28737 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
28738 rather than an rtx.
28739 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
28740 Handle SVE vector and predicate modes. Accept VL-based constants
28741 that need only one temporary register, and VL offsets that require
28742 no temporary registers.
28743 (aarch64_conditional_register_usage): Mark the predicate registers
28744 as fixed if SVE isn't available.
28745 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
28746 Return true for SVE vector and predicate modes.
28747 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
28748 rather than an unsigned int. Handle SVE modes.
28749 (aarch64_preferred_simd_mode): Update call accordingly. Handle
28751 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
28753 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
28754 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
28755 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
28756 (aarch64_sve_float_mul_immediate_p): New functions.
28757 (aarch64_sve_valid_immediate): New function.
28758 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
28759 Explicitly reject structure modes. Check for INDEX constants.
28760 Handle PTRUE and PFALSE constants.
28761 (aarch64_check_zero_based_sve_index_immediate): New function.
28762 (aarch64_simd_imm_zero_p): Delete.
28763 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
28764 vector modes. Accept constants in the range of CNT[BHWD].
28765 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
28766 ask for an Advanced SIMD mode.
28767 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
28768 (aarch64_simd_vector_alignment): Handle SVE predicates.
28769 (aarch64_vectorize_preferred_vector_alignment): New function.
28770 (aarch64_simd_vector_alignment_reachable): Use it instead of
28772 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
28773 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
28775 (MAX_VECT_LEN): Delete.
28776 (expand_vec_perm_d): Add a vec_flags field.
28777 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
28778 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
28779 (aarch64_evpc_ext): Don't apply a big-endian lane correction
28781 (aarch64_evpc_rev): Rename to...
28782 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
28783 (aarch64_evpc_rev_global): New function.
28784 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
28785 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
28787 (aarch64_evpc_sve_tbl): New function.
28788 (aarch64_expand_vec_perm_const_1): Update after rename of
28789 aarch64_evpc_rev. Handle SVE permutes too, trying
28790 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
28791 than aarch64_evpc_tbl.
28792 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
28793 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
28794 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
28795 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
28796 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
28797 (aarch64_expand_sve_vcond): New functions.
28798 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
28799 of aarch64_vector_mode_p.
28800 (aarch64_dwarf_poly_indeterminate_value): New function.
28801 (aarch64_compute_pressure_classes): Likewise.
28802 (aarch64_can_change_mode_class): Likewise.
28803 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
28804 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
28805 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
28806 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
28807 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
28808 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
28809 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
28810 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
28812 (Dn, Dl, Dr): Accept const as well as const_vector.
28813 (Dz): Likewise. Compare against CONST0_RTX.
28814 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
28815 of "vector" where appropriate.
28816 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
28817 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
28818 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
28819 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
28820 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
28821 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
28822 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
28823 (v_int_equiv): Extend to SVE modes.
28824 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
28826 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
28827 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
28828 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
28829 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
28830 (SVE_COND_FP_CMP): New int iterators.
28831 (perm_hilo): Handle the new unpack unspecs.
28832 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
28834 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
28835 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
28836 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
28837 (aarch64_equality_operator, aarch64_constant_vector_operand)
28838 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
28839 (aarch64_sve_nonimmediate_operand): Likewise.
28840 (aarch64_sve_general_operand): Likewise.
28841 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
28842 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
28843 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
28844 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
28845 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
28846 (aarch64_sve_float_arith_immediate): Likewise.
28847 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
28848 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
28849 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
28850 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
28851 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
28852 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
28853 (aarch64_sve_float_arith_operand): Likewise.
28854 (aarch64_sve_float_arith_with_sub_operand): Likewise.
28855 (aarch64_sve_float_mul_operand): Likewise.
28856 (aarch64_sve_vec_perm_operand): Likewise.
28857 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
28858 (aarch64_mov_operand): Accept const_poly_int and const_vector.
28859 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
28860 as well as const_vector.
28861 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
28862 in file. Use CONST0_RTX and CONSTM1_RTX.
28863 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
28864 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
28865 Use aarch64_simd_imm_zero.
28866 * config/aarch64/aarch64-sve.md: New file.
28867 * config/aarch64/aarch64.md: Include it.
28868 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
28869 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
28870 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
28871 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
28872 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
28873 (sve): New attribute.
28874 (enabled): Disable instructions with the sve attribute unless
28876 (movqi, movhi): Pass CONST_POLY_INT operaneds through
28877 aarch64_expand_mov_immediate.
28878 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
28879 CNT[BHSD] immediates.
28880 (movti): Split CONST_POLY_INT moves into two halves.
28881 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
28882 Split additions that need a temporary here if the destination
28883 is the stack pointer.
28884 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
28885 (*add<mode>3_poly_1): New instruction.
28886 (set_clobber_cc): New expander.
28888 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28890 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
28891 parameter and use it instead of GET_MODE_SIZE (innermode). Use
28892 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
28893 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
28894 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
28895 Change innermode from fixed_mode_size to machine_mode.
28896 (simplify_subreg): Update call accordingly. Handle a constant-sized
28897 subreg of a variable-length CONST_VECTOR.
28899 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28900 Alan Hayward <alan.hayward@arm.com>
28901 David Sherwood <david.sherwood@arm.com>
28903 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
28904 (add_offset_to_base): New function, split out from...
28905 (create_mem_ref): ...here. When handling a scale other than 1,
28906 check first whether the address is valid without the offset.
28907 Add it into the base if so, leaving the index and scale as-is.
28909 2018-01-12 Jakub Jelinek <jakub@redhat.com>
28912 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
28913 fold_for_warn before checking if arg2 is INTEGER_CST.
28915 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
28917 * config/rs6000/predicates.md (load_multiple_operation): Delete.
28918 (store_multiple_operation): Delete.
28919 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
28920 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
28921 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
28922 guarded by TARGET_STRING.
28923 (rs6000_output_load_multiple): Delete.
28924 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28925 OPTION_MASK_STRING / TARGET_STRING handling.
28926 (print_operand) <'N', 'O'>: Add comment that these are unused now.
28927 (const rs6000_opt_masks) <"string">: Change mask to 0.
28928 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
28929 (MASK_STRING): Delete.
28930 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
28932 (load_multiple): Delete.
28939 (store_multiple): Delete.
28946 (movmemsi_8reg): Delete.
28947 (corresponding unnamed define_insn): Delete.
28948 (movmemsi_6reg): Delete.
28949 (corresponding unnamed define_insn): Delete.
28950 (movmemsi_4reg): Delete.
28951 (corresponding unnamed define_insn): Delete.
28952 (movmemsi_2reg): Delete.
28953 (corresponding unnamed define_insn): Delete.
28954 (movmemsi_1reg): Delete.
28955 (corresponding unnamed define_insn): Delete.
28956 * config/rs6000/rs6000.opt (mno-string): New.
28957 (mstring): Replace by deprecation warning stub.
28958 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
28960 2018-01-12 Jakub Jelinek <jakub@redhat.com>
28962 * regrename.c (regrename_do_replace): If replacing the same
28963 reg multiple times, try to reuse last created gen_raw_REG.
28966 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
28967 main to workaround a bug in GDB.
28969 2018-01-12 Tom de Vries <tom@codesourcery.com>
28972 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
28974 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
28976 PR rtl-optimization/80481
28977 * ira-color.c (get_cap_member): New function.
28978 (allocnos_conflict_by_live_ranges_p): Use it.
28979 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
28980 (setup_slot_coalesced_allocno_live_ranges): Ditto.
28982 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
28985 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
28986 (*saddl_se_1): Ditto.
28987 (*ssubsi_1): Ditto.
28988 (*ssubl_se_1): Ditto.
28990 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
28992 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
28993 rather than wi::to_widest for DR_INITs.
28994 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
28995 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
28996 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
28998 (vect_analyze_group_access_1): Note that here.
29000 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29002 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
29003 polynomial type sizes.
29005 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29007 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
29008 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
29009 (gimple_add_tmp_var): Likewise.
29011 2018-01-12 Martin Liska <mliska@suse.cz>
29013 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
29014 (gimple_alloc_sizes): Likewise.
29015 (dump_gimple_statistics): Use PRIu64 in printf format.
29016 * gimple.h: Change uint64_t to int.
29018 2018-01-12 Martin Liska <mliska@suse.cz>
29020 * tree-core.h: Use uint64_t instead of int.
29021 * tree.c (tree_node_counts): Likewise.
29022 (tree_node_sizes): Likewise.
29023 (dump_tree_statistics): Use PRIu64 in printf format.
29025 2018-01-12 Martin Liska <mliska@suse.cz>
29027 * Makefile.in: As qsort_chk is implemented in vec.c, add
29028 vec.o to linkage of gencfn-macros.
29029 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
29030 passing the info to record_node_allocation_statistics.
29031 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
29033 * ggc-common.c (struct ggc_usage): Add operator== and use
29034 it in operator< and compare function.
29035 * mem-stats.h (struct mem_usage): Likewise.
29036 * vec.c (struct vec_usage): Remove operator< and compare
29037 function. Can be simply inherited.
29039 2018-01-12 Martin Jambor <mjambor@suse.cz>
29042 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
29043 * tree-ssa-math-opts.c: Include domwalk.h.
29044 (convert_mult_to_fma_1): New function.
29045 (fma_transformation_info): New type.
29046 (fma_deferring_state): Likewise.
29047 (cancel_fma_deferring): New function.
29048 (result_of_phi): Likewise.
29049 (last_fma_candidate_feeds_initial_phi): Likewise.
29050 (convert_mult_to_fma): Added deferring logic, split actual
29051 transformation to convert_mult_to_fma_1.
29052 (math_opts_dom_walker): New type.
29053 (math_opts_dom_walker::after_dom_children): New method, body moved
29054 here from pass_optimize_widening_mul::execute, added deferring logic
29056 (pass_optimize_widening_mul::execute): Moved most of code to
29057 math_opts_dom_walker::after_dom_children.
29058 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
29059 * config/i386/i386.c (ix86_option_override_internal): Added
29060 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
29062 2018-01-12 Richard Biener <rguenther@suse.de>
29065 * dwarf2out.c (gen_variable_die): Do not reset old_die for
29066 inline instance vars.
29068 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
29071 * config/rx/rx.c (rx_is_restricted_memory_address):
29072 Handle SUBREG case.
29074 2018-01-12 Richard Biener <rguenther@suse.de>
29076 PR tree-optimization/80846
29077 * target.def (split_reduction): New target hook.
29078 * targhooks.c (default_split_reduction): New function.
29079 * targhooks.h (default_split_reduction): Declare.
29080 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
29081 target requests first reduce vectors by combining low and high
29083 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
29084 (get_vectype_for_scalar_type_and_size): Export.
29085 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
29086 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
29087 * doc/tm.texi: Regenerate.
29088 * config/i386/i386.c (ix86_split_reduction): Implement
29089 TARGET_VECTORIZE_SPLIT_REDUCTION.
29091 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29094 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
29095 in PIC mode except for TARGET_VXWORKS_RTP.
29096 * config/sparc/sparc.c: Include cfgrtl.h.
29097 (TARGET_INIT_PIC_REG): Define.
29098 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
29099 (sparc_pic_register_p): New predicate.
29100 (sparc_legitimate_address_p): Use it.
29101 (sparc_legitimize_pic_address): Likewise.
29102 (sparc_delegitimize_address): Likewise.
29103 (sparc_mode_dependent_address_p): Likewise.
29104 (gen_load_pcrel_sym): Remove 4th parameter.
29105 (load_got_register): Adjust call to above. Remove obsolete stuff.
29106 (sparc_expand_prologue): Do not call load_got_register here.
29107 (sparc_flat_expand_prologue): Likewise.
29108 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
29109 (sparc_use_pseudo_pic_reg): New function.
29110 (sparc_init_pic_reg): Likewise.
29111 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
29112 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
29114 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
29116 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29117 Add item for branch_cost.
29119 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29121 PR rtl-optimization/83565
29122 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
29123 not extend the result to a larger mode for rotate operations.
29124 (num_sign_bit_copies1): Likewise.
29126 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29129 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
29131 Use values-Xc.o for -pedantic.
29132 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
29134 2018-01-12 Martin Liska <mliska@suse.cz>
29137 * ipa-devirt.c (final_warning_record::grow_type_warnings):
29139 (possible_polymorphic_call_targets): Use it.
29140 (ipa_devirt): Likewise.
29142 2018-01-12 Martin Liska <mliska@suse.cz>
29144 * profile-count.h (enum profile_quality): Use 0 as invalid
29145 enum value of profile_quality.
29147 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
29149 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
29150 -mext-string options.
29152 2018-01-12 Richard Biener <rguenther@suse.de>
29154 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
29155 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
29156 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
29158 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
29160 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29162 * configure.ac (--with-long-double-format): Add support for the
29163 configuration option to change the default long double format on
29165 * config.gcc (powerpc*-linux*-*): Likewise.
29166 * configure: Regenerate.
29167 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
29168 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
29169 used without modification.
29171 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29173 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
29174 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
29175 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
29176 MISC_BUILTIN_SPEC_BARRIER.
29177 (rs6000_init_builtins): Likewise.
29178 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
29180 (speculation_barrier): New define_insn.
29181 * doc/extend.texi: Document __builtin_speculation_barrier.
29183 2018-01-11 Jakub Jelinek <jakub@redhat.com>
29186 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
29187 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
29188 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
29190 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
29191 integral modes instead of "ss" and "sd".
29192 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
29193 vectors with 32-bit and 64-bit elements.
29194 (vecdupssescalarmodesuffix): New mode attribute.
29195 (vec_dup<mode>): Use it.
29197 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
29200 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
29201 frame if argument is passed on stack.
29203 2018-01-11 Jakub Jelinek <jakub@redhat.com>
29206 * ree.c (combine_reaching_defs): Optimize also
29207 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
29208 reg2=any_extend(exp); reg1=reg2;, formatting fix.
29210 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
29212 PR middle-end/83189
29213 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
29215 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
29217 PR middle-end/83718
29218 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
29219 after they are computed.
29221 2018-01-11 Bin Cheng <bin.cheng@arm.com>
29223 PR tree-optimization/83695
29224 * gimple-loop-linterchange.cc
29225 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
29226 reset cached scev information after interchange.
29227 (pass_linterchange::execute): Remove call to scev_reset_htab.
29229 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29231 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
29232 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
29233 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
29234 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
29235 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
29236 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
29237 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
29238 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
29239 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
29240 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
29241 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
29242 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
29243 (V_lane_reg): Likewise.
29244 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
29246 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
29247 (vfmal_lane_low<mode>_intrinsic,
29248 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
29249 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
29250 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
29251 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
29252 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
29253 vfmsl_lane_high<mode>_intrinsic): New define_insns.
29255 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29257 * config/arm/arm-cpus.in (fp16fml): New feature.
29258 (ALL_SIMD): Add fp16fml.
29259 (armv8.2-a): Add fp16fml as an option.
29260 (armv8.3-a): Likewise.
29261 (armv8.4-a): Add fp16fml as part of fp16.
29262 * config/arm/arm.h (TARGET_FP16FML): Define.
29263 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
29265 * config/arm/arm-modes.def (V2HF): Define.
29266 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
29267 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
29268 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
29269 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
29270 vfmsl_low, vfmsl_high): New set of builtins.
29271 * config/arm/iterators.md (PLUSMINUS): New code iterator.
29272 (vfml_op): New code attribute.
29273 (VFMLHALVES): New int iterator.
29274 (VFML, VFMLSEL): New mode attributes.
29275 (V_reg): Define mapping for V2HF.
29276 (V_hi, V_lo): New mode attributes.
29277 (VF_constraint): Likewise.
29278 (vfml_half, vfml_half_selector): New int attributes.
29279 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
29281 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
29282 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
29284 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
29285 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
29286 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
29287 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
29289 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
29290 Document new effective target and option set.
29292 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29294 * config/arm/arm-cpus.in (armv8_4): New feature.
29295 (ARMv8_4a): New fgroup.
29296 (armv8.4-a): New arch.
29297 * config/arm/arm-tables.opt: Regenerate.
29298 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
29299 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
29300 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
29301 Add matching rules for -march=armv8.4-a and extensions.
29302 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
29304 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
29307 * config/rx/rx.md (BW): New mode attribute.
29308 (sync_lock_test_and_setsi): Add mode suffix to insn output.
29310 2018-01-11 Richard Biener <rguenther@suse.de>
29312 PR tree-optimization/83435
29313 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
29314 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
29315 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
29317 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29318 Alan Hayward <alan.hayward@arm.com>
29319 David Sherwood <david.sherwood@arm.com>
29321 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
29323 (aarch64_classify_address): Initialize it. Track polynomial offsets.
29324 (aarch64_print_address_internal): Use it to check for a zero offset.
29326 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29327 Alan Hayward <alan.hayward@arm.com>
29328 David Sherwood <david.sherwood@arm.com>
29330 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
29331 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
29332 Return a poly_int64 rather than a HOST_WIDE_INT.
29333 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
29334 rather than a HOST_WIDE_INT.
29335 * config/aarch64/aarch64.h (aarch64_frame): Protect with
29336 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
29337 hard_fp_offset, frame_size, initial_adjust, callee_offset and
29338 final_offset from HOST_WIDE_INT to poly_int64.
29339 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
29340 to_constant when getting the number of units in an Advanced SIMD
29342 (aarch64_builtin_vectorized_function): Check for a constant number
29344 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
29346 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
29347 attribute instead of GET_MODE_NUNITS.
29348 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
29349 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
29350 GET_MODE_SIZE for fixed-size registers.
29351 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
29352 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
29353 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
29354 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
29355 (aarch64_print_operand, aarch64_print_address_internal)
29356 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
29357 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
29358 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
29359 Handle polynomial GET_MODE_SIZE.
29360 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
29361 wider than SImode without modification.
29362 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
29363 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
29364 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
29365 passing and returning SVE modes.
29366 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
29367 rather than GEN_INT.
29368 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
29369 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
29370 (aarch64_allocate_and_probe_stack_space): Likewise.
29371 (aarch64_layout_frame): Cope with polynomial offsets.
29372 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
29373 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
29374 polynomial offsets.
29375 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
29376 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
29377 poly_int64 rather than a HOST_WIDE_INT.
29378 (aarch64_get_separate_components, aarch64_process_components)
29379 (aarch64_expand_prologue, aarch64_expand_epilogue)
29380 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
29381 (aarch64_anchor_offset): New function, split out from...
29382 (aarch64_legitimize_address): ...here.
29383 (aarch64_builtin_vectorization_cost): Handle polynomial
29384 TYPE_VECTOR_SUBPARTS.
29385 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
29387 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
29388 number of elements from the PARALLEL rather than the mode.
29389 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
29390 rather than GET_MODE_BITSIZE.
29391 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
29392 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
29393 (aarch64_expand_vec_perm_const_1): Handle polynomial
29394 d->perm.length () and d->perm elements.
29395 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
29396 Apply to_constant to d->perm elements.
29397 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
29398 polynomial CONST_VECTOR_NUNITS.
29399 (aarch64_move_pointer): Take amount as a poly_int64 rather
29401 (aarch64_progress_pointer): Avoid temporary variable.
29402 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
29403 the mode attribute instead of GET_MODE.
29405 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29406 Alan Hayward <alan.hayward@arm.com>
29407 David Sherwood <david.sherwood@arm.com>
29409 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
29410 x exists before using it.
29411 (aarch64_add_constant_internal): Rename to...
29412 (aarch64_add_offset_1): ...this. Replace regnum with separate
29413 src and dest rtxes. Handle the case in which they're different,
29414 including when the offset is zero. Replace scratchreg with an rtx.
29415 Use 2 additions if there is no spare register into which we can
29416 move a 16-bit constant.
29417 (aarch64_add_constant): Delete.
29418 (aarch64_add_offset): Replace reg with separate src and dest
29419 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
29420 Use aarch64_add_offset_1.
29421 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
29422 an rtx rather than an int. Take the delta as a poly_int64
29423 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
29424 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
29425 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
29426 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
29427 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
29428 and aarch64_add_sp.
29429 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
29430 aarch64_add_constant.
29432 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29434 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
29435 Use scalar_float_mode.
29437 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29439 * config/aarch64/aarch64-simd.md
29440 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
29441 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
29442 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
29443 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
29444 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
29445 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
29446 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
29447 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
29448 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
29449 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
29451 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29454 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
29455 targ_options->x_arm_arch_string is non NULL.
29457 2018-01-11 Tamar Christina <tamar.christina@arm.com>
29459 * config/aarch64/aarch64.h
29460 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
29462 2018-01-11 Sudakshina Das <sudi.das@arm.com>
29465 * expmed.c (emit_store_flag_force): Swap if const op0
29466 and change VOIDmode to mode of op0.
29468 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29470 PR rtl-optimization/83761
29471 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
29472 than bytes to mode_for_size.
29474 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29476 PR middle-end/83189
29477 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
29478 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
29481 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29483 PR middle-end/83575
29484 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
29485 when in layout mode.
29486 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
29487 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
29490 2018-01-10 Michael Collison <michael.collison@arm.com>
29492 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
29493 * config/aarch64/aarch64-option-extension.def: Add
29494 AARCH64_OPT_EXTENSION of 'fp16fml'.
29495 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29496 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
29497 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
29498 * config/aarch64/constraints.md (Ui7): New constraint.
29499 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
29500 (VFMLA_SEL_W): Ditto.
29503 (VFMLA16_LOW): New int iterator.
29504 (VFMLA16_HIGH): Ditto.
29505 (UNSPEC_FMLAL): New unspec.
29506 (UNSPEC_FMLSL): Ditto.
29507 (UNSPEC_FMLAL2): Ditto.
29508 (UNSPEC_FMLSL2): Ditto.
29509 (f16mac): New code attribute.
29510 * config/aarch64/aarch64-simd-builtins.def
29511 (aarch64_fmlal_lowv2sf): Ditto.
29512 (aarch64_fmlsl_lowv2sf): Ditto.
29513 (aarch64_fmlalq_lowv4sf): Ditto.
29514 (aarch64_fmlslq_lowv4sf): Ditto.
29515 (aarch64_fmlal_highv2sf): Ditto.
29516 (aarch64_fmlsl_highv2sf): Ditto.
29517 (aarch64_fmlalq_highv4sf): Ditto.
29518 (aarch64_fmlslq_highv4sf): Ditto.
29519 (aarch64_fmlal_lane_lowv2sf): Ditto.
29520 (aarch64_fmlsl_lane_lowv2sf): Ditto.
29521 (aarch64_fmlal_laneq_lowv2sf): Ditto.
29522 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
29523 (aarch64_fmlalq_lane_lowv4sf): Ditto.
29524 (aarch64_fmlsl_lane_lowv4sf): Ditto.
29525 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
29526 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
29527 (aarch64_fmlal_lane_highv2sf): Ditto.
29528 (aarch64_fmlsl_lane_highv2sf): Ditto.
29529 (aarch64_fmlal_laneq_highv2sf): Ditto.
29530 (aarch64_fmlsl_laneq_highv2sf): Ditto.
29531 (aarch64_fmlalq_lane_highv4sf): Ditto.
29532 (aarch64_fmlsl_lane_highv4sf): Ditto.
29533 (aarch64_fmlalq_laneq_highv4sf): Ditto.
29534 (aarch64_fmlsl_laneq_highv4sf): Ditto.
29535 * config/aarch64/aarch64-simd.md:
29536 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
29537 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
29538 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
29539 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
29540 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
29541 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
29542 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
29543 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
29544 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
29545 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
29546 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
29547 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
29548 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
29549 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
29550 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
29551 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
29552 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
29553 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
29554 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
29555 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
29556 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
29557 (vfmlsl_low_u32): Ditto.
29558 (vfmlalq_low_u32): Ditto.
29559 (vfmlslq_low_u32): Ditto.
29560 (vfmlal_high_u32): Ditto.
29561 (vfmlsl_high_u32): Ditto.
29562 (vfmlalq_high_u32): Ditto.
29563 (vfmlslq_high_u32): Ditto.
29564 (vfmlal_lane_low_u32): Ditto.
29565 (vfmlsl_lane_low_u32): Ditto.
29566 (vfmlal_laneq_low_u32): Ditto.
29567 (vfmlsl_laneq_low_u32): Ditto.
29568 (vfmlalq_lane_low_u32): Ditto.
29569 (vfmlslq_lane_low_u32): Ditto.
29570 (vfmlalq_laneq_low_u32): Ditto.
29571 (vfmlslq_laneq_low_u32): Ditto.
29572 (vfmlal_lane_high_u32): Ditto.
29573 (vfmlsl_lane_high_u32): Ditto.
29574 (vfmlal_laneq_high_u32): Ditto.
29575 (vfmlsl_laneq_high_u32): Ditto.
29576 (vfmlalq_lane_high_u32): Ditto.
29577 (vfmlslq_lane_high_u32): Ditto.
29578 (vfmlalq_laneq_high_u32): Ditto.
29579 (vfmlslq_laneq_high_u32): Ditto.
29580 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
29581 (AARCH64_FL_FOR_ARCH8_4): New.
29582 (AARCH64_ISA_F16FML): New ISA flag.
29583 (TARGET_F16FML): New feature flag for fp16fml.
29584 (doc/invoke.texi): Document new fp16fml option.
29586 2018-01-10 Michael Collison <michael.collison@arm.com>
29588 * config/aarch64/aarch64-builtins.c:
29589 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
29590 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29591 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
29592 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
29593 (AARCH64_ISA_SHA3): New ISA flag.
29594 (TARGET_SHA3): New feature flag for sha3.
29595 * config/aarch64/iterators.md (sha512_op): New int attribute.
29596 (CRYPTO_SHA512): New int iterator.
29597 (UNSPEC_SHA512H): New unspec.
29598 (UNSPEC_SHA512H2): Ditto.
29599 (UNSPEC_SHA512SU0): Ditto.
29600 (UNSPEC_SHA512SU1): Ditto.
29601 * config/aarch64/aarch64-simd-builtins.def
29602 (aarch64_crypto_sha512hqv2di): New builtin.
29603 (aarch64_crypto_sha512h2qv2di): Ditto.
29604 (aarch64_crypto_sha512su0qv2di): Ditto.
29605 (aarch64_crypto_sha512su1qv2di): Ditto.
29606 (aarch64_eor3qv8hi): Ditto.
29607 (aarch64_rax1qv2di): Ditto.
29608 (aarch64_xarqv2di): Ditto.
29609 (aarch64_bcaxqv8hi): Ditto.
29610 * config/aarch64/aarch64-simd.md:
29611 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
29612 (aarch64_crypto_sha512su0qv2di): Ditto.
29613 (aarch64_crypto_sha512su1qv2di): Ditto.
29614 (aarch64_eor3qv8hi): Ditto.
29615 (aarch64_rax1qv2di): Ditto.
29616 (aarch64_xarqv2di): Ditto.
29617 (aarch64_bcaxqv8hi): Ditto.
29618 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
29619 (vsha512h2q_u64): Ditto.
29620 (vsha512su0q_u64): Ditto.
29621 (vsha512su1q_u64): Ditto.
29622 (veor3q_u16): Ditto.
29623 (vrax1q_u64): Ditto.
29624 (vxarq_u64): Ditto.
29625 (vbcaxq_u16): Ditto.
29626 * config/arm/types.md (crypto_sha512): New type attribute.
29627 (crypto_sha3): Ditto.
29628 (doc/invoke.texi): Document new sha3 option.
29630 2018-01-10 Michael Collison <michael.collison@arm.com>
29632 * config/aarch64/aarch64-builtins.c:
29633 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
29634 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29635 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
29636 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
29637 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
29638 (AARCH64_ISA_SM4): New ISA flag.
29639 (TARGET_SM4): New feature flag for sm4.
29640 * config/aarch64/aarch64-simd-builtins.def
29641 (aarch64_sm3ss1qv4si): Ditto.
29642 (aarch64_sm3tt1aq4si): Ditto.
29643 (aarch64_sm3tt1bq4si): Ditto.
29644 (aarch64_sm3tt2aq4si): Ditto.
29645 (aarch64_sm3tt2bq4si): Ditto.
29646 (aarch64_sm3partw1qv4si): Ditto.
29647 (aarch64_sm3partw2qv4si): Ditto.
29648 (aarch64_sm4eqv4si): Ditto.
29649 (aarch64_sm4ekeyqv4si): Ditto.
29650 * config/aarch64/aarch64-simd.md:
29651 (aarch64_sm3ss1qv4si): Ditto.
29652 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
29653 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
29654 (aarch64_sm4eqv4si): Ditto.
29655 (aarch64_sm4ekeyqv4si): Ditto.
29656 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
29657 (sm3part_op): Ditto.
29658 (CRYPTO_SM3TT): Ditto.
29659 (CRYPTO_SM3PART): Ditto.
29660 (UNSPEC_SM3SS1): New unspec.
29661 (UNSPEC_SM3TT1A): Ditto.
29662 (UNSPEC_SM3TT1B): Ditto.
29663 (UNSPEC_SM3TT2A): Ditto.
29664 (UNSPEC_SM3TT2B): Ditto.
29665 (UNSPEC_SM3PARTW1): Ditto.
29666 (UNSPEC_SM3PARTW2): Ditto.
29667 (UNSPEC_SM4E): Ditto.
29668 (UNSPEC_SM4EKEY): Ditto.
29669 * config/aarch64/constraints.md (Ui2): New constraint.
29670 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
29671 * config/arm/types.md (crypto_sm3): New type attribute.
29672 (crypto_sm4): Ditto.
29673 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
29674 (vsm3tt1aq_u32): Ditto.
29675 (vsm3tt1bq_u32): Ditto.
29676 (vsm3tt2aq_u32): Ditto.
29677 (vsm3tt2bq_u32): Ditto.
29678 (vsm3partw1q_u32): Ditto.
29679 (vsm3partw2q_u32): Ditto.
29680 (vsm4eq_u32): Ditto.
29681 (vsm4ekeyq_u32): Ditto.
29682 (doc/invoke.texi): Document new sm4 option.
29684 2018-01-10 Michael Collison <michael.collison@arm.com>
29686 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
29687 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
29688 (AARCH64_FL_FOR_ARCH8_4): New.
29689 (AARCH64_FL_V8_4): New flag.
29690 (doc/invoke.texi): Document new armv8.4-a option.
29692 2018-01-10 Michael Collison <michael.collison@arm.com>
29694 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29695 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
29696 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
29697 * config/aarch64/aarch64-option-extension.def: Add
29698 AARCH64_OPT_EXTENSION of 'sha2'.
29699 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
29700 (crypto): Disable sha2 and aes if crypto disabled.
29701 (crypto): Enable aes and sha2 if enabled.
29702 (simd): Disable sha2 and aes if simd disabled.
29703 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
29705 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
29706 (TARGET_SHA2): New feature flag for sha2.
29707 (TARGET_AES): New feature flag for aes.
29708 * config/aarch64/aarch64-simd.md:
29709 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
29710 conditional on TARGET_AES.
29711 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
29712 (aarch64_crypto_sha1hsi): Make pattern conditional
29714 (aarch64_crypto_sha1hv4si): Ditto.
29715 (aarch64_be_crypto_sha1hv4si): Ditto.
29716 (aarch64_crypto_sha1su1v4si): Ditto.
29717 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
29718 (aarch64_crypto_sha1su0v4si): Ditto.
29719 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
29720 (aarch64_crypto_sha256su0v4si): Ditto.
29721 (aarch64_crypto_sha256su1v4si): Ditto.
29722 (doc/invoke.texi): Document new aes and sha2 options.
29724 2018-01-10 Martin Sebor <msebor@redhat.com>
29726 PR tree-optimization/83781
29727 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
29730 2018-01-11 Martin Sebor <msebor@gmail.com>
29731 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29733 PR tree-optimization/83501
29734 PR tree-optimization/81703
29736 * tree-ssa-strlen.c (get_string_cst): Rename...
29737 (get_string_len): ...to this. Handle global constants.
29738 (handle_char_store): Adjust.
29740 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
29741 Jim Wilson <jimw@sifive.com>
29743 * config/riscv/riscv-protos.h (riscv_output_return): New.
29744 * config/riscv/riscv.c (struct machine_function): New naked_p field.
29745 (riscv_attribute_table, riscv_output_return),
29746 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
29747 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
29748 (riscv_compute_frame_info): Only compute frame->mask if not a naked
29750 (riscv_expand_prologue): Add early return for naked function.
29751 (riscv_expand_epilogue): Likewise.
29752 (riscv_function_ok_for_sibcall): Return false for naked function.
29753 (riscv_set_current_function): New.
29754 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
29755 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
29756 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
29757 * doc/extend.texi (RISC-V Function Attributes): New.
29759 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
29761 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
29762 check for 128-bit long double before checking TCmode.
29763 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
29764 128-bit long doubles before checking TFmode or TCmode.
29765 (FLOAT128_IBM_P): Likewise.
29767 2018-01-10 Martin Sebor <msebor@redhat.com>
29769 PR tree-optimization/83671
29770 * builtins.c (c_strlen): Unconditionally return zero for the empty
29772 Use -Warray-bounds for warnings.
29773 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
29774 for non-constant array indices with COMPONENT_REF, arrays of
29775 arrays, and pointers to arrays.
29776 (gimple_fold_builtin_strlen): Determine and set length range for
29777 non-constant character arrays.
29779 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
29781 PR middle-end/81897
29782 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
29785 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
29787 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
29789 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
29792 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
29793 VECTOR_MEM_ALTIVEC_OR_VSX_P.
29794 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
29795 indexed_or_indirect_operand predicate.
29796 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
29797 (*vsx_le_perm_load_v8hi): Likewise.
29798 (*vsx_le_perm_load_v16qi): Likewise.
29799 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
29800 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
29801 (*vsx_le_perm_store_v8hi): Likewise.
29802 (*vsx_le_perm_store_v16qi): Likewise.
29803 (eight unnamed splitters): Likewise.
29805 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
29807 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
29808 * config/rs6000/emmintrin.h: Likewise.
29809 * config/rs6000/mmintrin.h: Likewise.
29810 * config/rs6000/xmmintrin.h: Likewise.
29812 2018-01-10 David Malcolm <dmalcolm@redhat.com>
29815 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
29817 * tree.c (tree_nop_conversion): Return true for location wrapper
29819 (maybe_wrap_with_location): New function.
29820 (selftest::check_strip_nops): New function.
29821 (selftest::test_location_wrappers): New function.
29822 (selftest::tree_c_tests): Call it.
29823 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
29824 (maybe_wrap_with_location): New decl.
29825 (EXPR_LOCATION_WRAPPER_P): New macro.
29826 (location_wrapper_p): New inline function.
29827 (tree_strip_any_location_wrapper): New inline function.
29829 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
29832 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
29833 stack_realign_offset for the largest alignment of stack slot
29835 (ix86_find_max_used_stack_alignment): New function.
29836 (ix86_finalize_stack_frame_flags): Use it. Set
29837 max_used_stack_alignment if we don't realign stack.
29838 * config/i386/i386.h (machine_function): Add
29839 max_used_stack_alignment.
29841 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
29843 * config/arm/arm.opt (-mbranch-cost): New option.
29844 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
29847 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
29850 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
29851 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
29853 2018-01-10 Richard Biener <rguenther@suse.de>
29856 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
29857 early out so it also covers the case where we have a non-NULL
29860 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
29862 PR tree-optimization/83753
29863 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
29864 for non-strided grouped accesses if the number of elements is 1.
29866 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29869 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
29870 * i386.h (TARGET_USE_GATHER): Define.
29871 * x86-tune.def (X86_TUNE_USE_GATHER): New.
29873 2018-01-10 Martin Liska <mliska@suse.cz>
29876 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
29877 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
29879 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
29880 CLEANUP_NO_PARTITIONING is not set.
29882 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
29884 * doc/rtl.texi: Remove documentation of (const ...) wrappers
29885 for vectors, as a partial revert of r254296.
29886 * rtl.h (const_vec_p): Delete.
29887 (const_vec_duplicate_p): Don't test for vector CONSTs.
29888 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
29889 * expmed.c (make_tree): Likewise.
29892 * common.md (E, F): Use CONSTANT_P instead of checking for
29894 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
29895 checking for CONST_VECTOR.
29897 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
29899 PR middle-end/83575
29900 * predict.c (force_edge_cold): Handle in more sane way edges
29901 with no prediction.
29903 2018-01-09 Carl Love <cel@us.ibm.com>
29905 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
29907 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
29908 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
29909 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
29910 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
29911 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
29912 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
29913 * config/rs6000/rs6000-protos.h: Add extern defition for
29914 rs6000_generate_float2_double_code.
29915 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
29917 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
29918 (float2_v2df): Add define_expand.
29920 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
29923 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
29924 op_mode in the force_to_mode call.
29926 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
29928 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
29929 instead of checking each element individually.
29930 (aarch64_evpc_uzp): Likewise.
29931 (aarch64_evpc_zip): Likewise.
29932 (aarch64_evpc_ext): Likewise.
29933 (aarch64_evpc_rev): Likewise.
29934 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
29935 instead of checking each element individually. Return true without
29937 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
29938 whether all selected elements come from the same input, instead of
29939 checking each element individually. Remove calls to gen_rtx_REG,
29940 start_sequence and end_sequence and instead assert that no rtl is
29943 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
29945 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
29946 order of HIGH and CONST checks.
29948 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
29950 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
29951 if the destination isn't an SSA_NAME.
29953 2018-01-09 Richard Biener <rguenther@suse.de>
29955 PR tree-optimization/83668
29956 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
29958 (canonicalize_loop_form): ... here, renamed from ...
29959 (canonicalize_loop_closed_ssa_form): ... this and amended to
29960 swap successor edges for loop exit blocks to make us use
29961 the RPO order we need for initial schedule generation.
29963 2018-01-09 Joseph Myers <joseph@codesourcery.com>
29965 PR tree-optimization/64811
29966 * match.pd: When optimizing comparisons with Inf, avoid
29967 introducing or losing exceptions from comparisons with NaN.
29969 2018-01-09 Martin Liska <mliska@suse.cz>
29972 * asan.c (shadow_mem_size): Add gcc_assert.
29974 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
29976 Don't save registers in main().
29979 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
29980 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
29981 * config/avr/avr.c (avr_set_current_function): Don't error if
29982 naked, OS_task or OS_main are specified at the same time.
29983 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
29985 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
29987 * common/config/avr/avr-common.c (avr_option_optimization_table):
29988 Switch on -mmain-is-OS_task for optimizing compilations.
29990 2018-01-09 Richard Biener <rguenther@suse.de>
29992 PR tree-optimization/83572
29993 * graphite.c: Include cfganal.h.
29994 (graphite_transform_loops): Connect infinite loops to exit
29995 and remove fake edges at the end.
29997 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
29999 * ipa-inline.c (edge_badness): Revert accidental checkin.
30001 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30004 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
30005 symbols; not inline clones.
30007 2018-01-09 Jakub Jelinek <jakub@redhat.com>
30010 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
30011 hard registers. Formatting fixes.
30013 PR preprocessor/83722
30014 * gcc.c (try_generate_repro): Pass
30015 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
30016 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
30019 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
30020 Kito Cheng <kito.cheng@gmail.com>
30022 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
30023 (riscv_leaf_function_p): Delete.
30024 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
30026 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30028 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
30030 (do_ifelse): New function.
30031 (do_isel): New function.
30032 (do_sub3): New function.
30033 (do_add3): New function.
30034 (do_load_mask_compare): New function.
30035 (do_overlap_load_compare): New function.
30036 (expand_compare_loop): New function.
30037 (expand_block_compare): Call expand_compare_loop() when appropriate.
30038 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
30039 option description.
30040 (-mblock-compare-inline-loop-limit): New option.
30042 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30045 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
30046 Reverse order of second and third operands in first alternative.
30047 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
30048 of first and second elements in UNSPEC_VPERMR vector.
30049 (altivec_expand_vec_perm_le): Likewise.
30051 2018-01-08 Jeff Law <law@redhat.com>
30053 PR rtl-optimizatin/81308
30054 * tree-switch-conversion.c (cfg_altered): New file scoped static.
30055 (process_switch): If group_case_labels makes a change, then set
30057 (pass_convert_switch::execute): If a switch is converted, then
30058 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
30060 PR rtl-optimization/81308
30061 * recog.c (split_all_insns): Conditionally cleanup the CFG after
30064 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
30066 PR target/83663 - Revert r255946
30067 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
30068 generation for cases where splatting a value is not useful.
30069 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
30070 across a vec_duplicate and a paradoxical subreg forming a vector
30071 mode to a vec_concat.
30073 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30075 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
30076 -march=armv8.3-a variants.
30077 * config/arm/t-multilib: Likewise.
30078 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
30080 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30082 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
30084 (cceq_ior_compare_complement): Give it a name so I can use it, and
30085 change boolean_or_operator predicate to boolean_operator so it can
30086 be used to generate a crand.
30087 (eqne): New code iterator.
30088 (bd/bd_neg): New code_attrs.
30089 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
30090 a single define_insn.
30091 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
30092 decrement (bdnzt/bdnzf/bdzt/bdzf).
30093 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
30094 with the new names of the branch decrement patterns, and added the
30095 names of the branch decrement conditional patterns.
30097 2018-01-08 Richard Biener <rguenther@suse.de>
30099 PR tree-optimization/83563
30100 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
30103 2018-01-08 Richard Biener <rguenther@suse.de>
30105 PR middle-end/83713
30106 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
30108 2018-01-08 Richard Biener <rguenther@suse.de>
30110 PR tree-optimization/83685
30111 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
30112 references to abnormals.
30114 2018-01-08 Richard Biener <rguenther@suse.de>
30117 * dwarf2out.c (output_indirect_strings): Handle empty
30118 skeleton_debug_str_hash.
30119 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
30121 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
30123 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
30124 (emit_store_direct): Likewise.
30125 (arc_trampoline_adjust_address): Likewise.
30126 (arc_asm_trampoline_template): New function.
30127 (arc_initialize_trampoline): Use asm_trampoline_template.
30128 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
30129 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
30130 * config/arc/arc.md (flush_icache): Delete pattern.
30132 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
30134 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
30135 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
30138 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
30141 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
30142 by not USED_FOR_TARGET.
30143 (make_pass_resolve_sw_modes): Likewise.
30145 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
30147 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
30150 2018-01-08 Richard Biener <rguenther@suse.de>
30152 PR middle-end/83580
30153 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
30155 2018-01-08 Richard Biener <rguenther@suse.de>
30157 PR middle-end/83517
30158 * match.pd ((t * 2) / 2) -> t): Add missing :c.
30160 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
30162 PR middle-end/81897
30163 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
30164 basic blocks with a small number of successors.
30165 (convert_control_dep_chain_into_preds): Improve handling of
30167 (dump_predicates): Split apart into...
30168 (dump_pred_chain): ...here...
30169 (dump_pred_info): ...and here.
30170 (can_one_predicate_be_invalidated_p): Add debugging printfs.
30171 (can_chain_union_be_invalidated_p): Improve check for invalidation
30173 (uninit_uses_cannot_happen): Avoid unnecessary if
30174 convert_control_dep_chain_into_preds yielded nothing.
30176 2018-01-06 Martin Sebor <msebor@redhat.com>
30178 PR tree-optimization/83640
30179 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
30180 subtracting negative offset from size.
30181 (builtin_access::overlap): Adjust offset bounds of the access to fall
30182 within the size of the object if possible.
30184 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
30186 PR rtl-optimization/83699
30187 * expmed.c (extract_bit_field_1): Restrict the vector usage of
30188 extract_bit_field_as_subreg to cases in which the extracted
30189 value is also a vector.
30191 * lra-constraints.c (process_alt_operands): Test for the equivalence
30192 substitutions when detecting a possible reload cycle.
30194 2018-01-06 Jakub Jelinek <jakub@redhat.com>
30197 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
30198 by default if flag_selective_schedling{,2}. Formatting fixes.
30200 PR rtl-optimization/83682
30201 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
30202 if it has non-VECTOR_MODE element mode.
30203 (vec_duplicate_p): Likewise.
30205 PR middle-end/83694
30206 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
30207 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
30209 2018-01-05 Jakub Jelinek <jakub@redhat.com>
30212 * config/i386/i386-builtin.def
30213 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
30214 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
30215 Require also OPTION_MASK_ISA_AVX512F in addition to
30216 OPTION_MASK_ISA_GFNI.
30217 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
30218 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
30219 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
30220 to OPTION_MASK_ISA_GFNI.
30221 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
30222 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
30223 OPTION_MASK_ISA_AVX512BW.
30224 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
30225 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
30226 addition to OPTION_MASK_ISA_GFNI.
30227 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
30228 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
30229 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
30230 to OPTION_MASK_ISA_GFNI.
30231 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
30232 a requirement for all ISAs rather than any of them with a few
30234 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
30236 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
30237 bitmasks to be enabled with 3 exceptions, instead of requiring any
30238 enabled ISA with lots of exceptions.
30239 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
30240 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
30241 Change avx512bw in isa attribute to avx512f.
30242 * config/i386/sgxintrin.h: Add license boilerplate.
30243 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
30244 to __AVX512F__ and __AVX512VL to __AVX512VL__.
30245 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
30246 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
30248 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
30249 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
30250 temporarily sse2 rather than sse if not enabled already.
30253 * config/i386/sse.md (VI248_VLBW): Rename to ...
30254 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
30255 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
30256 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
30257 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
30258 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
30259 mode iterator instead of VI248_VLBW.
30261 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
30263 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
30264 (record_modified): Skip clobbers; add debug output.
30265 (param_change_prob): Use sreal frequencies.
30267 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
30269 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
30270 punt for user-aligned variables.
30272 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
30274 * tree-chrec.c (chrec_contains_symbols): Return true for
30277 2018-01-05 Sudakshina Das <sudi.das@arm.com>
30280 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
30281 of (x|y) == x for BICS pattern.
30283 2018-01-05 Jakub Jelinek <jakub@redhat.com>
30285 PR tree-optimization/83605
30286 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
30287 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
30290 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
30292 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
30293 * config/epiphany/rtems.h: New file.
30295 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30296 Uros Bizjak <ubizjak@gmail.com>
30299 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
30300 QIreg_operand instead of register_operand predicate.
30301 * config/i386/i386.c (ix86_rop_should_change_byte_p,
30302 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
30303 comments instead of -fmitigate[-_]rop.
30305 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30308 * cgraphunit.c (symbol_table::compile): Switch to text_section
30309 before calling assembly_start debug hook.
30310 * run-rtl-passes.c (run_rtl_passes): Likewise.
30313 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30315 * tree-vrp.c (extract_range_from_binary_expr_1): Check
30316 range_int_cst_p rather than !symbolic_range_p before calling
30317 extract_range_from_multiplicative_op_1.
30319 2018-01-04 Jeff Law <law@redhat.com>
30321 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
30322 redundant test in assertion.
30324 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30326 * doc/rtl.texi: Document machine_mode wrapper classes.
30328 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30330 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
30331 using tree_to_uhwi.
30333 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30335 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
30336 the VEC_PERM_EXPR fold to fail.
30338 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30341 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
30342 to switched_sections.
30344 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30347 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
30348 test for d.testing.
30350 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
30353 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
30354 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
30356 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30359 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
30360 is BLKmode and bitpos not zero or mode change is needed.
30362 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30365 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
30368 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
30371 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
30372 instead of MULT rtx. Update all corresponding splitters.
30373 (*saddl_se): Ditto.
30374 (*ssub<modesuffix>): Ditto.
30375 (*ssubl_se): Ditto.
30376 (*cmp_sadd_di): Update split patterns.
30377 (*cmp_sadd_si): Ditto.
30378 (*cmp_sadd_sidi): Ditto.
30379 (*cmp_ssub_di): Ditto.
30380 (*cmp_ssub_si): Ditto.
30381 (*cmp_ssub_sidi): Ditto.
30382 * config/alpha/predicates.md (const23_operand): New predicate.
30383 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
30384 Look for ASHIFT, not MULT inner operand.
30385 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
30387 2018-01-04 Martin Liska <mliska@suse.cz>
30389 PR gcov-profile/83669
30390 * gcov.c (output_intermediate_file): Add version to intermediate
30392 * doc/gcov.texi: Document new field 'version' in intermediate
30393 file format. Fix location of '-k' option of gcov command.
30395 2018-01-04 Martin Liska <mliska@suse.cz>
30398 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
30400 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30402 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
30404 2018-01-03 Martin Sebor <msebor@redhat.com>
30406 PR tree-optimization/83655
30407 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
30408 checking calls with invalid arguments.
30410 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30412 * tree-vect-stmts.c (vect_get_store_rhs): New function.
30413 (vectorizable_mask_load_store): Delete.
30414 (vectorizable_call): Return false for masked loads and stores.
30415 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
30416 instead of gimple_assign_rhs1.
30417 (vectorizable_load): Handle IFN_MASK_LOAD.
30418 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
30420 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30422 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
30424 (vectorizable_mask_load_store): ...here.
30425 (vectorizable_load): ...and here.
30427 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30429 * tree-vect-stmts.c (vect_build_all_ones_mask)
30430 (vect_build_zero_merge_argument): New functions, split out from...
30431 (vectorizable_load): ...here.
30433 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30435 * tree-vect-stmts.c (vect_check_store_rhs): New function,
30437 (vectorizable_mask_load_store): ...here.
30438 (vectorizable_store): ...and here.
30440 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30442 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
30444 (vectorizable_mask_load_store): ...here.
30446 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30448 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
30449 (vect_model_store_cost): Take a vec_load_store_type instead of a
30451 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
30452 (vect_model_store_cost): Take a vec_load_store_type instead of a
30454 (vectorizable_mask_load_store): Update accordingly.
30455 (vectorizable_store): Likewise.
30456 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
30458 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30460 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
30461 IFN_MASK_LOAD calls here rather than...
30462 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
30464 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30465 Alan Hayward <alan.hayward@arm.com>
30466 David Sherwood <david.sherwood@arm.com>
30468 * expmed.c (extract_bit_field_1): For vector extracts,
30469 fall back to extract_bit_field_as_subreg if vec_extract
30472 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30473 Alan Hayward <alan.hayward@arm.com>
30474 David Sherwood <david.sherwood@arm.com>
30476 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
30477 they are variable or constant sized.
30478 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
30479 slots for constant-sized data.
30481 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30482 Alan Hayward <alan.hayward@arm.com>
30483 David Sherwood <david.sherwood@arm.com>
30485 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
30486 handling COND_EXPRs with boolean comparisons, try to find a better
30487 basis for the mask type than the boolean itself.
30489 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30491 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
30492 is calculated and how it can be overridden.
30493 * genmodes.c (max_bitsize_mode_any_mode): New variable.
30494 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
30496 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
30499 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30500 Alan Hayward <alan.hayward@arm.com>
30501 David Sherwood <david.sherwood@arm.com>
30503 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
30504 Remove the mode argument.
30505 (aarch64_simd_valid_immediate): Remove the mode and inverse
30507 * config/aarch64/iterators.md (bitsize): New iterator.
30508 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
30509 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
30510 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
30511 aarch64_simd_valid_immediate.
30512 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
30513 (aarch64_reg_or_bic_imm): Likewise.
30514 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
30515 with an insn_type enum and msl with a modifier_type enum.
30516 Replace element_width with a scalar_mode. Change the shift
30517 to unsigned int. Add constructors for scalar_float_mode and
30518 scalar_int_mode elements.
30519 (aarch64_vect_float_const_representable_p): Delete.
30520 (aarch64_can_const_movi_rtx_p)
30521 (aarch64_simd_scalar_immediate_valid_for_move)
30522 (aarch64_simd_make_constant): Update call to
30523 aarch64_simd_valid_immediate.
30524 (aarch64_advsimd_valid_immediate_hs): New function.
30525 (aarch64_advsimd_valid_immediate): Likewise.
30526 (aarch64_simd_valid_immediate): Remove mode and inverse
30527 arguments. Rewrite to use the above. Use const_vec_duplicate_p
30528 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
30529 and aarch64_float_const_representable_p on the result.
30530 (aarch64_output_simd_mov_immediate): Remove mode argument.
30531 Update call to aarch64_simd_valid_immediate and use of
30532 simd_immediate_info.
30533 (aarch64_output_scalar_simd_mov_immediate): Update call
30536 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30537 Alan Hayward <alan.hayward@arm.com>
30538 David Sherwood <david.sherwood@arm.com>
30540 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
30541 (mode_nunits): Likewise CONST_MODE_NUNITS.
30542 * machmode.def (ADJUST_NUNITS): Document.
30543 * genmodes.c (mode_data::need_nunits_adj): New field.
30544 (blank_mode): Update accordingly.
30545 (adj_nunits): New variable.
30546 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
30548 (emit_mode_size_inline): Set need_bytesize_adj for all modes
30549 listed in adj_nunits.
30550 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
30551 listed in adj_nunits. Don't emit case statements for such modes.
30552 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
30553 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
30554 nothing if adj_nunits is nonnull.
30555 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
30556 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
30557 (emit_mode_fbit): Update use of print_maybe_const_decl.
30558 (emit_move_size): Likewise. Treat the array as non-const
30560 (emit_mode_adjustments): Handle adj_nunits.
30562 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30564 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
30565 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
30566 (VECTOR_MODES): Use it.
30567 (make_vector_modes): Take the prefix as an argument.
30569 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30570 Alan Hayward <alan.hayward@arm.com>
30571 David Sherwood <david.sherwood@arm.com>
30573 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
30574 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
30575 for MODE_VECTOR_BOOL.
30576 * machmode.def (VECTOR_BOOL_MODE): Document.
30577 * genmodes.c (VECTOR_BOOL_MODE): New macro.
30578 (make_vector_bool_mode): New function.
30579 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
30581 * lto-streamer-in.c (lto_input_mode_table): Likewise.
30582 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
30584 * stor-layout.c (int_mode_for_mode): Likewise.
30585 * tree.c (build_vector_type_for_mode): Likewise.
30586 * varasm.c (output_constant_pool_2): Likewise.
30587 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
30588 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
30589 for MODE_VECTOR_BOOL.
30590 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
30591 of mode class checks.
30592 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
30593 instead of a list of mode class checks.
30594 (expand_vector_scalar_condition): Likewise.
30595 (type_for_widest_vector_mode): Handle BImode as an inner mode.
30597 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30598 Alan Hayward <alan.hayward@arm.com>
30599 David Sherwood <david.sherwood@arm.com>
30601 * machmode.h (mode_size): Change from unsigned short to
30603 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
30604 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
30605 or if measurement_type is not polynomial.
30606 (fixed_size_mode::includes_p): Check for constant-sized modes.
30607 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
30608 return a poly_uint16 rather than an unsigned short.
30609 (emit_mode_size): Change the type of mode_size from unsigned short
30610 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
30611 (emit_mode_adjustments): Cope with polynomial vector sizes.
30612 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30614 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30616 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
30617 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
30618 * caller-save.c (setup_save_areas): Likewise.
30619 (replace_reg_with_saved_mem): Likewise.
30620 * calls.c (emit_library_call_value_1): Likewise.
30621 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
30622 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
30623 (gen_lowpart_for_combine): Likewise.
30624 * convert.c (convert_to_integer_1): Likewise.
30625 * cse.c (equiv_constant, cse_insn): Likewise.
30626 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
30627 (cselib_subst_to_values): Likewise.
30628 * dce.c (word_dce_process_block): Likewise.
30629 * df-problems.c (df_word_lr_mark_ref): Likewise.
30630 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
30631 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
30632 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
30633 (rtl_for_decl_location): Likewise.
30634 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
30635 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
30636 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
30637 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
30638 (expand_expr_real_1): Likewise.
30639 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
30640 (pad_below): Likewise.
30641 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
30642 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
30643 * ira.c (get_subreg_tracking_sizes): Likewise.
30644 * ira-build.c (ira_create_allocno_objects): Likewise.
30645 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
30646 (ira_sort_regnos_for_alter_reg): Likewise.
30647 * ira-costs.c (record_operand_costs): Likewise.
30648 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
30649 (resolve_simple_move): Likewise.
30650 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
30651 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
30652 (lra_constraints): Likewise.
30653 (CONST_POOL_OK_P): Reject variable-sized modes.
30654 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
30655 (add_pseudo_to_slot, lra_spill): Likewise.
30656 * omp-low.c (omp_clause_aligned_alignment): Likewise.
30657 * optabs-query.c (get_best_extraction_insn): Likewise.
30658 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
30659 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
30660 (expand_mult_highpart, valid_multiword_target_p): Likewise.
30661 * recog.c (offsettable_address_addr_space_p): Likewise.
30662 * regcprop.c (maybe_mode_change): Likewise.
30663 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
30664 * regrename.c (build_def_use): Likewise.
30665 * regstat.c (dump_reg_info): Likewise.
30666 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
30667 (find_reloads, find_reloads_subreg_address): Likewise.
30668 * reload1.c (eliminate_regs_1): Likewise.
30669 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
30670 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
30671 (simplify_binary_operation_1, simplify_subreg): Likewise.
30672 * targhooks.c (default_function_arg_padding): Likewise.
30673 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
30674 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
30675 (verify_gimple_assign_ternary): Likewise.
30676 * tree-inline.c (estimate_move_cost): Likewise.
30677 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30678 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
30679 (get_address_cost_ainc): Likewise.
30680 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30681 (vect_supportable_dr_alignment): Likewise.
30682 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30683 (vectorizable_reduction): Likewise.
30684 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
30685 (vectorizable_operation, vectorizable_load): Likewise.
30686 * tree.c (build_same_sized_truth_vector_type): Likewise.
30687 * valtrack.c (cleanup_auto_inc_dec): Likewise.
30688 * var-tracking.c (emit_note_insn_var_location): Likewise.
30689 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
30690 (ADDR_VEC_ALIGN): Likewise.
30692 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30693 Alan Hayward <alan.hayward@arm.com>
30694 David Sherwood <david.sherwood@arm.com>
30696 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
30698 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
30699 or if measurement_type is polynomial.
30700 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
30701 * combine.c (make_extraction): Likewise.
30702 * dse.c (find_shift_sequence): Likewise.
30703 * dwarf2out.c (mem_loc_descriptor): Likewise.
30704 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
30705 (extract_bit_field, extract_low_bits): Likewise.
30706 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
30707 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
30708 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
30709 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
30710 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
30711 * reload.c (find_reloads): Likewise.
30712 * reload1.c (alter_reg): Likewise.
30713 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
30714 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
30715 * tree-if-conv.c (predicate_mem_writes): Likewise.
30716 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
30717 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
30718 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
30719 * valtrack.c (dead_debug_insert_temp): Likewise.
30720 * varasm.c (mergeable_constant_section): Likewise.
30721 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
30723 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30724 Alan Hayward <alan.hayward@arm.com>
30725 David Sherwood <david.sherwood@arm.com>
30727 * expr.c (expand_assignment): Cope with polynomial mode sizes
30728 when assigning to a CONCAT.
30730 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30731 Alan Hayward <alan.hayward@arm.com>
30732 David Sherwood <david.sherwood@arm.com>
30734 * machmode.h (mode_precision): Change from unsigned short to
30736 (mode_to_precision): Return a poly_uint16 rather than an unsigned
30738 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
30739 or if measurement_type is not polynomial.
30740 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
30741 in which the mode is already known to be a scalar_int_mode.
30742 * genmodes.c (emit_mode_precision): Change the type of mode_precision
30743 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
30745 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30746 for GET_MODE_PRECISION.
30747 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30748 for GET_MODE_PRECISION.
30749 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
30751 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
30752 (expand_field_assignment, make_extraction): Likewise.
30753 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
30754 (get_last_value): Likewise.
30755 * convert.c (convert_to_integer_1): Likewise.
30756 * cse.c (cse_insn): Likewise.
30757 * expr.c (expand_expr_real_1): Likewise.
30758 * lra-constraints.c (simplify_operand_subreg): Likewise.
30759 * optabs-query.c (can_atomic_load_p): Likewise.
30760 * optabs.c (expand_atomic_load): Likewise.
30761 (expand_atomic_store): Likewise.
30762 * ree.c (combine_reaching_defs): Likewise.
30763 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
30764 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
30765 * tree.h (type_has_mode_precision_p): Likewise.
30766 * ubsan.c (instrument_si_overflow): Likewise.
30768 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30769 Alan Hayward <alan.hayward@arm.com>
30770 David Sherwood <david.sherwood@arm.com>
30772 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
30773 polynomial numbers of units.
30774 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
30775 (valid_vector_subparts_p): New function.
30776 (build_vector_type): Remove temporary shim and take the number
30777 of units as a poly_uint64 rather than an int.
30778 (build_opaque_vector_type): Take the number of units as a
30779 poly_uint64 rather than an int.
30780 * tree.c (build_vector_from_ctor): Handle polynomial
30781 TYPE_VECTOR_SUBPARTS.
30782 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
30783 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
30784 (build_vector_from_val): If the number of units is variable,
30785 use build_vec_duplicate_cst for constant operands and
30786 VEC_DUPLICATE_EXPR otherwise.
30787 (make_vector_type): Remove temporary is_constant ().
30788 (build_vector_type, build_opaque_vector_type): Take the number of
30789 units as a poly_uint64 rather than an int.
30790 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
30792 * cfgexpand.c (expand_debug_expr): Likewise.
30793 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
30794 (store_constructor, expand_expr_real_1): Likewise.
30795 (const_scalar_mask_from_tree): Likewise.
30796 * fold-const-call.c (fold_const_reduction): Likewise.
30797 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
30798 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
30799 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
30800 (fold_relational_const): Likewise.
30801 (native_interpret_vector): Likewise. Change the size from an
30802 int to an unsigned int.
30803 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
30804 TYPE_VECTOR_SUBPARTS.
30805 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
30806 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
30807 duplicating a non-constant operand into a variable-length vector.
30808 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
30809 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
30810 * ipa-icf.c (sem_variable::equals): Likewise.
30811 * match.pd: Likewise.
30812 * omp-simd-clone.c (simd_clone_subparts): Likewise.
30813 * print-tree.c (print_node): Likewise.
30814 * stor-layout.c (layout_type): Likewise.
30815 * targhooks.c (default_builtin_vectorization_cost): Likewise.
30816 * tree-cfg.c (verify_gimple_comparison): Likewise.
30817 (verify_gimple_assign_binary): Likewise.
30818 (verify_gimple_assign_ternary): Likewise.
30819 (verify_gimple_assign_single): Likewise.
30820 * tree-pretty-print.c (dump_generic_node): Likewise.
30821 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30822 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
30823 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
30824 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
30825 (vect_shift_permute_load_chain): Likewise.
30826 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
30827 (expand_vector_condition, optimize_vector_constructor): Likewise.
30828 (lower_vec_perm, get_compute_type): Likewise.
30829 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30830 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
30831 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
30832 (vect_recog_mask_conversion_pattern): Likewise.
30833 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
30834 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
30835 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30836 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
30837 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
30838 (vectorizable_shift, vectorizable_operation, vectorizable_store)
30839 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
30840 (supportable_widening_operation): Likewise.
30841 (supportable_narrowing_operation): Likewise.
30842 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
30844 * varasm.c (output_constant): Likewise.
30846 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30847 Alan Hayward <alan.hayward@arm.com>
30848 David Sherwood <david.sherwood@arm.com>
30850 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
30851 so that both the length == 3 and length != 3 cases set up their
30852 own permute vectors. Add comments explaining why we know the
30853 number of elements is constant.
30854 (vect_permute_load_chain): Likewise.
30856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30857 Alan Hayward <alan.hayward@arm.com>
30858 David Sherwood <david.sherwood@arm.com>
30860 * machmode.h (mode_nunits): Change from unsigned char to
30862 (ONLY_FIXED_SIZE_MODES): New macro.
30863 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
30864 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
30865 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
30867 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
30868 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
30869 or if measurement_type is not polynomial.
30870 * genmodes.c (ZERO_COEFFS): New macro.
30871 (emit_mode_nunits_inline): Make mode_nunits_inline return a
30873 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
30874 Use ZERO_COEFFS when emitting initializers.
30875 * data-streamer.h (bp_pack_poly_value): New function.
30876 (bp_unpack_poly_value): Likewise.
30877 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30878 for GET_MODE_NUNITS.
30879 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30880 for GET_MODE_NUNITS.
30881 * tree.c (make_vector_type): Remove temporary shim and make
30882 the real function take the number of units as a poly_uint64
30883 rather than an int.
30884 (build_vector_type_for_mode): Handle polynomial nunits.
30885 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
30886 * emit-rtl.c (const_vec_series_p_1): Likewise.
30887 (gen_rtx_CONST_VECTOR): Likewise.
30888 * fold-const.c (test_vec_duplicate_folding): Likewise.
30889 * genrecog.c (validate_pattern): Likewise.
30890 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
30891 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
30892 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
30893 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
30894 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
30895 * rtlanal.c (subreg_get_info): Likewise.
30896 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30897 (vect_grouped_load_supported): Likewise.
30898 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
30899 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
30900 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
30901 (simplify_const_unary_operation, simplify_binary_operation_1)
30902 (simplify_const_binary_operation, simplify_ternary_operation)
30903 (test_vector_ops_duplicate, test_vector_ops): Likewise.
30904 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
30905 instead of CONST_VECTOR_NUNITS.
30906 * varasm.c (output_constant_pool_2): Likewise.
30907 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
30908 explicit-encoded elements in the XVEC for variable-length vectors.
30910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30912 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
30914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30915 Alan Hayward <alan.hayward@arm.com>
30916 David Sherwood <david.sherwood@arm.com>
30918 * coretypes.h (fixed_size_mode): Declare.
30919 (fixed_size_mode_pod): New typedef.
30920 * builtins.h (target_builtins::x_apply_args_mode)
30921 (target_builtins::x_apply_result_mode): Change type to
30922 fixed_size_mode_pod.
30923 * builtins.c (apply_args_size, apply_result_size, result_vector)
30924 (expand_builtin_apply_args_1, expand_builtin_apply)
30925 (expand_builtin_return): Update accordingly.
30927 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30929 * cse.c (hash_rtx_cb): Hash only the encoded elements.
30930 * cselib.c (cselib_hash_rtx): Likewise.
30931 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
30932 CONST_VECTOR encoding.
30934 2018-01-03 Jakub Jelinek <jakub@redhat.com>
30935 Jeff Law <law@redhat.com>
30938 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
30939 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
30940 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
30941 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
30944 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
30945 explicitly probe *sp in a noreturn function if there were any callee
30946 register saves or frame pointer is needed.
30948 2018-01-03 Jakub Jelinek <jakub@redhat.com>
30951 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
30952 BLKmode for ternary, binary or unary expressions.
30955 * var-tracking.c (delete_vta_debug_insn): New inline function.
30956 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
30957 insns from get_insns () to NULL instead of each bb separately.
30958 Use delete_vta_debug_insn. No longer static.
30959 (vt_debug_insns_local, variable_tracking_main_1): Adjust
30960 delete_vta_debug_insns callers.
30961 * rtl.h (delete_vta_debug_insns): Declare.
30962 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
30963 instead of variable_tracking_main.
30965 2018-01-03 Martin Sebor <msebor@redhat.com>
30967 PR tree-optimization/83603
30968 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
30969 arguments past the endof the argument list in functions declared
30970 without a prototype.
30971 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
30972 Avoid checking when arguments are null.
30974 2018-01-03 Martin Sebor <msebor@redhat.com>
30977 * doc/extend.texi (attribute const): Fix a typo.
30978 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
30979 issuing -Wsuggest-attribute for void functions.
30981 2018-01-03 Martin Sebor <msebor@redhat.com>
30983 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
30984 offset_int::from instead of wide_int::to_shwi.
30985 (maybe_diag_overlap): Remove assertion.
30986 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
30987 * gimple-ssa-sprintf.c (format_directive): Same.
30988 (parse_directive): Same.
30989 (sprintf_dom_walker::compute_format_length): Same.
30990 (try_substitute_return_value): Same.
30992 2018-01-03 Jeff Law <law@redhat.com>
30994 PR middle-end/83654
30995 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
30996 non-constant residual for zero at runtime and avoid probing in
30997 that case. Reorganize code for trailing problem to mirror handling
31000 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31002 PR tree-optimization/83501
31003 * tree-ssa-strlen.c (get_string_cst): New.
31004 (handle_char_store): Call get_string_cst.
31006 2018-01-03 Martin Liska <mliska@suse.cz>
31008 PR tree-optimization/83593
31009 * tree-ssa-strlen.c: Include tree-cfg.h.
31010 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
31011 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
31012 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
31014 (strlen_dom_walker::before_dom_children): Call
31015 gimple_purge_dead_eh_edges. Dump tranformation with details
31017 (strlen_dom_walker::before_dom_children): Update call by adding
31018 new argument cleanup_eh.
31019 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
31021 2018-01-03 Martin Liska <mliska@suse.cz>
31024 * cif-code.def (VARIADIC_THUNK): New enum value.
31025 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
31028 2018-01-03 Jan Beulich <jbeulich@suse.com>
31030 * sse.md (mov<mode>_internal): Tighten condition for when to use
31031 vmovdqu<ssescalarsize> for TI and OI modes.
31033 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31035 Update copyright years.
31037 2018-01-03 Martin Liska <mliska@suse.cz>
31040 * ipa-visibility.c (function_and_variable_visibility): Skip
31041 functions with noipa attribure.
31043 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31045 * gcc.c (process_command): Update copyright notice dates.
31046 * gcov-dump.c (print_version): Ditto.
31047 * gcov.c (print_version): Ditto.
31048 * gcov-tool.c (print_version): Ditto.
31049 * gengtype.c (create_file): Ditto.
31050 * doc/cpp.texi: Bump @copying's copyright year.
31051 * doc/cppinternals.texi: Ditto.
31052 * doc/gcc.texi: Ditto.
31053 * doc/gccint.texi: Ditto.
31054 * doc/gcov.texi: Ditto.
31055 * doc/install.texi: Ditto.
31056 * doc/invoke.texi: Ditto.
31058 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31060 * vector-builder.h (vector_builder::m_full_nelts): Change from
31061 unsigned int to poly_uint64.
31062 (vector_builder::full_nelts): Update prototype accordingly.
31063 (vector_builder::new_vector): Likewise.
31064 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
31065 (vector_builder::operator ==): Likewise.
31066 (vector_builder::finalize): Likewise.
31067 * int-vector-builder.h (int_vector_builder::int_vector_builder):
31068 Take the number of elements as a poly_uint64 rather than an
31070 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
31071 from unsigned int to poly_uint64.
31072 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
31073 (vec_perm_indices::new_vector): Likewise.
31074 (vec_perm_indices::length): Likewise.
31075 (vec_perm_indices::nelts_per_input): Likewise.
31076 (vec_perm_indices::input_nelts): Likewise.
31077 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
31078 number of elements per input as a poly_uint64 rather than an
31079 unsigned int. Use the original encoding for variable-length
31080 vectors, rather than clamping each individual element.
31081 For the second and subsequent elements in each pattern,
31082 clamp the step and base before clamping their sum.
31083 (vec_perm_indices::series_p): Handle polynomial element counts.
31084 (vec_perm_indices::all_in_range_p): Likewise.
31085 (vec_perm_indices_to_tree): Likewise.
31086 (vec_perm_indices_to_rtx): Likewise.
31087 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
31088 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
31089 (tree_vector_builder::new_binary_operation): Handle polynomial
31090 element counts. Return false if we need to know the number
31091 of elements at compile time.
31092 * fold-const.c (fold_vec_perm): Punt if the number of elements
31093 isn't known at compile time.
31095 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31097 * vec-perm-indices.h (vec_perm_builder): Change element type
31098 from HOST_WIDE_INT to poly_int64.
31099 (vec_perm_indices::element_type): Update accordingly.
31100 (vec_perm_indices::clamp): Handle polynomial element_types.
31101 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
31102 (vec_perm_indices::all_in_range_p): Likewise.
31103 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
31105 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
31106 polynomial vec_perm_indices element types.
31107 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
31108 * fold-const.c (fold_vec_perm): Likewise.
31109 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
31110 * tree-vect-generic.c (lower_vec_perm): Likewise.
31111 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
31112 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
31113 element type to HOST_WIDE_INT.
31115 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31116 Alan Hayward <alan.hayward@arm.com>
31117 David Sherwood <david.sherwood@arm.com>
31119 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
31120 rather than an int. Use plus_constant.
31121 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
31122 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
31124 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31125 Alan Hayward <alan.hayward@arm.com>
31126 David Sherwood <david.sherwood@arm.com>
31128 * calls.c (emit_call_1, expand_call): Change struct_value_size from
31129 a HOST_WIDE_INT to a poly_int64.
31131 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31132 Alan Hayward <alan.hayward@arm.com>
31133 David Sherwood <david.sherwood@arm.com>
31135 * calls.c (load_register_parameters): Cope with polynomial
31136 mode sizes. Require a constant size for BLKmode parameters
31137 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
31138 forces a parameter to be padded at the lsb end in order to
31139 fill a complete number of words, require the parameter size
31140 to be ordered wrt UNITS_PER_WORD.
31142 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31143 Alan Hayward <alan.hayward@arm.com>
31144 David Sherwood <david.sherwood@arm.com>
31146 * reload1.c (spill_stack_slot_width): Change element type
31147 from unsigned int to poly_uint64_pod.
31148 (alter_reg): Treat mode sizes as polynomial.
31150 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31151 Alan Hayward <alan.hayward@arm.com>
31152 David Sherwood <david.sherwood@arm.com>
31154 * reload.c (complex_word_subreg_p): New function.
31155 (reload_inner_reg_of_subreg, push_reload): Use it.
31157 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31158 Alan Hayward <alan.hayward@arm.com>
31159 David Sherwood <david.sherwood@arm.com>
31161 * lra-constraints.c (process_alt_operands): Reject matched
31162 operands whose sizes aren't ordered.
31163 (match_reload): Refer to this check here.
31165 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31166 Alan Hayward <alan.hayward@arm.com>
31167 David Sherwood <david.sherwood@arm.com>
31169 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
31170 that the mode size is in the set {1, 2, 4, 8, 16}.
31172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31173 Alan Hayward <alan.hayward@arm.com>
31174 David Sherwood <david.sherwood@arm.com>
31176 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
31177 Use plus_constant instead of gen_rtx_PLUS.
31179 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31180 Alan Hayward <alan.hayward@arm.com>
31181 David Sherwood <david.sherwood@arm.com>
31183 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
31184 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
31185 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
31186 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
31187 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
31188 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
31189 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
31190 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
31191 * config/i386/i386.c (ix86_push_rounding): ...this new function.
31192 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
31194 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
31195 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
31196 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
31197 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
31198 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
31199 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
31200 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
31201 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
31202 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
31203 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
31205 * expr.c (emit_move_resolve_push): Treat the input and result
31206 of PUSH_ROUNDING as a poly_int64.
31207 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
31208 (emit_push_insn): Likewise.
31209 * lra-eliminations.c (mark_not_eliminable): Likewise.
31210 * recog.c (push_operand): Likewise.
31211 * reload1.c (elimination_effects): Likewise.
31212 * rtlanal.c (nonzero_bits1): Likewise.
31213 * calls.c (store_one_arg): Likewise. Require the padding to be
31214 known at compile time.
31216 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31217 Alan Hayward <alan.hayward@arm.com>
31218 David Sherwood <david.sherwood@arm.com>
31220 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
31221 Use plus_constant instead of gen_rtx_PLUS.
31223 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31224 Alan Hayward <alan.hayward@arm.com>
31225 David Sherwood <david.sherwood@arm.com>
31227 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
31228 rather than an int.
31230 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31231 Alan Hayward <alan.hayward@arm.com>
31232 David Sherwood <david.sherwood@arm.com>
31234 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
31235 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
31236 via stack temporaries. Treat the mode size as polynomial too.
31238 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31239 Alan Hayward <alan.hayward@arm.com>
31240 David Sherwood <david.sherwood@arm.com>
31242 * expr.c (expand_expr_real_2): When handling conversions involving
31243 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
31244 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
31245 as a poly_uint64 too.
31247 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31248 Alan Hayward <alan.hayward@arm.com>
31249 David Sherwood <david.sherwood@arm.com>
31251 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
31253 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31254 Alan Hayward <alan.hayward@arm.com>
31255 David Sherwood <david.sherwood@arm.com>
31257 * combine.c (can_change_dest_mode): Handle polynomial
31258 REGMODE_NATURAL_SIZE.
31259 * expmed.c (store_bit_field_1): Likewise.
31260 * expr.c (store_constructor): Likewise.
31261 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
31262 and polynomial REGMODE_NATURAL_SIZE.
31263 (gen_lowpart_common): Likewise.
31264 * reginfo.c (record_subregs_of_mode): Likewise.
31265 * rtlanal.c (read_modify_subreg_p): Likewise.
31267 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31268 Alan Hayward <alan.hayward@arm.com>
31269 David Sherwood <david.sherwood@arm.com>
31271 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
31272 numbers of elements.
31274 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31275 Alan Hayward <alan.hayward@arm.com>
31276 David Sherwood <david.sherwood@arm.com>
31278 * match.pd: Cope with polynomial numbers of vector elements.
31280 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31281 Alan Hayward <alan.hayward@arm.com>
31282 David Sherwood <david.sherwood@arm.com>
31284 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
31285 in a POINTER_PLUS_EXPR.
31287 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31288 Alan Hayward <alan.hayward@arm.com>
31289 David Sherwood <david.sherwood@arm.com>
31291 * omp-simd-clone.c (simd_clone_subparts): New function.
31292 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
31293 (ipa_simd_modify_function_body): Likewise.
31295 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31296 Alan Hayward <alan.hayward@arm.com>
31297 David Sherwood <david.sherwood@arm.com>
31299 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
31300 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
31301 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
31302 (expand_vector_condition, vector_element): Likewise.
31303 (subparts_gt): New function.
31304 (get_compute_type): Use subparts_gt.
31305 (count_type_subparts): Delete.
31306 (expand_vector_operations_1): Use subparts_gt instead of
31307 count_type_subparts.
31309 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31310 Alan Hayward <alan.hayward@arm.com>
31311 David Sherwood <david.sherwood@arm.com>
31313 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
31314 (vect_compile_time_alias): ...this new function. Do the calculation
31315 on poly_ints rather than trees.
31316 (vect_prune_runtime_alias_test_list): Update call accordingly.
31318 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31319 Alan Hayward <alan.hayward@arm.com>
31320 David Sherwood <david.sherwood@arm.com>
31322 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
31324 (vect_schedule_slp_instance): Likewise.
31326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31327 Alan Hayward <alan.hayward@arm.com>
31328 David Sherwood <david.sherwood@arm.com>
31330 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
31331 constant and extern definitions for variable-length vectors.
31332 (vect_get_constant_vectors): Note that the number of units
31333 is known to be constant.
31335 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31336 Alan Hayward <alan.hayward@arm.com>
31337 David Sherwood <david.sherwood@arm.com>
31339 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
31340 of units as polynomial. Choose between WIDE and NARROW based
31343 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31344 Alan Hayward <alan.hayward@arm.com>
31345 David Sherwood <david.sherwood@arm.com>
31347 * tree-vect-stmts.c (simd_clone_subparts): New function.
31348 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
31350 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31351 Alan Hayward <alan.hayward@arm.com>
31352 David Sherwood <david.sherwood@arm.com>
31354 * tree-vect-stmts.c (vectorizable_call): Treat the number of
31355 vectors as polynomial. Use build_index_vector for
31356 IFN_GOMP_SIMD_LANE.
31358 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31359 Alan Hayward <alan.hayward@arm.com>
31360 David Sherwood <david.sherwood@arm.com>
31362 * tree-vect-stmts.c (get_load_store_type): Treat the number of
31363 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
31364 for variable-length vectors.
31365 (vectorizable_mask_load_store): Treat the number of units as
31366 polynomial, asserting that it is constant if the condition has
31367 already been enforced.
31368 (vectorizable_store, vectorizable_load): Likewise.
31370 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31371 Alan Hayward <alan.hayward@arm.com>
31372 David Sherwood <david.sherwood@arm.com>
31374 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
31375 of units as polynomial. Punt if we can't tell at compile time
31376 which vector contains the final result.
31378 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31379 Alan Hayward <alan.hayward@arm.com>
31380 David Sherwood <david.sherwood@arm.com>
31382 * tree-vect-loop.c (vectorizable_induction): Treat the number
31383 of units as polynomial. Punt on SLP inductions. Use an integer
31384 VEC_SERIES_EXPR for variable-length integer reductions. Use a
31385 cast of such a series for variable-length floating-point
31388 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31389 Alan Hayward <alan.hayward@arm.com>
31390 David Sherwood <david.sherwood@arm.com>
31392 * tree.h (build_index_vector): Declare.
31393 * tree.c (build_index_vector): New function.
31394 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
31395 of units as polynomial, forcibly converting it to a constant if
31396 vectorizable_reduction has already enforced the condition.
31397 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
31398 to create a {1,2,3,...} vector.
31399 (vectorizable_reduction): Treat the number of units as polynomial.
31400 Choose vectype_in based on the largest scalar element size rather
31401 than the smallest number of units. Enforce the restrictions
31404 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31405 Alan Hayward <alan.hayward@arm.com>
31406 David Sherwood <david.sherwood@arm.com>
31408 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
31409 number of units as polynomial.
31411 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31412 Alan Hayward <alan.hayward@arm.com>
31413 David Sherwood <david.sherwood@arm.com>
31415 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
31416 * target.def (autovectorize_vector_sizes): Return the vector sizes
31417 by pointer, using vector_sizes rather than a bitmask.
31418 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
31419 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
31420 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
31422 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
31423 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
31424 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
31425 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
31426 * omp-general.c (omp_max_vf): Likewise.
31427 * omp-low.c (omp_clause_aligned_alignment): Likewise.
31428 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
31429 * tree-vect-loop.c (vect_analyze_loop): Likewise.
31430 * tree-vect-slp.c (vect_slp_bb): Likewise.
31431 * doc/tm.texi: Regenerate.
31432 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
31434 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
31435 the vector size as a poly_uint64 rather than an unsigned int.
31436 (current_vector_size): Change from an unsigned int to a poly_uint64.
31437 (get_vectype_for_scalar_type): Update accordingly.
31438 * tree.h (build_truth_vector_type): Take the size and number of
31439 units as a poly_uint64 rather than an unsigned int.
31440 (build_vector_type): Add a temporary overload that takes
31441 the number of units as a poly_uint64 rather than an unsigned int.
31442 * tree.c (make_vector_type): Likewise.
31443 (build_truth_vector_type): Take the number of units as a poly_uint64
31444 rather than an unsigned int.
31446 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31447 Alan Hayward <alan.hayward@arm.com>
31448 David Sherwood <david.sherwood@arm.com>
31450 * target.def (get_mask_mode): Take the number of units and length
31451 as poly_uint64s rather than unsigned ints.
31452 * targhooks.h (default_get_mask_mode): Update accordingly.
31453 * targhooks.c (default_get_mask_mode): Likewise.
31454 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
31455 * doc/tm.texi: Regenerate.
31457 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31458 Alan Hayward <alan.hayward@arm.com>
31459 David Sherwood <david.sherwood@arm.com>
31461 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
31462 * omp-general.c (omp_max_vf): Likewise.
31463 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
31464 (expand_omp_simd): Handle polynomial safelen.
31465 * omp-low.c (omplow_simd_context): Add a default constructor.
31466 (omplow_simd_context::max_vf): Change from int to poly_uint64.
31467 (lower_rec_simd_input_clauses): Update accordingly.
31468 (lower_rec_input_clauses): Likewise.
31470 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31471 Alan Hayward <alan.hayward@arm.com>
31472 David Sherwood <david.sherwood@arm.com>
31474 * tree-vectorizer.h (vect_nunits_for_cost): New function.
31475 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
31476 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
31477 (vect_analyze_slp_cost): Likewise.
31478 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
31479 (vect_model_load_cost): Likewise.
31481 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31482 Alan Hayward <alan.hayward@arm.com>
31483 David Sherwood <david.sherwood@arm.com>
31485 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
31486 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
31487 from an unsigned int * to a poly_uint64_pod *.
31488 (calculate_unrolling_factor): New function.
31489 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
31491 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31492 Alan Hayward <alan.hayward@arm.com>
31493 David Sherwood <david.sherwood@arm.com>
31495 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
31496 from an unsigned int to a poly_uint64.
31497 (_loop_vec_info::slp_unrolling_factor): Likewise.
31498 (_loop_vec_info::vectorization_factor): Change from an int
31500 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
31501 (vect_get_num_vectors): New function.
31502 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
31503 (vect_get_num_copies): Use vect_get_num_vectors.
31504 (vect_analyze_data_ref_dependences): Change max_vf from an int *
31505 to an unsigned int *.
31506 (vect_analyze_data_refs): Change min_vf from an int * to a
31508 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
31509 than an unsigned HOST_WIDE_INT.
31510 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
31511 (vect_analyze_data_ref_dependence): Change max_vf from an int *
31512 to an unsigned int *.
31513 (vect_analyze_data_ref_dependences): Likewise.
31514 (vect_compute_data_ref_alignment): Handle polynomial vf.
31515 (vect_enhance_data_refs_alignment): Likewise.
31516 (vect_prune_runtime_alias_test_list): Likewise.
31517 (vect_shift_permute_load_chain): Likewise.
31518 (vect_supportable_dr_alignment): Likewise.
31519 (dependence_distance_ge_vf): Take the vectorization factor as a
31520 poly_uint64 rather than an unsigned HOST_WIDE_INT.
31521 (vect_analyze_data_refs): Change min_vf from an int * to a
31523 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
31524 vfm1 as a poly_uint64 rather than an int. Make the same change
31525 for the returned bound_scalar.
31526 (vect_gen_vector_loop_niters): Handle polynomial vf.
31527 (vect_do_peeling): Likewise. Update call to
31528 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
31529 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
31531 * tree-vect-loop.c (vect_determine_vectorization_factor)
31532 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
31533 (vect_get_known_peeling_cost): Likewise.
31534 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
31535 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
31536 (vect_transform_loop): Likewise. Use the lowest possible VF when
31537 updating the upper bounds of the loop.
31538 (vect_min_worthwhile_factor): Make static. Return an unsigned int
31539 rather than an int.
31540 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
31541 polynomial unroll factors.
31542 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
31543 (vect_make_slp_decision): Likewise.
31544 (vect_supported_load_permutation_p): Likewise, and polynomial
31546 (vect_analyze_slp_cost): Handle polynomial vf.
31547 (vect_slp_analyze_node_operations): Likewise.
31548 (vect_slp_analyze_bb_1): Likewise.
31549 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
31550 than an unsigned HOST_WIDE_INT.
31551 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
31552 (vectorizable_load): Handle polynomial vf.
31553 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
31555 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
31557 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31558 Alan Hayward <alan.hayward@arm.com>
31559 David Sherwood <david.sherwood@arm.com>
31561 * match.pd: Handle bit operations involving three constants
31562 and try to fold one pair.
31564 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31566 * tree-vect-loop-manip.c: Include gimple-fold.h.
31567 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
31568 niters_maybe_zero parameters. Handle other cases besides a step of 1.
31569 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
31570 Add a path that uses a step of VF instead of 1, but disable it
31572 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
31573 and niters_no_overflow parameters. Update calls to
31574 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
31575 Create a new SSA name if the latter choses to use a ste other
31576 than zero, and return it via niters_vector_mult_vf_var.
31577 * tree-vect-loop.c (vect_transform_loop): Update calls to
31578 vect_do_peeling, vect_gen_vector_loop_niters and
31579 slpeel_make_loop_iterate_ntimes.
31580 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
31581 (vect_gen_vector_loop_niters): Update declarations after above changes.
31583 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
31585 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
31586 128-bit round to integer instructions.
31587 (ceil<mode>2): Likewise.
31588 (btrunc<mode>2): Likewise.
31589 (round<mode>2): Likewise.
31591 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31593 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
31594 unaligned VSX load/store on P8/P9.
31595 (expand_block_clear): Allow the use of unaligned VSX
31596 load/store on P8/P9.
31598 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31600 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
31602 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
31603 swap associated with both a load and a store.
31605 2018-01-02 Andrew Waterman <andrew@sifive.com>
31607 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
31608 * config/riscv/riscv.md (clear_cache): Use it.
31610 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
31612 * web.c: Remove out-of-date comment.
31614 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31616 * expr.c (fixup_args_size_notes): Check that any existing
31617 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
31618 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
31619 (emit_single_push_insn): ...here.
31621 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31623 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
31624 (const_vector_encoded_nelts): New function.
31625 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
31626 (const_vector_int_elt, const_vector_elt): Declare.
31627 * emit-rtl.c (const_vector_int_elt_1): New function.
31628 (const_vector_elt): Likewise.
31629 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
31630 of CONST_VECTOR_ELT.
31632 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31634 * expr.c: Include rtx-vector-builder.h.
31635 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
31636 directly on the tree encoding.
31637 (const_vector_from_tree): Likewise.
31638 * optabs.c: Include rtx-vector-builder.h.
31639 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
31640 sequence of "u" values.
31641 * vec-perm-indices.c: Include rtx-vector-builder.h.
31642 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
31643 directly on the vec_perm_indices encoding.
31645 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31647 * doc/rtl.texi (const_vector): Describe new encoding scheme.
31648 * Makefile.in (OBJS): Add rtx-vector-builder.o.
31649 * rtx-vector-builder.h: New file.
31650 * rtx-vector-builder.c: Likewise.
31651 * rtl.h (rtx_def::u2): Add a const_vector field.
31652 (CONST_VECTOR_NPATTERNS): New macro.
31653 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
31654 (CONST_VECTOR_DUPLICATE_P): Likewise.
31655 (CONST_VECTOR_STEPPED_P): Likewise.
31656 (CONST_VECTOR_ENCODED_ELT): Likewise.
31657 (const_vec_duplicate_p): Check for a duplicated vector encoding.
31658 (unwrap_const_vec_duplicate): Likewise.
31659 (const_vec_series_p): Check for a non-duplicated vector encoding.
31660 Say that the function only returns true for integer vectors.
31661 * emit-rtl.c: Include rtx-vector-builder.h.
31662 (gen_const_vec_duplicate_1): Delete.
31663 (gen_const_vector): Call gen_const_vec_duplicate instead of
31664 gen_const_vec_duplicate_1.
31665 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
31666 (gen_const_vec_duplicate): Use rtx_vector_builder.
31667 (gen_const_vec_series): Likewise.
31668 (gen_rtx_CONST_VECTOR): Likewise.
31669 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
31670 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
31671 Build a new vector rather than modifying a CONST_VECTOR in-place.
31672 (handle_special_swappables): Update call accordingly.
31673 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
31674 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
31675 Build a new vector rather than modifying a CONST_VECTOR in-place.
31676 (handle_special_swappables): Update call accordingly.
31678 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31680 * simplify-rtx.c (simplify_const_binary_operation): Use
31681 CONST_VECTOR_ELT instead of XVECEXP.
31683 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31685 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
31686 the selector elements to be different from the data elements
31687 if the selector is a VECTOR_CST.
31688 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
31689 ssizetype for the selector.
31691 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31693 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
31694 before testing each element individually.
31695 * tree-vect-generic.c (lower_vec_perm): Likewise.
31697 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31699 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
31700 * selftest-run-tests.c (selftest::run_tests): Call it.
31701 * vector-builder.h (vector_builder::operator ==): New function.
31702 (vector_builder::operator !=): Likewise.
31703 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
31704 (vec_perm_indices::all_from_input_p): New function.
31705 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
31706 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
31707 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
31708 instead of reading the VECTOR_CST directly. Detect whether both
31709 vector inputs are the same before constructing the vec_perm_indices,
31710 and update the number of inputs argument accordingly. Use the
31711 utility functions added above. Only construct sel2 if we need to.
31713 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31715 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
31716 the broadcast of the low byte.
31717 (expand_mult_highpart): Use an explicit encoding for the permutes.
31718 * optabs-query.c (can_mult_highpart_p): Likewise.
31719 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
31720 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31721 (vectorizable_bswap): Likewise.
31722 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
31723 explicit encoding for the power-of-2 permutes.
31724 (vect_permute_store_chain): Likewise.
31725 (vect_grouped_load_supported): Likewise.
31726 (vect_permute_load_chain): Likewise.
31728 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31730 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
31731 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
31732 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
31733 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
31734 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
31735 (vect_gen_perm_mask_any): Likewise.
31737 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31739 * int-vector-builder.h: New file.
31740 * vec-perm-indices.h: Include int-vector-builder.h.
31741 (vec_perm_indices): Redefine as an int_vector_builder.
31742 (auto_vec_perm_indices): Delete.
31743 (vec_perm_builder): Redefine as a stand-alone class.
31744 (vec_perm_indices::vec_perm_indices): New function.
31745 (vec_perm_indices::clamp): Likewise.
31746 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
31747 (vec_perm_indices::new_vector): New function.
31748 (vec_perm_indices::new_expanded_vector): Update for new
31749 vec_perm_indices class.
31750 (vec_perm_indices::rotate_inputs): New function.
31751 (vec_perm_indices::all_in_range_p): Operate directly on the
31752 encoded form, without computing elided elements.
31753 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
31754 encoding. Update for new vec_perm_indices class.
31755 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
31756 the given vec_perm_builder.
31757 (expand_vec_perm_var): Update vec_perm_builder constructor.
31758 (expand_mult_highpart): Use vec_perm_builder instead of
31759 auto_vec_perm_indices.
31760 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
31761 vec_perm_indices instead of auto_vec_perm_indices. Use a single
31762 or double series encoding as appropriate.
31763 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
31764 vec_perm_indices instead of auto_vec_perm_indices.
31765 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31766 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31767 (vect_permute_store_chain): Likewise.
31768 (vect_grouped_load_supported): Likewise.
31769 (vect_permute_load_chain): Likewise.
31770 (vect_shift_permute_load_chain): Likewise.
31771 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
31772 (vect_transform_slp_perm_load): Likewise.
31773 (vect_schedule_slp_instance): Likewise.
31774 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31775 (vectorizable_mask_load_store): Likewise.
31776 (vectorizable_bswap): Likewise.
31777 (vectorizable_store): Likewise.
31778 (vectorizable_load): Likewise.
31779 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
31780 vec_perm_indices instead of auto_vec_perm_indices. Use
31781 tree_to_vec_perm_builder to read the vector from a tree.
31782 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
31783 vec_perm_builder instead of a vec_perm_indices.
31784 (have_whole_vector_shift): Use vec_perm_builder and
31785 vec_perm_indices instead of auto_vec_perm_indices. Leave the
31786 truncation to calc_vec_perm_mask_for_shift.
31787 (vect_create_epilog_for_reduction): Likewise.
31788 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
31789 from auto_vec_perm_indices to vec_perm_indices.
31790 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
31791 instead of changing individual elements.
31792 (aarch64_vectorize_vec_perm_const): Use new_vector to install
31793 the vector in d.perm.
31794 * config/arm/arm.c (expand_vec_perm_d::perm): Change
31795 from auto_vec_perm_indices to vec_perm_indices.
31796 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
31797 instead of changing individual elements.
31798 (arm_vectorize_vec_perm_const): Use new_vector to install
31799 the vector in d.perm.
31800 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
31801 Update vec_perm_builder constructor.
31802 (rs6000_expand_interleave): Likewise.
31803 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
31804 (rs6000_expand_interleave): Likewise.
31806 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31808 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
31809 to qimode could truncate the indices.
31810 * optabs.c (expand_vec_perm_var): Likewise.
31812 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31814 * Makefile.in (OBJS): Add vec-perm-indices.o.
31815 * vec-perm-indices.h: New file.
31816 * vec-perm-indices.c: Likewise.
31817 * target.h (vec_perm_indices): Replace with a forward class
31819 (auto_vec_perm_indices): Move to vec-perm-indices.h.
31820 * optabs.h: Include vec-perm-indices.h.
31821 (expand_vec_perm): Delete.
31822 (selector_fits_mode_p, expand_vec_perm_var): Declare.
31823 (expand_vec_perm_const): Declare.
31824 * target.def (vec_perm_const_ok): Replace with...
31825 (vec_perm_const): ...this new hook.
31826 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
31827 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
31828 * doc/tm.texi: Regenerate.
31829 * optabs.def (vec_perm_const): Delete.
31830 * doc/md.texi (vec_perm_const): Likewise.
31831 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
31832 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
31833 expand_vec_perm for constant permutation vectors. Assert that
31834 the mode of variable permutation vectors is the integer equivalent
31835 of the mode that is being permuted.
31836 * optabs-query.h (selector_fits_mode_p): Declare.
31837 * optabs-query.c: Include vec-perm-indices.h.
31838 (selector_fits_mode_p): New function.
31839 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
31840 is defined, instead of checking whether the vec_perm_const_optab
31841 exists. Use targetm.vectorize.vec_perm_const instead of
31842 targetm.vectorize.vec_perm_const_ok. Check whether the indices
31843 fit in the vector mode before using a variable permute.
31844 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
31845 vec_perm_indices instead of an rtx.
31846 (expand_vec_perm): Replace with...
31847 (expand_vec_perm_const): ...this new function. Take the selector
31848 as a vec_perm_indices rather than an rtx. Also take the mode of
31849 the selector. Update call to shift_amt_for_vec_perm_mask.
31850 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
31851 Use vec_perm_indices::new_expanded_vector to expand the original
31852 selector into bytes. Check whether the indices fit in the vector
31853 mode before using a variable permute.
31854 (expand_vec_perm_var): Make global.
31855 (expand_mult_highpart): Use expand_vec_perm_const.
31856 * fold-const.c: Includes vec-perm-indices.h.
31857 * tree-ssa-forwprop.c: Likewise.
31858 * tree-vect-data-refs.c: Likewise.
31859 * tree-vect-generic.c: Likewise.
31860 * tree-vect-loop.c: Likewise.
31861 * tree-vect-slp.c: Likewise.
31862 * tree-vect-stmts.c: Likewise.
31863 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
31865 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
31866 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
31867 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
31868 (aarch64_vectorize_vec_perm_const): ...this new function.
31869 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
31870 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31871 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
31872 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
31873 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
31874 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31875 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
31877 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
31878 check for NEON modes.
31879 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
31880 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
31881 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
31882 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
31884 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
31885 the old VEC_PERM_CONST conditions.
31886 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
31887 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
31888 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
31889 (ia64_vectorize_vec_perm_const_ok): Merge into...
31890 (ia64_vectorize_vec_perm_const): ...this new function.
31891 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
31892 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
31893 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
31894 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
31895 * config/mips/mips.c (mips_expand_vec_perm_const)
31896 (mips_vectorize_vec_perm_const_ok): Merge into...
31897 (mips_vectorize_vec_perm_const): ...this new function.
31898 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
31899 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
31900 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
31901 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
31902 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
31903 (rs6000_expand_vec_perm_const): Delete.
31904 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
31906 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31907 (altivec_expand_vec_perm_const_le): Take each operand individually.
31908 Operate on constant selectors rather than rtxes.
31909 (altivec_expand_vec_perm_const): Likewise. Update call to
31910 altivec_expand_vec_perm_const_le.
31911 (rs6000_expand_vec_perm_const): Delete.
31912 (rs6000_vectorize_vec_perm_const_ok): Delete.
31913 (rs6000_vectorize_vec_perm_const): New function.
31914 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
31915 an element count and rtx array.
31916 (rs6000_expand_extract_even): Update call accordingly.
31917 (rs6000_expand_interleave): Likewise.
31918 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
31919 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
31920 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
31921 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
31922 (rs6000_expand_vec_perm_const): Delete.
31923 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
31924 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31925 (altivec_expand_vec_perm_const_le): Take each operand individually.
31926 Operate on constant selectors rather than rtxes.
31927 (altivec_expand_vec_perm_const): Likewise. Update call to
31928 altivec_expand_vec_perm_const_le.
31929 (rs6000_expand_vec_perm_const): Delete.
31930 (rs6000_vectorize_vec_perm_const_ok): Delete.
31931 (rs6000_vectorize_vec_perm_const): New function. Remove stray
31932 reference to the SPE evmerge intructions.
31933 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
31934 an element count and rtx array.
31935 (rs6000_expand_extract_even): Update call accordingly.
31936 (rs6000_expand_interleave): Likewise.
31937 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
31938 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
31940 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31942 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31944 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
31945 vector mode and that that mode matches the mode of the data
31947 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
31948 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
31949 directly using expand_vec_perm_1 when forcing selectors into
31951 (expand_vec_perm_var): New function, split out from expand_vec_perm.
31953 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31955 * optabs-query.h (can_vec_perm_p): Delete.
31956 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
31957 * optabs-query.c (can_vec_perm_p): Split into...
31958 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
31959 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
31960 particular selector is valid.
31961 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31962 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31963 (vect_grouped_load_supported): Likewise.
31964 (vect_shift_permute_load_chain): Likewise.
31965 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
31966 (vect_transform_slp_perm_load): Likewise.
31967 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31968 (vectorizable_bswap): Likewise.
31969 (vect_gen_perm_mask_checked): Likewise.
31970 * fold-const.c (fold_ternary_loc): Likewise. Don't take
31971 implementations of variable permutation vectors into account
31972 when deciding which selector to use.
31973 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
31974 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
31975 with a false third argument.
31976 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
31977 to test whether the constant selector is valid and can_vec_perm_var_p
31978 to test whether a variable selector is valid.
31980 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31982 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
31983 * optabs-query.c (can_vec_perm_p): Likewise.
31984 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
31985 instead of vec_perm_indices.
31986 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
31987 (vect_gen_perm_mask_checked): Likewise,
31988 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
31989 (vect_gen_perm_mask_checked): Likewise,
31991 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31993 * optabs-query.h (qimode_for_vec_perm): Declare.
31994 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
31995 (qimode_for_vec_perm): ...this new function.
31996 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
31998 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32000 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
32001 does not have a conditional at the top.
32003 2018-01-02 Richard Biener <rguenther@suse.de>
32005 * ipa-inline.c (big_speedup_p): Fix expression.
32007 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32010 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
32013 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32017 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
32018 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
32019 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
32020 cond_taken_branch_cost 3->4.
32022 2018-01-01 Jakub Jelinek <jakub@redhat.com>
32024 PR tree-optimization/83581
32025 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
32026 TODO_cleanup_cfg if any changes have been made.
32028 PR middle-end/83608
32029 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
32030 convert_modes if target mode has the right side, but different mode
32033 PR middle-end/83609
32034 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
32035 last argument when extracting from CONCAT. If either from_real or
32036 from_imag is NULL, use expansion through memory. If result is not
32037 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
32038 the parts directly to inner mode, if even that fails, use expansion
32041 PR middle-end/83623
32042 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
32043 check for bswap in mode rather than HImode and use that in expand_unop
32046 Copyright (C) 2018 Free Software Foundation, Inc.
32048 Copying and distribution of this file, with or without modification,
32049 are permitted in any medium without royalty provided the copyright
32050 notice and this notice are preserved.