1 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
2 Kwok Cheung Yeung <kcy@codesourcery.com>
4 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
5 int_mode_for_size fails.
6 (vectorizable_load): Likewise.
8 2018-11-15 David Malcolm <dmalcolm@redhat.com>
10 * doc/ux.texi (Group logically-related diagnostics): Move
11 discussion of auto_diagnostic_group into this new subsection.
12 Give an example of where this grouping is used.
14 2018-11-15 David Malcolm <dmalcolm@redhat.com>
17 * Makefile.in (OBJS): Move json.o to...
18 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
19 * common.opt (fdiagnostics-format=): New option.
20 (diagnostics_output_format): New enum.
21 * diagnostic-format-json.cc: New file.
22 * diagnostic.c (default_diagnostic_final_cb): New function, taken
23 from start of diagnostic_finish.
24 (diagnostic_initialize): Initialize final_cb to
25 default_diagnostic_final_cb.
26 (diagnostic_finish): Move "being treated as errors" messages to
27 default_diagnostic_final_cb. Call any final_cb.
28 (default_diagnostic_finalizer): Add diagnostic_t param.
29 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
30 diagnostic_finalizer callback.
31 * diagnostic.h (enum diagnostics_output_format): New enum.
32 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
33 (struct diagnostic_context): Add "final_cb".
34 (default_diagnostic_finalizer): Add diagnostic_t param.
35 (diagnostic_output_format_init): New decl.
36 * doc/invoke.texi (-fdiagnostics-format): New option.
37 * dwarf2out.c (gen_producer_string): Ignore
38 OPT_fdiagnostics_format_.
39 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
40 * lto-wrapper.c (append_diag_options): Ignore it.
41 * opts.c (common_handle_option): Handle it.
43 2018-11-15 David Malcolm <dmalcolm@redhat.com>
45 PR tree-optimization/88015
46 * graphite-isl-ast-to-gimple.c
47 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
49 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
51 2018-11-15 Richard Biener <rguenther@suse.de>
54 * gimple.c (gimple_call_flags): Union flags from decl, type
56 * trans-mem.c (is_tm_pure_call): Simplify.
58 2018-11-15 Richard Biener <rguenther@suse.de>
60 PR tree-optimization/88031
61 * tree-vect-loop.c (vectorizable_reduction): Move check
62 for multiple types earlier so we get the expected dump.
63 Simplify calls to vectorizable_condition.
64 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
65 * tree-vect-stmts.c (vectorizable_condition): Instead of
66 reduc_def and reduc_index take just a flag. Simplify
67 code-generation now that we can rely on the defs being set up.
68 (vectorizable_comparison): Remove unused argument.
70 2018-11-15 Richard Biener <rguenther@suse.de>
72 PR tree-optimization/88030
73 * tree-complex.c (need_eh_cleanup): New global.
74 (update_complex_assignment): Mark blocks that need EH update.
75 (expand_complex_comparison): Likewise.
76 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
77 perform EH cleanup and schedule CFG cleanup if that did anything.
79 2018-11-15 Jakub Jelinek <jakub@redhat.com>
81 PR rtl-optimization/88018
82 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
83 on the presence of fallthru edge, rather than if it is a USE or not.
85 2018-11-15 Richard Biener <rguenther@suse.de>
88 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
89 analyze_subscript_affine_affine properly.
91 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
93 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
94 (GLIBC_DYNAMIC_LINKER): Define.
95 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
97 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
99 PR tree-optimization/84648
100 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
101 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
102 by calling adjust_cond_for_loop_until_wrap.
104 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
107 * doc/extend.texi (Common Function Attributes): Clarify linker
108 restrictions on "aligned" attribute.
109 (Common Variable Attributes): Likewise. Mention that linker
110 restrictions don't apply to stack-allocated variables.
112 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
114 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
117 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
121 * doc/invoke.texi (Optimize Options): Remove bad example about
122 interaction between -flto and -O. Replace it with a note that
123 you need to compile with -O and not just link. Copy-edit -flto
124 discussion to reduce verbiage and improve flow.
126 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
129 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
130 also suppress many optimizations. Alphabetize option lists for
131 -O1, -O2, and -Os. Add list of options disabled with -Og, and
132 correct documentation for those options to say that.
133 * opts.c (default_options_table): Sort table by level and option
134 name, to make it easier to correlate to the manual.
136 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
138 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
139 printing of the names of segment registers.
141 2018-11-14 Nathan Sidwell <nathan@acm.org>
145 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
148 2018-11-14 David Malcolm <dmalcolm@redhat.com>
150 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
152 2018-11-14 Jakub Jelinek <jakub@redhat.com>
154 P1236R1 - Signed integers are two's complement
155 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
158 * hash-map.h (hash_map::iterator::reference_pair): New class.
159 (hash_map::iterator::operator*): Return it rather than std::pair.
161 2018-11-14 Jeff Law <law@redhat.com>
163 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
164 iff the operand is a constant.
166 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
168 * gimple-ssa-evrp-analyze.c
169 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
170 ignore_equivs_equal_p to equal_p.
171 * ipa-cp.c (meet_with_1): Use equal_p instead of
172 ignore_equivs_equal_p.
173 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
174 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
175 (value_range::operator==): Remove.
176 (value_range::operator!=): Remove.
177 (vrp_prop::visit_stmt): Use equal_p.
178 * tree-vrp.h (value_range): Remove operator==, operator!=,
179 ignore_equivs_equal_p.
180 * vr-values.c (update_value_range): Use equal_p.
182 2018-11-14 Michael Matz <matz@suse.de>
185 * gimplify.c (collect_fallthrough_labels): Add new argument,
186 return location via that, don't modify statements.
187 (warn_implicit_fallthrough_r): Adjust call, don't use
188 statement location directly.
190 2018-11-14 Richard Biener <rguenther@suse.de>
193 * tree-data-ref.c (split_constant_offset): Add wrapper
194 allocating a cache hash-map.
195 (split_constant_offset_1): Cache results of expanding
196 expressions from SSA def stmts.
198 2018-11-14 Richard Biener <rguenther@suse.de>
201 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
203 (lambda_vector_mult_const): Likewise.
204 (lambda_matrix_right_hermite): Use lambda_int temporaries.
206 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
207 Jackson Woodruff <jackson.woodruff@arm.com>
209 PR 71026/tree-optimization
210 * match.pd: Simplify floating point comparisons.
212 2018-11-14 Jakub Jelinek <jakub@redhat.com>
214 PR rtl-optimization/87817
215 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
216 last argument has low 8 bits clear, fold to 0.
218 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
220 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
221 MAX_OFILE_ALIGNMENT as needed.
223 2018-11-14 Richard Biener <rguenther@suse.de>
225 PR tree-optimization/88019
226 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
227 COND_EXPR arguments to non-trapping overflow.
229 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
231 * config/arc/arc.h (reg_class): Reorder registers classes, remove
232 unused register classes.
233 (REG_CLASS_NAMES): Likewise.
234 (REG_CLASS_CONTENTS): Likewise.
235 (FIXED_REGISTERS): Make lp_count fixed.
236 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
237 (PROGRAM_COUNTER_REGNO): Remove.
238 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
239 register classes, use constants for register numbers, remove
240 reg_class_contents references.
241 (arc_process_double_reg_moves): Add asserts.
242 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
243 lra_in_progress predicate.
244 (arc_init_reg_tables): Remove unused register classes.
245 (arc_register_move_cost): Likewise.
246 (arc_preferred_reload_class): Likewise.
247 (hwloop_optimize): Update rtx patterns involving lp_count register.
248 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
249 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
250 CHEAP_CORE_REGS. Former one will be used for LRA.
251 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
252 one will be used for LRA.
253 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
254 one will be used for LRA.
255 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
256 Former one will be used for LRA.
257 ("f"): Delete constraint.
260 ("l"): Change it from register constraint to constraint.
261 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
262 (movhi_insn): Likewise.
263 (movsi_insn): Update pattern.
266 (stack_tie): Remove 'b' constraint letter.
268 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
269 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
270 (R57_REG, R59_REG, PCL_REG): Likewise.
271 (ILINK1_REGNUM): Renamed to ILINK1_REG.
272 (ILINK2_REGNUM): Renamed to ILINK2_REG.
275 (Rcw): Remove unused reg classes.
276 * config/arc/predicates.md (dest_reg_operand): Just default on
277 register_operand predicate.
278 (mpy_dest_reg_operand): Likewise.
279 (move_dest_operand): Use macros instead of constants.
280 * config/arc/arc.opt (mlra): Switch to lra as default.
282 2018-11-14 Richard Biener <rguenther@suse.de>
284 PR tree-optimization/87974
285 * tree-vect-loop.c (vectorizable_reduction): When computing
286 the vectorized reduction PHI vector type ignore constant
289 2018-11-14 Jakub Jelinek <jakub@redhat.com>
291 PR tree-optimization/87977
292 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
293 stmt, build a new one and replace the old one with it. Formatting fix.
294 Call release_ssa_name (x) if !has_other_use and !delete_div.
295 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
296 verify lhs of stmt is still def.
298 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
300 PR rtl-optimization/87507
301 * lower-subreg.c (operand_for_swap_move_operator): New function.
302 (simple_move): Strip simple operators.
303 (find_pseudo_copy): Likewise.
304 (resolve_operand_for_swap_move_operator): New function.
305 (resolve_simple_move): Strip simple operators and swap operands.
307 2018-11-13 Jakub Jelinek <jakub@redhat.com>
309 PR tree-optimization/87898
310 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
311 (ipa_simd_modify_function_body): Remove debug stmts where the first
312 argument was changed into a non-decl.
314 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
316 PR rtl-optimization/87899
317 * lra-lives.c (start_living): Update white space in comment.
318 (enum point_type): New.
319 (sparseset_contains_pseudos_p): New function.
320 (update_pseudo_point): Likewise.
321 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
322 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
323 handling. Move early exit after adding conflicts.
324 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
325 if regno is already live. Remove all handling of program points.
326 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
327 after adding conflicts. Remove all handling of program points and
328 ignore_reg_for_conflicts.
329 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
330 and do not guard call to mark_pseudo_live.
331 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
332 and do not guard call to mark_pseudo_dead.
333 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
334 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
335 Use new function update_pseudo_point. Handle register copies by
336 removing the source register from the live set. Handle INOUT operands.
337 Update to the next program point using the unused_set, dead_set and
339 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
341 2018-11-13 David Malcolm <dmalcolm@redhat.com>
343 * builtins.c: Replace "source_location" with "location_t".
344 * diagnostic-show-locus.c: Likewise.
345 * diagnostic.c: Likewise.
346 * dumpfile.c: Likewise.
347 * gcc-rich-location.h: Likewise.
348 * genmatch.c: Likewise.
349 * gimple.h: Likewise.
350 * gimplify.c: Likewise.
352 * input.h: Likewise. Eliminate the typedef.
353 * omp-expand.c: Likewise.
354 * selftest.h: Likewise.
355 * substring-locations.h (get_source_location_for_substring):
357 (get_location_within_string): ...this.
358 * tree-cfg.c: Replace "source_location" with "location_t".
359 * tree-cfgcleanup.c: Likewise.
360 * tree-diagnostic.c: Likewise.
361 * tree-into-ssa.c: Likewise.
362 * tree-outof-ssa.c: Likewise.
363 * tree-parloops.c: Likewise.
364 * tree-phinodes.c: Likewise.
365 * tree-phinodes.h: Likewise.
366 * tree-ssa-loop-ivopts.c: Likewise.
367 * tree-ssa-loop-manip.c: Likewise.
368 * tree-ssa-phiopt.c: Likewise.
369 * tree-ssa-phiprop.c: Likewise.
370 * tree-ssa-threadupdate.c: Likewise.
371 * tree-ssa.c: Likewise.
372 * tree-ssa.h: Likewise.
373 * tree-vect-loop-manip.c: Likewise.
375 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
377 * doc/extend.texi: Fix typo in the weakref description.
379 2018-11-13 Richard Biener <rguenther@suse.de>
381 PR tree-optimization/86991
382 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
383 group building until we have successfully detected the SLP
385 (vect_is_simple_reduction): Remove fixup code here.
387 2018-11-13 David Malcolm <dmalcolm@redhat.com>
389 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
390 (dump_gimple_stmt): Use it.
391 (dump_gimple_stmt_loc): Likewise.
392 (dump_gimple_expr): Likewise.
393 (dump_gimple_expr_loc): Likewise.
394 (dump_generic_expr): Likewise.
395 (dump_generic_expr_loc): Likewise.
396 (dump_printf): Likewise.
397 (dump_printf_loc): Likewise.
398 (dump_dec): Likewise.
399 (dump_dec): Likewise.
400 (dump_hex): Likewise.
401 (dump_symtab_node): Likewise.
402 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
403 Guard dump call with dump_enabled_p.
404 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
405 * graphite-optimize-isl.c (optimize_isl): Likewise.
406 * graphite.c (graphite_transform_loops): Likewise.
407 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
408 * tree-parloops.c (parallelize_loops): Likewise.
409 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
410 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
411 (vect_prune_runtime_alias_test_list): Likewise.
412 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
413 (vect_estimate_min_profitable_iters): Likewise.
414 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
415 (vect_build_slp_tree_2): Likewise.
416 (vect_supported_load_permutation_p): Likewise.
417 (vect_slp_analyze_operations): Likewise.
418 (vect_slp_analyze_bb_1): Likewise.
419 (vect_slp_bb): Likewise.
420 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
421 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
422 (pass_slp_vectorize::execute): Likewise.
423 (increase_alignment): Likewise.
425 2018-11-13 David Malcolm <dmalcolm@redhat.com>
428 * ipa-inline.c (report_inline_failed_reason): Guard calls to
429 cl_target_option_print_diff and cl_optimization_print_diff with
432 2018-11-13 David Malcolm <dmalcolm@redhat.com>
434 * doc/invoke.texi (-fsave-optimization-record): Note that the
435 output is compressed.
436 * optinfo-emit-json.cc: Include <zlib.h>.
437 (optrecord_json_writer::write): Compress the output.
439 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
441 * tree-vrp.c (value_range_base::dump): Dump type.
442 Do not use INF nomenclature for 1-bit types.
443 (dump_value_range): Group all variants to common dumping code.
444 (debug): New overloaded functions for value_ranges.
445 (value_range_base::dump): Remove no argument version.
446 (value_range::dump): Same.
448 2018-11-13 Richard Biener <rguenther@suse.de>
450 PR tree-optimization/87931
451 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
452 nested cycles we support to latch computations vectorizable_reduction
455 2018-11-13 Martin Liska <mliska@suse.cz>
457 PR tree-optimization/87885
458 * cfghooks.c (account_profile_record): Rename
460 (profile_record_check_consistency): ... this.
461 Calculate missing num_mismatched_freq_in.
462 (profile_record_account_profile): New function
463 that calculates time and size of a function.
464 * cfghooks.h (struct profile_record): Remove
466 (struct cfg_hooks): Remove after_pass flag.
467 (account_profile_record): Rename to ...
468 (profile_record_check_consistency): ... this.
469 (profile_record_account_profile): New.
470 * cfgrtl.c (rtl_account_profile_record): Remove
472 * passes.c (check_profile_consistency): Do only
474 (account_profile): Calculate size and time of
476 (pass_manager::dump_profile_report): Reformat
478 (execute_one_ipa_transform_pass): Call
479 consistency check before clean upand call account_profile
480 after a clean up is done.
481 (execute_one_pass): Call check_profile_consistency and
482 account_profile instead of using after_pass flag..
483 * tree-cfg.c (gimple_account_profile_record): Likewise.
485 2018-11-13 Martin Liska <mliska@suse.cz>
488 * config/i386/i386.c (ix86_option_override_internal): Error
489 about usage -mabi=ms and -fsanitize={,kernel-}address.
491 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
493 * config/aarch64/aarch64.c
494 (aarch64_vectorize_preferred_vector_alignment): Change return type to
496 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
497 alignment being a poly int.
498 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
499 return type to poly_uint64.
500 * target.def (default_preferred_vector_alignment): Likewise.
501 * targhooks.c (default_preferred_vector_alignment): Likewise.
502 * targhooks.h (default_preferred_vector_alignment): Likewise.
503 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
504 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
506 (vect_update_misalignment_for_peel): Likewise.
507 (vect_enhance_data_refs_alignment): Likewise.
508 (vect_find_same_alignment_drs): Likewise.
509 (vect_duplicate_ssa_name_ptr_info): Likewise.
510 (vect_setup_realignment): Likewise.
511 (vect_can_force_dr_alignment_p): Change alignment parameter type to
513 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
514 mask with a compile time variable vector alignment.
515 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
517 (vect_do_peeling): Exit early if vector alignment is not constant.
518 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
520 (vectorizable_store): Likewise.
521 (vectorizable_load): Likweise.
522 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
524 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
526 (vect_can_force_dr_alignment_p): Change alignment parameter type to
529 2018-11-13 Richard Biener <rguenther@suse.de>
531 PR tree-optimization/87962
532 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
533 detect outer reduction for disqualifying in-loop uses.
535 2018-11-13 Richard Biener <rguenther@suse.de>
537 PR tree-optimization/87967
538 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
539 for constants for the scalar loop.
541 2018-11-13 Alan Modra <amodra@gmail.com>
543 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
544 integer constants when -mcmodel=medium.
546 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
548 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
549 * config/arc/elf.h (PROFILE_HOOK): Define.
550 * config/arc/linux.h (PROFILE_HOOK): Likewise.
552 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
554 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
555 jump table data in the text section.
557 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
559 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
560 to fit the eh_return pattern.
561 * config/arc/arc.md (eh_return): Define.
562 (VUNSPEC_ARC_EH_RETURN): Likewise.
563 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
565 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
568 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
570 * common/config/arc/arc-common.c (arc_option_optimization_table):
571 Millicode optimization is default on for size optimizations.
572 * config/arc/arc-protos.h (arc_check_multi): New function.
573 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
574 (ENTER_LEAVE_START_REG): Define.
575 (ENTER_LEAVE_END_REG): Likewise.
576 (arc_override_options): Disable millicode when long calls option
578 (arc_frame_info): Change it from int to bool.
579 (arc_compute_frame_size): Clean up.
580 (arc_save_restore): Remove.
581 (frame_save_reg): New function.
582 (frame_restore_reg): Likewise.
583 (arc_enter_leave_p): Likewise.
584 (arc_save_callee_saves): Likewise.
585 (arc_restore_callee_saves): Likewise.
586 (arc_save_callee_enter): Likewise.
587 (arc_restore_callee_leave): Likewise.
588 (arc_save_callee_milli): Likewise.
589 (arc_restore_callee_milli): Likewise.
590 (arc_expand_prologue): Reimplement to emit enter/leave
592 (arc_expand_epilogue): Likewise.
593 (arc_check_multi): New function.
594 * config/arc/arc.md (push_multi_fp): New pattern.
595 (push_multi_fp_blink): Likewise.
596 (pop_multi_fp): Likewise.
597 (pop_multi_fp_blink): Likewise.
598 (pop_multi_fp_ret): Likewise.
599 (pop_multi_fp_blink_ret): Likewise.
600 * config/arc/arc.opt (mmillicode): Update option.
601 (mcode-density-frame): New option.
602 * config/arc/predicates.md (push_multi_operand): New predicate.
603 (pop_multi_operand): Likewise.
604 * doc/invoke.texi (ARC): Update ARC options information.
607 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
609 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
610 * config/arc/arc.c (operands_ok_ldd_std): New function.
611 (mem_ok_for_ldd_std): Likewise.
612 (gen_operands_ldd_std): Likewise.
613 * config/arc/arc.md: Add peephole2 rules for std/ldd.
615 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
617 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
618 test on stack_usage_file.
619 (lang_dependent_init): Do not open the .su file if generating LTO.
621 2018-11-13 Jakub Jelinek <jakub@redhat.com>
623 PR rtl-optimization/87918
624 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
625 simplify_gen_relational rather than simplify_gen_binary.
627 2018-11-13 Richard Biener <rguenther@suse.de>
629 * tree-ssanames.h (set_range_info): Use value_range_base.
630 (get_range_info): Likewise.
631 * tree-ssanames.c (set_range_info): Likewise.
632 (get_range_info): Likewise.
633 * tree-vrp.c (value_range_base::union_helper): Split
634 out common parts of value_range[_base]::union_.
635 (value_range_base::union_): Update.
636 (value_range::union_): Likewise.
637 (determine_value_range_1): Use value_range_base.
638 (determine_value_range): Likewise.
639 * tree-vrp.h (value_range_base::union_helper): Move ...
640 (value_range::union_helper): ... from here.
642 2018-11-13 Alan Modra <amodra@gmail.com>
644 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
646 (rs6000_secondary_reload_gpr): Don't call find_replacement.
648 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
651 * doc/invoke.texi (Optimize Options): Clarify that the
652 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
653 apply to data cache size.
655 2018-11-13 Alan Modra <amodra@gmail.com>
657 * config/rs6000/predicates.md (logical_const_operand),
658 (logical_operand): Correct comment.
659 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
661 2018-11-13 Alan Modra <amodra@gmail.com>
663 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
665 2018-11-13 Alan Modra <amodra@gmail.com>
667 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
669 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
670 covered by alternative.
671 (movcc_internal1): Ignore h for register preference.
672 (mov<mode>_hardfloat64): Likewise.
673 (mov<mode>_softfloat): Ignore c, l, h for register preference.
675 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
677 PR preprocessor/47823
678 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
681 * doc/extend.texi (Pragmas): Note additional pragmas documented
684 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
688 * tree-scalar-evolution.c (expression_expensive_p): Make
689 BUILTIN POPCOUNT as expensive when backend does not define it.
691 2018-11-12 Fredrik Noring <noring@nocrew.org>
693 * config.gcc: Update with-llsc defaults for MIPS r5900.
695 2018-11-12 Martin Liska <mliska@suse.cz>
697 PR gcov-profile/87442
698 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
700 * doc/invoke.texi: Document them.
701 * tree-profile.c (parse_profile_filter): New.
702 (parse_profile_file_filtering): Likewise.
703 (release_profile_file_filtering): Likewise.
704 (include_source_file_for_profile): Likewise.
705 (tree_profiling): Filter source files based on the
708 2018-11-12 Martin Liska <mliska@suse.cz>
711 * doc/extend.texi: Add missing values for __builtin_cpu_is and
712 __builtin_cpu_supports for x86 target.
714 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
717 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
719 (Arithmetic): Likewise for "lo_sum".
721 2018-11-12 Renlin Li <renlin.li@arm.com>
724 * dse.c (get_stored_val): Add check for compile-time constantness
727 2018-11-12 Sudakshina Das <sudi.das@arm.com>
729 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
730 (ARMv8_5a): New fgroup.
731 (armv8.5-a): New arch.
732 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
734 * config/arm/arm-tables.opt: Regenerate.
735 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
736 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
737 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
738 Add matching rules for -march=armv8.5-a and extensions.
739 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
740 Add sb and predres to all armv8-a except armv8.5-a.
742 2018-11-12 Richard Biener <rguenther@suse.de>
744 * tree-vrp.h (value_range[_base]::set): Make public. Provide
745 overload for single value.
746 (value_range[_base]::set_nonnull): New.
747 (value_range[_base]::set_null): Likewise.
748 (value_range): Document bitmap copying behavior, mark
749 copy constructor and assignment operator deleted.
750 (value_range::move): New.
751 (value_range::set_and_canonicalize): Default bitmap to zero.
752 (set_value_range_to_nonnull): Remove.
753 (set_value_range_to_null): Likewise.
754 (set_value_range): Likewise.
755 (set_value_range_to_value): Likewise.
756 (extract_range_from_unary_expr): Work on value_range_base.
757 (extract_range_from_binary_expr_1): Likewise. Rename to...
758 (extract_range_from_binary_expr): ... this.
759 * tree-vrp.c (value_range::update): Clear equiv bitmap
761 (value_range::move): New, move equiv bitmap.
762 (value_range_base::set_undefined): Avoid assignment.
763 (value_range::set_undefined): Likewise.
764 (value_range_base::set_varying): Likewise.
765 (value_range::set_varying): Likewise.
766 (set_value_range): Remove.
767 (value_range_base::set): New overload for value.
768 (value_range::set): Likewise.
769 (set_value_range_to_nonnull): Remove.
770 (value_range_base::set_nonnull): New.
771 (value_range::set_nonnull): Likewise.
772 (set_value_range_to_null): Remove.
773 (value_range_base::set_null): New.
774 (value_range::set_null): Likewise.
775 (range_is_null): Work on value_range_base.
776 (range_is_nonnull): Likewise.
777 (ranges_from_anti_range): Likewise.
778 (extract_range_into_wide_ints): Likewise.
779 (extract_range_from_multiplicative_op): Likewise.
780 (extract_range_from_binary_expr): Likewise. Update for API changes.
781 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
783 (value_range::intersect_helper): Avoid copy and assignment.
784 (value_range::union_helper): Likewise.
785 (determine_value_range_1): Adjust.
786 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
787 Avoid assignment by using move.
788 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
789 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
791 * tree-ssanames.c (get_range_info): Likewise.
792 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
793 * vr-values.c (vr_values::get_value_range): Adjust.
794 (vr_values::update_value_range): Likewise.
795 (symbolic_range_based_on_p): Work on value_range_base.
796 (vr_values::extract_range_from_binary_expr): Use value_range_base.
797 (vr_values::extract_range_from_unary_expr): Likewise.
798 (vr_values::extract_range_from_cond_expr): Avoid assignment.
799 (vr_values::extract_range_from_comparison): Adjust.
800 (vr_values::check_for_binary_op_overflow): Use value_range_base.
801 (vr_values::extract_range_basic): Adjust.
802 (vr_values::adjust_range_with_scev): Likewise.
803 (vr_values::vrp_visit_assignment_or_call): Likewise.
804 (vr_values::get_vr_for_comparison): Change API to avoid
805 assignment and copy construction.
806 (vr_values::compare_name_with_value): Adjust accordingly.
807 (vr_values::compare_names): Likewise.
808 (vr_values::extract_range_from_phi_node): Avoid assignment and
809 bogus in-place modify of equiv bitmap.
810 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
811 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
812 for extract_range_from_unary_expr API change.
813 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
815 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
817 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
819 2018-11-12 Richard Biener <rguenther@suse.de>
821 * tree-vrp.h (value_range_base::symbolic_p,
822 value_range_base::constant_p, value_range_base::zero_p,
823 value_range_base::singleton_p): Move from value_range.
824 (value_range::dump): Add.
825 * gimple-ssa-evrp-analyze.c
826 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
827 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
828 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
830 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
831 (value_range::constant_p): Likewise.
832 (value_range::singleton_p): Likewise.
833 (value_range_base::dump): Add.
834 (set_value_range_to_undefined): Remove.
835 (set_value_range_to_varying): Likewise.
836 (range_int_cst_p): Take value_range_base argument.
837 (range_int_cst_singleton_p): Likewise.
838 (value_range_constant_singleton): Likewise.
839 (vrp_set_zero_nonzero_bits): Likewise.
840 (extract_range_from_multiplicative_op): Use set_varying.
841 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
842 (extract_range_from_unary_expr): Likewise.
843 (dump_value_range_base): Change to overload of dump_value_range.
844 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
845 (vrp_prop::visit_stmt): Likewise.
846 (value_range::intersect_helper): Likewise.
847 (value_range::union_helper): Likewise.
848 (determine_value_range_1): Likewise.
850 2018-11-12 Richard Biener <rguenther@suse.de>
852 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
853 (set_value_range_to_null): Likewise.
854 * vr-values.c (vr_values::extract_range_from_comparison):
855 Clear equiv for constant singleton ranges.
857 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
859 * config/i386/sse.md: Combine VFIXUPIMM* patterns
860 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
861 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
862 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
863 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
864 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
865 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
867 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
870 * doc/extend.texi (Common Type Attributes): For the align type
871 attribute, copy language about decreasing alignment from the
872 corresponding variable attribute.
874 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
876 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
879 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
882 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
883 -frtti and -fno-rtti code.
885 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
888 * doc/extend.texi (Other Builtins): Document probability associated
889 with __builtin_expect.
891 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
894 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
895 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
896 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
897 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
899 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
901 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
903 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
905 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
907 2018-11-11 Richard Biener <rguenther@suse.de>
909 * tree-vrp.h (class value_range_base): New base class for
910 value_range containing all but the m_equiv member.
911 (dump_value_range_base): Add.
912 (range_includes_zero_p): Work on value_range_base.
913 * tree-vrp.c (value_range_base::set): Split out base handling
915 (value_range::set): this.
916 (value_range::set_equiv): New.
917 (value_range_base::value_range_base): New constructors.
918 (value_range_base::check): Split out base handling from...
919 (value_range::check): this.
920 (value_range::equal_p): Refactor in terms of
921 ignore_equivs_equal_p which is now member of the base.
922 (value_range_base::set_undefined): New.
923 (value_range_base::set_varying): Likewise.
924 (value_range_base::dump):Split out base handling from...
925 (value_range::dump): this.
926 (value_range_base::set_and_canonicalize): Split out base handling
928 (value_range::set_and_canonicalize): this.
929 (value_range_base::union_): New.
930 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
932 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
933 instead of value_range everywhere.
934 (ipcp_vr_lattice::print): Use dump_value_range_base.
935 (ipcp_vr_lattice::meet_with): Adjust.
936 (ipcp_vr_lattice::meet_with_1): Likewise.
937 (ipa_vr_operation_and_type_effects): Likewise.
938 (propagate_vr_across_jump_function): Likewise.
939 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
940 (ipa_get_value_range): Likewise.
941 (ipa_set_jfunc_vr): Likewise.
942 (ipa_compute_jump_functions_for_edge): Likewise.
944 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
947 * doc/invoke.texi (Optimize Options): Add @opindex entries
948 for the positive forms of -fno-xxx and -mno-xxx options
949 that were lacking them.
951 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
953 * combine.c (make_more_copies): Only make an intermediate copy if the
954 dest of a move is a pseudo.
956 2018-11-09 Maya Rashish <coypu@sdf.org>
959 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
960 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
962 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
966 * doc/invoke.texi (Optimize Options): Clarify default behavior
967 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
969 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
971 PR tree-optimization/87940
972 * expr.c (string_constant): Don't strip NOPS in subexpressions.
973 Fold PLUS_EXPR correctly.
975 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
978 * config/s390/s390.c (s390_safe_relative_long_p): New function.
979 (annotate_constant_pool_refs): Skip insns which support
981 (annotate_constant_pool_refs_1): New helper function.
982 (find_constant_pool_ref): Skip insns which support relative
984 (find_constant_pool_ref_1): New helper function.
985 (replace_constant_pool_ref): Skip insns which support
987 (replace_constant_pool_ref_1): New helper function.
988 (s390_mainpool_start): Adapt to the new signature.
989 (s390_mainpool_finish): Likewise.
990 (s390_chunkify_start): Likewise.
991 (s390_chunkify_finish): Likewise.
992 (pass_s390_early_mach::execute): Likewise.
993 (s390_prologue_plus_offset): Likewise.
994 (s390_emit_prologue): Likewise.
995 (s390_emit_epilogue): Likewise.
997 2018-11-09 Jakub Jelinek <jakub@redhat.com>
999 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
1000 but unsupported lastprivate with conditional modifier.
1002 2018-11-09 Jeff Law <law@redhat.com>
1004 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
1005 unused argument better. Add gcc_unreachable to silence warning.
1007 2018-11-09 Martin Sebor <msebor@redhat.com>
1010 * attribs.c (has_attribute): New helper function.
1011 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
1012 * attribs.h (decls_mismatched_attributes): Declare.
1013 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
1014 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
1015 * common.opt (-Wattribute-alias): Take an argument.
1016 (-Wno-attribute-alias): New option.
1017 * doc/extend.texi (Common Function Attributes): Document copy.
1018 (Common Variable Attributes): Same.
1019 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
1020 (-Wattribute-alias): Document new option argument.
1022 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
1024 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
1025 matching criteria. Remove unused array initializer.
1027 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
1028 Jinsong Ji <jji@us.ibm.com>
1030 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
1031 constraints by introducing a new temporary.
1032 (_mm_cvtss_si64): Likewise.
1034 2018-11-09 Martin Liska <mliska@suse.cz>
1036 * common.opt: Add -fipa-stack-alignment flag.
1037 * doc/invoke.texi: Document it.
1038 * final.c (rest_of_clean_state): Guard stack
1039 shrinking with flag.
1041 2018-11-09 Martin Liska <mliska@suse.cz>
1043 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
1045 (ipa_discover_variable_flags): ... this.
1046 * common.opt: Come up with new flag -fipa-reference-addressable.
1047 * doc/invoke.texi: Document it.
1048 * ipa-reference.c (propagate): Call the renamed fn.
1049 * ipa-visibility.c (whole_program_function_and_variable_visibility):
1051 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
1053 (ipa_discover_variable_flags): ... this. Discover
1054 non-addressable variables only with the newly added flag.
1055 * opts.c: Enable the newly added flag with -O1 and higher
1058 2018-11-09 David Malcolm <dmalcolm@redhat.com>
1060 * json.cc (selftest::test_writing_literals): Fix comment.
1062 2018-11-09 Martin Liska <mliska@suse.cz>
1064 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
1065 string to a stack buffer.
1066 (aarch64_parse_cpu): Likewise.
1067 (aarch64_parse_tune): Likewise.
1069 2018-11-09 Richard Biener <rguenther@suse.de>
1071 PR tree-optimization/87953
1072 * tree-vect-loop.c (vectorizable_reduction): For analysis
1073 always pass ops[0] to vectorizable_condition.
1075 2018-11-09 Stafford Horne <shorne@gmail.com>
1076 Richard Henderson <rth@twiddle.net>
1077 Joel Sherrill <joel@rtems.org>
1079 * common/config/or1k/or1k-common.c: New file.
1080 * config/or1k/*: New.
1081 * config.gcc (or1k*-*-*): New.
1082 * configure.ac (or1k*-*-*): New test for openrisc tls.
1083 * configure: Regenerated.
1084 * doc/install.texi: Document OpenRISC triplets.
1085 * doc/invoke.texi: Document OpenRISC arguments.
1086 * doc/md.texi: Document OpenRISC.
1088 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
1090 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
1091 (arm7tdmi-s): Delete CPU.
1092 (arm710t): Add aliases for arm720t and arm740t.
1093 (arm720t, arm740t): Delete CPUs.
1094 (arm920t): Add aliases for arm920, arm922t and arm940t.
1095 (arm920, arm922t, arm940t): Delete CPUs.
1096 (arm10tdmi): Add alias for arm1020t.
1097 (arm1020t): Delete CPU.
1098 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
1099 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
1100 (arm10e): Add aliases for arm1020e and arm1022e.
1101 (arm1020e, arm1022e): Delete CPU.
1102 * config/arm/arm.md (generic_sched): Remove entries that are now
1104 (generic_vfp): Likewise.
1105 * config/arm/arm1020e.md: Simplify tuning selection based on alias
1107 * config/arm/arm-tune.md: Regenerated.
1108 * config/arm/arm-tables.opt: Regenerated.
1110 2018-11-09 Richard Biener <rguenther@suse.de>
1112 PR tree-optimization/87621
1113 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
1114 op with only phi inputs.
1115 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
1116 (ch_base::copy_headers): Run CSE on copied loop headers.
1117 (pass_ch_vect::process_loop_p): Simplify.
1119 2018-11-09 Alexandre Oliva <oliva@adacore.com>
1121 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
1122 for non-w64 x86_64 biarch.
1124 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
1126 PR rtl-optimization/86438
1127 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
1128 of in_b for the compare if in_b is SET_DEST.
1131 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
1132 non-toplevel UNSPEC.
1134 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1136 * tree-vrp.c (value_range::check): Do not access internals
1138 (value_range::singleton_p): Same.
1139 (value_range::type): Same.
1140 (vrp_finalize): Use value_range API.
1142 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1144 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
1146 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1148 * vr-values.c (vr_values::get_value_range): Use value_range API
1149 instead of piecing together ranges.
1150 (vr_values::update_value_range): Same.
1152 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1154 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
1155 of performing ad-hoc calculations.
1156 * tree-ssanames.c (set_range_info): New overloaded function
1157 accepting value_range &.
1158 (get_range_info): Same.
1159 * tree-ssanames.h (set_range_info_raw): Remove.
1160 (set_range_info): New prototype.
1161 (get_range_info): Same.
1162 * tree-vrp.h (value_range::null_p): Rename to zero_p.
1163 * tree-vrp.c (value_range::null_p): Same.
1165 2018-11-09 Jan Hubicka <jh@suse.cz>
1167 * tree.c (fld_type_variant_equal_p): Test user align flag.
1168 (flt_type_variant): Copy user align flag.
1169 (fld_incomplete_type_of): Clear it.
1171 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1173 * config/arm/neon.md (div<mode>3): New pattern.
1175 2018-11-08 Andi Kleen <ak@linux.intel.com>
1177 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
1178 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
1179 (ix86_handle_option): Handle OPT_mptwrite.
1180 * config/i386/cpuid.h (bit_PTWRITE): Add.
1181 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
1182 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
1183 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
1184 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
1185 (ix86_option_override_internal): Handle PTA_PTWRITE.
1186 (ix86_valid_target_attribute_inner_p): Define ptwrite.
1187 (def_builtin2): Force UINT64 to be 64bit only.
1188 * config/i386/i386.h (TARGET_PTWRITE): Add.
1189 (TARGET_PTWRITE_P): Add.
1191 * config/i386/i386.md: Define ptwrite.
1192 * config/i386/i386.opt: Add -mptwrite.
1193 * config/i386/immintrin.h (_ptwrite64): Add.
1195 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
1196 * doc/invoke.texi: Document -mptwrite.
1198 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
1200 PR rtl-optimization/87600
1201 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
1202 * lra-constraints.c (process_alt_operands): Skip illegal hard
1203 register usage. Prefer reloading non hard register operands.
1205 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
1208 * doc/invoke.texi (Optimize Options): Clarify default behavior
1209 for -fno-sched-interblock and -fno-sched-spec.
1211 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
1213 * collect2.c (linker_select): Add USE_LLD_LD.
1214 (ld_suffixes): Add ld.lld.
1215 (main): Handle -fuse-ld=lld.
1216 * common.opt (-fuse-ld=lld): New option.
1217 * doc/invoke.texi (-fuse-ld=lld): Document.
1218 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
1220 2018-11-08 Paul Koning <ni1d@arrl.net>
1222 * config/pdp11/constraints.md: Add "Z" series constraints for use
1223 with pre-dec and post-inc addressing.
1224 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
1225 (pdp11_expand_operands): Add int argument (word count).
1226 (pdp11_sp_frame_offset): Delete.
1227 (pdp11_cmp_length): New function.
1228 (pushpop_regeq): New function.
1229 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
1231 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
1233 (pdp11_initial_elimination_offset): Ditto.
1234 (pdp11_expand_operands): Add word count argument. Bugfixes.
1235 (output_move_multiple): Change how pointer adjustment is done.
1236 (pdp11_gen_int_label): Correct format.
1237 (output_ascii): Ditto.
1238 (pdp11_asm_output_var): Add code for DEC assembler case.
1239 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
1241 (legitimate_const_double_p): Ditto.
1242 (pdp11_register_move_cost): Adjust for new register classes.
1243 (pdp11_regno_reg_class): Ditto.
1244 (expand_block_move): Delete.
1245 (pushpop_regeq): New function.
1246 (pdp11_legitimate_address_p): Bugfix in check for constant
1248 (pdp11_sp_frame_offset): Delete.
1249 (pdp11_reg_save_size): New helper function for new frame layout.
1250 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
1251 (pdp11_expand_shift): Bugfix in check for constant shift count.
1252 (pdp11_shift_length): Ditto.
1253 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
1254 (pdp11_cmp_length): New function.
1255 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
1256 some compile options.
1257 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
1258 (CALL_USED_REGISTERS): Ditto.
1259 (ELIMINABLE_REGS): Ditto.
1260 (REGISTER_NAMES): Ditto.
1261 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
1263 (REG_CLASS_NAMES): Ditto.
1264 (REG_CLASS_CONTENTS): Ditto. Also remove
1265 HARD_FRAME_POINTER_REGNUM.
1266 (CPU_REG_CLASS): New macro.
1267 (CLASS_MAX_NREGS): Adjust for new register classes.
1268 (FUNCTION_PROFILER): Make no-op.
1269 (may_call_alloca): Remove unused declaration.
1270 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
1271 (ASM_OUTPUT_SKIP): Fix format.
1272 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
1273 (HARD_FRAME_POINTER_REGNUM): Remove.
1275 (*rts): Rename. Remove epilogue related checks.
1276 (cmpsi, cmpdi): New insn.
1277 (cbranch<mode>4): Change to apply to SI and DI modes as well.
1278 (mov<mode>): Change constraints to enforce that push/pop
1279 destination cannot use the same register as source.
1280 (*mov<mode><cc_cc>): Ditto.
1281 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
1282 at assembly output rather than as RTL expander.
1283 (zero_extendqihi2): Bugfix in check for same registers.
1284 (adddi3_nocc): Bugfix in check for constant operand.
1285 (addsi3_nocc): Ditto.
1286 (subdi3_nocc): Ditto.
1287 (subsi3_nocc): Ditto.
1288 (negdi2_nocc): Copy input to pdp11_expand_operands.
1289 (negsi2_nocc): Ditto.
1290 (bswap2_nocc): Ditto.
1291 * config/pdp11/pdp11.opt (mlra): Fix documentation.
1292 * config/pdp11/t-pdp11: Use -Os.
1294 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
1296 * config/arm/parsecpu.awk (/alias/): New parsing rule.
1297 (/begin cpu/): Check that the cpu name hasn't been previously defined.
1298 (gen_comm_data): Print out CPU alias tables.
1299 (check_cpu): Match aliases when checking the CPU name.
1300 * config/arm/arm-protos.h (cpu_alias): New structure.
1301 (cpu_option): Add entry for aliases.
1302 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
1303 strongarm1100 and strongarm1110.
1304 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
1305 (config/arm/arm-generic.md): Remove redundant references to
1306 strongarm110, strongarm1100 and strongarm1110.
1307 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
1308 Scan aliases for additional hints.
1309 (arm_parse_cpu_option_name): Also match a cpu name against the list
1311 * config/arm/arm-tables.opt: Regenerated.
1312 * config/arm/arm-tune.md: Regenerated.
1314 2018-11-08 Jakub Jelinek <jakub@redhat.com>
1316 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
1317 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
1318 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
1319 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
1320 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
1321 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
1322 * gengtype.c (open_base_files): Add omp-general.h.
1323 * gimple.c (gimple_build_omp_critical):
1324 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
1325 gimple_omp_taskgroup_set_clauses.
1326 (gimple_build_omp_atomic_load): Add mo argument, call
1327 gimple_omp_atomic_set_memory_order.
1328 (gimple_build_omp_atomic_store): Likewise.
1329 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
1330 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
1332 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
1333 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
1334 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
1335 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
1336 different value for GF_OMP_ATOMIC_NEED_VALUE.
1337 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
1339 (struct gimple_statement_omp_single_layout): And remove here.
1340 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
1341 than gimple_statement_omp_single_layout.
1342 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
1344 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
1345 (gimple_omp_subcode): Formatting fix.
1346 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
1347 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
1348 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
1349 gimple_omp_teams_host, gimple_omp_teams_set_host,
1350 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
1351 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
1352 gimple_omp_taskgroup_set_clauses): New inline functions.
1353 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
1354 (gimple_build_omp_atomic_store): Likewise.
1355 (gimple_omp_atomic_seq_cst_p): Remove.
1356 (gimple_omp_atomic_memory_order): New function.
1357 (gimple_omp_atomic_set_seq_cst): Remove.
1358 (gimple_omp_atomic_set_memory_order): New function.
1359 (gimple_build_omp_taskgroup): Add clauses argument.
1360 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
1361 (dump_gimple_omp_task): Print taskwait with depend clauses.
1362 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
1363 dump_omp_atomic_memory_order.
1364 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
1365 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
1366 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
1367 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
1368 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
1369 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
1370 ORT_UNTIED_TASKLOOP enumerators.
1371 (enum gimplify_defaultmap_kind): New.
1372 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
1373 target_map_pointers_as_0len_arrays members, add defaultmap.
1374 (new_omp_context): Initialize defaultmap member.
1375 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1376 (maybe_fold_stmt): Don't fold even in host teams regions.
1377 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
1378 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
1379 ctx->omp_firstprivatize_variable.
1380 (omp_add_variable): Don't add private/firstprivate for VLAs in
1382 (omp_default_clause): Print "taskloop" rather than "task" if
1384 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1385 Handle new defaultmap clause kinds.
1386 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
1387 iterator to be lastprivate or private. Fix up diagnostics if linear
1388 is used on collapse>1 simd iterator.
1389 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1390 (gimplify_omp_depend): New function.
1391 (gimplify_scan_omp_clauses): Add shared clause on parallel for
1392 combined parallel master taskloop{, simd} if taskloop has
1393 firstprivate, lastprivate or reduction clause. Handle
1394 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
1395 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
1396 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
1397 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
1398 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
1400 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
1401 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
1402 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
1403 OMP_CLAUSE_{TASK,IN}_REDUCTION.
1404 (gimplify_omp_task): Handle taskwait with depend clauses.
1405 (gimplify_omp_for): Add shared clause on parallel for combined
1406 parallel master taskloop{, simd} if taskloop has firstprivate,
1407 lastprivate or reduction clause. Use ORT_TASKLOOP or
1408 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
1409 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
1410 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
1411 __for_end and __for_range temporaries on OMP_PARALLEL for
1412 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
1413 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
1414 sandwiched in between two taskloops.
1415 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
1416 instead of ctx->omp_firstprivatize_variable.
1417 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
1418 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
1419 host teams, use gimplify_and_return_first etc. for body like
1420 for target or target data constructs, and at the end call
1421 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
1422 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
1423 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
1424 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
1425 gimple_omp_atomic_set_seq_cst calls.
1426 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
1427 case, handle taskgroup clauses.
1428 * lto-streamer-out.c (hash_tree): Handle
1429 OMP_CLAUSE_{TASK,IN}_REDUCTION.
1430 * Makefile.in (GTFILES): Add omp-general.h.
1431 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
1432 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
1433 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
1434 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
1435 BUILT_IN_GOMP_LOOP_DOACROSS_START,
1436 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
1437 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
1438 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
1439 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
1440 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
1441 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
1442 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
1443 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
1444 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
1445 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
1446 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
1447 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
1448 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
1449 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
1450 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
1451 * omp-expand.c (workshare_safe_to_combine_p): Return false for
1452 non-worksharing loops.
1453 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
1454 (determine_parallel_type): Don't combine parallel with worksharing
1455 which has _reductemp_ clause.
1456 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
1457 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
1458 if there is nonmonotonic modifier or if there is no modifier and no
1459 ordered clause. For dynamic and guided schedule without monotonic
1460 and nonmonotonic modifier, default to nonmonotonic.
1461 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
1462 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
1464 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
1465 there are any reduction clauses.
1466 (expand_taskwait_call): New function.
1467 (expand_teams_call): New function.
1468 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
1469 expand_teams_call for it. Formatting fix. Handle taskwait with
1471 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
1472 of worksharing loops with task reductions.
1473 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
1474 expansion of worksharing loops with task reductions.
1475 (expand_omp_sections): Handle expansion of sections with task
1477 (expand_omp_synch): For host teams call expand_omp_taskreg.
1478 (omp_memory_order_to_memmodel): New function.
1479 (expand_omp_atomic_load, expand_omp_atomic_store,
1480 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
1481 instead of gimple_omp_atomic_seq_cst_p.
1482 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
1483 depend clauses as a standalone directive.
1484 * omp-general.c (enum omp_requires): New variable.
1485 (omp_extract_for_data): Initialize have_reductemp member. Allow
1486 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
1487 GT_EXPR loops depending on incr sign. Formatting fixes.
1488 * omp-general.h (struct omp_for_data): Add have_reductemp member.
1489 (enum omp_requires): New enum.
1490 (omp_requires_mask): Declare.
1491 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
1493 * omp-low.c (struct omp_context): Add task_reductions and
1494 task_reduction_map fields.
1495 (is_host_teams_ctx): New function.
1496 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
1497 (use_pointer_for_field): Use is_global_var instead of
1498 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
1500 (build_outer_var_ref): Ignore taskgroup outer contexts.
1501 (delete_omp_context): Release task_reductions and task_reduction_map.
1502 (scan_sharing_clauses): Don't add any fields for reduction clause on
1503 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
1504 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
1505 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
1506 Handle OMP_CLAUSE_NONTEMPORAL.
1507 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
1509 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
1510 clauses with task modifier.
1511 (scan_omp_task): Handle taskwait with depend clauses.
1512 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
1513 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
1514 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
1515 (scan_omp_for): Fix comment formatting.
1516 (scan_omp_teams): Handle host teams constructs.
1517 (check_omp_nesting_restrictions): Allow teams with no outer
1518 OpenMP context. Adjust diagnostics for teams strictly nested into
1519 some explicit OpenMP construct other than target. Allow OpenMP atomics
1520 inside of simd regions.
1521 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
1522 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
1523 taskreg_nesting_level while scanning host teams construct.
1524 (task_reduction_read): New function.
1525 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
1526 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
1527 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
1528 second argument create_tmp_var if it is NULL. Don't ignore shared
1529 clauses in is_host_teams_ctx contexts. Handle
1530 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
1532 (lower_reduction_clauses): Ignore reduction clauses with task
1533 modifier. Remove second argument create_tmp_var if it is NULL.
1534 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
1535 (lower_send_clauses): Ignore reduction clauses with task modifier.
1536 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
1537 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
1538 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
1539 rely that it is the last stmt in body so far. Ignore outer taskgroup
1541 (omp_task_reductions_find_first, omp_task_reduction_iterate,
1542 lower_omp_task_reductions): New functions.
1543 (lower_omp_sections): Handle reduction clauses with taskgroup
1544 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
1545 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
1546 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
1548 (lower_omp_taskgroup): Handle taskgroup reductions.
1549 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
1550 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
1551 (lower_depend_clauses): If there are any
1552 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
1553 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
1554 seen, assume lowering is done already and return early. Set kind
1555 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
1556 (lower_omp_taskreg): Handle reduction clauses with task modifier on
1557 parallel construct. Handle reduction clause on taskloop construct.
1558 Handle taskwait with depend clauses.
1559 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
1560 for host teams constructs.
1561 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
1562 nontemporal and _reductemp_ clause entries.
1563 (omp_clause_code_name): Likewise.
1564 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
1565 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
1566 * tree-core.h (enum omp_clause_code): Add
1567 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
1568 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
1569 (struct tree_base): Add omp_atomic_memory_order field into union.
1570 Remove OMP_ATOMIC_SEQ_CST comment.
1571 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
1572 and OMP_CLAUSE_DEPEND_DEPOBJ.
1573 (struct tree_omp_clause): Add subcode.defaultmap_kind.
1574 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
1575 OpenMP constructs with body and clauses operands.
1576 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
1577 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
1578 (OMP_TASKGROUP_CLAUSES): Define.
1579 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
1580 OMP_CLAUSE__LOOPTEMP_.
1581 (OMP_ATOMIC_SEQ_CST): Remove.
1582 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
1583 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
1584 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
1585 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
1586 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
1587 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
1588 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
1589 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
1590 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
1591 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
1593 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
1594 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
1595 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
1596 (convert_local_omp_clauses): Likewise. Remove useless test.
1597 * tree-parloops.c (create_call_for_reduction_1): Pass
1598 OMP_MEMORY_ORDER_RELAXED as new argument to
1599 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
1600 * tree-pretty-print.c (dump_omp_iterators): New function.
1601 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
1602 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
1603 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
1604 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
1605 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
1606 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
1607 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
1608 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
1609 (dump_omp_atomic_memory_order): New function.
1610 (dump_generic_node): Use it. Print taskgroup clauses. Print
1611 taskwait with depend clauses.
1612 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
1613 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
1614 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
1615 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
1616 write_ts_omp_clause_tree_pointers): Likewise.
1618 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1621 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
1622 "-missed", "-note", and "-all" sub-options.
1623 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
1625 (can_early_inline_edge_p): Likewise.
1626 (want_early_inline_function_p): Likewise.
1627 (want_inline_self_recursive_call_p): Likewise.
1628 (recursive_inlining): Likewise.
1629 (inline_small_functions): Likewise.
1630 (flatten_function): Likewise.
1631 (ipa_inline): Likewise.
1632 (inline_always_inline_functions): Likewise.
1633 (early_inline_small_functions): Likewise.
1634 (early_inliner): Likewise.
1635 * tree-inline.c (expand_call_inline): Likewise.
1637 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1639 * pretty-print.c (pp_format): Handle %f.
1640 (selftest::test_pp_format): Add test of %f.
1641 * pretty-print.h (pp_double): New macro.
1643 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1645 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
1646 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
1647 (dump_pretty_printer::decode_format): Implement "%C" for
1649 (selftest::test_capture_of_dump_calls): Rename "where" to
1650 "stmt_loc". Convert test_decl to a function decl and set its
1651 location. Add a symbol_table_test RAII instance and a
1652 cgraph_node, using it to test "%C" and dump_symtab_node.
1654 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
1657 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
1659 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1661 * cgraph.c: Include "selftest.h".
1662 (saved_symtab): New variable.
1663 (selftest::symbol_table_test::symbol_table_test): New ctor.
1664 (selftest::symbol_table_test::~symbol_table_test): New dtor.
1665 (selftest::test_symbol_table_test): New test.
1666 (selftest::cgraph_c_tests): New.
1667 * cgraph.h (saved_symtab): New decl.
1668 (selftest::symbol_table_test): New class.
1669 * selftest-run-tests.c (selftest::run_tests): Call
1670 selftest::cgraph_c_tests.
1671 * selftest.h (selftest::cgraph_c_tests): New decl.
1673 2018-11-08 Richard Biener <rguenther@suse.de>
1675 * tree-data-ref.h (lambda_int): New typedef.
1676 (lambda_vector_gcd): Adjust.
1677 (lambda_vector_new): Likewise.
1678 (lambda_matrix_new): Likewise.
1679 * tree-data-ref.c (print_lambda_vector): Adjust.
1681 2018-11-08 Richard Biener <rguenther@suse.de>
1683 PR tree-optimization/87929
1684 * tree-complex.c (expand_complex_comparison): Clean EH.
1686 2018-11-08 Martin Liska <mliska@suse.cz>
1688 * doc/extend.texi: Reword.
1689 * predict.c (expr_expected_value_1): Likewise.
1691 2018-11-08 Richard Biener <rguenther@suse.de>
1693 PR tree-optimization/87913
1694 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
1695 of extreme values to ordered comparisons.
1697 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1700 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
1702 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1704 * doc/invoke.texi: Remove leading dash from @opindex entries
1705 throughout the file.
1707 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1710 * doc/invoke.texi (Option Summary): Add -e and --entry.
1711 (Link Options): Likewise.
1713 2018-11-07 Nathan Sidwell <nathan@acm.org>
1716 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
1717 --disable-checking bootstrap.
1719 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
1721 * configure: Regenerated.
1723 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1726 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
1727 to the mode of the widest field iff the widest field has mode class
1728 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
1731 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
1733 * common.opt: Add -Wattribute-warning.
1734 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
1735 * expr.c (expand_expr_real_1): Add new attribute to warning_at
1736 call to allow user configure behavior of "warning" attribute.
1738 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1740 * target.def: Put @: after every vs., e.g., and i.e. where it is
1741 followed by whitespace.
1742 * doc/extend.texi: Ditto.
1743 * doc/fragments.texi: Ditto.
1744 * doc/gimple.texi: Ditto.
1745 * doc/implement-c.texi: Ditto.
1746 * doc/install.texi: Ditto.
1747 * doc/invoke.texi: Ditto.
1748 * doc/md.texi: Ditto.
1749 * doc/plugins.texi: Ditto.
1750 * doc/rtl.texi: Ditto.
1751 * doc/sourcebuild.texi: Ditto.
1752 * doc/tm.texi.in: Ditto.
1753 * doc/ux.texi: Ditto.
1754 * doc/tm.texi: Regenerate.
1756 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1758 * config/arm/arm-cpus.in (ares): New entry.
1759 * config/arm/arm-tables.opt: Regenerate.
1760 * config/arm/arm-tune.md: Likewise.
1761 * doc/invoke.texi (ARM Options): Document ares.
1763 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1765 * config/aarch64/aarch64-cores.def (ares): Define.
1766 * config/aarch64/aarch64-tune.md: Regenerate.
1767 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
1769 2018-11-07 Jan Hubicka <jh@suse.cz>
1771 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
1772 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
1773 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
1776 2018-11-07 Richard Biener <rguenther@suse.de>
1778 PR tree-optimization/87914
1779 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
1781 (vectorizable_reduction): Handle shifts and rotates by dispatching
1782 to vectorizable_shift.
1783 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
1784 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
1786 (vectorizable_shift): Export and handle being called as
1788 (vect_analyze_stmt): Call vectorizable_shift after
1789 vectorizable_reduction.
1790 * tree-vectorizer.h (vectorizable_shift): Declare.
1792 2018-11-07 Jan Hubicka <jh@suse.cz>
1794 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
1795 than const decls in TREE_VALUE of enum.
1796 (dump_type_inheritance_graph): Improve duplicate dumping.
1797 (free_enum_values): New.
1798 (build_type_inheritance_graph): Use it.
1799 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
1800 which are not main variants or not ODR types.
1801 (verify_type_variant): Expect variants to have no TYPE_VALUES.
1803 2018-11-07 Richard Biener <rguenther@suse.de>
1805 * ipa-inline.c (want_inline_small_function_p): Compute
1806 big_speedup_p lazily and last.
1808 2018-11-07 Jan Hubicka <jh@suse.cz>
1810 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
1811 building incomplete variant of complete type.
1812 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
1813 variant of complete type.
1815 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1817 * config/mips/mips.c: Fix typo in documentation of
1818 mips_loongson_ext2_prefetch_cookie.
1819 (mips_option_override): fix brain twister logical.
1820 * config/mips/mips.h: Fix typo in documentation of
1821 ISA_HAS_CTZ_CTO and define pattern.
1822 * config/mips/mips.md (prefetch): Hoist EXT2 above
1824 (prefetch_indexed): Hoist EXT2 above the EXT block.
1826 2018-11-07 Jan Hubicka <jh@suse.cz>
1828 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
1829 return and parameter types of function and method types.
1830 (free_lang_data_in_cgraph): Update.
1832 2018-11-07 Martin Liska <mliska@suse.cz>
1834 PR rtl-optimization/87868
1835 * postreload-gcse.c (eliminate_partially_redundant_load): Set
1836 threshold to max_count if we would overflow.
1837 * profile-count.h: Make max_count a public constant.
1839 2018-11-07 Martin Liska <mliska@suse.cz>
1841 * mem-stats.h: Fix GNU coding style.
1843 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1845 * config/mips/gs264e.md: New.
1846 * config/mips/mips-cpus.def: Define gs264e.
1847 * config/mips/mips-tables.opt: Regenerate.
1848 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
1850 (mips_issue_rate): Add support for gs264e.
1851 (mips_multipass_dfa_lookahead): Likewise.
1852 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
1853 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
1854 (MIPS_ASE_MSA_SPEC): New.
1855 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
1856 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
1857 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
1858 * config/mips/mips.md: Include gs264e.md.
1859 (processor): Add gs264e.
1860 * config/mips/mips.opt (MSA): Use Mask instead of Var.
1861 * doc/invoke.texi: Add gs264e to supported architectures.
1863 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1865 * config/mips/gs464e.md: New.
1866 * config/mips/mips-cpus.def: Define gs464e.
1867 * config/mips/mips-tables.opt: Regenerate.
1868 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
1870 (mips_issue_rate): Add support for gs464e.
1871 (mips_multipass_dfa_lookahead): Likewise.
1872 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
1873 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
1874 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
1875 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
1876 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
1877 * config/mips/mips.md: Include gs464e.md.
1878 (processor): Add gs464e.
1879 * doc/invoke.texi: Add gs464e to supported architectures.
1881 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1883 * config/mips/loongson3a.md: Rename to ...
1884 * config/mips/gs464.md: ... here.
1885 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
1886 as an alias of gs464 processor.
1887 * config/mips/mips-tables.opt: Regenerate.
1888 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
1889 instead of PROCESSOR_LOONGSON_3A.
1890 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
1892 (mips_option_override): Enable MMI and EXT for gs464.
1893 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
1894 Rename TUNE_LOONGSON_3A to TUNE_GS464.
1895 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
1896 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
1897 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
1899 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
1900 (processor): Add gs464;
1901 * doc/invoke.texi: Add gs464 to supported architectures.
1903 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1905 * config/mips/mips-protos.h
1906 (mips_loongson_ext2_prefetch_cookie): New prototype.
1907 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
1908 (mips_option_override): Enable TARGET_LOONGSON_EXT when
1909 TARGET_LOONGSON_EXT2 is true.
1910 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
1911 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
1912 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
1913 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
1914 TARGET_LOONGSON_EXT2.
1915 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
1916 (define_insn "ctz<mode>2"): New insn pattern.
1917 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
1918 (define_insn "prefetch_indexed_<mode>"): Include
1919 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
1920 * config/mips/mips.opt (-mloongson-ext2): Add option.
1921 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
1923 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1925 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
1926 __mips_loongson_ext.
1927 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
1928 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
1930 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
1931 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
1932 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
1933 instead of TARGET_LOONGSON_3A.
1934 * config/mips/mips.opt (-mloongson-ext): Add option.
1935 * gcc/doc/invoke.texi (-mloongson-ext): Document.
1937 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1939 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
1940 * config/mips/loongson.md: Move to ...
1941 * config/mips/loongson-mmi.md: here; Adjustment.
1942 * config/mips/loongson.h: Move to ...
1943 State as deprecated. Include loongson-mmiintrin.h for back
1944 compatibility and warning.
1945 * config/mips/loongson-mmiintrin.h: ... here.
1946 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
1947 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
1948 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
1949 (mips_option_override): Make sure MMI use hard float;
1950 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
1951 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
1952 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
1953 TARGET_LOONGSON_VECTORS.
1954 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
1955 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
1956 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
1957 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
1959 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
1960 TARGET_LOONGSON_VECTORS.
1961 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
1962 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
1963 (Loongson MMI patterns): Include loongson-mmi.md instead of
1965 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
1966 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
1968 2018-11-07 Richard Biener <rguenther@suse.de>
1971 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
1972 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
1974 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
1976 PR rtl-optimization/87874
1977 * lra.c (lra_substitute_pseudo): Do not create a subreg for
1980 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
1982 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
1983 if not in indexed or indirect form.
1984 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
1985 (bswap<mode>2_store): Ditto.
1987 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
1989 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
1992 2018-11-06 Richard Biener <rguenther@suse.de>
1994 PR tree-optimization/86850
1995 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
1996 instead of src.m_vec.
1998 2018-11-06 Jan Hubicka <jh@suse.cz>
2000 * tree.c (fld_simplified_type_name): Break out form ...
2001 (free_lang_data_in_type): ... here.
2002 (fld_type_variant_equal_p): Use it.
2004 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2006 * config/default-d.c: Include memmodel.h.
2008 * config/sol2-d.c: New file.
2009 * config/t-sol2 (sol2-d.o): New rule.
2010 * config.gcc <*-*-solaris2*>: Set d_target_objs,
2011 target_has_targetdm.
2013 2018-11-06 Jan Hubicka <jh@suse.cz>
2015 * tree.c (fld_type_variant): Also copy alignment; be sure that
2016 new variant is equal.
2018 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
2021 * config/s390/s390.md: Add relative_long attribute.
2023 2018-11-06 Jan Hubicka <jh@suse.cz>
2025 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
2026 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
2027 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
2028 stream TYPE_NEEDS_CONSTRUCTING.
2029 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
2030 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
2032 2018-11-06 Richard Biener <rguenther@suse.de>
2034 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
2036 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
2038 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2040 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
2041 REG_CLASS_CONTENTS[GEN_REGS].
2042 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
2044 2018-11-06 Jan Hubicka <jh@suse.cz>
2046 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
2048 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2051 * config/sol2.h (ASAN_CC1_SPEC): Define.
2052 (LD_WHOLE_ARCHIVE_OPTION): Define.
2053 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
2054 (ASAN_REJECT_SPEC): Provide default.
2055 (LIBASAN_EARLY_SPEC): Define.
2056 (LIBTSAN_EARLY_SPEC): Define.
2057 (LIBLSAN_EARLY_SPEC): Define.
2058 * config/i386/sol2.h (CC1_SPEC): Redefine.
2059 (ASAN_REJECT_SPEC): Define.
2061 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
2062 (TARGET_ASAN_SHADOW_OFFSET): Define.
2063 (sparc_asan_shadow_offset): New function.
2064 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
2065 (ASAN_REJECT_SPEC): Define.
2067 2018-11-06 Jan Hubicka <jh@suse.cz>
2069 * tree.c (fld_type_variant): Copy canonical type.
2070 (fld_incomplete_type_of): Check that canonical types looks sane;
2071 copy canonical type.
2072 (verify_type): Accept when incomplete type has complete canonical type.
2074 2018-11-06 Jan Hubicka <jh@suse.cz>
2076 * tree.c (free_lang_data): Reset overwite_assembler_name,
2077 print_xnode, print_decl, print_type and print_identifier of
2080 2018-11-06 Richard Biener <rguenther@suse.de>
2082 PR tree-optimization/87889
2083 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
2084 Do nothing if old and new arg are the same
2086 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
2089 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
2090 attributes for operands 3 and 4.
2092 2018-11-06 Richard Biener <rguenther@suse.de>
2095 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
2096 matching bitfield insertion.
2098 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
2100 * auto-inc-dec.c: Include valtrack.h. Improve comments.
2101 (reg_next_debug_use): New.
2102 (attempt_change): Propagate adjusted expression into affected
2104 (merge_in_block): Track uses in debug insns.
2105 (pass_inc_dec::execute): Allocate and release
2108 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
2110 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
2111 (_mm512_fixupimm_round_pd): Update parameters and builtin.
2112 (_mm512_maskz_fixupimm_round_pd): Ditto.
2113 (_mm512_fixupimm_round_ps): Ditto.
2114 (_mm512_maskz_fixupimm_round_ps): Ditto.
2115 (_mm_fixupimm_round_sd): Ditto.
2116 (_mm_maskz_fixupimm_round_sd): Ditto.
2117 (_mm_fixupimm_round_ss): Ditto.
2118 (_mm_maskz_fixupimm_round_ss): Ditto.
2119 (_mm512_fixupimm_pd): Ditto.
2120 (_mm512_maskz_fixupimm_pd): Ditto.
2121 (_mm512_fixupimm_ps): Ditto.
2122 (_mm512_maskz_fixupimm_ps): Ditto.
2123 (_mm_fixupimm_sd): Ditto.
2124 (_mm_maskz_fixupimm_sd): Ditto.
2125 (_mm_fixupimm_ss): Ditto.
2126 (_mm_maskz_fixupimm_ss): Ditto.
2127 (_mm512_mask_fixupimm_round_pd): Update builtin.
2128 (_mm512_mask_fixupimm_round_ps): Ditto.
2129 (_mm_mask_fixupimm_round_sd): Ditto.
2130 (_mm_mask_fixupimm_round_ss): Ditto.
2131 (_mm512_mask_fixupimm_pd): Ditto.
2132 (_mm512_mask_fixupimm_ps): Ditto.
2133 (_mm_mask_fixupimm_sd): Ditto.
2134 (_mm_mask_fixupimm_ss): Ditto.
2135 * config/i386/avx512vlintrin.h:
2136 (_mm256_fixupimm_pd): Update parameters and builtin.
2137 (_mm256_maskz_fixupimm_pd): Ditto.
2138 (_mm256_fixupimm_ps): Ditto.
2139 (_mm256_maskz_fixupimm_ps): Ditto.
2140 (_mm_fixupimm_pd): Ditto.
2141 (_mm_maskz_fixupimm_pd): Ditto.
2142 (_mm_fixupimm_ps): Ditto.
2143 (_mm_maskz_fixupimm_ps): Ditto.
2144 (_mm256_mask_fixupimm_pd): Update builtin.
2145 (_mm256_mask_fixupimm_ps): Ditto.
2146 (_mm_mask_fixupimm_pd): Ditto.
2147 (_mm_mask_fixupimm_ps): Ditto.
2148 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
2149 * config/i386/i386-builtin.def: Update builtin definitions.
2150 * config/i386/i386.c: Handle new builtin types and remove useless ones.
2151 * config/i386/sse.md: Update VFIXUPIMM* patterns.
2152 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2153 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2154 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
2155 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2156 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2157 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
2158 * config/i386/subst.md:
2159 (round_saeonly_sd_mask_operand4): Add new subst_attr.
2160 (round_saeonly_sd_mask_op4): Ditto.
2161 (round_saeonly_expand_operand5): Ditto.
2162 (round_saeonly_expand): Update.
2164 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
2166 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
2168 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
2170 PR rtl-optimization/87871
2171 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
2173 2018-11-05 Paul Koning <ni1d@arrl.net>
2175 * doc/sourcebuild.texi (target attributes): Document new "inf"
2176 effective target keyword.
2178 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2180 * config/s390/s390.c (s390_register_move_cost): Increase costs for
2181 moves involving the CC reg.
2183 2018-11-05 Richard Biener <rguenther@suse.de>
2185 PR tree-optimization/87873
2186 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
2188 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
2189 * tree-vect-loop.c (vect_transform_loop): When splitting the
2190 loop exit also create forwarder PHIs for constants.
2191 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
2192 Handle constant to_arg, add extra checking we match up the correct
2195 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2197 * config/s390/s390.md: QImode and HImode for load on condition.
2199 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2201 * config/s390/predicates.md: Fix typo.
2202 * config/s390/s390.md: Allow immediates for load on condition.
2204 2018-11-05 Martin Liska <mliska@suse.cz>
2206 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
2207 * bitmap.h (struct bitmap_usage): Likewise.
2208 * ggc-common.c (SCALE): Remove.
2210 (struct ggc_usage): Use SIZE_AMOUNT. And update
2212 * ggc-page.c (SCALE): Remove.
2213 (STAT_LABEL): Remove.
2214 (ggc_print_statistics): Use SIZE_AMOUNT.
2215 * gimple.h (SCALE): Remove.
2217 * input.c (ONE_K): Remove.
2220 (STAT_LABEL): Likewise.
2221 (FORMAT_AMOUNT): Likewise.
2222 (dump_line_table_statistics): Use SIZE_AMOUNT.
2223 * mem-stats.h (struct mem_usage): Likewise.
2224 * rtl.c (dump_rtx_statistics): Likewise.
2225 (rtx_alloc_counts): Change type to size_t.
2226 (rtx_alloc_sizes): Likewise.
2227 (rtx_count_cmp): New.
2228 (dump_rtx_statistics): Sort first based on counts.
2229 * tree.c (tree_nodes_cmp): New.
2230 (tree_codes_cmp): New.
2231 (dump_tree_statistics): Sort first based on counts.
2232 * system.h (ONE_K): New.
2234 (SIZE_SCALE): Likewise.
2235 (SIZE_LABEL): Likewise.
2236 (SIZE_AMOUNT): Likewise.
2237 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
2238 * tree-dfa.c (dump_dfa_stats): Likewise.
2239 * tree-phinodes.c (phinodes_print_statistics): Likewise.
2240 * tree-ssanames.c (ssanames_print_statistics): Likewise.
2241 * tree.c (dump_tree_statistics): Likewise.
2242 * vec.c (struct vec_usage): Likewise.
2243 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
2244 trigger a -Werror=format-overflow with
2245 --enable-gather-detailed-stats.
2247 2018-11-05 Martin Liska <mliska@suse.cz>
2249 * mem-stats.h (mem_alloc_description::release_instance_overhead):
2251 * vec.c (struct vec_usage): Register m_element_size.
2252 (vec_prefix::register_overhead): New arguments: elements and
2254 (vec_prefix::release_overhead): Subtract elements.
2255 * vec.h (struct vec_prefix): Change signature.
2256 (va_heap::reserve): Pass proper arguments.
2257 (va_heap::release): Likewise.
2259 2018-11-05 Martin Liska <mliska@suse.cz>
2261 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
2265 2018-11-05 Richard Biener <rguenther@suse.de>
2267 * tree-scalar-evolution.h (final_value_replacement_loop): Update
2269 * tree-scalar-evolution.c (final_value_replacement_loop): Return
2270 whether anything was done.
2271 (scev_const_prop): Remove constant propagation part, fold
2273 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
2274 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
2277 2018-11-05 Jakub Jelinek <jakub@redhat.com>
2279 PR tree-optimization/87859
2280 * gimple-ssa-store-merging.c (struct merged_store_group): Add
2281 only_constants and first_nonmergeable_order members.
2282 (merged_store_group::merged_store_group): Initialize them.
2283 (merged_store_group::do_merge): Clear only_constants member if
2284 adding something other than INTEGER_CST store.
2285 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
2286 stores with order >= first_nonmergeable_order. Use
2287 merged_store->only_constants instead of always recomputing it.
2288 Set merged_store->first_nonmergeable_order if we've skipped any
2289 stores. Attempt to merge overlapping INTEGER_CST stores that
2290 we would otherwise skip.
2293 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
2295 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
2298 * config/i386/emmintrin.h (__v16qs): New to cope with option
2300 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
2301 (_mm_cmplt_epi8): Likewise.
2302 (_mm_cmpgt_epi8): Likewise.
2304 2018-11-05 Richard Biener <rguenther@suse.de>
2306 PR rtl-optimization/87852
2307 * fwprop.c (use_killed_between): Only consider single-defs of the
2308 use whose definition statement dominates the use.
2310 2018-11-05 Martin Liska <mliska@suse.cz>
2313 * doc/invoke.texi: Remove options that are
2314 not disabled with -Os.
2316 2018-11-05 Martin Liska <mliska@suse.cz>
2319 * doc/extend.texi: Update constrain about the last argument
2320 of __builtin_expect_with_probability.
2322 2018-11-05 Martin Liska <mliska@suse.cz>
2325 * predict.c (expr_expected_value_1): Verify
2326 that last argument is a real constants and emit
2329 2018-11-05 Martin Liska <mliska@suse.cz>
2331 PR gcov-profile/77698
2332 * ipa-profile.c (ipa_profile): Adjust hotness threshold
2335 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2337 PR tree-optimization/86572
2338 * builtins.c (c_strlen): Handle negative offsets in a safe way.
2340 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2342 PR tree-optimization/87672
2343 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
2344 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
2346 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
2349 * cfgexpand.c (pass_expand::execute): Move the call to
2350 finish_eh_generation in front of the call to expand_stack_alignment.
2352 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2354 * common/config/i386/i386-common.c (processor_alias_table): Add
2356 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
2357 (case ${target}): Add znver2.
2358 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
2359 -march=native recognize znver2 processors.
2360 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
2361 * config/i386/i386.c (m_znver2): New definition.
2362 (m_ZNVER): New definition.
2363 (m_AMD_MULTIPLE): Includes m_znver2.
2364 (processor_cost_table): Add znver2 entry.
2365 (processor_target_table): Add znver2 entry.
2366 (get_builtin_code_for_version): Set priority for
2368 (processor_model): Add M_AMDFAM17H_ZNVER2.
2369 (arch_names_table): Ditto.
2370 (ix86_reassociation_width): Include znver2.
2371 * config/i386/i386.h (TARGET_znver2): New definition.
2372 (struct ix86_size_cost): Add TARGET_ZNVER2.
2373 (enum processor_type): Add PROCESSOR_ZNVER2.
2374 * config/i386/i386.md (define_attr "cpu"): Add znver2.
2375 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
2376 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
2377 (ix86_adjust_cost): Add znver2.
2378 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
2379 * gcc/doc/extend.texi: Add details about znver2.
2380 * gcc/doc/invoke.texi: Add details about znver2.
2382 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
2386 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
2389 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2391 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
2392 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
2394 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
2396 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
2397 shrink-wrapping when -mtrack-speculation.
2399 2018-11-02 Richard Biener <rguenther@suse.de>
2401 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
2402 (pop_cost_one_pair): Do not free pair.
2403 (pop_best_coalesce): Likewise.
2404 (create_coalesce_list): Initialize obstack.
2405 (delete_coalesce_list): Free obstack.
2406 (find_coalesce_pair): Obstack-allocate coalesce pairs.
2407 (add_cost_one_coalesce): Likewise.
2408 (struct live_track): Remove bitmap pointer indirections.
2409 (new_live_track): Adjust.
2410 (delete_live_track): Likewise.
2411 (live_track_remove_partition): Likewise.
2412 (live_track_add_partition): Likewise.
2413 (live_track_live_p): Likewise.
2414 (live_track_process_def): Likewise.
2415 (live_track_clear_base_vars): Likewise.
2417 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2419 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
2420 * configure: Regenerate.
2421 * config.in: Regenerate.
2422 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
2423 HAVE_AS_REGISTER_PSEUDO_OP guard.
2424 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
2425 !HAVE_AS_REGISTER_PSEUDO_OP support.
2426 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
2429 2018-11-02 Richard Biener <rguenther@suse.de>
2431 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
2434 2018-11-02 Richard Biener <rguenther@suse.de>
2436 PR tree-optimization/87776
2437 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
2438 executable when iterating but running into rpo-vn-max-loop-depth
2439 and not eliding the iteration.
2441 2018-11-30 Jan Hubicka <jh@suse.cz>
2443 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
2446 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
2448 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
2450 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
2451 Combine with rs6000_address_for_fpconvert.
2452 (rs6000_address_for_fpconvert): Combine with
2453 rs6000_force_indexed_or_indirect_mem.
2454 (rs6000_expand_vector_init): Change function call from
2455 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
2456 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
2457 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
2458 (floatsi<mode>2_lfiwax_mem): Ditto.
2459 (floatunssi<mode>2_lfiwzx): Ditto.
2460 (floatunssi<mode>2_lfiwzx_mem): Ditto.
2461 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
2462 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
2463 (fix_trunc<mode>si2_stfiwx): Ditto.
2464 (fixuns_trunc<mode>si2_stfiwx): Ditto.
2465 (float_<mode>si2_hw): Ditto.
2466 (floatuns_<mode>si2_hw): Ditto.
2467 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
2468 (vsx_splat_<mode>): Ditto.
2470 2018-11-01 Joseph Myers <joseph@codesourcery.com>
2472 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
2473 at end of assembler input text.
2474 * configure: Regenerate.
2476 2018-11-01 Jakub Jelinek <jakub@redhat.com>
2478 PR tree-optimization/87826
2479 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
2480 negative or larger or equal to type's precision.
2482 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
2484 * opts.c (default_options_table): Do not enable
2485 OPT_fdelayed_branch at -Og.
2486 * doc/invoke.texi (-fdelayed-branch): Document it.
2488 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
2490 * optabs-libfuncs.c (build_libfunc_function_visibility):
2491 New, split out from...
2492 (build_libfunc_function): ... here.
2493 (init_one_libfunc_visibility): New, split out from ...
2494 (init_one_libfunc): ... here.
2496 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
2497 scratch register need not be early-clobber. Document the reason
2498 why we cannot use ST<OP>.
2500 2018-10-31 Joseph Myers <joseph@codesourcery.com>
2503 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
2504 line for second argument of AC_DEFINE_UNQUOTED.
2505 * doc/install.texi (Tools/packages necessary for modifying GCC):
2506 Update to autoconf 2.69 and automake 1.15.1.
2507 * aclocal.m4, config.in, configure: Regenerate.
2509 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
2511 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
2514 2018-10-31 Martin Liska <mliska@suse.cz>
2517 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
2518 Add new argument invalid_extension.
2519 (aarch64_get_all_extension_candidates): New function.
2520 (aarch64_rewrite_selected_cpu): Add NULL to function call.
2521 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
2523 (aarch64_get_all_extension_candidates): New function.
2524 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
2525 argument invalid_extension.
2526 (aarch64_parse_cpu): Likewise.
2527 (aarch64_print_hint_for_extensions): New function.
2528 (aarch64_validate_mcpu): Provide hint about invalid extension.
2529 (aarch64_validate_march): Likewise.
2530 (aarch64_handle_attr_arch): Pass new argument.
2531 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
2532 (aarch64_handle_attr_isa_flags): Likewise.
2534 2018-10-31 Richard Biener <rguenther@suse.de>
2538 * tree-outof-ssa.c (insert_backedge_copies): Restrict
2539 copy generation to useful cases. Place the copy before
2540 the definition of the backedge value when possible.
2542 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2544 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
2545 * config/arc/arc.c (arc_active_insn): New function.
2546 (check_store_cacheline_hazard): Likewise.
2547 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
2548 (arc_override_options): Disable delay slot scheduler for older
2550 (arc_store_addr_hazard_p): New implementation, old one renamed to
2552 (arc_store_addr_hazard_internal_p): Renamed.
2553 (arc_reorg): Don't combine into brcc instructions which are part
2554 of hardware hazard solution.
2555 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
2556 (tune_arc700): Likewise.
2557 * config/arc/arc.opt (arc7xx): New tune value.
2558 * config/arc/arc700.md: Improve A7 scheduler.
2560 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2562 * config/arc/arc.c (arc_override_options): Remove
2563 TARGET_COMPACT_CASESI.
2564 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
2565 (CASE_VECTOR_MODE): Likewise.
2566 (CASE_VECTOR_PC_RELATIVE): Likewise.
2567 (CASE_VECTOR_SHORTEN_MODE): Likewise.
2568 (CASE_VECTOR_SHORTEN_MODE1): Delete.
2569 (ADDR_VEC_ALIGN): Update.
2570 (ASM_OUTPUT_CASE_LABEL): Undefine.
2571 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
2572 (TARGET_BI_BIH): Define.
2573 (DEFAULT_BRANCH_INDEX): Likewise.
2574 * config/arc/arc.md (casesi): Rework to accept BI/BIH
2575 instructions, remove compact_casesi use case.
2576 (casesi_compact_jump): Remove.
2577 (casesi_dispatch): New pattern.
2578 * config/arc/arc.opt: Add mbranch-index option. Deprecate
2579 compact_casesi option.
2580 * doc/invoke.texi: Document mbranch-index option.
2582 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2584 * config/arc/arc.c (arc_get_tp): Remove function.
2585 (arc_emit_call_tls_get_addr): Likewise.
2586 (arc_call_tls_get_addr): New function.
2587 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
2588 * config/arc/arc.md (tls_load_tp_soft): Remove.
2589 (tls_gd_get_addr): Likewise.
2591 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2593 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
2594 (mulsi3_600_lib): Remove pattern.
2595 (umulsi3_highpart_600_lib_le): Likewise.
2596 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
2597 (umulsidi3): Remove call to umulsidi3_600_lib.
2598 (umulsidi3_600_lib): Remove pattern.
2599 (peephole2): Remove peephole using the above deprecated patterns.
2601 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2604 * config/arm/arm.c (arm_option_check_internal): Disable the combined
2605 use of -mslow-flash-data and -mword-relocations.
2606 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
2607 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
2609 * doc/invoke.texi (-mword-relocations): Mention conflict with
2611 (-mslow-flash-data): Reciprocally.
2613 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
2615 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
2616 16-byte modes held in GP registers to use an even regno.
2618 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
2619 (aarch64_atomic_ldop_supported_p): Remove.
2620 (aarch64_gen_atomic_ldop): Remove.
2621 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
2622 Fully expand LSE operations here.
2623 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
2624 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
2625 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
2626 and use ATOMIC_LDOP instead; use register_operand for the input;
2627 drop the split and emit insns directly.
2628 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
2629 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
2630 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
2632 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
2633 (aarch64_gen_atomic_ldop): Don't call it.
2634 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
2635 Use aarch64_reg_or_zero.
2636 (aarch64_atomic_exchange<ALLI>): Likewise.
2637 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
2638 operand 0; use aarch64_reg_or_zero for input; merge ...
2639 (@aarch64_atomic_swp<ALLI>): ... this and remove.
2641 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
2642 (aarch64_split_compare_and_swap): Use it.
2643 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
2644 test oldval against the proper predicate.
2645 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
2646 Use nonmemory_operand for expected.
2647 (cas_short_expected_pred): New.
2648 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
2649 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
2650 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
2651 (aarch64_plushi_operand): New.
2653 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
2654 Force oldval into the rval register for TARGET_LSE; emit the compare
2655 during initial expansion so that it may be deleted if unused.
2656 (aarch64_gen_atomic_cas): Remove.
2657 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
2658 Change =&r to +r for operand 0; use match_dup for operand 2;
2659 remove is_weak and mod_f operands as unused. Drop the split
2661 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
2662 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
2663 (@aarch64_atomic_cas<GPI>): Similarly.
2665 2018-10-31 Richard Biener <rguenther@suse.de>
2667 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
2670 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
2672 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
2673 * config/aarch64/aarch64.md: Include saphira.md
2674 * config/aarch64/saphira.md: New file for pipeline description.
2676 2018-10-30 Martin Sebor <msebor@redhat.com>
2679 * gimple-ssa-sprintf.c (format_directive): Use %G to include
2681 (sprintf_dom_walker::compute_format_length):
2682 Avoid setting POSUNDER4K here.
2683 (get_destination_size): Handle null argument values.
2684 (get_user_idx_format): New function.
2685 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
2686 functions, including user-defined with attribute format printf.
2687 Use %G to include inlining context.
2688 Set POSUNDER4K here.
2690 2018-10-30 Jan Hubicka <jh@suse.cz>
2692 * params.def (lto-partitions): Bump from 32 to 128.
2694 2018-10-30 Jan Hubicka <jh@suse.cz>
2697 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
2699 (free_lang_data_in_type): Forward declare.
2700 (fld_type_variant_equal_p): New function.
2701 (fld_type_variant): New function
2702 (fld_incomplete_types): New hash.
2703 (fld_incomplete_type_of): New function
2704 (fld_simplfied-type): New function.
2705 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
2706 (free_lang_data): Allocate and free fld_incomplete_type; update call
2707 of free_lang_data_in_decl.
2709 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
2711 * gcov.c (output_lines): Remove duplicate line.
2713 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
2715 * config/rs6000/rs6000.md (bswapdi2): Force address into register
2716 if not in indexed or indirect form.
2717 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
2718 (bswapdi2_store): Ditto.
2719 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
2721 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
2722 Prototype for helper function.
2724 2018-10-30 Martin Sebor <msebor@redhat.com>
2726 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
2727 (target, pragma GCC optimize, pragma GCC target): Ditto.
2729 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
2731 * doc/extend.texi: Fix prototype and description of
2732 __builtin_expect_with_probability.
2734 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2736 * cgraph.h (clone_function_name_1): Replaced by new
2737 clone_function_name_numbered that takes name as string; for
2738 privatize_symbol_name_1 use only.
2739 (clone_function_name): Renamed to
2740 clone_function_name_numbered to be explicit about numbering.
2741 (clone_function_name): New two-argument function that does
2742 not number its output.
2743 (clone_function_name): New three-argument function that
2744 takes a number to append to its output.
2745 * cgraphclones.c (duplicate_thunk_for_node):
2746 (clone_function_name_1): Renamed.
2747 (clone_function_name_numbered): Two new functions.
2748 (clone_function_name): Improved documentation.
2749 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
2750 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
2751 * final.c (final_scan_insn_1): Use the new clone_function_name
2753 * multiple_target.c (create_dispatcher_calls): Ditto.
2754 (create_target_clone): Ditto.
2755 * omp-expand.c (grid_expand_target_grid_body): Ditto.
2756 * omp-low.c (create_omp_child_function_name): Ditto.
2757 * omp-simd-clone.c (simd_clone_create): Ditto.
2758 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
2759 new clone_function_name without numbering.
2761 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
2763 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
2764 Assert that the allocation size is not zero.
2766 2018-10-30 Richard Biener <rguenther@suse.de>
2768 PR tree-optimization/87800
2769 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
2770 non-induction or reduction PHIs.
2772 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
2774 * config/aarch64/falkor-tag-collision-avoidance.c
2775 (execute_tag_collision_avoidance): Call df_note_add_problem.
2777 2018-10-30 Martin Liska <mliska@suse.cz>
2779 * doc/extend.texi: Fix typo in documentation
2780 of __builtin_expect_with_probability.
2782 2018-10-29 David Malcolm <dmalcolm@redhat.com>
2785 * input.c (get_substring_ranges_for_loc): Detect if
2786 linemap_resolve_location gives us a NULL map, and reject
2789 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
2791 * config.gcc (xstormy16-*-elf): Set tm_d_file.
2793 2018-10-29 David Malcolm <dmalcolm@redhat.com>
2794 Martin Sebor <msebor@redhat.com>
2795 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2797 * doc/ux.texi (Quoting): New subsection, adapted from material at
2798 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
2799 MartinSebor and ManuelLopezIbanez.
2800 (Fix-it hints): Note that fix-it hints shouldn't be marked for
2803 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
2806 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
2809 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
2811 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
2813 2018-10-29 Paul Koning <ni1d@arrl.net>
2815 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
2817 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
2819 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
2820 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
2821 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
2822 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
2823 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
2824 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
2825 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
2826 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
2827 Change 'vector' to '__vector'.
2828 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
2829 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
2830 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
2831 _mm_avg_pu16): Likewise. And, whitespace corrections.
2833 2018-10-29 Richard Biener <rguenther@suse.de>
2835 PR tree-optimization/87785
2836 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
2839 2018-10-29 Olivier Hainque <hainque@adacore.com>
2841 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
2843 2018-10-29 Olivier Hainque <hainque@adacore.com>
2845 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
2846 ports configuration macro, defaults to "gnu".
2847 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
2848 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
2850 2018-10-29 Olivier Hainque <hainque@adacore.com>
2852 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
2854 2018-10-29 Richard Biener <rguenther@suse.de>
2856 PR tree-optimization/87790
2857 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
2858 (vect_make_slp_decision): Adjust.
2859 (vect_slp_analyze_bb_1): Likewise.
2860 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
2863 2018-10-29 Richard Biener <rguenther@suse.de>
2865 PR tree-optimization/87785
2866 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
2868 (vect_build_slp_tree): Likewise.
2869 (vect_gather_slp_loads): New function.
2870 (vect_analyze_slp_instance): Gather loads separately from the
2873 2018-10-29 Martin Liska <mliska@suse.cz>
2875 * Makefile.in: Make dependency to json.o.
2876 * doc/gcov.texi: Document new JSON format, remove
2877 old intermediate format documentation.
2878 * gcov.c (struct function_info): Come up with m_name and
2880 (function_info::function_info): Initialize it.
2881 (function_info::~function_info): Release it.
2882 (main): Rename flag_intermediate_format to flag_json_format.
2883 (print_usage): Describe --json-format.
2884 (process_args): Set flag_json_format.
2885 (output_intermediate_line): Remove.
2886 (output_intermediate_json_line): Likewise.
2887 (get_gcov_intermediate_filename): Return new extension
2889 (output_intermediate_file): Implement JSON emission.
2890 (output_json_intermediate_file): Implement JSON emission.
2891 (generate_results): Use ::get_name for function name.
2892 Handle JSON output file.
2893 (read_graph_file): Use ::get_name instead of cplus_demangle.
2894 (read_count_file): Likewise.
2895 (solve_flow_graph): Likewise.
2896 (add_line_counts): Likewise.
2897 (accumulate_line_counts): Use new flag_json_format.
2898 (output_function_details): Use ::get_name instead of cplus_demangle.
2899 (output_lines): Likewise.
2900 * json.cc (test_writing_literals): Add new tests.
2901 * json.h (class literal): Add new boolean constructor.
2903 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
2905 PR rtl-optimization/87701
2906 PR rtl-optimization/87780
2907 * combine.c (make_more_copies): Rewrite.
2909 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
2911 * doc/generic.texi (ABSU_EXPR): Document.
2912 * match.pd (absu(x)*absu(x) -> x*x): Handle.
2913 (absu(absu(X)) -> absu(X)): Likewise.
2914 (absu(-X) -> absu(X)): Likewise.
2915 (absu(X) where X is nonnegative -> X): Likewise.
2917 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
2919 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
2920 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
2921 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
2922 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
2923 (s-tm-texi): Also check timestamp on d-target.def.
2924 (generated_files): Add TM_D_H and d-target-hooks-def.h.
2925 (build/genhooks.o): Also depend on D_TARGET_DEF.
2926 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
2928 * config/aarch64/aarch64-d.c: New file.
2929 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
2931 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
2933 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
2934 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
2935 * config/arm/arm-d.c: New file.
2936 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
2937 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
2938 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2939 * config/arm/t-arm (arm-d.o): New rule.
2940 * config/default-d.c: New file.
2941 * config/glibc-d.c: New file.
2942 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2943 * config/i386/i386-d.c: New file.
2944 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
2945 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
2946 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2947 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
2948 * config/i386/t-i386 (i386-d.o): New rule.
2949 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2950 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2951 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
2952 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2953 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2954 * config/mips/mips-d.c: New file.
2955 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
2956 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
2957 * config/mips/t-mips (mips-d.o): New rule.
2958 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2959 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2960 * config/powerpcspe/powerpcspe-d.c: New file.
2961 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
2963 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
2964 Support GNU D by using 0 as the language type.
2965 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
2966 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
2967 * config/riscv/riscv-d.c: New file.
2968 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
2970 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
2971 * config/riscv/t-riscv (riscv-d.o): New rule.
2972 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2973 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2974 * config/rs6000/rs6000-d.c: New file.
2975 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
2977 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
2978 Support GNU D by using 0 as the language type.
2979 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
2980 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
2981 * config/s390/s390-d.c: New file.
2982 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
2983 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
2984 * config/s390/t-s390 (s390-d.o): New rule.
2985 * config/sparc/sparc-d.c: New file.
2986 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
2988 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
2989 * config/sparc/t-sparc (sparc-d.o): New rule.
2990 * config/t-glibc (glibc-d.o): New rule.
2991 * configure: Regenerated.
2992 * configure.ac (tm_d_file): New variable.
2993 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
2994 * doc/contrib.texi (Contributors): Add self for the D frontend.
2995 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
2996 * doc/install.texi (Configuration): Mention libphobos as an option for
2997 --enable-shared. Mention d as an option for --enable-languages.
2998 (Testing): Mention check-d as a target.
2999 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
3000 name suffixes. Mention d as a -x option.
3001 * doc/sourcebuild.texi (Top Level): Mention libphobos.
3002 * doc/standards.texi (Standards): Add section on D language.
3003 * doc/tm.texi: Regenerated.
3004 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
3005 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
3006 * dwarf2out.c (is_dlang): New function.
3007 (gen_compile_unit_die): Use DW_LANG_D for D.
3008 (declare_in_namespace): Return module die for D, instead of adding
3009 extra declarations into the namespace.
3010 (gen_namespace_die): Generate DW_TAG_module for D.
3011 (gen_decl_die): Handle CONST_DECLSs for D.
3012 (dwarf2out_decl): Likewise.
3013 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
3014 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
3015 kinds of aggregates.
3016 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
3017 * genhooks.c: Include d/d-target.def.
3019 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
3022 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
3023 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
3024 (STACK_DYNAMIC_OFFSET): Likewise.
3026 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
3029 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
3032 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
3034 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
3035 return type and other typos.
3037 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
3039 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
3040 a shorter sequence with fewer branches.
3041 (emit_final_str_compare_gpr): Ditto.
3043 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3045 * config/rs6000/tmmintrin.h: New file.
3046 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
3048 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3050 * config/rs6000/mmintrin.h: Enable 32bit compilation.
3051 * config/rs6000/xmmintrin.h: Likewise.
3053 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3055 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
3057 2018-10-26 Richard Biener <rguenther@suse.de>
3059 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
3061 (vect_mark_slp_stmts_relevant): Likewise.
3062 (vect_detect_hybrid_slp_stmts): Likewise.
3063 (vect_bb_slp_scalar_cost): Likewise.
3064 (vect_remove_slp_scalar_calls): Likewise.
3066 2018-10-26 Jan Hubicka <jh@suse.cz>
3068 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
3069 (warn_types_mismatch): Fix walk of DECL_NAME.
3070 (odr_types_equivalent_p): Fix overactive assert.
3072 2018-10-26 Richard Biener <rguenther@suse.de>
3074 PR tree-optimization/87105
3075 * tree-vectorizer.h (_slp_tree::refcnt): New member.
3076 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
3078 (vect_create_new_slp_node): Initialize refcnt to one.
3080 (scalar_stmts_set_t, bst_fail): Remove.
3081 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
3082 (vect_build_slp_tree): Add bst_map argument and lookup
3083 already created SLP nodes.
3084 (vect_print_slp_tree): Handle a SLP graph, print SLP node
3086 (vect_slp_rearrange_stmts): Handle a SLP graph.
3087 (vect_analyze_slp_instance): Adjust and free SLP nodes from
3088 the CSE map. Fix indenting.
3089 (vect_schedule_slp_instance): Add short-cut.
3091 2018-10-26 Martin Liska <mliska@suse.cz>
3094 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
3095 addr_expr and not with pointers.
3097 2018-10-26 Jan Hubicka <jh@suse.cz>
3099 * tree.c (free_lang_data_in_type): Only check main variants.
3100 * ipa-devirt.c (warn_odr): Make static.
3101 (types_same_for_odr): Drop strict variant.
3102 (types_odr_comparable): Likewise.
3103 (odr_or_derived_type_p): Look for main variants.
3104 (odr_name_hasher::equal): Cleanup comment.
3105 (odr_subtypes_equivalent): Add warn and warned arguments; check main
3107 (type_variants_equivalent_p): break out from ...
3108 (odr_types_equivalent): ... here; go for main variants where needed.
3109 (warn_odr): ... here; turn static.
3110 (warn_types_mismatch): Compare mangled names of main variants.
3111 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
3112 (type_with_linkage_p): Sanity check that we look at main variant.
3113 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
3114 * tree.h (types_same_for_odr): Drop strict argument.
3116 2018-10-26 Richard Biener <rguenther@suse.de>
3118 PR tree-optimization/87746
3119 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
3120 Simplify and fix WRT strided store groups with size not
3121 equal to step in element count.
3122 (vect_analyze_group_access_1): Dump the whole group.
3124 2018-10-25 Carl Love <cel@us.ibm.com>
3126 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
3127 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
3128 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
3129 precicion entry for each overloaded builtin.
3130 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
3131 VSCEDPUO): Rename overloaded name.
3132 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
3133 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
3134 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
3135 define_expand for xscmpexqp instruction.
3136 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
3138 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3139 Jinsong Ji <jji@us.ibm.com>
3141 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
3142 function with vec_sl.
3143 (_mm_slli_epi32): Likewise.
3144 (_mm_slli_epi64): Likewise.
3145 (_mm_srai_epi16): Replace deprecated function with vec_sra.
3146 (_mm_srai_epi32): Likewise.
3147 (_mm_srli_epi16): Replace deprecated function with vec_sr.
3148 (_mm_srli_epi32): Likewise.
3149 (_mm_srli_epi64): Likewise.
3150 (_mm_sll_epi16): Replace deprecated function with vec_sl.
3151 (_mm_sll_epi32): Likewise.
3152 (_mm_sll_epi64): Likewise.
3153 (_mm_sra_epi16): Replace deprecated function with vec_sra.
3154 (_mm_sra_epi32): Likewise.
3155 (_mm_srl_epi16): Replace deprecated function with vec_sr.
3156 (_mm_srl_epi32): Likewise.
3157 (_mm_srl_epi64): Likewise.
3159 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3160 Jinsong Ji <jji@us.ibm.com>
3162 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
3163 comparison operators with vec_cmp* for compatibility due to
3164 unfortunate history; clean up formatting and use types more
3166 (_mm_sll_epi32): Likewise.
3167 (_mm_sll_epi64): Likewise.
3168 (_mm_srl_epi16): Likewise.
3169 (_mm_srl_epi32): Likewise.
3170 (_mm_srl_epi64): Likewise.
3172 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3173 Jinsong Ji <jji@us.ibm.com>
3175 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
3176 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
3177 __vector __bool int. Use vec_cmpgt in preference to deprecated
3178 function vec_vcmpgtfp.
3179 (_mm_max_ps): Likewise.
3181 2018-10-25 Jeff Law <law@redhat.com>
3183 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
3186 2018-10-25 Martin Sebor <msebor@redhat.com>
3188 * doc/extend.texi (aligned): Expand attribute description.
3189 (Alignment): Rename section. Discuss function arguments.
3191 2018-10-25 Jan Hubicka <jh@suse.cz>
3193 * ipa-devirt.c (main_odr_variant): Remove.
3194 (hash_odr_name, types_same_for_odr, types_odr_comparable,
3195 odr_name_hasher::equal, odr_subtypes_equivalent_p):
3196 Drop use of main_odr_variant.
3197 (add_type_duplicate): Silence confused warnings on integer types.
3198 (get_odr_type): Always look for main variant.
3199 (register_odr_type): Simplify.
3201 2018-10-25 Richard Biener <rguenther@suse.de>
3203 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
3204 Initialize ng to silence error with release checking bootstrap.
3206 2018-10-25 Richard Biener <rguenther@suse.de>
3208 * tree-if-conv.c: Include tree-ssa-sccvn.h.
3209 (tree_if_conversion): Run CSE on the if-converted loop body.
3211 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
3213 * config/s390/constraints.md (ZL): New constraint.
3214 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
3216 * config/s390/s390.md (movdi_larl): Remove.
3217 (movdi_64): Add the LARL alternative.
3219 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
3222 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
3223 (RTX_CODE_HWINT_P): New macro.
3224 (rtx_code_size): Use RTX_CODE_HWINT_P ().
3226 2018-10-25 Jan Hubicka <jh@suse.cz>
3228 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
3231 2018-10-25 Richard Biener <rguenther@suse.de>
3233 PR tree-optimization/87665
3234 PR tree-optimization/87745
3235 * tree-vectorizer.h (get_earlier_stmt): Remove.
3236 (get_later_stmt): Pick up UID from the original non-pattern stmt.
3238 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
3240 * options.texi (Deprecated): Move list to Var section.
3242 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
3243 Jinsong Ji <jji@us.ibm.com>
3245 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
3246 __vector long to __vector long long.
3247 (_mm_cvtpd_ps): Likewise.
3248 (_mm_cvttpd_epi32): Likewise.
3249 (_mm_cvtpi32_pd): Likewise.
3250 (_mm_unpackhi_epi64): Likewise.
3251 (_mm_unpacklo_epi64): Likewise.
3253 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
3255 PR rtl-optimization/87720
3256 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
3258 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
3260 * gimple-ssa-isolate-paths.c
3261 (find_implicit_erroneous_behavior): Do not change code if the
3262 pass is running for warnings only.
3263 (find_explicit_erroneous_behavior): Likewise.
3265 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
3267 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
3268 Define as rs6000_mangle_decl_assembler_name.
3269 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
3270 long double to IEEE long double, switch the names of the long
3271 double built-in functions to be <func>f128 instead of <func>l.
3273 2018-10-24 Martin Sebor <msebor@redhat.com>
3275 * doc/extend.texi (nonnull): List no-argument form. Reference
3276 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
3278 2018-10-24 Richard Biener <rguenther@suse.de>
3280 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
3282 2018-10-24 Martin Liska <mliska@suse.cz>
3284 PR tree-optimization/84436
3285 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
3287 (switch_conversion::contains_linear_function_p): New.
3288 (switch_conversion::build_one_array): Support linear
3289 transformation on input.
3290 * tree-switch-conversion.h (struct switch_conversion): Add
3291 contains_linear_function_p declaration.
3293 2018-10-24 Richard Biener <rguenther@suse.de>
3295 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
3296 if its argument is CONSTANT_CLASS_P.
3298 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
3300 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
3301 it is wrong for forward declarations.
3303 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
3305 * config/s390/s390.c (s390_check_qrst_address): Add the missing
3306 SYMBOL_REF_P () check.
3308 2018-10-24 Richard Biener <rguenther@suse.de>
3310 PR tree-optimization/87105
3311 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
3312 dump classification.
3313 (vect_analyze_data_ref_accesses): Handle duplicate loads and
3314 stores by splitting the affected group after the fact.
3315 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
3316 fail the SLP build because of size constraints.
3318 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3320 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
3321 * configure: Regenerate.
3322 * config.in: Regenerate.
3323 * varasm.c (mergeable_string_section): Use readonly_data_section
3324 if linker doesn't support SHF_MERGE with alignment > 8.
3325 (mergeable_constant_section): Likewise.
3327 2018-10-24 Richard Biener <rguenther@suse.de>
3329 PR tree-optimization/84013
3330 * tree-ssa-structalias.c (struct msdi_data): New struct for
3331 marshalling data to walk_stmt_load_store_ops.
3332 (maybe_set_dependence_info): Refactor as callback for
3333 walk_stmt_load_store_ops.
3334 (compute_dependence_clique): Set restrict info on all stmt kinds.
3336 2018-10-24 Martin Liska <mliska@suse.cz>
3338 * cgraph.c (cgraph_node::dump):
3339 Remove reduntant dumps and make tp_first_run dump more compact.
3341 2018-10-24 Richard Biener <rguenther@suse.de>
3343 PR tree-optimization/87665
3344 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
3347 2018-10-12 Jeff Law <law@redhat.com>
3349 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
3352 2018-10-23 Richard Biener <rguenther@suse.de>
3354 * tree-vrp.c (add_assert_info): Guard dump_printf with
3356 * gimple-ssa-evrp-analyze.c
3357 (evrp_range_analyzer::record_ranges_from_incoming_edge):
3358 Use value_range::ignore_equivs_equal_p.
3360 2018-10-23 Richard Biener <rguenther@suse.de>
3362 PR tree-optimization/87105
3363 PR tree-optimization/87608
3364 * passes.def (pass_all_early_optimizations): Add early phi-opt
3366 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
3367 addition to debug stmts.
3368 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
3369 and abs replacement early.
3370 * tree-cfg.c (gimple_empty_block_p): Likewise.
3372 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
3375 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
3376 specified to configure.
3377 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
3379 2018-10-23 Richard Biener <rguenther@suse.de>
3381 PR tree-optimization/87700
3382 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
3384 2018-10-23 Jakub Jelinek <jakub@redhat.com>
3387 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
3388 second argument from __mmask16 to __mmask8.
3389 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
3390 _mm_mask_packs_epi32): Likewise.
3391 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
3393 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
3395 2018-10-23 Richard Biener <rguenther@suse.de>
3397 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
3399 2018-10-23 Richard Biener <rguenther@suse.de>
3401 PR tree-optimization/86144
3402 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
3403 over simd attribute.
3405 2018-10-23 Richard Biener <rguenther@suse.de>
3407 PR tree-optimization/87693
3408 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
3409 the case we do not find the taken edge.
3411 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
3412 Jinsong Ji <jji@us.ibm.com>
3414 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
3415 (_mm_store_pd): Use unaligned vector type for pointer cast.
3416 (_mm_maskmoveu_si128): Likewise.
3417 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
3418 (_mm_store_ps): Use unaligned vector type for pointer cast.
3420 2018-10-22 Paul Koning <ni1d@arrl.net>
3422 * symtab.c (symtab_node::increase_alignment): Correct max
3425 2018-10-22 Yury Gribov <tetra2005@gmail.com>
3427 PR tree-optimization/87633
3428 * match.pd: Do not generate unordered integer comparisons.
3430 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
3432 PR rtl-optimization/87600
3433 * combine.c: Add include of expr.h.
3434 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
3435 register to a pseudo.
3436 (make_more_copies): New function, add a copy to a new pseudo after
3437 the moves from hard registers into pseudos.
3438 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
3439 later. Call make_more_copies.
3441 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
3443 * lra-constraints.c (process_alt_operands): New local array,
3444 matching_early_clobber. Check matching_early_clobber before
3445 decrementing reject, and set matching_early_clobber after.
3447 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
3450 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
3451 call output_addr_const and hope for the best.
3453 2018-10-22 Richard Biener <rguenther@suse.de>
3455 * gimple-ssa-evrp-analyze.c
3456 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
3457 smarter about what ranges to use.
3458 * tree-vrp.c (add_assert_info): Dump here.
3459 (register_edge_assert_for_2): Instead of here at multiple but
3462 * gcc.dg/tree-ssa/evrp12.c: New testcase.
3463 * gcc.dg/predict-6.c: Adjust.
3464 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
3465 * gcc.dg/tree-ssa/vrp02.c: Likewise.
3466 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
3468 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
3469 Richard Biener <rguenther@suse.de>
3471 * bitmap.h: Update data structure documentation, including a
3472 description of bitmap views as either linked-lists or splay trees.
3473 (struct bitmap_element_def): Update comments for splay tree bitmaps.
3474 (struct bitmap_head_def): Likewise.
3475 (bitmap_list_view, bitmap_tree_view): New prototypes.
3476 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
3478 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
3479 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
3480 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
3481 released bitmap element here.
3482 (bitmap_element_free): Remove.
3483 (bitmap_elt_clear_from): Work on splay tree bitmaps.
3484 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
3485 this function similar ones such that linked-list bitmap implementation
3486 functions are grouped.
3487 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
3488 and moved for grouping.
3489 (bitmap_list_insert_element_after): Renamed from
3490 bitmap_elt_insert_after, and moved for grouping.
3491 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
3492 (bitmap_tree_link_left, bitmap_tree_link_right,
3493 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
3494 bitmap_tree_link_element, bitmap_tree_unlink_element,
3495 bitmap_tree_find_element): New functions for splay-tree bitmap
3497 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
3498 Renamed and moved, see above entries.
3499 (bitmap_tree_listify_from): New function to convert part of a splay
3500 tree bitmap to a linked-list bitmap.
3501 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
3502 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
3503 (bitmap_find_bit): Remove.
3504 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
3505 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
3506 Handle splay tree bitmaps.
3507 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
3508 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
3509 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
3510 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
3511 bitmap_intersect_compl_p, bitmap_ior_and_compl,
3512 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
3513 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
3514 corresponding changes to use linked-list specific bitmap_element
3515 manipulation functions as applicable for efficiency.
3516 (bitmap_tree_to_vec): New function.
3517 (debug_bitmap_elt_file): New function split out from ...
3518 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
3519 (bitmap_print): Likewise.
3521 PR tree-optimization/63155
3522 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
3524 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
3527 2018-10-22 Martin Liska <mliska@suse.cz>
3529 PR tree-optimization/87686
3531 2018-08-29 Martin Liska <mliska@suse.cz>
3533 * tree-switch-conversion.c (switch_conversion::expand):
3534 Strenghten assumption about gswitch statements.
3536 2018-10-22 Martin Liska <mliska@suse.cz>
3538 * ipa-icf.c (sem_item::compare_attributes): Remove.
3539 (sem_item::compare_referenced_symbol_properties): Use
3540 attribute_list_equal instead.
3541 (sem_function::equals_wpa): Likewise.
3542 * ipa-icf.h: Remove compare_attributes.
3544 2018-10-22 Richard Biener <rguenther@suse.de>
3547 * mem-stats.h (mem_usage::operator==): Fix pasto.
3549 2018-10-22 Richard Biener <rguenther@suse.de>
3551 PR tree-optimization/87640
3552 * tree-vrp.c (set_value_range_with_overflow): Decompose
3554 (extract_range_from_binary_expr_1): Adjust.
3556 2018-10-22 Martin Jambor <mjambor@suse.cz>
3558 * tree-eh.h (stmt_could_throw_p): Add function parameter.
3559 (stmt_can_throw_external): Likewise.
3560 (stmt_can_throw_internal): Likewise.
3561 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
3562 (lower_eh_constructs_2): Likewise.
3563 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
3564 (stmt_can_throw_external): Likewise.
3565 (stmt_can_throw_internal): Likewise.
3566 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
3567 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
3568 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
3569 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
3570 (pass_lower_eh_dispatch::execute): Pass cfun to
3571 stmt_can_throw_external.
3572 (cleanup_empty_eh): Likewise.
3573 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
3574 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
3575 stmt_can_throw_external instead of pushing it to cfun.
3576 (symbol_table::create_edge): Likewise.
3577 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
3578 stmt_can_throw_internal.
3579 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
3580 to stmt_could_throw_p.
3581 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
3582 stmt_can_throw_internal.
3583 (pass_store_merging::execute): Likewise.
3584 * gimple-ssa-strength-reduction.c
3585 (find_candidates_dom_walker::before_dom_children): Pass cfun to
3587 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
3588 stmt_can_throw_internal.
3589 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
3590 to stmt_can_throw_external.
3591 (check_stmt): Pass cfun to stmt_could_throw_p.
3592 (check_stmt): Pass cfun to stmt_can_throw_external.
3593 (pass_nothrow::execute): Likewise.
3594 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
3595 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
3596 stmt_can_throw_internal.
3597 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
3598 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
3599 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
3600 * tree-complex.c (expand_complex_libcall): Pass cfun to
3601 stmt_could_throw_p and to stmt_can_throw_internal.
3602 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
3603 * tree-inline.c (copy_edges_for_bb): Likewise.
3604 (maybe_move_debug_stmts_to_successors): Likewise.
3605 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
3607 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
3608 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
3609 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
3610 stmt_can_throw_internal.
3611 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
3612 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
3614 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
3615 stmt_can_throw_internal.
3616 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
3617 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
3619 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
3620 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
3621 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3622 (convert_mult_to_fma_1): Likewise.
3623 (convert_to_divmod): Likewise.
3624 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
3625 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
3626 * tree-ssa-propagate.c
3627 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
3628 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
3629 (maybe_optimize_range_tests): Likewise.
3630 (linearize_expr_tree): Likewise.
3631 (reassociate_bb): Likewise.
3632 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
3633 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
3634 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
3635 (handle_char_store): Likewise.
3636 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
3637 stmt_can_throw_internal.
3638 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
3640 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
3641 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
3642 (vectorizable_simd_clone_call): Likewise.
3643 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
3644 (gimple_stringop_fixed_value): Likewise.
3646 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
3648 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
3649 literal pool references.
3650 (s390_check_qrst_address): Adapt to the new behavior of
3651 s390_loadrelative_operand_p ().
3653 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3656 * config/i386/sse.md (*andnot<mode>3_bcst): New.
3658 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3661 * config/i386/sse.md (*<code><mode>3_bcst): New.
3663 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3666 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
3667 V4DI, V16SI and V8DI.
3668 (*sub<mode>3<mask_name>_bcst): New.
3669 (*add<mode>3<mask_name>_bcst): Likewise.
3671 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
3672 Jinsong Ji <jji@us.ibm.com>
3674 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
3675 __m64 with __vector unsigned long long for compatibility.
3676 (_mm_movemask_epi8): Likewise.
3677 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
3678 (_mm_cvttps_pi32): Likewise.
3679 (_mm_cvtpi32_ps): Likewise.
3680 (_mm_cvtps_pi16): Likewise.
3681 (_mm_loadh_pi): Likewise.
3682 (_mm_storeh_pi): Likewise.
3683 (_mm_movehl_ps): Likewise.
3684 (_mm_movelh_ps): Likewise.
3685 (_mm_loadl_pi): Likewise.
3686 (_mm_storel_pi): Likewise.
3687 (_mm_movemask_ps): Likewise.
3688 (_mm_shuffle_pi16): Likewise.
3690 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3693 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
3694 __builtin_ia32_vfnmsubpd512_mask.
3695 (_mm512_mask_fnmsub_round_pd): Likewise.
3696 (_mm512_fnmsub_pd): Likewise.
3697 (_mm512_mask_fnmsub_pd): Likewise.
3698 (_mm512_maskz_fnmsub_round_pd): Use
3699 __builtin_ia32_vfnmsubpd512_maskz.
3700 (_mm512_maskz_fnmsub_pd): Likewise.
3701 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
3702 (_mm512_mask_fnmsub_round_ps): Likewise.
3703 (_mm512_fnmsub_ps): Likewise.
3704 (_mm512_mask_fnmsub_ps): Likewise.
3705 (_mm512_maskz_fnmsub_round_ps): Use
3706 __builtin_ia32_vfnmsubps512_maskz.
3707 (_mm512_maskz_fnmsub_ps): Likewise.
3708 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
3709 __builtin_ia32_vfnmsubpd256_mask.
3710 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
3711 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
3712 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
3713 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
3714 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
3715 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
3716 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
3717 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
3718 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
3719 __builtin_ia32_vfnmsubpd.
3720 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
3721 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
3722 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
3723 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
3724 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
3725 * config/i386/i386-builtin.def: Add
3726 __builtin_ia32_vfnmsubpd256_mask,
3727 __builtin_ia32_vfnmsubpd256_maskz,
3728 __builtin_ia32_vfnmsubpd128_mask,
3729 __builtin_ia32_vfnmsubpd128_maskz,
3730 __builtin_ia32_vfnmsubps256_mask,
3731 __builtin_ia32_vfnmsubps256_maskz,
3732 __builtin_ia32_vfnmsubps128_mask,
3733 __builtin_ia32_vfnmsubps128_maskz,
3734 __builtin_ia32_vfnmsubpd512_mask,
3735 __builtin_ia32_vfnmsubpd512_maskz,
3736 __builtin_ia32_vfnmsubps512_mask,
3737 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
3738 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
3739 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
3740 __builtin_ia32_vfnmsubpd256.
3741 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
3742 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
3743 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
3745 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
3747 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
3749 (fmai_vmfnmsub_<mode><round_name>): Likewise.
3751 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3754 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
3755 __builtin_ia32_vfnmaddpd512_mask.
3756 (_mm512_mask_fnmadd_round_pd): Likewise.
3757 (_mm512_fnmadd_pd): Likewise.
3758 (_mm512_mask_fnmadd_pd): Likewise.
3759 (_mm512_maskz_fnmadd_round_pd): Use
3760 __builtin_ia32_vfnmaddpd512_maskz.
3761 (_mm512_maskz_fnmadd_pd): Likewise.
3762 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
3763 (_mm512_mask_fnmadd_round_ps): Likewise.
3764 (_mm512_fnmadd_ps): Likewise.
3765 (_mm512_mask_fnmadd_ps): Likewise.
3766 (_mm512_maskz_fnmadd_round_ps): Use
3767 __builtin_ia32_vfnmaddps512_maskz.
3768 (_mm512_maskz_fnmadd_ps): Likewise.
3769 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
3770 __builtin_ia32_vfnmaddpd256_mask.
3771 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
3772 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
3773 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
3774 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
3775 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
3776 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
3777 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
3778 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
3779 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
3780 __builtin_ia32_vfnmaddpd.
3781 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
3782 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
3783 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
3784 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
3785 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
3786 * config/i386/i386-builtin.def: Add
3787 __builtin_ia32_vfnmaddpd256_mask,
3788 __builtin_ia32_vfnmaddpd256_maskz,
3789 __builtin_ia32_vfnmaddpd128_mask,
3790 __builtin_ia32_vfnmaddpd128_maskz,
3791 __builtin_ia32_vfnmaddps256_mask,
3792 __builtin_ia32_vfnmaddps256_maskz,
3793 __builtin_ia32_vfnmaddps128_mask,
3794 __builtin_ia32_vfnmaddps128_maskz,
3795 __builtin_ia32_vfnmaddpd512_mask,
3796 __builtin_ia32_vfnmaddpd512_maskz,
3797 __builtin_ia32_vfnmaddps512_mask,
3798 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
3799 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
3800 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
3801 __builtin_ia32_vfnmaddpd256.
3802 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
3803 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
3804 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
3806 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
3808 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
3810 (fmai_vmfnmadd_<mode><round_name>): Likewise.
3812 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3815 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
3816 __builtin_ia32_vfmsubpd512_mask.
3817 (_mm512_mask_fmsub_round_pd): Likewise.
3818 (_mm512_fmsub_pd): Likewise.
3819 (_mm512_mask_fmsub_pd): Likewise.
3820 (_mm512_maskz_fmsub_round_pd): Use
3821 __builtin_ia32_vfmsubpd512_maskz.
3822 (_mm512_maskz_fmsub_pd): Likewise.
3823 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
3824 (_mm512_mask_fmsub_round_ps): Likewise.
3825 (_mm512_fmsub_ps): Likewise.
3826 (_mm512_mask_fmsub_ps): Likewise.
3827 (_mm512_maskz_fmsub_round_ps): Use
3828 __builtin_ia32_vfmsubps512_maskz.
3829 (_mm512_maskz_fmsub_ps): Likewise.
3830 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
3831 __builtin_ia32_vfmsubpd256_mask.
3832 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
3833 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
3834 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
3835 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
3836 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
3837 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
3838 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
3839 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
3840 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
3841 __builtin_ia32_vfmsubpd.
3842 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
3843 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
3844 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
3845 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
3846 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
3847 * config/i386/i386-builtin.def: Add
3848 __builtin_ia32_vfmsubpd256_mask,
3849 __builtin_ia32_vfmsubpd256_maskz,
3850 __builtin_ia32_vfmsubpd128_mask,
3851 __builtin_ia32_vfmsubpd128_maskz,
3852 __builtin_ia32_vfmsubps256_mask,
3853 __builtin_ia32_vfmsubps256_maskz,
3854 __builtin_ia32_vfmsubps128_mask,
3855 __builtin_ia32_vfmsubps128_maskz,
3856 __builtin_ia32_vfmsubpd512_mask,
3857 __builtin_ia32_vfmsubpd512_maskz,
3858 __builtin_ia32_vfmsubps512_mask,
3859 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
3860 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
3861 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
3862 __builtin_ia32_vfmsubpd256.
3863 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
3864 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
3865 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
3867 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
3869 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
3871 (fmai_vmfmsub_<mode><round_name>): Likewise.
3873 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3875 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
3876 Remove plus. Renamed to ...
3877 (*sub<mode>3<mask_name>_bcst): This.
3878 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
3879 (*add<mode>3<mask_name>_bcst): This.
3881 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3884 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
3886 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3889 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
3890 (_mm_or_epi32): Likewise.
3891 (_mm256_xor_epi32): Likewise.
3892 (_mm_xor_epi32): Likewise.
3893 (_mm256_or_epi64): Likewise.
3894 (_mm_or_epi64): Likewise.
3895 (_mm256_xor_epi64): Likewise.
3896 (_mm_xor_epi64): Likewise.
3898 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
3901 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
3903 2018-10-20 Jakub Jelinek <jakub@redhat.com>
3906 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
3908 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
3910 * doc/ux.texi: Move @section directly after @node.
3912 2018-10-19 Jakub Jelinek <jakub@redhat.com>
3916 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
3917 depend closely nested inside of loop with ordered clause with
3920 2018-10-19 David Malcolm <dmalcolm@redhat.com>
3922 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
3923 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
3924 * doc/ux.texi: New file.
3926 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3928 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
3929 be the first CR field allocated.
3931 2018-10-19 Richard Biener <rguenther@suse.de>
3934 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
3935 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
3937 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
3940 * config/i386/sse.md
3941 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
3942 (*add<mode>3<mask_name>_bcst_2): Likewise.
3944 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
3946 * config/i386/sse.md
3947 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
3948 Replace nonimmediate_operand with register_operand.
3949 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
3951 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
3954 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
3956 PR rtl-optimization/87596
3957 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
3958 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
3959 for instructions in FROM..TO range.
3961 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
3963 * cfgexpand.c (expand_one_var): Use specific wording in error message
3964 for non-local frame variables.
3965 * stor-layout.c (layout_decl): Do not issue a warning for them.
3967 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
3969 * haifa-sched.c (priority): Add force_recompute parameter.
3970 (apply_replacement): Call priority () with force_recompute = true.
3971 (restore_pattern): Likewise.
3973 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3975 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
3976 HOST_BITS_PER_WIDE_INT.
3977 (test_vector_ops_duplicate): Likewise.
3979 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3982 * config/i386/sse.md (VF_AVX512): New.
3983 (avx512bcst): Likewise.
3984 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
3986 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
3988 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
3991 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
3993 * doc/invoke.texi (-dumpversion): Improve grammar.
3994 (-dumpfullversion): Make more consistent with -dumpversion.
3996 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
3998 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
3999 Set JUMP_LABEL to the jump insn.
4000 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
4001 Predict emitted jump and add label to jump insn.
4003 2018-10-18 David Malcolm <dmalcolm@redhat.com>
4005 PR tree-optimization/87562
4006 * input.c (get_substring_ranges_for_loc): Use
4007 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
4008 getting the linemap for the endpoint. Verify that it's either
4009 in the same linemap as the start point's spelling location, or
4010 at least in the same file.
4012 2018-10-18 Richard Biener <rguenther@suse.de>
4014 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
4015 feed width-specific load/store costs through ix86_vec_cost.
4016 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
4017 (k8_cost): Likewise.
4018 (bdver_cost): Likewise.
4019 (znver1_cost): Likewise.
4020 (btver1_cost): Likewise.
4021 (btver2_cost): Likewise.
4023 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4025 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
4026 to simplify subreg of vec_merge.
4028 2018-10-18 Richard Biener <rguenther@suse.de>
4030 * config/i386/i386.c: Fix costing of vector FMA.
4032 2018-10-18 Richard Biener <rguenther@suse.de>
4034 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
4036 (ix86_builtin_vectorization_cost): For vec_construct properly
4037 cost insertion into SSE regs.
4038 (...): Adjust calls to ix86_vec_cost.
4040 2018-10-18 Richard Biener <rguenther@suse.de>
4044 2018-02-07 Richard Biener <rguenther@suse.de>
4046 PR tree-optimization/84204
4047 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
4050 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4053 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
4055 (test_vector_ops_duplicate): Add test for a scalar subreg of a
4056 VEC_MERGE of a VEC_DUPLICATE.
4058 2018-10-17 Joseph Myers <joseph@codesourcery.com>
4060 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
4061 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
4062 * doc/standards.texi (C Language): Document C2X.
4063 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
4064 (rl78_option_override): Handle "GNU C2X" language name.
4066 2018-10-17 Joseph Myers <joseph@codesourcery.com>
4068 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
4069 Document C17 as published in 2018.
4071 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
4074 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
4075 bail out if both sides do not have the same storage order.
4077 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4079 * bitmap.c (bitmap_head::dump): New.
4080 * bitmap.h (bitmap_head): Add dump().
4081 * gimple-ssa-evrp-analyze.c
4082 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
4083 (evrp_range_analyzer::set_ssa_range_info): Same.
4084 (evrp_range_analyzer::record_ranges_from_phis): Same.
4085 (evrp_range_analyzer::record_ranges_from_stmt): Same.
4086 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
4087 * gimple-ssa-sprintf.c (get_int_range): Same.
4088 (format_integer): Same.
4089 (sprintf_dom_walker::handle_gimple_call): Same.
4090 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
4091 (ipcp_vr_lattice::top_p): Same.
4092 (ipcp_vr_lattice::bottom_p): Same.
4093 (ipcp_vr_lattice::set_to_bottom): Same.
4094 (ipa_vr_operation_and_type_effects): Same.
4095 (propagate_vr_across_jump_function): Same.
4096 (ipcp_store_vr_results): Same.
4097 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
4098 (ipa_print_node_jump_functions_for_edge): Same.
4099 (ipa_get_value_range): Same.
4100 (ipa_compute_jump_functions_for_edge): Same.
4101 (ipa_write_jump_function): Same.
4102 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
4103 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
4105 * vr-values.c (set_value_range_to_nonnegative): Same.
4106 (set_value_range_to_truthvalue): Same.
4107 (vr_values::get_value_range): Same.
4108 (vr_values::set_defs_to_varying): Same.
4109 (vr_values::update_value_range): Same.
4110 (symbolic_range_based_on_p): Same.
4111 (vr_values::op_with_boolean_value_range_p): Same.
4112 (vr_values::extract_range_for_var_from_comparison_expr): Same.
4113 (vr_values::extract_range_from_ssa_name): Same.
4114 (vr_values::extract_range_from_binary_expr): Same.
4115 (vr_values::extract_range_from_unary_expr): Same.
4116 (vr_values::extract_range_from_cond_expr): Same.
4117 (vr_values::extract_range_from_comparison): Same.
4118 (vr_values::check_for_binary_op_overflow): Same.
4119 (vr_values::extract_range_basic): Same.
4120 (vr_values::extract_range_from_assignment): Same.
4121 (compare_ranges): Same.
4122 (compare_range_with_value): Same.
4123 (vr_values::adjust_range_with_scev): Same.
4124 (vrp_valueize): Same.
4125 (vrp_valueize_1): Same.
4126 (vr_values::get_vr_for_comparison): Same.
4127 (vr_values::compare_name_with_value): Same.
4128 (vr_values::compare_names): Same.
4129 (vr_values::vrp_evaluate_conditional): Same.
4130 (find_case_label_ranges): Same.
4131 (vr_values::vrp_visit_switch_stmt): Same.
4132 (vr_values::extract_range_from_phi_node): Same.
4133 (vr_values::simplify_div_or_mod_using_ranges): Same.
4134 (vr_values::simplify_bit_ops_using_ranges): Same.
4135 (test_for_singularity): Same.
4136 (range_fits_type_p): Same.
4137 (vr_values::simplify_cond_using_ranges_1): Same.
4138 (vr_values::simplify_switch_using_ranges): Same.
4139 (vr_values::simplify_float_conversion_using_ranges): Same.
4140 (vr_values::two_valued_val_range_p): Same.
4141 (vr_values::add_equivalence): Move to value_range::equiv_add.
4142 * vr-values.h (vr_values::add_equivalence): Remove.
4143 (VR_INITIALIZER): Remove.
4144 * tree-vrp.c (value_range::set): New.
4145 (value_range::equiv_add): New.
4146 (value_range::value_range): New.
4147 (value_range::deep_copy): New.
4148 (value_range::check): New.
4149 (value_range::equal_p): New.
4150 (value_range::ignore_equivs_equal_p): New.
4151 (value_range::operator==): New.
4152 (value_range::operator!=): New.
4153 (value_range::symbolic_p): New.
4154 (value_range::numeric_p): New.
4155 (value_range::set_undefined): New.
4156 (value_range::set_varying): New.
4157 (value_range::may_contain_p): New.
4158 (value_range::equiv_clear): New.
4159 (value_range::singleton_p): New.
4160 (value_range::intersect): New.
4161 (value_range::dump): New.
4162 (value_range::set_and_canonicalize): New.
4163 (set_value_range): Adjust for value_range API.
4164 (set_value_range_to_undefined): Same.
4165 (set_value_range_to_varying): Same.
4166 (set_and_canonicalize_value_range): Same.
4167 (set_value_range_to_nonnull): Same.
4168 (set_value_range_to_null): Same.
4169 (range_is_null): Same.
4170 (range_is_nonnull): Same.
4171 (range_int_cst_p): Same.
4172 (range_int_cst_singleton_p): Same.
4173 (symbolic_range_p): Same.
4174 (range_includes_zero_p): Same.
4175 (value_range_constant_singleton): Same.
4176 (vrp_set_zero_nonzero_bits): Same.
4177 (ranges_from_anti_range): Same.
4178 (extract_range_into_wide_ints): Same.
4179 (extract_range_from_multiplicative_op): Same.
4180 (set_value_range_with_overflow): Same.
4181 (extract_range_from_binary_expr_1): Same.
4182 (extract_range_from_unary_expr): Same.
4183 (dump_value_range): Same.
4184 (debug_value_range): Same.
4185 (vrp_prop::check_array_ref): Same.
4186 (vrp_prop::check_mem_ref): Same.
4187 (vrp_prop::vrp_initialize): Same.
4188 (vrp_prop::visit_stmt): Same.
4189 (intersect_ranges): Same.
4190 (vrp_prop::visit_phi): Same.
4191 (vrp_prop::vrp_finalize): Same.
4192 (determine_value_range_1): Same.
4193 (determine_value_range): Same.
4194 (vrp_intersect_ranges_1): Rename to...
4195 (vrp_intersect_1): this.
4196 (vrp_intersect_ranges): Rename to...
4197 (value_range::intersect_helper): ...this.
4198 (vrp_meet_1): Rename to...
4199 (value_range::union_helper): ...this.
4200 (vrp_meet): Rename to...
4201 (value_range::union_): ...this.
4202 (copy_value_range): Remove.
4203 * tree-vrp.h (struct value_range): Rewrite into a proper class.
4204 (value_range::vrtype): New.
4205 (value_range::type): New.
4206 (value_range::equiv): New.
4207 (value_range::min): New.
4208 (value_range::max): New.
4209 (value_range::varying_p): New.
4210 (value_range::undefined_p): New.
4211 (value_range::null_p): New.
4212 (value_range::equiv_add): New.
4213 (copy_value_range): Remove.
4215 2018-10-17 David Malcolm <dmalcolm@redhat.com>
4217 * Makefile.in (SELFTEST_TARGETS): New.
4218 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
4219 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
4220 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
4222 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
4223 (selftest-c++-gdb, selftest-c++-valgrind): Move to
4225 * configure: Regenerate.
4226 * configure.ac (selftest_languages): New.
4228 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4230 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
4231 overflow wraps argument.
4232 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
4233 wide_int_range_multiplicative_op.
4234 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
4235 overflow wraps argument.
4236 (wide_int_range_multiplicative_op): Same.
4237 (wide_int_range_lshift): Same.
4238 (wide_int_range_div): Same.
4239 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
4240 (wide_int_range_lshift): Same.
4241 (wide_int_range_div): Same.
4243 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4245 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
4246 use sign as argument.
4247 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
4248 wide_int_range_shift_undefined_p.
4250 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4252 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
4254 (@despeculate_copy<ALLI_TI:mode>): ... This.
4255 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
4258 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4260 * config.gcc: Obsolete *-*-solaris2.10*.
4261 * doc/install.texi (Specific, *-*-solaris2*): Document it.
4263 2018-10-12 Jeff Law <law@redhat.com>
4265 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
4266 reg + sym +- const_int addressing modes.
4268 2018-10-15 David Malcolm <dmalcolm@redhat.com>
4270 * common.opt (fdiagnostics-minimum-margin-width=): New option.
4271 * diagnostic-show-locus.c (layout::layout): Apply the minimum
4273 (layout::start_annotation_line): Only print up to 3 of the
4274 margin character, to avoid touching the left-hand side.
4275 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
4276 minimum margin width, as set by test_diagnostic_context's ctor.
4277 (selftest::test_fixit_insert_containing_newline): Likewise.
4278 (selftest::test_fixit_insert_containing_newline_2): Likewise.
4279 (selftest::test_line_numbers_multiline_range): Clear
4280 dc.min_margin_width.
4281 * diagnostic.c (diagnostic_initialize): Initialize
4283 * diagnostic.h (struct diagnostic_context): Add field
4285 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
4286 * opts.c (common_handle_option): Handle
4287 OPT_fdiagnostics_minimum_margin_width_.
4288 * selftest-diagnostic.c
4289 (selftest::test_diagnostic_context::test_diagnostic_context):
4290 Initialize min_margin_width to 6.
4291 * toplev.c (general_init): Initialize global_dc->min_margin_width.
4293 2018-10-15 David Malcolm <dmalcolm@redhat.com>
4295 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
4296 Fix usage of "error_at_rich_loc" in the comment.
4298 2018-10-15 Renlin Li <renlin.li@arm.com>
4301 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
4302 if-conversioned loop when it contains ifn with types not
4303 supported by backend.
4304 * internal-fn.c (expand_direct_optab_fn): Add an assert.
4305 (direct_internal_fn_supported_p): New helper function.
4306 * internal-fn.h (direct_internal_fn_supported_p): Declare.
4308 2018-10-15 Jakub Jelinek <jakub@redhat.com>
4311 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4312 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4313 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4315 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
4317 PR tree-optimization/87022
4318 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
4319 bits in dist vector rather than the first one.
4321 2018-10-15 Richard Biener <rguenther@suse.de>
4324 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
4325 (visit_loadstore): When a used restrict tag escaped verify that
4326 the points-to solution of "other" pointers do not include
4328 (compute_dependence_clique): If a used restrict tag escaped
4329 communicated that down to visit_loadstore.
4331 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
4333 * config/s390/s390.c (s390_expand_vec_init): Force vector element
4334 into reg if it isn't a general operand.
4336 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4339 * config/i386/sse.md (*vec_dupv2di): Add register source to
4342 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4345 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4346 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
4347 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4348 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4350 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
4352 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
4353 (notice_args_size): Set it in the current trace if no insn that can
4354 throw internally has been seen yet.
4355 (connect_traces): When connecting args_size between traces, allow the
4356 incoming values not to match if there is an insn setting it before the
4357 first insn that can throw internally; in that case, force the creation
4358 of a CFI note on this latter insn.
4360 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
4362 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
4363 the base class, because using the injected-class-name was not clearly
4364 specified until DR 176.
4366 2018-10-12 Paul Koning <ni1d@arrl.net>
4368 * config/pdp11/pdp11.md (doloop_end): New expander.
4369 (doloop_end_insn): renamed from "doloop_end".
4370 (addqi3): New pattern.
4371 (subqi3): New pattern.
4372 * config/pdp11/predicates.md (incdec_operand): New predicate.
4374 2018-10-12 Yury Gribov <tetra2005@gmail.com>
4377 * real.c (format_helper::can_represent_integral_type_p): New function
4378 * real.h (format_helper::can_represent_integral_type_p): Ditto.
4379 * match.pd: New pattern.
4381 2018-10-12 Alexandre Oliva <oliva@adacore.com>
4383 * configure.ac: Introduce --enable-large-address-aware
4384 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
4385 * doc/install.texi: Document it.
4386 * configure, config.in: Rebuilt.
4387 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
4388 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
4389 (LINK_SPEC): Insert it.
4390 * config/i386/mingw-w64.h: Likewise.
4392 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
4394 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
4396 PR rtl-optimization/87600
4397 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
4399 2018-10-12 Paul Koning <ni1d@arrl.net>
4401 * doc/md.texi (doloop_end): Document that the pattern code may
4402 need to check operand mode.
4404 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
4406 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
4407 to zero-extend between int and floating-point registers.
4408 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
4409 ldp into floating-point registers. Add type and arch attributes.
4410 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
4411 Use f_loads for type attribute.
4413 2018-10-11 Martin Sebor <msebor@redhat.com>
4415 * doc/extend.texi (attribute packed): Correct typos.
4417 2018-10-11 Martin Sebor <msebor@redhat.com>
4419 * doc/extend.texi (attribute flatten): Mention interaction with
4422 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
4425 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
4426 Set new_decl virtual flag to zero.
4428 2018-10-11 Martin Sebor <msebor@redhat.com>
4431 * doc/extend.texi (attribute format_arg): Discuss using multiple
4432 attributes on a single function.
4434 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
4436 PR tree-optimization/86829
4437 * match.pd (sin (atan (x))): New simplification rules.
4438 (cos (atan (x))): Likewise.
4439 * real.c (build_sinatan_real): New function.
4440 * real.h (build_sinatan_real): Prototype.
4442 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4444 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
4446 (fold_mergeeo_helper): New helper function.
4447 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
4448 intrinsics. Correct some whitespace indentation issues.
4450 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
4453 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
4454 Use HOST_WIDE_INT_1U for shift.
4456 2018-10-11 Doug Rupp <rupp@adacore.com>
4457 Olivier Hainque <hainque@adacore.com>
4459 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
4460 Pass --relax to the linker for RTPs.
4461 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
4463 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
4464 Jan Hubicka <jh@suse.cz>
4465 Martin Jambor <mjambor@suse.cz>
4467 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
4468 the same elements are repeated rather than printing all of them.
4469 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
4470 "repeated" elements.
4471 * read-rtl-function.c (test_loading_repeat): New function.
4472 (read_rtl_function_c_tests): Call test_loading_repeat.
4473 * rtl-tests.c (test_dumping_repeat): New function.
4474 (rtl_tests_c_tests): Call test_dumping_repeat.
4476 2018-10-11 Richard Biener <rguenther@suse.de>
4478 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
4479 bdver?_cost): Unify to ...
4480 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
4481 * config/i386/i386.c (processor_cost_table): Adjust.
4483 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
4486 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
4487 the thunk when expanding to GIMPLE.
4489 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
4491 * varasm.c (mergeable_string_section): Don't try to move zero-length
4492 strings to the merge section.
4494 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
4497 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
4499 2018-10-10 Jakub Jelinek <jakub@redhat.com>
4502 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
4503 to special_args set.
4505 2018-10-10 Richard Biener <rguenther@suse.de>
4507 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
4508 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
4509 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
4510 and recursing and pattern terminating the recursion on SSE
4511 vector width using ix86_expand_reduc.
4512 (reduc_sminmax_scal_<mode>): Split into part reducing to half
4513 width and recursing and SSE2 vector variant doing the final
4514 reduction with ix86_expand_reduc.
4515 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
4516 with terminating the recursion at AVX level, splitting that
4519 2018-10-09 David Malcolm <dmalcolm@redhat.com>
4521 * genmatch.c (error_cb): Rename to...
4522 (diagnostic_cb): ...this, converting int params to enums.
4523 (fatal_at): Update for renaming.
4524 (warning_at): Likewise.
4526 * input.c (selftest::ebcdic_execution_charset::apply):
4527 Update for renaming of...
4528 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
4530 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
4531 converting level and reason to enums.
4532 (class selftest::lexer_error_sink): Rename to...
4533 (class selftest::lexer_test_options): ...this, renaming field
4534 "m_errors" to "m_diagnostics".
4535 (selftest::lexer_test_options::apply): Update for renaming of...
4536 (selftest::lexer_test_options::on_error): ...this, renaming to...
4537 (selftest::lexer_test_options::on_diagnostic): ...this
4538 converting level and reason to enums.
4539 (selftest::test_lexer_string_locations_raw_string_unterminated):
4540 Update for renamings.
4541 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
4544 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
4546 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
4547 * config/rs6000/pmmintrin.h: New file.
4549 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
4551 PR tree-optimization/86659
4552 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
4554 2018-10-09 Richard Biener <rguenther@suse.de>
4556 PR tree-optimization/63155
4557 * tree-ssa-structalias.c: Include tree-ssa.h.
4558 (get_constraint_for_ssa_var): For undefs return nothing_id.
4559 (find_func_aliases): Cleanup PHI handling.
4561 2018-10-09 Richard Biener <rguenther@suse.de>
4563 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
4566 2018-10-09 Martin Liska <mliska@suse.cz>
4568 * asan.c (asan_emit_stack_protection): If a stack variable
4569 is located in a same file as current function, then emit
4570 line info into variable definition string.
4572 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4574 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
4577 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4579 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
4582 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4584 PR tree-optimization/86659
4585 * gimple-match.h (struct gimple_match_op): Add reverse field.
4586 (gimple_match_op::set_op): New overloaded method.
4587 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
4588 the REF_REVERSE_STORAGE_ORDER flag on the value.
4589 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
4590 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
4592 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
4595 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
4596 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
4598 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
4601 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
4602 Defined with __builtin_ia32_vfmaddsubpd512_mask.
4604 2018-10-08 Richard Biener <rguenther@suse.de>
4606 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
4607 cost the same as AVX128 ones.
4609 2018-10-08 Paul Koning <ni1d@arrl.net>
4611 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
4612 (expand_block_move): New function.
4613 * config/pdp11/pdp11.c (output_block_move): Remove.
4614 (expand_block_move): New function.
4615 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
4616 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
4617 (*movmemhi1): Remove.
4619 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
4621 * config/s390/2827.md: Increase latencies for some FP instructions.
4623 2018-10-08 Richard Biener <rguenther@suse.de>
4625 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
4627 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
4628 * tree-vectorizer.h (dump_stmt_cost): Adjust.
4629 (add_stmt_cost): Dump return value of the hook.
4631 2018-10-08 Richard Biener <rguenther@suse.de>
4633 PR tree-optimization/63155
4634 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
4635 (ssa_propagation_engine::ssa_propagate): Remove redundant
4636 bitmap bit clearing.
4638 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
4640 PR rtl-optimization/86939
4641 PR rtl-optimization/87479
4642 * ira.h (non_conflicting_reg_copy_p): New prototype.
4643 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
4644 (make_hard_regno_dead): Don't add conflicts for register
4645 ignore_reg_for_conflicts.
4646 (make_object_dead): Likewise.
4647 (non_conflicting_reg_copy_p): New function.
4648 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
4649 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
4650 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
4651 (make_hard_regno_dead): Don't add conflicts for register
4652 ignore_reg_for_conflicts. Remove special conflict handling of
4653 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
4654 check_pic_pseudo_p and update callers.
4655 (mark_pseudo_dead): Don't add conflicts for register
4656 ignore_reg_for_conflicts.
4657 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
4659 2018-10-05 Andrew Waterman <andrew@sifive.com>
4660 Jim Wilson <jimw@sifive.com>
4662 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
4663 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
4664 new pattern using HONOR_SNANS that emits one extra instruction.
4666 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4668 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
4669 patterns): Merge SI and DI patterns to a GPR pattern.
4670 (unnamed define_insn and define_split for record form of that): Merge
4671 to a single define_insn_and_split pattern.
4673 2018-10-05 David Malcolm <dmalcolm@redhat.com>
4676 * input.c (expand_location_to_spelling_point): Add param "aspect"
4677 and use rather than hardcoding LOCATION_ASPECT_CARET.
4678 (get_substring_ranges_for_loc): Handle the case of a single token
4679 within a macro expansion.
4680 * input.h (expand_location_to_spelling_point): Add "aspect" param,
4681 defaulting to LOCATION_ASPECT_CARET.
4683 2018-10-05 Paul Koning <ni1d@arrl.net>
4685 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
4686 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
4687 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
4688 (pdp11_guard_type): New function.
4690 2018-10-05 Paul Koning <ni1d@arrl.net>
4692 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
4693 * config/pdp11/pdp11.opt (mfloat32): Remove.
4695 * doc/invoke.texi (pdp11 -mfloat32): Remove:
4696 (pdp11 -mfloat64): Remove.
4698 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4700 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
4701 (*cmp<mode>_cc_i387): Ditto.
4702 (*cmpu<mode>_cc_i387): Ditto.
4703 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4704 * config/i386/i386.c (ix86_expand_fp_compare): Remove
4706 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
4707 Emit x86_sahf_1 pattern.
4708 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
4709 (ix86_expand_carry_flag_compare): Ditto.
4711 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4713 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
4714 to reg_or_0_operand. Add "C" constraint.
4715 (*cmpxf_cc_i387): Ditto.
4716 (*cmp<mode>_i387): Change operand 2 predicate
4717 to nonimm_or_0_operand. Add "C" constraint.
4718 (*cmp<mode>_cc_i387): Ditto.
4719 (*cmp<mode>_0_i387): Remove insn pattern.
4720 (*cmp<mode>_0_cc_i387): Ditto.
4722 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4724 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
4725 * config/i386/predicates.md (nonimm_or_0_operand): Rename
4726 from vector_move_operand. Update all uses.
4728 2018-10-05 Martin Sebor <msebor@redhat.com>
4730 PR tree-optimization/87490
4731 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
4734 2018-10-05 Richard Biener <rguenther@suse.de>
4736 PR tree-optimization/63155
4737 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
4738 vertical space in dumpfiles.
4739 * tree-ssa-propagate.h
4740 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
4741 * tree-ssa-propagate.c (cfg_blocks_back): New global.
4742 (ssa_edge_worklist_back): Likewise.
4743 (curr_order): Likewise.
4744 (cfg_blocks_get): Remove abstraction.
4745 (cfg_blocks_add): Likewise.
4746 (cfg_blocks_empty_p): Likewise.
4747 (add_ssa_edge): Add to current or next worklist based on
4749 (add_control_edge): Likewise.
4750 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
4752 (ssa_propagation_engine::ssa_propagate): ... here. Unify
4753 iteration from CFG and SSA edge worklist so we process
4754 everything in RPO order, prioritizing forward progress
4756 (ssa_prop_init): Allocate new worklists, do not dump
4758 (ssa_prop_fini): Free new worklists.
4760 2018-10-05 Richard Biener <rguenther@suse.de>
4762 * tree-core.h (tree_block::abstract_flag): Remove.
4763 (tree_block::block_num): Make full 32bits.
4764 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
4765 * tree.h (BLOCK_ABSTRACT): Remove.
4766 * dwarf2out.c (gen_lexical_block_die): Remove dead code
4767 resulting from BLOCK_ABSTRACT being always false.
4768 (gen_inlined_subroutine_die): Likewise.
4769 (gen_block_die): Likewise.
4770 * tree.c (block_ultimate_origin): Likewise.
4771 * tree-pretty-print.c (dump_block_node): Remove code dealing
4772 with BLOCK_ABSTRACT.
4773 * tree-ssa-live.c (dump_scope_block): Likewise.
4774 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
4775 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
4777 2018-10-05 Richard Biener <rguenther@suse.de>
4779 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
4780 is asked for initialize mode to the component mode of the
4783 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
4786 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
4787 assembler for -mavx.
4788 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
4790 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4793 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
4795 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
4796 to be DImode. When using mffscrn, force the operand to a register.
4798 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
4800 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
4801 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
4802 X87MODEF mode iterator.
4803 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
4804 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
4805 X87MODEF mode iterator.
4807 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
4809 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
4810 -Wno-prio-ctor-dtor.
4812 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4814 * Makefile.in (OBJS): Add opt-problem.o.
4815 * dump-context.h: Include "selftest.h.
4816 (selftest::temp_dump_context): New forward decl.
4817 (class dump_context): Make friend of class
4818 selftest::temp_dump_context.
4819 (dump_context::dump_loc_immediate): New decl.
4820 (class dump_pretty_printer): Move here from dumpfile.c.
4821 (class temp_dump_context): Move to namespace selftest.
4822 (temp_dump_context::temp_dump_context): Add param
4823 "forcibly_enable_dumping".
4824 (selftest::verify_dumped_text):
4825 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
4826 (selftest::verify_item):
4827 (ASSERT_IS_TEXT): Move here from dumpfile.c.
4828 (ASSERT_IS_TREE): Likewise.
4829 (ASSERT_IS_GIMPLE): Likewise.
4830 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
4832 (dump_context::dump_loc_immediate): ...this new function.
4833 (class dump_pretty_printer): Move to dump-context.h.
4834 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
4835 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
4836 (temp_dump_context::temp_dump_context): Move to "selftest"
4837 namespace. Add param "forcibly_enable_dumping", and use it to
4838 conditionalize the use of m_pp;
4839 (selftest::verify_dumped_text): Make non-static.
4840 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
4841 (selftest::verify_item): Make non-static.
4842 (ASSERT_IS_TEXT): Move to dump-context.h.
4843 (ASSERT_IS_TREE): Likewise.
4844 (ASSERT_IS_GIMPLE): Likewise.
4845 (selftest::test_capture_of_dump_calls): Pass "true" for new
4846 param of temp_dump_context.
4847 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
4848 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
4850 * opt-problem.cc: New file.
4851 * opt-problem.h: New file.
4852 * optinfo-emit-json.cc
4853 (selftest::test_building_json_from_dump_calls): Pass "true" for
4854 new param of temp_dump_context.
4855 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
4856 (optinfo::emit_for_opt_problem): New function.
4857 (optinfo::emit): Clarity which emit_item is used.
4858 * optinfo.h (optinfo::get_dump_location): New accessor.
4859 (optinfo::emit_for_opt_problem): New decl.
4860 (optinfo::emit): Make const.
4861 * selftest-run-tests.c (selftest::run_tests): Call
4862 selftest::opt_problem_cc_tests.
4863 * selftest.h (selftest::opt_problem_cc_tests): New decl.
4864 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
4865 bool to opt_result, converting fprintf messages to
4866 opt_result::failure_at calls. Add "stmt" param for use by the
4868 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
4869 (runtime_alias_check_p): Convert return type from bool to
4870 opt_result, converting dump_printf calls to
4871 opt_result::failure_at, using the statement DDR_A for their
4873 (find_data_references_in_stmt): Convert return type from bool to
4874 opt_result, converting "return false" to opt_result::failure_at
4876 * tree-data-ref.h: Include "opt-problem.h".
4877 (dr_analyze_innermost): Convert return type from bool to opt_result,
4878 and add a const gimple * param.
4879 (find_data_references_in_stmt): Convert return type from bool to
4881 (runtime_alias_check_p): Likewise.
4882 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
4883 dr_analyze_innermost.
4884 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
4885 Convert return type from bool to opt_result, adding a message for
4886 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
4887 (vect_analyze_data_ref_dependence): Convert return type from bool
4888 to opt_result. Change sense of return type from "false"
4889 effectively meaning "no problems" to "false" meaning a problem,
4890 so that "return false" becomes "return opt_result::success".
4891 Convert "return true" calls to opt_result::failure_at, using
4892 the location of statement A rather than vect_location.
4893 (vect_analyze_data_ref_dependences): Convert return type from bool
4895 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
4896 calls to opt_result::failure_at, using the stmt location rather
4898 (vect_verify_datarefs_alignment): Convert return type from bool
4900 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
4901 into multiple more-tightly-scoped copies.
4902 (vect_analyze_data_refs_alignment): Convert return type from bool
4904 (vect_analyze_data_ref_accesses): Likewise, converting a
4905 "return false" to a "return opt_result::failure_at", adding a
4907 (vect_prune_runtime_alias_test_list): Convert return type from
4908 bool to opt_result, converting dump_printf_loc to
4909 opt_result::failure_at. Add a %G to show the pertinent statement,
4910 and use the stmt's location rather than vect_location.
4911 (vect_find_stmt_data_reference): Convert return type from
4912 bool to opt_result, converting dump_printf_loc to
4913 opt_result::failure_at, using stmt's location.
4914 (vect_analyze_data_refs): Convert return type from bool to
4915 opt_result. Convert "return false" to "return
4916 opt_result::failure_at", adding messages as needed.
4917 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
4918 type from bool to opt_result.
4919 (vect_determine_vf_for_stmt): Likewise.
4920 (vect_determine_vectorization_factor): Likewise, converting
4921 dump_printf_loc to opt_result::failure_at, using location of phi
4922 rather than vect_location.
4923 (vect_analyze_loop_form_1): Convert return type from bool to
4924 opt_result, converting dump_printf_loc calls, retaining the use of
4926 (vect_analyze_loop_form): Convert return type from loop_vec_info
4927 to opt_loop_vec_info.
4928 (vect_analyze_loop_operations): Convert return type from bool to
4929 opt_result, converting dump_printf_loc calls, using the location
4930 of phi/stmt rather than vect_location where available. Convert
4931 various "return false" to "return opt_result::failure_at" with
4932 "unsupported phi" messages.
4933 (vect_get_datarefs_in_loop): Convert return type from bool to
4934 opt_result. Add a message for the
4935 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
4936 (vect_analyze_loop_2): Convert return type from bool to
4937 opt_result. Ensure "ok" is set to a opt_result::failure_at before
4938 each "goto again;", adding new messages where needed.
4939 Add "unsupported grouped {store|load}" messages.
4940 (vect_analyze_loop): Convert return type from loop_vec_info to
4942 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
4944 * tree-vect-stmts.c (process_use): Likewise, converting
4945 dump_printf_loc call and using stmt location, rather than
4947 (vect_mark_stmts_to_be_vectorized): Likeise.
4948 (vect_analyze_stmt): Likewise, adding a %G.
4949 (vect_get_vector_types_for_stmt): Convert return type from bool to
4950 opt_result, converting dump_printf_loc calls and using stmt
4951 location, rather than vect_location.
4952 (vect_get_mask_type_for_stmt): Convert return type from tree to
4953 opt_tree, converting dump_printf_loc calls and using stmt location.
4954 * tree-vectorizer.c: Include "opt-problem.h.
4955 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
4956 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
4957 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
4958 enabled, use it to report at the top level "couldn't vectorize
4959 loop" followed by the problem.
4960 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
4961 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
4963 (vect_analyze_stmt): Likewise.
4964 (vect_get_vector_types_for_stmt): Likewise.
4965 (tree vect_get_mask_type_for_stmt): Likewise.
4966 (vect_analyze_data_ref_dependences): Likewise.
4967 (vect_enhance_data_refs_alignment): Likewise.
4968 (vect_analyze_data_refs_alignment): Likewise.
4969 (vect_verify_datarefs_alignment): Likewise.
4970 (vect_analyze_data_ref_accesses): Likewise.
4971 (vect_prune_runtime_alias_test_list): Likewise.
4972 (vect_find_stmt_data_reference): Likewise.
4973 (vect_analyze_data_refs): Likewise.
4974 (vect_analyze_loop): Convert return type from loop_vec_info to
4976 (vect_analyze_loop_form): Likewise.
4977 (vect_analyze_slp): Convert return type from bool to opt_result.
4979 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4981 * doc/invoke.texi (-fopt-info): Document new "internals"
4983 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
4984 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
4986 (optinfo_verbosity_options): Add "internals".
4987 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
4988 (dump_context::apply_dump_filter_p): New member function.
4989 (dump_context::dump_loc): Use apply_dump_filter_p rather than
4990 explicitly masking the dump_kind.
4991 (dump_context::begin_scope): Increment the scope depth first. Use
4992 apply_dump_filter_p rather than explicitly masking the dump_kind.
4993 (dump_context::emit_item): Use apply_dump_filter_p rather than
4994 explicitly masking the dump_kind.
4995 (dump_dec): Likewise.
4996 (dump_hex): Likewise.
4997 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
4998 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
4999 (opt_info_switch_p): Update handling of default
5000 MSG_OPTIMIZED_LOCATIONS to cope with default of
5001 MSG_PRIORITY_USER_FACING.
5002 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
5003 masking the dump_kind.
5004 (selftest::test_capture_of_dump_calls): Update test_dump_context
5005 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
5006 than MSG_ALL. Generalize scope test to be run at all four
5007 combinations of with/without MSG_PRIORITY_USER_FACING and
5008 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
5009 for each of the two values.
5010 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
5011 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
5012 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
5013 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
5014 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
5015 with MSG_PRIORITY_*.
5016 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
5017 dump messages as MSG_PRIORITY_USER_FACING.
5018 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
5019 about the interaction with MSG_PRIORITY_*.
5021 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5023 * varasm.c (output_constant): Add new parameter merge_strings.
5024 Make strings properly zero terminated in merge string sections.
5025 (mergeable_string_section): Don't fail if the last char is non-zero.
5026 (assemble_variable_contents): Handle merge string sections.
5027 (assemble_variable): Likewise.
5028 (assemble_constant_contents): Likewise.
5029 (output_constant_def_contents): Likewise.
5030 (output_constructor_array_range,
5031 output_constructor_regular_field): Adjust call to output_constant.
5032 (output_object_block): Adjust call to assemble_constant_contents
5033 and assemble_variable_contents.
5035 2018-10-04 Martin Liska <mliska@suse.cz>
5038 * cgraphunit.c (process_function_and_variable_attributes):
5039 Warn about a function with alias attribute and a body.
5041 2018-10-04 Martin Liska <mliska@suse.cz>
5044 * multiple_target.c (redirect_to_specific_clone): New function.
5045 (ipa_target_clone): Use it.
5046 * tree-inline.c: Fix comment.
5048 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5050 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
5052 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
5053 (gcc::dump_manager::register_pass): New member function, adapted
5054 from loop body in gcc::pass_manager::register_pass, adding a
5055 call to update_dfi_for_opt_info.
5056 (gcc::dump_manager::opt_info_enable_passes): Store the
5057 -fopt-info options into the new fields. Move the loop
5059 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
5061 * dumpfile.h (struct opt_pass): New forward decl.
5062 (gcc::dump_manager::register_pass): New decl.
5063 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
5064 (class gcc::dump_manager): Add fields "m_optgroup_flags",
5065 "m_optinfo_flags", and "m_optinfo_filename".
5066 * passes.c (gcc::pass_manager::register_pass): Move all of the
5067 dump-handling code to gcc::dump_manager::register_pass.
5069 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
5071 PR rtl-optimization/87466
5072 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
5073 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
5074 * doc/tm.texi: Regenerate.
5075 * ira-lives.c (process_bb_node_lives): Use the new target hook.
5076 * lra-lives.c (process_bb_lives): Likewise.
5077 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
5080 2018-10-04 Tamar Christina <tamar.christina@arm.com>
5082 * params.c (add_params): Fix initialization.
5084 2018-10-04 Martin Liska <mliska@suse.cz>
5086 PR gcov-profile/84107
5087 * tree-profile.c (init_ic_make_global_vars):
5088 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
5089 Come up with new ic_tuple* variables. Emit
5090 __gcov_indirect_call{,_topn} variables.
5091 (gimple_gen_ic_profiler): Access the variable
5093 (gimple_gen_ic_func_profiler): Access
5094 __gcov_indirect_call.callee field.
5095 (gimple_init_gcov_profiler): Use ptr_type_node.
5096 * value-prof.c (gimple_ic): Use ptr_type_node.
5098 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5100 PR tree-optimization/85787
5101 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
5102 malloc_candidate_p into this function and add support for
5103 detecting multiple phis.
5104 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
5106 2018-10-04 Martin Liska <mliska@suse.cz>
5109 * ipa-inline.c (inline_to_all_callers_1):
5110 Call ultimate_alias_target for node being inlined.
5112 2018-10-03 Jeff Law <law@redhat.com>
5114 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
5116 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
5117 * tree.h (get_typenode_from_name): Prototype.
5119 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
5121 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5122 Change operand 2 predicate to nonimmediate_operand.
5123 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5125 2018-10-03 Martin Sebor <msebor@redhat.com>
5126 Jeff Law <law@redhat.com>
5128 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
5130 (get_string_length): Detect unterminated arrays.
5131 (format_string): Same.
5132 (format_directive): Warn about unterminated arrays.
5133 (handle_gimple_call): Mark statements with no_warning as needed.
5135 2018-10-03 Jim Wilson <jimw@sifive.com>
5137 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
5138 also define __riscv_abi_rve. Delete trailing white space.
5140 2018-10-03 Paul Koning <ni1d@arrl.net>
5142 Enable LRA register allocator for PDP11.
5143 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
5146 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
5147 * config/pdp11/pdp11.opt (-mlra): New option.
5148 * doc/invoke.texi (PDP-11 Options): Document -mlra.
5150 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
5152 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
5153 (*<absneg:code>extend<mode>xf2): Ditto.
5155 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
5157 PR tree-optimization/87415
5158 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
5161 2018-10-02 Jeff Law <law@redhat.com>
5163 * gimple-fold.c (get_range_strlen): Only set *nonstr when
5164 an unterminated string is discovered. Bubble up range
5165 even for unterminated strings.
5166 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
5167 indicates the string was not terminated via NONSTR.
5169 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
5171 * tree-vrp.c (extract_range_from_unary_expr): Special case all
5172 pointer conversions.
5173 Do not do anything special for anti-ranges.
5175 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
5177 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
5178 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
5180 2018-10-03 Martin Liska <mliska@suse.cz>
5182 PR gcov-profile/86109
5183 * coverage.c (coverage_begin_function): Do not
5184 mark lambdas as artificial.
5185 * tree-core.h (struct GTY): Remove tm_clone_flag
5186 and introduce new lambda_function.
5187 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
5189 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
5192 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
5193 P8_VECTOR and VSX are enabled.
5195 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
5197 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5198 0x3907 as CPU model number.
5200 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
5202 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
5203 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
5204 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
5205 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
5206 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
5207 * config/s390/s390.md: Likewise. Rename also the cpu attribute
5208 value from arch12 to z14.
5210 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
5212 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
5214 (isinf<mode>2): Ditto.
5216 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
5218 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
5219 before emitting fxam. Perform calculations in XFmode.
5221 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
5223 * match.pd (((X /[ex] A) +- B) * A): New transformation.
5225 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
5228 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
5229 * tree.c (signed_or_unsigned_type_for): Handle complex.
5231 2018-10-02 Jeff Law <law@redhat.com>
5233 * gimple-fold.c (get_range_strlen): Remove dead code.
5235 2018-10-02 Martin Sebor <msebor@redhat.com>
5236 Jeff Law <law@redhat.com>
5238 * builtins.c (unterminated_array): Add new arguments.
5239 If argument is not terminated, bubble up size and exact
5241 (expand_builtin_strnlen): Detect, avoid expanding
5242 and diagnose unterminated arrays.
5243 (c_strlen): Fill in offset of start of unterminated strings.
5244 * builtins.h (unterminated_array): Update prototype.
5246 2018-10-02 Richard Biener <rguenther@suse.de>
5248 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
5249 of haddv4df, first reduce to SSE width and exploit the fact
5250 that we only need element zero with the reduction result.
5251 (reduc_plus_scal_v2df): Likewise.
5253 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
5255 * dojump.h (do_jump): Delete.
5256 (do_jump_1): Likewise.
5257 (split_comparison): Move around.
5258 * dojump.c (do_jump): Make static.
5259 (do_jump_1): Likewise.
5260 (jumpifnot): Move around.
5261 (jumpifnot_1): Likewise.
5263 (jumpif_1): Likewise.
5264 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
5266 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
5268 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
5269 insns in the delay slot and add_insn_after for the jump insn.
5271 2018-10-02 Richard Biener <rguenther@suse.de>
5273 * tree-inline.c (expand_call_inline): Use the location of
5274 the callee declaration for the inline-entry marker.
5275 * final.c (notice_source_line): Remove special-casing of
5276 NOTE_INSN_INLINE_ENTRY.
5278 2018-10-01 Carl Love <cel@us.ibm.com>
5281 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
5282 (__builtin_mtfsb0): New.
5283 (__builtin_mtfsb1): New.
5284 ( __builtin_set_fpscr_rn): New.
5285 (__builtin_set_fpscr_drn): New.
5286 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
5287 (rs6000_expand_set_fpscr_rn_builtin): Add.
5288 (rs6000_expand_set_fpscr_drn_builtin): Add.
5289 (rs6000_expand_builtin): Add case statement entries for
5290 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
5291 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
5292 RS6000_BUILTIN_MFFSL.
5293 (rs6000_init_builtins): Add ftype initialization and def_builtin
5294 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
5295 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
5296 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
5297 rs6000_mffscdrn): Add define_insn.
5298 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
5299 * doc/extend.texi: Add documentation for the builtins.
5301 2018-10-01 Richard Biener <rguenther@suse.de>
5303 PR tree-optimization/87465
5304 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
5305 causing branch miscounts.
5307 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5309 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
5310 aarch64_option_default_param): New.
5311 (params.h): Include.
5312 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
5313 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
5314 stack-clash protection validation code.
5316 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5318 * params.c (validate_param): New.
5319 (add_params): Use it.
5320 (set_param_value): Refactor param validation into validate_param.
5321 (diagnostic.h): Include.
5322 * diagnostic.h (diagnostic_ready_p): New.
5324 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5326 * params.c (set_param_value):
5327 Add index of parameter being validated.
5328 * common/common-target.def (option_validate_param): New.
5329 * common/common-targhooks.h (default_option_validate_param): New.
5330 * common/common-targhooks.c (default_option_validate_param): New.
5331 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
5332 * doc/tm.texi: Regenerate.
5334 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5337 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5338 Add validation for stack-clash parameters and set defaults.
5340 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5343 * configure.ac: Add stack-clash-protection-guard-size.
5344 * doc/install.texi: Document it.
5345 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
5346 * params.def: Update comment for guard-size.
5347 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
5348 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
5349 * configure: Regenerate.
5351 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5354 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
5355 STACK_DYNAMIC_OFFSET): New.
5356 * config/aarch64/aarch64.c (aarch64_layout_frame):
5357 Update outgoing args size.
5358 (aarch64_stack_clash_protection_alloca_probe_range,
5359 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5361 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5364 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
5366 * target.def (stack_clash_protection_alloca_probe_range): New.
5367 (stack_clash_protection_final_dynamic_probe): Remove.
5368 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
5369 (default_stack_clash_protection_final_dynamic_probe): Remove.
5370 * targhooks.c: Likewise.
5371 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5372 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
5373 * doc/tm.texi: Regenerate.
5375 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5378 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
5379 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
5380 aarch64_clamp_to_uimm12_shift): New.
5381 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
5382 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
5384 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5387 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
5389 2018-10-01 Jeff Law <law@redhat.com>
5390 Richard Sandiford <richard.sandiford@linaro.org>
5391 Tamar Christina <tamar.christina@arm.com>
5394 * config/aarch64/aarch64.md
5395 (probe_stack_range): Add k (SP) constraint.
5396 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
5397 STACK_CLASH_MAX_UNROLL_PAGES): New.
5398 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
5399 stack probes for stack clash.
5400 (aarch64_allocate_and_probe_stack_space): New.
5401 (aarch64_expand_prologue): Use it.
5402 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
5403 (aarch64_sub_sp): Add emit_move_imm optional param.
5405 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
5407 PR tree-optimization/87261
5408 * match.pd: Remove trailing whitespace.
5409 Add (x & y) | ~(x | y) -> ~(x ^ y),
5410 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
5412 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
5414 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
5416 (ashlsi3_insn): Update instruction constraints.
5417 (ashrsi3_insn): Likewise.
5418 (rotrsi3): Likewise.
5419 (add_shift): Likewise.
5420 * config/arc/constraints.md (Csz): New 32 bit constraint. It
5421 avoids placing in the limm field small constants which, otherwise,
5422 could end into a small instruction.
5424 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
5426 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
5427 destination register is not odd-even.
5428 (umaddsidi4_split): Likewise.
5430 2018-10-01 Richard Biener <rguenther@suse.de>
5432 * tree-inline.c (expand_call_inline): Store origin of fn
5433 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
5434 * tree.c (block_ultimate_origin): Simplify and do some
5437 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
5439 * config/i386/mmx.md (EMMS): New int iterator.
5440 (emms): New int attribute.
5441 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
5442 EMMS int iterator. Explicitly declare clobbers.
5443 (mmx_emms): Remove expander.
5445 * config/i386/predicates.md (emms_operation): Remove predicate.
5446 (vzeroall_pattern): New predicate.
5447 (vzeroupper_pattern): Rename from vzeroupper_operation.
5448 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
5449 vzeroupper_pattern and vzeroall_pattern predicates.
5451 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
5453 PR rtl-optimization/86939
5454 * ira-lives.c (make_hard_regno_born): Rename from this...
5455 (make_hard_regno_live): ... to this. Remove update to conflict
5456 information. Update function comment.
5457 (make_hard_regno_dead): Add conflict information update. Update
5459 (make_object_born): Rename from this...
5460 (make_object_live): ... to this. Remove update to conflict information.
5461 Update function comment.
5462 (make_object_dead): Add conflict information update. Update function
5464 (mark_pseudo_regno_live): Call make_object_live.
5465 (mark_pseudo_regno_subword_live): Likewise.
5466 (mark_hard_reg_dead): Update function comment.
5467 (mark_hard_reg_live): Call make_hard_regno_live.
5468 (process_bb_node_lives): Likewise.
5469 * lra-lives.c (make_hard_regno_born): Rename from this...
5470 (make_hard_regno_live): ... to this. Remove update to conflict
5471 information. Remove now uneeded check_pic_pseudo_p argument.
5472 Update function comment.
5473 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
5474 to conflict information. Update function comment.
5475 (mark_pseudo_live): Remove update to conflict information. Update
5477 (mark_pseudo_dead): Add conflict information update.
5478 (mark_regno_live): Call make_hard_regno_live.
5479 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
5480 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
5482 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
5485 * config/i386/i386.c (construct_container): Use TImode for
5486 BLKmode values in 2 integer registers.
5488 2018-09-29 Jeff Law <law@redhat.com>
5490 * builtins.c (unterminated_array): Pass in c_strlen_data * to
5491 c_strlen rather than just a tree *.
5492 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
5493 Update recursive calls appropriately. If caller did not provide a
5494 suitable data pointer, create a local one. When a non-terminated
5495 string is discovered, bubble up information about the string via the
5496 c_strlen_data object.
5497 * builtins.h (c_strlen): Update prototype.
5498 (c_strlen_data): New structure.
5499 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
5500 For a type 2 call, if c_strlen indicates a non-terminated string
5501 use the length of the non-terminated string.
5502 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
5504 2018-09-29 Jakub Jelinek <jakub@redhat.com>
5507 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
5508 __m512d type for __A argument rather than __m512.
5510 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
5512 * match.pd (simple_comparison): Don't optimize if either operand is
5513 a function pointer when target needs function pointer canonicalization.
5515 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5517 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
5518 power5 .. power9 to remove indirection.
5519 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
5520 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
5521 ASM_CPU_476_SPEC): Delete.
5522 (ASM_CPU_SPEC): Adjust.
5523 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
5524 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
5526 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5528 * config.in: Delete HAVE_AS_DCI.
5529 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
5530 * config/rs6000/rs6000.h: Ditto.
5531 * configure.ac: Delete HAVE_AS_DCI.
5532 * configure: Regenerate.
5534 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5536 * config.in (HAVE_AS_LWSYNC): Delete.
5537 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
5538 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
5540 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
5541 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
5543 * configure.ac: Delete HAVE_AS_LWSYNC.
5544 * configure: Regenerate.
5546 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
5547 Pierre-Marie de Rodat <derodat@adacore.com>
5549 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
5550 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
5551 (cgraph_node::create_thunk): Add indirect_offset parameter.
5552 (thunk_adjust): Likewise.
5553 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
5554 and initialize the corresponding field with it.
5555 (cgraph_node::dump): Dump indirect_offset field.
5556 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
5557 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
5558 (thunk_adjust): Add indirect_offset parameter and deal with it.
5559 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
5560 pass it to thunk_adjust. Do not call the target hook if it's non-zero
5561 or if the thunk is external or local. Fix formatting. Do not chain
5562 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
5563 if any, in the GIMPLE representation.
5564 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
5565 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
5566 (input_node): Read indirect_offset field.
5567 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
5568 call to thunk_adjust.
5569 * tree-nested.c (struct nesting_info): Add thunk_p field.
5570 (create_nesting_tree): Set it.
5571 (convert_all_function_calls): Copy static chain from targets to thunks.
5572 (finalize_nesting_tree_1): Return early for thunks.
5573 (unnest_nesting_tree_1): Do not finalize thunks.
5574 (gimplify_all_functions): Do not gimplify thunks.
5576 2018-09-28 David Malcolm <dmalcolm@redhat.com>
5578 * opt-suggestions.c (option_proposer::build_option_suggestions):
5579 Release "option_values".
5581 2018-09-28 David Malcolm <dmalcolm@redhat.com>
5583 * coverage.c (get_coverage_counts): Convert problem-reporting dump
5584 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
5585 * dumpfile.c (kind_as_string): New function.
5586 (dump_loc): Rather than a hardcoded prefix of "note: ", use
5587 kind_as_string to vary the prefix based on dump_kind.
5588 (selftest::test_capture_of_dump_calls): Update for above.
5590 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
5592 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
5593 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
5595 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
5597 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
5598 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
5599 INVALID_REGNUM instead of FPSR_REG.
5600 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
5601 * config/i386/i386.md: Update comment of FP compares.
5602 (fldenv): Do not clobber FPSR_REG.
5604 2018-09-28 Richard Biener <rguenther@suse.de>
5606 * tree.h (BLOCK_ORIGIN): New.
5607 * omp-expand.c (grid_expand_target_grid_body): Assign
5608 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
5609 * tree-inline.c (remap_block): Likewise.
5610 * auto-profile.c (get_function_decl_from_block): Simplify
5611 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
5612 * langhooks.c (lhd_print_error_function): Likewise.
5613 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
5615 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
5616 * tree.c (block_nonartificial_location): Likewise.
5617 (block_ultimate_origin): Likewise.
5618 * tree-pretty-print.c (percent_K_format): Likewise. Remove
5619 no longer needed LTO case.
5621 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
5622 Jan Hubicka <jh@suse.cz>
5623 Martin Jambor <mjambor@suse.cz>
5625 * simplify-rtx.c (simplify_merge_mask): New function.
5626 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
5627 same masks are used in op1 or op2.
5628 (test_vec_merge): New function.
5629 (test_vector_ops): Call test_vec_merge.
5631 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
5633 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
5634 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
5635 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
5636 (cypress_costs): Set it.
5637 (supersparc_costs): Likewise.
5638 (hypersparc_costs): Likewise.
5639 (leon_cost): Likewise.
5640 (leon3_costs): Likewise.
5641 (sparclet_costs): Likewise.
5642 (ultrasparc_costs): Likewise.
5643 (ultrasparc_costs): Likewise.
5644 (niagara_costs): Likewise.
5645 (niagara2_costs): Likewise.
5646 (niagara3_costs): Likewise.
5647 (niagara4_costs): Likewise.
5648 (niagara7_costs): Likewise.
5649 (m8_costs): Likewise.
5650 (TARGET_CAN_FOLLOW_JUMP): Define.
5651 (pass_work_around_errata::gate): Minor tweak.
5652 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
5653 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
5654 Automaitcally clear MASK_FSMULD mask for V7 processors.
5655 (sparc_can_follow_jump): New static function.
5656 (output_ubranch): Deal with CROSSING_JUMP_P.
5657 (sparc_use_sched_lookahead): Rewrite using switch statement.
5658 (sparc_issue_rate): Reorder.
5659 (sparc_branch_cost): New function.
5661 2018-09-27 Martin Sebor <msebor@redhat.com>
5663 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
5664 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
5665 (int_fits_type_p): Same.
5667 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
5669 * config/i386/i386.md (FPCR_REG): Remove.
5670 (UNSPEC_FLDCW): Remove.
5671 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
5672 (x86_fldcw_1): Remove insn pattern.
5673 (fnstenv): Do not clobber FPCR_REG.
5675 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
5676 (CALL_USED_REGISTERS): Ditto.
5677 (REG_ALLOC_ORDER): Ditto.
5678 (REG_CLASS_CONTENTS): Ditto.
5679 (HI_REGISTER_NAMES): Ditto.
5680 (ADDITIONAL_REGISTER_NAMES): Use defines instead
5681 of numerical constants.
5682 * config/i386/i386.c (regclass_map): Remove fpsr register.
5683 (dbx_register_map): Ditto.
5684 (dbx64_register_map): Ditto.
5685 (svr4_dbx_register_map): Ditto.
5686 (print_reg): Do not handle FPCR_REG.
5688 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
5691 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
5692 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
5693 Delete, always treat as true.
5694 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
5695 Ditto. Simplify remaining code.
5696 * config/powerpcspe/powerpcspe.h: Ditto.
5697 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
5698 Simplify remaining code.
5699 (rs6000_expand_builtin): Ditto.
5700 * config/rs6000/rs6000.h: Ditto.
5701 * configure.ac: Ditto.
5702 * configure: Regenerate.
5704 2018-09-27 Martin Liska <mliska@suse.cz>
5706 * coverage.c (get_coverage_counts): Revert the formatting
5707 of missing profile opt info.
5709 2018-09-27 Richard Biener <rguenther@suse.de>
5713 * dwarf2out.c (set_block_origin_self): Do not mark outermost
5714 block as we do not output that.
5715 (gen_inlined_subroutine_die): Elide the originally outermost
5716 block, matching what we do for concrete instances.
5717 (decls_for_scope): Add parameter specifying whether to recurse
5720 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
5721 Tom de Vries <tom@codesourcery.com>
5725 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
5726 STORE_FLAG_VALUE == 1.
5728 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
5730 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
5731 constant definitions.
5732 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
5733 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
5734 ("speculation_barrier"): New expander definition.
5736 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
5738 PR gcov-profile/86957
5739 * common.opt: New warning option -Wmissing-profile.
5740 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
5741 * doc/invoke.texi: Document -Wmissing-profile.
5743 2018-09-26 Jim Wilson <jimw@sifive.com>
5745 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
5746 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
5748 2018-09-26 Martin Sebor <msebor@redhat.com>
5750 * tree.c (zerop): Change return type to bool.
5751 (integer_zerop, integer_onep, integer_each_onep): Same.
5752 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
5753 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
5754 (real_onep, real_minus_onep, chain_index): Same.
5755 (print_type_hash_statistics, type_list_equal): Same.
5756 * tree.h (zerop): Same.
5757 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
5758 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
5759 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
5760 (real_onep, real_minus_onep, chain_index): Same.
5761 (print_type_hash_statistics, type_list_equal): Same.
5763 2018-09-26 Jim Wilson <jimw@sifive.com>
5765 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
5767 2018-09-26 Jakub Jelinek <jakub@redhat.com>
5770 * config/i386/i386.c: Include debug.h and dwarf2out.h.
5771 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
5774 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
5776 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
5778 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
5780 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
5781 and FP_SECOND_SSE_REGS.
5782 (REG_CLASS_NAMES): Ditto.
5783 (REG_CLASS_CONTENTS): Ditto.
5784 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
5785 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
5786 (ix86_preferred_output_reload_class): Ditto.
5787 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
5788 clobber constraint to "=&f".
5789 (fix_truncdi_i387): Ditto.
5790 (lrintxfdi2): Ditto.
5791 (fistdi2_<rounding>): Ditto.
5792 (fpremxf4_i387): Change "=u" constraint to "=f".
5793 (fprem1xf4_i387): Ditto.
5795 (fptanxf4_i387): Ditto.
5796 (fxtractxf3_i387): Ditto.
5797 (fscalexf4_i387): Ditto.
5798 (atan2xf3): Change "u" constraint to "f".
5799 (fyl2xxf3_i387): Ditto.
5800 (fyl2xp1xf3_i387): Ditto.
5802 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
5805 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
5806 for removed I387_MASK_PM entity.
5809 2018-09-26 Jeff Law <law@redhat.com>
5811 2018-09-26 Alexey Neyman <stilor@att.net>
5813 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
5814 headers are no longer pulled in by <isl/val.h>.
5816 2018-09-26 Richard Biener <rguenther@suse.de>
5819 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
5820 or concrete instance DIE to the tree. Create abstract origin
5821 attributes also for concrete instances.
5823 2018-09-26 Alexey Neyman <stilor@att.net>
5825 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
5826 headers are no longer pulled in by <isl/val.h>.
5828 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
5830 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
5831 Use new helper functions.
5832 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
5833 Use new helper functions.
5834 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
5835 aarch_mm_needs_release): New declarations.
5836 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
5837 aarch_mm_needs_release): New.
5839 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
5841 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
5842 (arm32_output_mi_thunk): Deal with long calls.
5844 2018-09-26 Richard Biener <rguenther@suse.de>
5848 * tree-inline.c (expand_call_inline): When the location
5849 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
5850 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
5851 the inserted BLOCK to make inlined_function_outer_scope_p
5853 * dwarf2out.c (add_call_src_coords_attributes): Do not add
5854 coords for reserved locations.
5856 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5858 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
5859 (*call_indirect_nonlocal_sysv<mode>): Ditto.
5860 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
5861 (*sibcall_nonlocal_sysv<mode>): Ditto.
5862 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
5863 (<bd>_<mode>): Ditto.
5864 (<bd>tf_<mode>): Ditto.
5866 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5868 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
5869 control string as a list of templates instead of as C code.
5870 (*altivec_movti): Ditto.
5871 * config/rs6000/darwin.md (movdf_low_di): Ditto.
5873 2018-09-25 Jim Wilson <jimw@sifive.com>
5875 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
5876 when target symbol is weak.
5878 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
5881 * builtins.c (unterminated_array): Simplify.
5882 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
5883 where pointer arithmetic is safe.
5885 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5888 * config/rs6000/altivec.md (altivec_vspltb): Use
5889 const_0_to_15_operand instead of u5bit_cint_operand.
5890 (*altivec_vspltb_internal): Ditto.
5891 (altivec_vspltb_direct): Ditto.
5892 (altivec_vsplth): Use const_0_to_7_operand instead of
5894 (*altivec_vsplth_internal): Ditto.
5895 (altivec_vsplth_direct): Ditto.
5896 (altivec_vspltw): Use const_0_to_3_operand instead of
5898 (*altivec_vspltw_internal): Ditto.
5899 (altivec_vspltw_direct): Ditto.
5900 (altivec_vspltsf): Ditto.
5901 (*altivec_vspltsf_internal): Ditto.
5902 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
5903 various splats with the proper size immediate. Reorder the various
5904 cases by ascending size of immediate, and put all such together.
5906 2018-09-25 Richard Biener <rguenther@suse.de>
5909 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
5911 (maybe_create_die_with_external_ref): Do not create
5912 DW_TAG_imported_unit here.
5913 (add_abstract_origin_attribute): Handle external BLOCK refs.
5914 (dwarf2out_abstract_function): Simplify LTO case.
5915 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
5916 rather than using maybe_create_die_with_external_ref.
5918 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
5921 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
5922 (frndintxf2_mask_pm_i387): Ditto.
5923 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
5924 Enable for !flag_trapping_math.
5925 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
5926 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
5927 Change operand 1 predicate to nonimmediate_operand.
5928 (attr "i387_cw"): Remove mask_pm.
5929 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
5930 (enum ix86_entity): Remove I387_MASK_PM.
5931 * config/i386/i386.c (ix86_i387_mode_needed): Do not
5932 handle I387_MASK_PM.
5933 (ix86_mode_needed): Ditto.
5934 (ix86_mode_after): Ditto.
5935 (ix86_mode_entry): Ditto.
5936 (ix86_mode_exit): Ditto.
5937 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
5939 2018-09-25 Jakub Jelinek <jakub@redhat.com>
5941 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
5942 to_update_switch_stmts to vNULL instead of calling create on them
5945 2018-09-25 Richard Biener <rguenther@suse.de>
5947 PR tree-optimization/87402
5948 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
5949 (visit_phi): Re-instantiate handling of supposed to be VARYING
5950 but non-VARYING backedge value.
5952 2018-09-25 Richard Biener <rguenther@suse.de>
5955 * dwarf2out.c (struct sym_off_pair): New.
5956 (external_die_map): New global.
5957 (lookup_decl_die): When in LTO create DIEs lazily from the
5959 (lookup_block_die): New function, create DIEs lazily in LTO.
5960 (equate_block_to_die): New function.
5961 (dwarf2out_die_ref_for_decl): During WPA get the association
5962 from the external DIE map.
5963 (dwarf2out_register_external_die): Record mapping into the
5965 (maybe_create_die_with_external_ref): New function split out from
5966 DIE generation part of old dwarf2out_register_external_die.
5967 (add_abstract_origin_attribute): Do not return the DIE. When
5968 in LTO reference externals directly.
5969 (dwarf2out_abstract_function): When in LTO ignore calls for
5970 decls with external DIEs (already present abstract instances).
5971 (gen_call_site_die): Adjust.
5972 (add_high_low_attributes): Likewise.
5973 (gen_lexical_block_die): Likewise.
5974 (gen_inlined_subroutine_die): Likewie.
5975 (gen_block_die): Likewise.
5976 (dwarf2out_inline_entry): Likewise.
5977 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
5980 2018-09-25 Martin Liska <mliska@suse.cz>
5982 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
5983 integers and not by a float value.
5985 2018-09-25 Martin Liska <mliska@suse.cz>
5988 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
5990 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
5991 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
5992 * godump.c (go_format_type): Remove extra parenthesis.
5994 2018-09-25 Martin Liska <mliska@suse.cz>
5996 * alias.c (set_dest_equal_p): Remove unused function.
5997 * config/i386/i386.c (def_builtin_pure2): Likewise.
5998 * diagnostic-show-locus.c (class layout): Remove
6000 (layout::layout): Likewise here.
6001 * dump-context.h (class temp_dump_context): Likewise.
6002 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
6003 (add_AT_loclistsptr): Likewise.
6004 (add_AT_offset): Likewise.
6005 (get_AT_hi_pc): Likewise.
6006 (is_comdat_die): Likewise.
6007 (type_is_enum): Likewise.
6008 (ceiling): Likewise.
6009 (add_AT_vms_delta): Likewise.
6010 (is_class_die): Likewise.
6011 * edit-context.c (class line_event): Remove unused field.
6012 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
6014 * ipa-cp.c (ipa_get_vr_lat): Likewise.
6015 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
6016 (ok_for_base_p_nonstrict): Likewise.
6017 * tree-chrec.c (is_not_constant_evolution): Likewise.
6018 (chrec_fold_poly_cst): Likewise.
6019 * tree-if-conv.c (has_pred_critical_p): Likewise.
6020 * tree-ssa-coalesce.c (print_exprs): Likewise.
6021 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
6022 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
6023 * tree-vrp.c (value_ranges_intersect_p): Likewise.
6024 (value_range_nonnegative_p): Likewise.
6026 2018-09-25 Martin Liska <mliska@suse.cz>
6028 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
6029 Do not handle "GNU Pascal".
6030 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
6032 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
6033 from documentation. Likewise.
6034 * dbxout.c (dbxout_range_type): Likewise.
6035 * doc/cpp.texi: Likewise.
6036 * doc/extend.texi: Likewise.
6037 * doc/frontends.texi: Likewise.
6038 * doc/invoke.texi: Remove Pascal entry.
6039 * tree.def (CLEANUP_POINT_EXPR): Likewise.
6040 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
6042 2018-09-25 Martin Liska <mliska@suse.cz>
6045 * doc/invoke.texi: Document all parameters and remove default
6048 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
6051 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
6052 contains HOST_WIDE_INTs when computing its size.
6054 2018-09-24 Jim Wilson <jimw@sifive.com>
6057 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
6059 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
6061 2018-09-24 Andrew Pinski <apinski@marvell.com>
6063 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
6064 access prev before checking it for NULLness in the
6065 AARCH64_FUSE_CMP_BRANCH case.
6067 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
6070 * config/i386/i386.c (rest_of_insert_endbranch): Set
6071 endbr_queued_at_entrance to true and don't insert ENDBR if
6072 x86_function_profiler will be called.
6073 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
6075 * config/i386/i386.h (machine_function): Add
6076 endbr_queued_at_entrance.
6078 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
6080 * genattrtab.c (mk_attr_alt): Use alternative_mask.
6081 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
6083 (check_attr_test): Use alternative_mask.
6084 (get_attr_value): Likewise.
6085 (compute_alternative_mask): Use alternative_mask and XWINT.
6086 (make_alternative_compare): Use alternative_mask.
6087 (attr_alt_subset_p): Use XWINT.
6088 (attr_alt_subset_of_compl_p): Likewise.
6089 (attr_alt_intersection): Use alternative_mask and XWINT.
6090 (attr_alt_union): Likewise.
6091 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
6092 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
6093 (simplify_test_exp): Use alternative_mask and XWINT.
6094 (write_test_expr): Use alternative_mask and XWINT, adjust bit
6095 number calculation to support 64 bits. Generate code that
6096 checks 64-bit masks.
6097 (main): Use alternative_mask.
6098 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
6100 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
6103 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
6104 RETURN+USE when returning via %r14.
6106 2018-09-24 Martin Liska <mliska@suse.cz>
6108 * gcov.c (output_lines): Print colorization legend
6109 for both flag_use_colors and flag_use_hotness_colors.
6112 2018-09-24 Martin Liska <mliska@suse.cz>
6114 * coverage.c (get_coverage_counts): Use warning_at
6115 with current_function_decl location. Use %qD in warning
6118 2018-09-24 Martin Liska <mliska@suse.cz>
6120 * memory-block.h (memory_block_pool::release): Annotate with
6121 valgrind that the memory is not accessible.
6123 2018-09-24 Martin Liska <mliska@suse.cz>
6126 * asan.c: Make asan_handled_variables extern.
6128 * cfgexpand.c (expand_stack_vars): Make sure
6129 a representative is unpoison if another
6130 variable in the partition is handled by
6131 use-after-scope sanitization.
6133 2018-09-24 Richard Biener <rguenther@suse.de>
6135 PR tree-optimization/63155
6136 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
6137 the worklist when the edge of the respective argument isn't
6140 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6142 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
6143 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
6144 (MASK_CLASS_P): Update for rename.
6145 (MAYBE_MASK_CLASS_P): Ditto.
6146 (REG_CLASS_NAMES): Update.
6147 (REG_CLASS_CONTENT): Update.
6148 * config/i386/i386.c (regclass_map): Update for MASK_REG
6149 and ALL_MASK_REGS rename.
6150 * config/i386/constraints.md (Yk): Update for rename.
6153 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6155 * config/i386/i386.h (enum reg_class): Remove
6156 EVEX_SSE_REGS and MOD4_SSE_REGS.
6157 (REG_CLASS_NAMES): Update.
6158 (REG_CLASS_CONTENT): Update.
6159 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
6160 registers as ALL_SSE_REGS.
6161 (ix86_additional_allocno_class_p): Remove.
6162 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
6163 (ix86_register_priority): Lower priority of EVEX SSE registers.
6164 Use IN_RANGE macro where appropriate.
6165 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
6166 AVX-5124VNNIW checks.
6167 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
6168 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
6169 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
6170 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
6171 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
6172 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
6173 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
6174 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
6175 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
6176 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
6177 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
6178 * config/i386/constraints.md (Yh): Remove.
6180 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6182 * config/i386/i386.c (regclass_map): Declare integer REX registers
6185 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
6187 * doc/service.texi (Service): Switch the fsf.org link to https.
6189 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
6192 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6193 Define to speculation_safe_value_not_needed.
6195 2018-09-21 Florian Weimer <fweimer@redhat.com>
6198 * doc/extend.texi (Common Function Attributes): Mention that
6199 noreturn suppresses tail call optimization.
6201 2018-09-21 Jeff Law <law@redhat.com>
6203 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
6204 vr_values::cleanup_edges_and_switches.
6205 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
6207 (identify_jump_threads): Remove EDGE_IGNORE handling.
6208 (execute_vrp): Move handling of to_remove_edges and
6209 to_update_switch_stmts into vr_values class member functions.
6210 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
6211 (to_update_switch_stmts): Likewise.
6212 * vr-values.c: Include cfghooks.h.
6213 (vr_values::vr_values): Initialize to_remove_edges and
6214 to_update_switch_stmts.
6215 (vr_values::~vr_values): Verify to_remove_edges and
6216 to_update_switch_stmts are empty.
6217 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
6218 (vr_values::cleanup_edges_and_switches): New member function.
6219 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
6220 function. Add new data members.
6222 2018-09-21 David Malcolm <dmalcolm@redhat.com>
6224 PR tree-optimization/87309
6225 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
6226 calls with pflags and alt_flags.
6227 (selftest::test_capture_of_dump_calls): Add test of interaction of
6228 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
6230 2018-09-21 Olivier Hainque <hainque@adacore.com>
6232 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
6234 2018-09-21 Olivier Hainque <hainque@adacore.com>
6236 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
6238 2018-09-21 Olivier Hainque <hainque@adacore.com>
6240 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
6241 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
6242 (VXWORKS_LIBS_RTP): Minor reordering.
6244 2018-09-21 Olivier Hainque <hainque@adacore.com>
6246 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
6247 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
6249 2018-09-21 Olivier Hainque <hainque@adacore.com>
6251 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
6252 (PTRDIFF_TYPE): Likewise.
6254 2018-09-21 Olivier Hainque <hainque@adacore.com>
6256 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
6257 triplet, similar to support for VxWorks7.
6258 * config/vxworks-dummy.h: Provide a default definition
6259 of TARGET_VXWORKS64 to 0.
6261 2018-09-21 Olivier Hainque <hainque@adacore.com>
6263 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
6264 * config/vxworks-dummy.h: here.
6266 2018-09-21 Olivier Hainque <hainque@adacore.com>
6268 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
6270 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
6271 Bo Zhou <zbo.zhou@hisilicon.com>
6273 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
6274 * config/aarch64/aarch64-tune.md: Regenerated.
6275 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
6276 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
6277 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
6279 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
6280 Julian Brown <julian@codesourcery.com>
6282 * builtins.c (get_builtin_sync_mem): Handle address spaces.
6284 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6286 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
6287 if the call takes a static chain.
6289 2018-09-21 Martin Liska <mliska@suse.cz>
6291 * auto-profile.c (autofdo_source_profile::read): Do not
6293 (read_profile): Do not add working sets.
6294 (read_autofdo_file): Remove sum_all.
6295 (afdo_callsite_hot_enough_for_early_inline): Remove const
6297 * coverage.c (struct counts_entry): Remove gcov_summary.
6298 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
6299 do not support GCOV_TAG_PROGRAM_SUMMARY.
6300 (get_coverage_counts): Remove summary and expected
6302 * coverage.h (get_coverage_counts): Likewise.
6303 * doc/gcov-dump.texi: Remove -w option.
6304 * gcov-dump.c (dump_working_sets): Remove.
6305 (main): Do not support '-w' option.
6306 (print_usage): Likewise.
6307 (tag_summary): Likewise.
6308 * gcov-io.c (gcov_write_summary): Do not dump
6310 (gcov_read_summary): Likewise.
6311 (gcov_histo_index): Remove.
6312 (gcov_histogram_merge): Likewise.
6313 (compute_working_sets): Likewise.
6314 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
6316 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
6317 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
6318 (GCOV_HISTOGRAM_SIZE): Remove.
6319 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
6320 (struct gcov_summary): Simplify rapidly just
6321 to runs and sum_max fields.
6322 (gcov_histo_index): Remove.
6323 (NUM_GCOV_WORKING_SETS): Likewise.
6324 (compute_working_sets): Likewise.
6325 * gcov-tool.c (print_overlap_usage_message): Remove
6326 trailing empty line.
6327 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
6328 (output_lines): Remove program related line.
6329 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
6330 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
6332 (input_profile_summary): Do not read it.
6333 (merge_profile_summaries): And do not merge it.
6334 (input_symtab): Do not call removed function.
6335 * modulo-sched.c (sms_schedule): Do not print sum_max.
6336 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
6337 removed when histogram method was invented.
6338 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
6340 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
6342 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
6343 and dump selected value.
6344 * profile.c (add_working_set): Remove.
6345 (get_working_sets): Likewise.
6346 (find_working_set): Likewise.
6347 (get_exec_counts): Do not work with working sets.
6348 (read_profile_edge_counts): Do not inform as sum_max is removed.
6349 (compute_branch_probabilities): Likewise.
6350 (compute_value_histograms): Remove argument for call of
6351 get_coverage_counts.
6352 * profile.h: Do not make gcov_summary const.
6354 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
6356 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
6358 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6360 PR tree-optimization/86990
6361 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
6362 Check that the entire merged store group is made of constants only for
6365 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
6367 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
6368 (VTABLE_VERIFICATION_SPEC): Likewise.
6369 (SANITIZER_EARLY_SPEC): Likewise.
6370 (SANITIZER_SPEC): Likewise.
6371 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
6372 * doc/invoke.texi (Link Options): Document -r.
6374 2018-09-20 Richard Biener <rguenther@suse.de>
6377 * gimplify.c (gimplify_expr): Retain alignment of
6378 addressable lvalue in dereference.
6380 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
6383 * configure.ac: Check for .loc is_stmt support.
6384 * configure, config.in: Rebuilt.
6385 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
6388 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6390 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
6391 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
6394 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6396 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
6398 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
6401 2018-09-20 Marek Polacek <polacek@redhat.com>
6403 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
6405 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
6407 PR tree-optimization/87288
6408 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
6409 into account when determining PEELING_FOR_NITERS.
6411 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
6413 PR tree-optimization/86877
6414 * tree-vect-loop.c (vect_analyze_loop_2): Call
6415 vect_verify_datarefs_alignment.
6417 2018-09-19 Marek Polacek <polacek@redhat.com>
6419 * doc/invoke.texi: Document -Wclass-conversion.
6421 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
6423 * config/pa/pa.c (pa_adjust_priority): Delete.
6424 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
6426 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
6427 (atomic_storehi): Likewise.
6428 (atomic_storesi): Likewise.
6429 (atomic_loaddi): Restore compare and swap exchange loop code.
6431 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6433 PR rtl-optimization/86902
6434 * combine.c (try_combine): When changing the CC mode used, don't change
6435 an unrelated mode in other_insn to that new CC mode.
6437 2018-09-19 David Malcolm <dmalcolm@redhat.com>
6439 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
6440 with %T in place of calls to dump_generic_expr.
6441 (prune_runtime_alias_test_list): Likewise.
6442 (create_runtime_alias_checks): Likewise.
6443 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
6444 (vect_analyze_data_ref_dependence): Likewise.
6445 (vect_slp_analyze_data_ref_dependence): Likewise.
6446 (vect_record_base_alignment): Likewise. Use %G in place of call
6447 to dump_gimple_stmt.
6448 (vect_compute_data_ref_alignment): Likewise.
6449 (verify_data_ref_alignment): Likewise.
6450 (vect_find_same_alignment_drs): Likewise.
6451 (vect_analyze_group_access_1): Likewise.
6452 (vect_analyze_data_ref_accesses): Likewise.
6453 (dependence_distance_ge_vf): Likewise.
6454 (dump_lower_bound): Likewise.
6455 (vect_prune_runtime_alias_test_list): Likewise.
6456 (vect_find_stmt_data_reference): Likewise.
6457 (vect_analyze_data_refs): Likewise.
6458 (vect_create_addr_base_for_vector_ref): Likewise.
6459 (vect_create_data_ref_ptr): Likewise.
6460 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
6461 (vect_can_advance_ivs_p): Likewise.
6462 (vect_update_ivs_after_vectorizer): Likewise.
6463 (vect_gen_prolog_loop_niters): Likewise.
6464 (vect_prepare_for_masked_peels): Likewise.
6465 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
6466 (vect_determine_vectorization_factor): Likewise.
6467 (vect_is_simple_iv_evolution): Likewise.
6468 (vect_analyze_scalar_cycles_1): Likewise.
6469 (vect_analyze_loop_operations): Likewise.
6470 (report_vect_op): Likewise.
6471 (vect_is_slp_reduction): Likewise.
6472 (check_reduction_path): Likewise.
6473 (vect_is_simple_reduction): Likewise.
6474 (vect_create_epilog_for_reduction): Likewise.
6475 (vect_finalize_reduction:): Likewise.
6476 (vectorizable_induction): Likewise.
6477 (vect_transform_loop_stmt): Likewise.
6478 (vect_transform_loop): Likewise.
6479 (optimize_mask_stores): Likewise.
6480 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
6481 (vect_split_statement): Likewise.
6482 (vect_recog_over_widening_pattern): Likewise.
6483 (vect_recog_average_pattern): Likewise.
6484 (vect_determine_min_output_precision_1): Likewise.
6485 (vect_determine_precisions_from_range): Likewise.
6486 (vect_determine_precisions_from_users): Likewise.
6487 (vect_mark_pattern_stmts): Likewise.
6488 (vect_pattern_recog_1): Likewise.
6489 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
6490 (vect_record_max_nunits): Likewise.
6491 (vect_build_slp_tree_1): Likewise.
6492 (vect_build_slp_tree_2): Likewise.
6493 (vect_print_slp_tree): Likewise.
6494 (vect_analyze_slp_instance): Likewise.
6495 (vect_detect_hybrid_slp_stmts): Likewise.
6496 (vect_detect_hybrid_slp_1): Likewise.
6497 (vect_slp_analyze_operations): Likewise.
6498 (vect_slp_analyze_bb_1): Likewise.
6499 (vect_transform_slp_perm_load): Likewise.
6500 (vect_schedule_slp_instance): Likewise.
6501 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
6502 (vect_mark_stmts_to_be_vectorized): Likewise.
6503 (vect_init_vector_1): Likewise.
6504 (vect_get_vec_def_for_operand): Likewise.
6505 (vect_finish_stmt_generation_1): Likewise.
6506 (vect_check_load_store_mask): Likewise.
6507 (vectorizable_call): Likewise.
6508 (vectorizable_conversion): Likewise.
6509 (vectorizable_operation): Likewise.
6510 (vectorizable_load): Likewise.
6511 (vect_analyze_stmt): Likewise.
6512 (vect_is_simple_use): Likewise.
6513 (vect_get_vector_types_for_stmt): Likewise.
6514 (vect_get_mask_type_for_stmt): Likewise.
6515 * tree-vectorizer.c (increase_alignment): Likewise.
6517 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
6519 * doc/rtl.texi: Adjust vec_select description.
6520 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
6521 non-constant selectors.
6523 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
6525 * config/aarch64/aarch64-protos.h
6526 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
6527 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
6528 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
6529 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
6530 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
6531 (AARCH64_FL_PROFILE): Move index so flags are ordered.
6532 (AARCH64_ISA_RCPC8_4): New flag.
6533 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
6534 to aarch64_offset_9bit_signed_unscaled_p.
6535 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
6537 * config/aarch64/constraints.md (Ust): New constraint.
6538 * config/aarch64/predicates.md.
6539 (aarch64_9bit_offset_memory_operand): New predicate.
6540 (aarch64_rcpc_memory_operand): New predicate.
6542 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
6544 PR rtl-optimization/87361
6545 * rtlanal.c (nonzero_bits1): Revert accidental change.
6547 2018-09-19 Richard Biener <rguenther@suse.de>
6549 PR tree-optimization/87349
6550 PR tree-optimization/87342
6551 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
6553 2018-09-18 Marek Polacek <polacek@redhat.com>
6555 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
6556 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
6558 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
6560 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
6562 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
6564 PR rtl-optimization/86882
6565 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
6567 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
6569 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
6570 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
6572 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
6575 * doc/invoke.texi (Link Options): Fix formatting and grammar.
6577 2018-09-18 Richard Biener <rguenther@suse.de>
6580 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
6581 (compute_samebase_partition_bases): Likewise.
6582 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
6583 (gimple_can_coalesce_p): Simplify.
6585 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
6587 Handle a library implementation of ffs calling __builtin_ffs.
6588 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
6589 (mmix_init_libfuncs): New function: make __builtin_ffs expand
6592 2018-09-17 David Malcolm <dmalcolm@redhat.com>
6594 * diagnostic-show-locus.c (class layout_range): Add field
6596 (layout_range::layout_range): Add "original_idx" param and use it
6597 to initialize new field.
6598 (make_range): Use 0 for original_idx.
6599 (layout::layout): Pass in index to calls to
6600 maybe_add_location_range.
6601 (layout::maybe_add_location_range): Add param "original_idx" and
6602 pass it on to layout_range.
6603 (layout::print_any_labels): Pass on range->m_original_idx to
6605 (gcc_rich_location::add_location_if_nearby): Use 0 for
6607 * gcc-rich-location.h (text_range_label::get_text): Update for new
6609 (range_label_for_type_mismatch::get_text): Likewise.
6611 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
6613 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
6615 2018-09-17 David Malcolm <dmalcolm@redhat.com>
6617 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
6618 format_string_diagnostic_t.
6619 (fmtwarn_n): Likewise.
6620 * substring-locations.c
6621 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
6622 (format_warning_n_va): Convert to...
6623 (format_string_diagnostic_t::emit_warning_n_va): ...this.
6624 (format_warning_va): Convert to...
6625 (format_string_diagnostic_t::emit_warning_va): ...this.
6626 (format_warning_at_substring): Convert to...
6627 (format_string_diagnostic_t::emit_warning): ...this.
6628 (format_warning_at_substring_n): Convert to...
6629 (format_string_diagnostic_t::emit_warning_n): ...this.
6630 * substring-locations.h (class format_string_diagnostic_t): New
6632 (format_warning_va): Convert to
6633 format_string_diagnostic_t::emit_warning_va.
6634 (format_warning_n_va): Convert to
6635 format_string_diagnostic_t::emit_warning_n_va.
6636 (format_warning_at_substring): Convert to
6637 format_string_diagnostic_t::emit_warning.
6638 (format_warning_at_substring_n): Convert to
6639 format_string_diagnostic_t::emit_warning_n.
6641 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
6642 Bernd Schmidt <bernds_cb1@t-online.de>
6644 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
6647 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
6649 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
6650 operand 0 predicate to nonimmediate operand.
6651 (rint<mode>2_frndint): Remove insn pattern.
6652 (rint<mode>2): Change operand 1 predicate to general_operand.
6653 Extend operand 1 to XFmode and generate rintxf2 insn.
6654 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
6655 Do not use X87MODEF mode macro.
6656 (frndintxf2_<rounding>_i387): Rename from
6657 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
6658 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
6659 to XFmode and generate significandxf3 insn.
6661 2018-09-17 Richard Biener <rguenther@suse.de>
6663 PR tree-optimization/87328
6664 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
6665 visiting unexecutable backedges when not iterating.
6666 (do_rpo_vn): Mark all edges not executable even when not
6669 2018-09-17 Martin Jambor <mjambor@suse.cz>
6672 * doc/invoke.texi (Warning Options): Likewise.
6674 2018-09-17 Richard Biener <rguenther@suse.de>
6676 PR tree-optimization/87301
6677 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
6678 clean EH info from leftover copy assignments.
6680 2018-09-17 Martin Liska <mliska@suse.cz>
6682 PR gcov-profile/85871
6683 * gcov.c (output_intermediate_file): Fix out of bounds
6686 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
6688 * config/arc/arc.c: Object attributes for core4 not reflected
6690 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
6693 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
6695 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
6697 2018-09-17 Martin Liska <mliska@suse.cz>
6699 * doc/gcov.texi: Document new option --use-hotness-colors.
6700 * gcov.c (struct source_info): Declare new field.
6701 (source_info::source_info): Set default for maximum_count.
6702 (print_usage): Add new -q option.
6703 (process_args): Process it.
6704 (accumulate_line_info): Save src->maximum_count.
6705 (output_line_beginning): Make color line number if
6706 flag_use_hotness_colors is set.
6707 (output_line_details): Pass default argument value.
6708 (output_lines): Pass src->maximum_count.
6710 2018-09-17 Martin Liska <mliska@suse.cz>
6712 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
6713 Use processor_names table.
6714 * config/i386/i386.c (ix86_default_align): Use
6715 processor_cost_table for alignment values.
6716 (ix86_option_override_internal): Use processor_names.
6717 (ix86_function_specific_print): Likewise.
6718 * config/i386/i386.h (struct processor_costs):
6719 Add alignment values.
6720 (struct ptt): Remove and replace with const char *.
6721 * config/i386/x86-tune-costs.h (struct processor_costs):
6722 Declare default alignments for all costs.
6724 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
6726 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
6727 symbolics or VR_VARYING ranges for ABS_EXPR.
6728 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
6729 when range will wrap.
6731 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
6734 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
6735 before and after a JUMP_TABLE_DATA.
6737 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
6740 * dojump.c (do_compare_and_jump): Canonicalize function pointers
6741 when one operand is a function pointer. Use POINTER_TYPE_P and
6742 FUNC_OR_METHOD_TYPE_P.
6743 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
6744 * fold-const.c (build_range_check): Likewise.
6745 * match.pd (simple_comparison): Likewise.
6747 2018-09-14 David Malcolm <dmalcolm@redhat.com>
6750 * spellcheck.c (get_edit_distance_cutoff): New function.
6751 (selftest::test_edit_distance_unit_test_oneway): Rename to...
6752 (selftest::test_get_edit_distance_one_way): ...this.
6753 (selftest::test_get_edit_distance_unit): Rename to...
6754 (selftest::test_get_edit_distance_both_ways): ...this.
6755 (selftest::test_edit_distances): Move tests to this new function,
6756 and test some more pairs of strings. Update for above renaming.
6757 (selftest::get_old_cutoff): New function.
6758 (selftest::test_get_edit_distance_cutoff): New function.
6759 (selftest::assert_suggested_for): New function.
6760 (ASSERT_SUGGESTED_FOR): New macro.
6761 (selftest::assert_not_suggested_for): New function.
6762 (ASSERT_NOT_SUGGESTED_FOR): New macro.
6763 (selftest::test_suggestions): New function.
6764 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
6765 tests to selftest::test_edit_distances and call it. Add calls to
6766 selftest::test_get_edit_distance_cutoff and
6767 selftest::test_suggestions.
6768 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
6769 (best_match::consider): Replace hard-coded cutoff calculation with
6771 (best_match::get_cutoff): New declaration.
6772 (best_match::get_best_meaningful_candidate): Likewise.
6774 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
6776 * builtins.c (fold_builtin_strlen): Remove TODO comment.
6778 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
6781 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
6783 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
6784 terminated string literal.
6786 2018-09-14 Martin Sebor <msebor@redhat.com>
6788 * builtins.c (unterminated_array): Handle ARRAY_REF.
6789 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
6790 * builtins.h (unterminated_array): Declare extern.
6791 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
6793 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
6796 2018-09-14 Martin Sebor <msebor@redhat.com>
6797 Jeff Law <law@redhat.com>
6799 * builtins.c (unterminated_array): New.
6800 (expand_builtin_strcpy): Adjust.
6801 (expand_builtin_strcpy_args): Detect unterminated arrays.
6802 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
6803 unterminated arrays.
6804 * gimple-fold.h (get_maxval_strlen): Add argument.
6805 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
6807 * gimple-fold.c (get_range_strlen): Add argument.
6808 (get_maxval_strlen): Adjust.
6809 * gimple-fold.h (get_range_strlen): Add argument.
6811 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
6813 * config/i386/movdirintrin.h: Fix copyright year.
6815 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
6817 * reg-stack.c: Include regs.h.
6818 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
6819 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
6820 FIRST_STACK_REG, not DFmode.
6821 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
6822 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
6823 (change stack): Default register mode to the reg_raw_mode of
6824 FIRST_STACK_REG, not DFmode.
6825 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
6826 (*swapxf): Rename from swapxf.
6828 2018-09-14 Carl Love <cel@us.ibm.com>
6830 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
6831 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
6833 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
6836 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
6839 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
6842 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
6844 2018-09-14 Jason Merrill <jason@redhat.com>
6846 Fix --enable-gather-detailed-mem-stats.
6847 * hash-table.c (hash_table_usage): Change from variable to function.
6848 * hash-table.h: Adjust.
6849 * Makefile.in: Add missing dependencies on hash-table.h.
6851 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6853 PR tree-optimization/87259
6855 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
6856 execute_cse_reciprocals_1 has tried transforming.
6858 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
6860 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
6861 VR_VARYING for PLUS/MINUS_EXPR.
6863 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
6865 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
6868 2018-09-14 Richard Biener <rguenther@suse.de>
6871 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
6872 bits for the merged partition.
6874 2018-09-13 Martin Sebor <msebor@redhat.com>
6875 Bernd Edlinger <bernd.edlinger@hotmail.de>
6877 * builtins.h (c_srlen): Add argument.
6878 * builtins.c (warn_string_no_nul): New function.
6879 (c_strlen): Add argument and use it. Update recursive calls.
6880 Pass DECL argument to string_constant to get info on non
6881 terminated strings. Update *NONSTR as needed.
6882 (fold_builtin_strlen): Add argument to calls to c_strlen.
6883 Warn for unterminated arrays.
6884 (warn_string_no_null): Add prototype.
6885 * expr.c (string_constant): Update arguments. Update recursive
6886 calls appropriately. Detect missing NUL terminator and outermost
6887 declaration its missing in.
6888 Improve checks for arrays with nonzero lower bound or elements
6889 that are not a single byte. Simplify offset computation.
6890 Simplify checks for non-NUL terminated strings.
6891 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
6892 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
6894 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
6896 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
6898 * fold-const.c (c_getstr): Fix function comment. Remove unused third
6899 argument. Fix range checks.
6900 * fold-const.h (c_getstr): Adjust protoype.
6901 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
6902 string is constant but contains no NUL byte.
6904 * expr.c (string_constant): Adjust function comment.
6905 Remove bogus check for zero termination.
6907 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
6909 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
6910 (get_constant_size): Don't make STRING_CSTs larger than they are.
6911 (check_string_literal): New check function for STRING_CSTs.
6912 (output_constant): Use it.
6914 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
6917 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
6919 2018-09-13 Richard Biener <rguenther@suse.de>
6921 PR tree-optimization/87263
6922 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
6923 (struct unwind_state): Add max_rpo field.
6924 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
6925 Compute max_rpo, the max RPO number a block can be backwards reached
6926 from. Re-write non-iterating mode to a RPO ordered worklist approach,
6927 separating it from the iterating mode.
6929 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
6931 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
6932 (rfs_decision): New scheduling decision.
6934 2018-09-13 Richard Biener <rguenther@suse.de>
6937 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
6938 (vn_nary_op_insert_pieces_predicated): Do not write useless
6939 valid_dominated_by_p entry outside of the allocated storage.
6941 2018-09-13 Omar Sandoval <osandov@osandov.com>
6942 Tom de Vries <tdevries@suse.de>
6945 * dwarf2out.c (is_c): New function.
6946 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
6948 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
6951 * config/aarch64/aarch64.md (*aarch64_bfxil):
6953 * config/aarch64/constraints.md (Ulc): Define.
6954 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
6956 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
6959 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
6961 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
6962 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
6963 aarch64_layout_frame call.
6964 (aarch64_expand_epilogue): Likewise.
6965 (aarch64_initial_elimination_offset): Likewise.
6966 (aarch64_get_separate_components): Likewise.
6967 (aarch64_use_return_insn_p): Likewise.
6968 (aarch64_layout_frame): Remove unneeded check.
6970 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6972 * configure.ac: Only append
6973 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
6974 gcc_config_arguments if it was never reconfigured or last reconfigure
6975 was with different arguments.
6976 * configure: Regenerated.
6978 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6979 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6982 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
6983 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
6985 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6987 PR tree-optimization/87287
6988 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
6989 X % C == 0 to X % (unsigned) C == 0 optimization to ...
6990 * match.pd (X % C == 0): ... here. New optimization.
6992 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6995 * expr.h (maybe_optimize_mod_cmp): Declare.
6996 * expr.c (mod_inv): New function.
6997 (maybe_optimize_mod_cmp): New function.
6998 (do_store_flag): Use it.
6999 * cfgexpand.c (expand_gimple_cond): Likewise.
7001 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
7002 Julian Brown <julian@codesourcery.com>
7005 * gimplify.c (gimplify_scan_omp_clauses): Set
7006 target_firstprivatize_array_bases in OpenACC parallel and kernels
7007 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
7008 OpenACC data regions.
7010 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
7012 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
7013 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
7015 2018-09-12 Richard Biener <rguenther@suse.de>
7017 PR tree-optimization/87280
7018 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
7019 edge but unreachable target.
7020 (do_rpo_vn): For conservatively handling a PHI only mark
7021 the backedge executable but not the block reachable.
7023 2018-09-12 Richard Biener <rguenther@suse.de>
7025 PR tree-optimization/87266
7026 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
7029 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
7031 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
7033 ("trunc<BFP:mode><DFP_ALL:mode>2")
7034 ("trunc<DFP_ALL:mode><BFP:mode>2")
7035 ("extend<BFP:mode><DFP_ALL:mode>2")
7036 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
7037 according to the target operand type.
7039 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7040 Andreas Krebbel <krebbel@linux.ibm.com>
7042 PR tree-optimization/86844
7043 * gimple-ssa-store-merging.c
7044 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
7045 there are any overlapping stores in between them, make sure they are
7046 also coalesced or we give up completely.
7048 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7051 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
7052 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
7054 2018-09-12 Tom de Vries <tdevries@suse.de>
7056 * common.opt (gdescribe-dies): Add option.
7057 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
7058 attribute for artifical and nameless decls.
7059 (dwarf2out_register_external_die): Add description attribute to
7060 external reference die.
7061 (add_desc_attribute): New functions.
7062 (gen_subprogram_die): Add description attribute to
7063 DW_TAG_call_site_parameter.
7064 * tree-pretty-print.c (print_generic_expr_to_str): New function.
7065 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
7066 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
7068 (@item -gdescribe-dies): Add.
7070 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
7072 * tree-vrp.c (vrp_shift_undefined_p): Remove.
7073 (extract_range_from_binary_expr_1: Call
7074 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
7075 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
7078 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
7080 * gimple-ssa-warn-alloca.c
7081 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
7082 field for ALLOCA_BOUND_*_LARGE.
7084 2018-09-11 Nathan Sidwell <nathan@acm.org>
7086 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
7088 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7090 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
7091 for clobbers. Remove obsolete comment.
7093 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7095 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
7096 mpxchk, mpxld and mpxst types.
7097 (define_attr length_immediate): Remove all processing of mpx types.
7098 (define_attr prefix_0f): Ditto.
7099 (define_attr memory): Ditto.
7101 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7103 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
7104 (log<mode>2): Change operand 1 predicate to general_operand.
7105 Extend operand 1 to XFmode and generate logxf3 insn.
7106 (log10<mode>2): Change operand 1 predicate to general_operand.
7107 Extend operand 1 to XFmode and generate log10xf3 insn.
7108 (log2<mode>2): Change operand 1 predicate to general_operand.
7109 Extend operand 1 to XFmode and generate log2xf3 insn.
7110 (fyl2xp1_extend<mode>xf3_i387): Remove.
7111 (log1p<mode>2): Change operand 1 predicate to general_operand.
7112 Extend operand 1 to XFmode and generate log1pxf3 insn.
7113 (fxtract_extend<mode>xf3_i387): Remove.
7114 (logb<mode>2): Change operand 1 predicate to general_operand.
7115 Extend operand 1 to XFmode and generate logbxf3 insn.
7116 (ilogb<mode>2): Change operand 1 predicate to general_operand.
7117 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
7118 (significand<mode>2): Change operand 1 predicate to general_operand.
7119 Extend operand 1 to XFmode and generate significandxf3 insn.
7121 2018-09-11 Nathan Sidwell <nathan@acm.org>
7123 * gcc.c (perror_with_name, pfatal_with_name): Delete.
7124 (load_specs): Use fatal_error.
7125 (DELETE_IF_ORDINARY, process_command): Use error.
7126 (execute, run_attempt): Use fatal_error.
7128 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
7130 * diagnostic-core.h (sorry_at): New prototype.
7131 * diagnostic.c (sorry_at): New function.
7133 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
7135 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
7136 by zero as VR_UNDEFINED.
7138 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
7140 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
7141 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
7142 (<sincos>mode2): New expander.
7143 (sincos_extend<mode>xf3_i387): Remove insn pattern.
7144 (sincos -> sin, cos splitters): Remove splitter patterns.
7145 (sincos<mode>3): Change operand 2 predicate to general_operand.
7146 Extend operand 2 to XFmode and generate sincosxf3 insn.
7147 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
7148 Change operand 3 predicate to const1_operand.
7149 (fptan_extend<mode>xf4_i387): Remove insn pattern.
7150 (tanxf2): Update operands in the call to fptanxf4_i387.
7151 (tan<mode>2): Change operand 1 predicate to general_operand.
7152 Extend operand 1 to XFmode and generate tanxf3 insn.
7153 (atan2xf3): Rename from *fpatanxf3_i387.
7154 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
7155 (atan2xf3): Remove expander.
7156 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
7157 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
7158 (atan<mode>2): Change operand 1 predicate to general_operand.
7159 Extend operand 1 to XFmode and generate atanxf3 insn.
7161 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
7163 * config/i386/i386.md (x87/SSE constant load splitter): Use
7164 memory_operand instead of nonimmediate_operand for input operand
7167 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
7169 * config/i386/i386.md (float partial SSE register stall splitter): Move
7170 splitter near its instruction pattern.
7171 (float_extend partial SSE register stall splitter): Ditto.
7172 (float_truncate partial SSE register stall splitter): Ditto.
7174 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
7177 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7178 to speculation_safe_value_not_needed.
7181 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
7182 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
7183 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
7184 leaf_function_p, instead use has_hard_reg_initial_val.
7186 2018-09-09 Nathan Sidwell <nathan@acm.org>
7188 * gcc.h (pfatal_with_name): Don't declare here.
7189 * gcc.c (pfatal_with_name): Make static.
7191 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
7193 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
7196 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
7198 PR rtl-optimization/85458
7199 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
7200 priority hook to reduce the priority of EXPR.
7202 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
7204 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
7205 DImode for x87 on 32bit targets. Conditionally disable x87 modes
7206 with X87_ENABLE_FLOAT. Remove preparation code.
7207 (*float<SWI48:mode><MODEF:mode>2): Rename from
7208 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
7209 math using "enabled" attribute.
7210 (*floatdi<MODEF:mode>2_i387): Rename from
7211 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
7212 enable for 32bit targets only.
7213 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
7215 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
7216 as operand 1 predicate. Rewrite as define_insn_and_split.
7217 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
7219 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
7221 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
7222 to fallthru to FLOAT case.
7224 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7227 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
7228 around folding of vec_sl to handle out of range shift values.
7230 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
7232 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
7233 Update callers to gen_fix_trunc<mode>_i387_fisttp
7234 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
7235 nonimmediate_operand.
7236 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
7237 and corresponding splitters.
7238 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
7239 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
7240 (fix_truncdi_i387_with_temp): Remove insn pattern
7241 and corresponding splitters.
7242 (fix_trunc<mode>_i387): Change operand 0 predicate to
7243 nonimmediate_operand.
7244 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
7245 and corresponding splitters.
7246 (*fistdi2_1): Remove.
7248 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
7249 (lrintxfdi2): Remove expander. Reimplement as define_insn.
7250 (*fist<mode>2_1): Remove.
7251 (fist<mode>2): Ditto.
7252 (fist<mode>2_with_temp): Remove insn pattern and corresponding
7254 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
7255 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
7256 (fistdi2_<rounding>): Change operand 0 predicate to
7257 nonimmediate_operand.
7258 (fistdi2_<rounding>_with_temp): Remove insn pattern
7259 and corresponding splitters.
7260 (fist<mode>2_<rounding>): Change operand 0 predicate to
7261 nonimmediate_operand.
7262 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
7263 and corresponding splitters.
7265 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
7267 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7269 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
7272 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7274 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7275 early gimple folding of vec_splat().
7276 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
7277 * gimple-fold.h: Add an extern define for tree_vec_extract().
7279 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7281 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
7282 wrappers around TREE_TYPE comparisons.
7284 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7287 * config/s390/predicates.md: Add nonsym_memory_operand.
7288 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
7289 contains a SYMBOL_REF, load it into an intermediate pseudo.
7290 (s390_emit_compare_and_swap): Legitimize operand.
7291 * config/s390/s390.md: Use the new nonsym_memory_operand
7292 with UNSPECV_CAS patterns.
7294 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7297 * config/s390/s390-passes.def: New file.
7298 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
7300 (make_pass_s390_early_mach): Add declaration.
7301 * config/s390/s390.c (make_pass_s390_early_mach):
7302 (s390_option_override): Remove dynamic registration.
7303 * config/s390/t-s390: Add s390-passes.def.
7305 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7307 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7308 Remove UNSPEC_LTREL_BASE check.
7309 (annotate_constant_pool_refs): Likewise.
7310 (find_constant_pool_ref): Likewise.
7311 (find_ltrel_base): Removed.
7312 (replace_ltrel_base): Removed.
7313 (s390_mainpool_finish): Remove replace_ltrel_base call.
7314 (s390_chunkify_start): Remove pending LTREL_BASE logic.
7315 (s390_chunkify_finish): Remove replace_ltrel_base call.
7316 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
7318 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
7321 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7322 to speculation_safe_value_not_needed.
7324 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
7325 Bernd Schmidt <bernds_cb1@t-online.de>
7327 * config/nvptx/nvptx-opts.h: New file.
7328 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
7329 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
7331 (TARGET_SM35): New macro.
7332 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
7334 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
7336 (misa=): New option.
7337 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
7339 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
7341 * config/i386/i386.md (truncdfsf2): Remove expander.
7342 (truncdfsf2_with_temp): Ditto.
7343 (truncxf<mode>2): Ditto.
7344 (*truncdfsf_fast_mixed): Remove insn pattern.
7345 (*truncdfsf_fast_i387): Ditto.
7346 (*truncdfsf_mixed): Ditto.
7347 (*truncdfsf_i387): Ditto.
7348 (*truncdfsf2_i387_1): Ditto.
7349 (*truncxfsf2_mixed): Ditto.
7350 (*truncxfdf2_mixed): Ditto.
7351 (*truncxf<mode>2_i387_noop): Ditto. Update callers
7352 to call gen_truncxf<mode>2 instead.
7353 (*truncxf<mode>2_i387): Remove.
7354 (reg->reg splitters): Remove splitter pattern.
7355 (reg->mem splitters): Ditto.
7357 (truncdfsf2): New insn pattern.
7358 (truncxf<mode>2): Ditto.
7360 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7362 * tree-ssa-math-opts.c (is_mult_by): New function.
7363 (is_square_of): Use the above.
7364 (optimize_recip_sqrt): New function.
7365 (pass_cse_reciprocals::execute): Use the above.
7367 2018-09-05 Richard Biener <rguenther@suse.de>
7370 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
7371 to zero-init the emplaced vec.
7373 2018-09-05 Martin Liska <mliska@suse.cz>
7375 PR tree-optimization/87205
7376 * tree-switch-conversion.c (pass_lower_switch::execute):
7377 Group cases for switch statements.
7379 2018-09-05 Richard Biener <rguenther@suse.de>
7381 PR tree-optimization/87217
7382 * tree-ssa-sccvn.c (vuse_valueize): New.
7383 (vn_reference_lookup_pieces): Use it.
7384 (vn_reference_lookup): Likewise.
7386 2018-09-05 Nathan Sidwell <nathan@acm.org>
7389 * stor-layout.c (place_field): Scan forwards to check last
7390 bitfield when ms_bitfield_placement is in effect.
7392 2018-09-05 Richard Biener <rguenther@suse.de>
7395 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
7398 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
7399 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
7401 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
7402 * config.gcc (extra_objs): Build it.
7403 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
7405 * config/aarch64/aarch64-passes.def
7406 (pass_tag_collision_avoidance): New pass.
7407 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
7408 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
7409 (aarch64_classify_address): Remove static qualifier.
7410 (aarch64_address_info, aarch64_address_type): Move to...
7411 * config/aarch64/aarch64-protos.h: ... here.
7412 (make_pass_tag_collision_avoidance): New function.
7413 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
7416 2018-09-05 Martin Liska <mliska@suse.cz>
7418 * doc/gcov.texi: Update documentation of humar
7420 * gcov.c (format_count): Print one decimal place, it provides
7421 more fine number of situations like '1G' vs. '1.4G'.
7423 2018-09-05 Martin Liska <mliska@suse.cz>
7426 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
7427 * optc-gen.awk: Allow 'Var' for Deprecated options in order
7428 to generate a MASK value.
7430 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
7433 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
7434 if frame pointer isn't used.
7435 (compute_frame_pointer_to_fb_displacement): Likewise.
7437 2018-09-04 Jakub Jelinek <jakub@redhat.com>
7440 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
7441 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
7442 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
7443 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
7444 and OPTION_MASK_ISA_XSAVEC_UNSET.
7446 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
7448 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
7449 XOR operations in NAND case.
7451 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
7453 * wide-int-range.cc (wide_int_range_convert): New.
7454 * wide-int-range.h (wide_int_range_convert): New.
7455 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
7456 code into wide_int_range_convert.
7457 (extract_range_into_wide_ints): Do not munge anti range constants
7458 into the entire domain. Just return the range back.
7460 2018-09-04 Martin Liska <mliska@suse.cz>
7462 * genmatch.c (output_line_directive): Add new argument
7464 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
7466 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
7468 * doc/invoke.texi (Option Summary): Add whitespace.
7470 * doc/invoke.texi (Option Summary): Add -Waligned-new.
7472 2018-09-04 Richard Biener <rguenther@suse.de>
7474 PR tree-optimization/87211
7475 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
7476 backedge value we're supposed to treat as VARYING also number
7477 the PHI to VARYING in case it got a different value-number already.
7479 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
7481 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
7482 (extract_range_from_binary_expr_1): Do not call
7483 vrp_can_optimize_bit_op.
7484 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
7486 (wide_int_range_get_mask_and_bounds): New.
7487 (wide_int_range_optimize_bit_op): New.
7488 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
7489 (wide_int_range_bit_and): Same.
7490 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
7491 (wide_int_range_optimize_bit_op): New.
7492 (wide_int_range_get_mask_and_bounds): New.
7494 2018-09-04 Richard Biener <rguenther@suse.de>
7496 PR tree-optimization/87176
7497 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
7498 variable. When value-numbering a virtual PHI node make sure
7499 to not value-number to the backedge value.
7501 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
7503 * doc/extend.texi (Long Long, Hex Floats): Document support for
7504 long long and hex floats in more recent versions of ISO C++.
7506 2018-09-03 Richard Biener <rguenther@suse.de>
7508 PR tree-optimization/87177
7509 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
7512 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7514 * bb-reorder.c (edge_order): Convert to C-qsort-style
7515 tri-state comparator.
7516 (reorder_basic_blocks_simple): Change std::stable_sort to
7519 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7521 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
7522 tri-state comparator.
7523 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
7525 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7527 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
7528 (mergesort): ... here as maximum count for using netsort.
7529 (gcc_qsort): Set nlim to 3 if stable sort is requested.
7530 (gcc_stablesort): New.
7531 * system.h (gcc_stablesort): Declare.
7533 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7535 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
7536 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
7537 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
7539 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
7541 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
7542 lxsdx and stxsdx alternatives.
7543 (*mov<mode>_hardfloat64): Ditto.
7544 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
7546 2018-09-03 Richard Biener <rguenther@suse.de>
7548 PR tree-optimization/87200
7549 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
7552 2018-09-03 Martin Liska <mliska@suse.cz>
7554 PR tree-optimization/87201
7555 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
7556 Fix parenthesis in an expression.
7558 2018-09-03 Richard Biener <rguenther@suse.de>
7560 PR tree-optimization/87197
7561 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
7562 visited. CSE the VN_INFO hashtable lookup.
7564 PR tree-optimization/87169
7565 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
7566 iterating make sure there's no extra backedges from irreducible
7567 regions feeding the header. Mark the destination block
7570 2018-09-03 Martin Liska <mliska@suse.cz>
7573 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
7574 * common/common-targhooks.c (default_get_valid_option_values):
7576 * common/common-targhooks.h (default_get_valid_option_values):
7578 * common/config/i386/i386-common.c: Move processor_target_table
7580 (ix86_get_valid_option_values): New function.
7581 (TARGET_GET_VALID_OPTION_VALUES): New macro.
7582 * config/i386/i386.c (struct ptt): Move to i386-common.c.
7583 (PTA_*): Move all defined masks into i386-common.c.
7584 (ix86_function_specific_restore): Use new processor_cost_table.
7585 * config/i386/i386.h (struct ptt): Moved from i386.c.
7586 (struct pta): Likewise.
7587 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
7588 * doc/tm.texi.in: Likewise.
7589 * opt-suggestions.c (option_proposer::suggest_option):
7590 Pass prefix to build_option_suggestions.
7591 (option_proposer::get_completions): Likewise.
7592 (option_proposer::build_option_suggestions): Use the new target
7594 * opts.c (struct option_help_tuple): New struct.
7595 (print_filtered_help): Use the new target hook.
7597 2018-09-03 Martin Liska <mliska@suse.cz>
7600 * predict.c (set_even_probabilities): Add likely_edges
7601 argument and handle cases where we have precisely one
7603 (combine_predictions_for_bb): Catch also likely_edges.
7604 (tree_predict_by_opcode): Handle gswitch statements.
7605 * tree-cfg.h (find_case_label_for_value): New declaration.
7606 (find_taken_edge_switch_expr): Likewise.
7607 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
7608 Find pivot in decision tree based on probabily, not by number of
7611 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
7613 * doc/standards.texi (Standards): Update Objective-C reference.
7615 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7617 * doc/install.texi (Prerequisites): Update link for MPC.
7619 2018-09-01 Michael Matz <matz@suse.de>
7621 PR tree-optimization/87074
7622 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
7623 PHIs for outer-loop uses.
7625 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7627 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
7628 * doc/invoke.texi (C Dialect Options): Ditto.
7630 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7632 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
7634 2018-08-31 Richard Biener <rguenther@suse.de>
7636 PR tree-optimization/87168
7637 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
7638 (rpo_elim::eliminate_avail): When OP was not visited it must
7641 2018-08-31 David Malcolm <dmalcolm@redhat.com>
7643 * tree-vrp.c (copy_value_range): Convert param "from" from
7644 "value_range *" to "const value_range *".
7645 (range_is_null): Likewise for param "vr".
7646 (range_int_cst_p): Likewise.
7647 (range_int_cst_singleton_p): Likewise.
7648 (symbolic_range_p): Likewise.
7649 (value_ranges_intersect_p): Likewise for both params.
7650 (value_range_nonnegative_p): Likewise for param "vr".
7651 (value_range_constant_singleton): Likewise.
7652 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
7653 (extract_range_into_wide_ints): Likewise for param "vr".
7654 (extract_range_from_multiplicative_op): Likewise for params "vr0"
7656 (vrp_can_optimize_bit_op): Likewise.
7657 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
7659 (extract_range_from_unary_expr): Likewise.
7660 (debug_value_range): Likewise for param "vr".
7661 (value_range::dump): Add "const" qualifier.
7662 (vrp_prop::check_array_ref): Convert local "vr" from
7663 "value_range *" to "const value_range *".
7664 (vrp_prop::check_mem_ref): Likewise.
7665 (vrp_prop::visit_stmt): Likewise for local "old_vr".
7666 (vrp_intersect_ranges_1): Likewise for param "vr_1".
7667 (vrp_intersect_ranges): Likewise.
7668 (simplify_stmt_for_jump_threading): Likewise for local "vr".
7669 (vrp_prop::vrp_finalize): Likewise.
7670 * tree-vrp.h (value_range::dump): Add "const" qualifier.
7671 (vrp_intersect_ranges): Add "const" qualifier to params as above.
7672 (extract_range_from_unary_expr): Likewise.
7673 (value_range_constant_singleton): Likewise.
7674 (symbolic_range_p): Likewise.
7675 (copy_value_range): Likewise.
7676 (extract_range_from_binary_expr_1): Likewise.
7677 (range_int_cst_p): Likewise.
7678 (vrp_set_zero_nonzero_bits): Likewise.
7679 (range_int_cst_singleton_p): Likewise.
7681 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
7683 * config/aarch64/arm_neon.h (vabsd_s64): New.
7684 (vnegd_s64): Likewise.
7686 2018-08-31 Martin Jambor <mjambor@suse.cz>
7688 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
7690 2018-08-31 Martin Liska <mliska@suse.cz>
7692 * ipa-icf.c (sem_item::add_type): Use
7693 sem_item::m_type_hash_cache.
7694 * ipa-icf.h: Move the cache from sem_item_optimizer
7697 2018-08-31 Nathan Sidwell <nathan@acm.org>
7699 * doc/extend.texi (Backwards Compatibility): Remove implicit
7700 extern C leeway of () being (...).
7702 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7704 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
7706 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
7710 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
7712 2018-08-31 Jakub Jelinek <jakub@redhat.com>
7715 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
7716 gen_int_mode. Formatting fixes.
7718 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
7720 * target.def (custom_function_descriptors): Improve documentation.
7721 * doc/tm.texi.in (Trampolines): Expand discussion of function
7722 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
7723 beginning of the section.
7724 * doc/tm.texi: Regenerated.
7726 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
7728 * cfg.h (class auto_edge_flag): Spell out the template-id of the
7729 base class in the initializer list. This is a workaround for
7730 building with older GCC.
7731 (class auto_bb_flag): Likewise.
7733 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
7735 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
7736 (altivec_vcmpequ<VI_char>_p): Remove star.
7737 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
7739 (expand_strncmp_vec_sequence): New function.
7740 (emit_final_str_compare_vec): New function.
7741 (expand_strn_compare): Add support for vector strncmp.
7742 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
7743 length specification to bytes.
7744 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
7745 (vcmpnezb_p): New pattern.
7746 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
7747 for option -mstring-compare-inline-limit.
7749 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
7751 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
7752 (PTA_SKYLAKE): Add PTA_AES.
7753 (PTA_GOLDMONT): Likewise.
7755 2018-08-29 Jan Hubicka <jh@suse.cz>
7758 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
7759 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
7761 2018-08-29 Jan Hubicka <jh@suse.cz>
7763 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
7765 (hash_tree): Do not visit TYPE_STUB_DECL.
7766 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
7767 stream TYPE_STUB_DECL.
7768 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
7769 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
7770 after free_lang_data.
7771 (type_in_anonymous_namespace_p): Likewise.
7773 2018-08-29 Jan Hubicka <jh@suse.cz>
7775 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
7776 comment that it has to be even number.
7777 (class sreal): Change m_sig type to int32_t.
7778 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
7779 int64_t for temporary calculations.
7780 (sreal_verify_basics): Drop one bit from minimum and maximum.
7782 2018-08-30 Richard Biener <rguenther@suse.de>
7784 PR tree-optimization/87147
7785 * tree-ssa-sccvn.c (SSA_VISITED): New function.
7786 (visit_phi): When the degenerate result is from the backedge and
7787 we didn't visit its definition yet drop to VARYING.
7788 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
7790 2018-08-29 Jan Hubicka <jh@suse.cz>
7792 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
7794 (hash_tree): Likewise.
7796 2018-08-29 Jan Hubicka <jh@suse.cz>
7798 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
7799 and TYPE_NEXT_REF_TO.
7801 2018-08-29 Jan Hubicka <jh@suse.cz>
7803 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
7804 comment that it has to be even number.
7805 (class sreal): Change m_sig type to int32_t.
7806 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
7807 int64_t for temporary calculations.
7808 (sreal_verify_basics): Drop one bit from minimum and maximum.
7810 2018-08-30 Tamar Christina <tamar.christina@arm.com>
7812 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
7814 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
7817 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
7818 if to_add is negative.
7820 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
7823 * builtins.c (c_strlen): Improve range checks.
7825 2018-08-29 Martin Sebor <msebor@redhat.com>
7826 Jeff Law <law@redhat.com>
7828 PR tree-optimization/86714
7829 PR tree-optimization/86711
7830 * builtins.c (c_strlen): Add arguments to call to string_constant.
7831 * expr.c (string_constant): Add argument. Detect missing nul
7832 terminator and outermost declaration it's missing in.
7833 * expr.h (string_constant): Add argument.
7834 * fold-const.c (read_from_constant_string): Add arguments to call to
7836 (c_getstr): Likewise.
7837 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
7839 * tree-ssa-strlen.c (get_stridx): Likewise.
7841 2018-08-29 Jan Hubicka <jh@suse.cz>
7843 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
7844 Do not stream DECL_VINDEX.
7845 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
7846 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
7847 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
7849 2018-08-29 Richard Biener <rguenther@suse.de>
7851 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
7852 virtual operands that are not default defs to honor region
7854 (rpo_vn_valueize): Remove ineffective code here.
7856 2018-08-29 Richard Biener <rguenther@suse.de>
7858 PR tree-optimization/87132
7859 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
7860 when skipping defs reachable over backedges.
7862 2018-08-29 Richard Biener <rguenther@suse.de>
7864 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
7865 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
7866 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
7867 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
7868 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
7869 (vn_reference_lookup_3): Remove use of const_parms.
7870 (free_rpo_vn): Do not free const_parms.
7871 (do_rpo_vn): Do not call init_const_parms.
7872 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
7873 SSA_NAME_POINTS_TO_READONLY_MEMORY.
7874 (call_may_clobber_ref_p_1): Likewise.
7876 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
7879 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
7881 (-ftree-scev-cprop): Document.
7883 2018-08-29 Jan Hubicka <jh@suse.cz>
7885 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
7887 (sreal constructor): Update.
7888 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
7889 sreal:operator/): Update.
7891 2018-08-29 Martin Liska <mliska@suse.cz>
7893 * tree-switch-conversion.c (switch_conversion::expand):
7894 Strenghten assumption about gswitch statements.
7896 2018-08-29 Richard Biener <rguenther@suse.de>
7898 PR tree-optimization/87117
7899 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
7900 re-value-number released SSA VDEFs.
7902 2018-08-29 Richard Biener <rguenther@suse.de>
7904 PR tree-optimization/87126
7905 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
7907 2018-08-28 Jim Wilson <jimw@sifive.com>
7909 * config/riscv/pic.md: Rewrite.
7910 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
7912 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
7913 (SOFTF, default_load, softload, softstore): New.
7915 2018-08-28 Jeff Law <law@redhat.com>
7917 * fold-const.c (fold_binary_loc): Remove recently added assert.
7919 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
7921 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
7922 to OP parmeter of generated function.
7924 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
7926 PR tree-optimization/87009
7927 * match.pd: Add boolean optimizations.
7929 2018-08-28 Martin Sebor <msebor@redhat.com>
7932 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
7933 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
7934 (pass_walloca::gate): Use it.
7935 (alloca_call_type): Same.
7936 (pass_walloca::execute): Same.
7937 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
7939 2018-08-28 David Malcolm <dmalcolm@redhat.com>
7941 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
7942 GCC_VERSION for usage of "__gcc_dump_printf__" format from
7945 2018-08-28 Richard Biener <rguenther@suse.de>
7947 PR tree-optimization/87124
7948 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
7949 constants before looking up avail.
7951 2018-08-28 Jakub Jelinek <jakub@redhat.com>
7954 * calls.c (maybe_warn_nonstring_arg): Punt early if
7955 warn_stringop_overflow is zero. Don't call get_range_strlen
7956 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
7957 Swap comparison operands to have constants on rhs. Only use
7958 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
7959 increment lenrng[0].
7961 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
7963 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
7964 use of tree_to_shwi. Remove duplicated test for the size being
7965 a whole number of bytes.
7967 2018-08-28 Richard Biener <rguenther@suse.de>
7969 PR tree-optimization/87117
7970 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
7971 Handle removed stmt without LHS (GIMPLE_NOP).
7973 2018-08-28 Richard Biener <rguenther@suse.de>
7975 PR tree-optimization/87117
7976 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
7977 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
7979 2018-08-28 Richard Biener <rguenther@suse.de>
7981 PR tree-optimization/87117
7982 * tree-ssa-pre.c (compute_avail): Do not make expressions
7983 with predicated values available.
7984 (get_expr_value_id): Assert we do not run into predicated value
7987 2018-08-28 Richard Biener <rguenther@suse.de>
7989 PR tree-optimization/87117
7990 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
7991 get virtual operands.
7992 (get_expr_operands): Handle STRING_CST like other decls.
7994 2018-08-28 Martin Liska <mliska@suse.cz>
7996 * tree.h: Update documentation of fndecl_built_in_p
8000 2018-08-27 Jeff Law <law@redhat.com>
8001 PR tree-optimization/87110
8002 * tree-ssa-dse.c (compute_trims): Handle non-constant
8005 2018-08-27 Martin Sebor <msebor@redhat.com>
8007 PR tree-optimization/86914
8008 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
8010 2018-08-27 Martin Sebor <msebor@redhat.com>
8012 PR tree-optimization/87112
8013 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
8014 the type of the bound argument.
8016 2018-08-27 Jeff Law <law@redhat.com>
8018 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
8019 type does not have a TYPE_SIZE_UNIT.
8021 2018-08-27 Steve Ellcey <sellcey@cavium.com>
8023 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
8024 with include of backend.h.
8026 2018-08-27 Richard Biener <rguenther@suse.de>
8028 PR tree-optimization/86927
8029 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
8030 use const cond reduction code.
8032 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
8034 PR tree-optimization/85758
8035 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
8037 2018-08-27 David Malcolm <dmalcolm@redhat.com>
8040 * diagnostic-show-locus.c (class layout_range): Update for
8041 conversion of show_caret_p to a tri-state.
8042 (layout_range::layout_range): Likewise.
8043 (make_range): Likewise.
8044 (layout::maybe_add_location_range): Likewise.
8045 (layout::should_print_annotation_line_p): Don't show annotation
8046 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
8047 (layout::get_state_at_point): Update for conversion of
8048 show_caret_p to a tri-state. Bail out early for
8049 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
8050 underlining or source colorization.
8051 (gcc_rich_location::add_location_if_nearby): Update for conversion
8052 of show_caret_p to a tri-state.
8053 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
8054 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
8056 (selftest::test_one_liner_labels): Likewise.
8057 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
8058 conversion of show_caret_p to a tri-state.
8059 * pretty-print.c (text_info::set_location): Likewise.
8060 * pretty-print.h (text_info::set_location): Likewise.
8061 * substring-locations.c (format_warning_n_va): Likewise.
8062 * tree-diagnostic.c (default_tree_printer): Likewise.
8063 * tree-pretty-print.c (newline_and_indent): Likewise.
8065 2018-08-27 David Malcolm <dmalcolm@redhat.com>
8068 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
8069 line above for line-insertion fix-it hints.
8070 (selftest::test_fixit_insert_containing_newline): Update the
8071 expected results, and add a test with line-numbering enabled.
8073 2018-08-27 Martin Liska <mliska@suse.cz>
8076 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
8077 params with DECL_HAS_VALUE_EXPR_P.
8079 2018-08-27 Martin Liska <mliska@suse.cz>
8081 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
8082 selected expansion strategy.
8084 2018-08-27 Martin Liska <mliska@suse.cz>
8086 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
8087 * builtins.c (is_builtin_fn): Likewise.
8088 * attribs.c (diag_attr_exclusions): Use new function
8089 fndecl_built_in_p and remove check for FUNCTION_DECL if
8091 (builtin_mathfn_code): Likewise.
8092 (fold_builtin_expect): Likewise.
8093 (fold_call_expr): Likewise.
8094 (fold_builtin_call_array): Likewise.
8095 (fold_call_stmt): Likewise.
8096 (set_builtin_user_assembler_name): Likewise.
8097 (is_simple_builtin): Likewise.
8098 * calls.c (gimple_alloca_call_p): Likewise.
8099 (maybe_warn_nonstring_arg): Likewise.
8100 * cfgexpand.c (expand_call_stmt): Likewise.
8101 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
8102 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
8103 (cgraph_node::verify_node): Likewise.
8104 * cgraphclones.c (build_function_decl_skip_args): Likewise.
8105 (cgraph_node::create_clone): Likewise.
8106 * config/arm/arm.c (arm_insert_attributes): Likewise.
8107 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
8108 * dse.c (scan_insn): Likewise.
8109 * expr.c (expand_expr_real_1): Likewise.
8110 * fold-const.c (operand_equal_p): Likewise.
8111 (fold_binary_loc): Likewise.
8112 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8113 * gimple-low.c (lower_stmt): Likewise.
8114 * gimple-pretty-print.c (dump_gimple_call): Likewise.
8115 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
8117 * gimple.c (gimple_build_call_from_tree): Likewise.
8118 (gimple_call_builtin_p): Likewise.
8119 (gimple_call_combined_fn): Likewise.
8120 * gimplify.c (gimplify_call_expr): Likewise.
8121 (gimple_boolify): Likewise.
8122 (gimplify_modify_expr): Likewise.
8123 (gimplify_addr_expr): Likewise.
8124 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
8125 * ipa-cp.c (determine_versionability): Likewise.
8126 * ipa-fnsummary.c (compute_fn_summary): Likewise.
8127 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
8128 * ipa-split.c (visit_bb): Likewise.
8129 (split_function): Likewise.
8130 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
8131 * lto-cgraph.c (input_node): Likewise.
8132 * lto-streamer-out.c (write_symbol): Likewise.
8133 * omp-low.c (setjmp_or_longjmp_p): Likewise.
8134 (lower_omp_1): Likewise.
8135 * predict.c (strip_predict_hints): Likewise.
8136 * print-tree.c (print_node): Likewise.
8137 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
8138 * trans-mem.c (is_tm_irrevocable): Likewise.
8139 (is_tm_load): Likewise.
8140 (is_tm_simple_load): Likewise.
8141 (is_tm_store): Likewise.
8142 (is_tm_simple_store): Likewise.
8143 (is_tm_abort): Likewise.
8144 (tm_region_init_1): Likewise.
8145 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
8146 * tree-cfg.c (verify_gimple_call): Likewise.
8147 (move_stmt_r): Likewise.
8148 (stmt_can_terminate_bb_p): Likewise.
8149 * tree-eh.c (lower_eh_constructs_2): Likewise.
8150 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
8151 * tree-inline.c (remap_gimple_stmt): Likewise.
8152 (copy_bb): Likewise.
8153 (estimate_num_insns): Likewise.
8154 (fold_marked_statements): Likewise.
8155 * tree-sra.c (scan_function): Likewise.
8156 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
8157 (optimize_stack_restore): Likewise.
8158 (pass_fold_builtins::execute): Likewise.
8159 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8160 (mark_all_reaching_defs_necessary_1): Likewise.
8161 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
8162 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
8163 (pass_forwprop::execute): Likewise.
8164 * tree-ssa-loop-im.c (stmt_cost): Likewise.
8165 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8166 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
8167 * tree-ssa-strlen.c (get_string_length): Likewise.
8168 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
8169 (find_func_aliases_for_call): Likewise.
8170 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
8171 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
8172 * tree-tailcall.c (find_tail_calls): Likewise.
8173 * tree.c (need_assembler_name_p): Likewise.
8174 (free_lang_data_in_decl): Likewise.
8175 (get_call_combined_fn): Likewise.
8176 * ubsan.c (is_ubsan_builtin_p): Likewise.
8177 * varasm.c (incorporeal_function_p): Likewise.
8178 * tree.h (DECL_BUILT_IN): Remove and replace with
8180 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
8181 (fndecl_built_in_p): New.
8183 2018-08-27 Martin Liska <mliska@suse.cz>
8185 PR tree-optimization/86847
8186 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
8187 Dump also subtree probability.
8188 (switch_decision_tree::do_jump_if_equal): New function.
8189 (switch_decision_tree::emit_case_nodes): Handle special
8190 situations in balanced tree that can be emitted much simpler.
8191 Fix calculation of probabilities that happen in tree expansion.
8192 * tree-switch-conversion.h (struct cluster): Add
8194 (struct simple_cluster): Likewise.
8195 (struct case_tree_node): Add new function has_child.
8196 (do_jump_if_equal): New.
8198 2018-08-27 Martin Liska <mliska@suse.cz>
8200 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
8201 Add new argument to bit_test_cluster constructor.
8202 (bit_test_cluster::emit): Set bits really number of values
8204 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
8205 probability argument.
8206 * tree-switch-conversion.h (struct bit_test_cluster):
8207 Add m_handles_entire_switch.
8209 2018-08-27 Martin Liska <mliska@suse.cz>
8211 PR tree-optimization/86702
8212 * tree-switch-conversion.c (jump_table_cluster::emit):
8213 Make probabilities even for values in jump table
8214 according to number of cases handled.
8215 (switch_decision_tree::compute_cases_per_edge): Pass
8216 argument to reset_out_edges_aux function.
8217 (switch_decision_tree::analyze_switch_statement): Likewise.
8218 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
8221 2018-08-27 Martin Liska <mliska@suse.cz>
8223 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
8224 cfun argument explicitly.
8225 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
8226 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
8227 function gimple_switch_default_bb.
8228 (convert_switch_statements):
8230 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
8231 * stmt.c (label_to_block_fn): Use label_to_block and pass
8232 cfun argument explicitly and use gimple_switch_label_bb.
8233 (expand_case): Likewise.
8234 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
8235 cfun argument explicitly. Likewise.
8236 (make_edges_bb): Likewise.
8237 (make_cond_expr_edges): Likewise.
8238 (get_cases_for_edge): Likewise.
8239 (make_gimple_switch_edges): Likewise.
8240 (label_to_block_fn): Likewise.
8241 (label_to_block): Likewise.
8242 (make_goto_expr_edges): Likewise.
8243 (make_gimple_asm_edges): Likewise.
8244 (main_block_label): Likewise.
8245 (group_case_labels_stmt): Likewise.
8246 (find_taken_edge_computed_goto): Likewise.
8247 (find_taken_edge_switch_expr): Likewise.
8248 (gimple_verify_flow_info): Likewise.
8249 (gimple_redirect_edge_and_branch): Likewise.
8250 (gimple_switch_label_bb): New function.
8251 (gimple_switch_default_bb): Likewise.
8252 (gimple_switch_edge): Likewise.
8253 (gimple_switch_default_edge): Likewise.
8254 * tree-cfg.h (label_to_block_fn): Remove and replace ...
8255 (label_to_block): ... with this.
8256 (gimple_switch_label_bb): New.
8257 (gimple_switch_default_bb): Likewise.
8258 (gimple_switch_edge): Likewise.
8259 (gimple_switch_default_edge): Likewise.
8260 * tree-cfgcleanup.c (convert_single_case_switch): Use
8261 new gimple functions and pass new argument to label_to_block.
8262 (cleanup_control_flow_bb):
8263 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
8264 cfun argument explicitly.
8265 (make_eh_edges): Likewise.
8266 (redirect_eh_dispatch_edge): Likewise.
8267 (lower_resx): Likewise.
8268 (lower_eh_dispatch): Likewise.
8269 (maybe_remove_unreachable_handlers): Likewise.
8270 (unsplit_eh): Likewise.
8271 (cleanup_empty_eh): Likewise.
8272 (verify_eh_edges): Likewise.
8273 (verify_eh_dispatch_edge): Likewise.
8274 * tree-ssa-dom.c (record_edge_info): Likewise.
8275 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
8276 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
8277 (thread_through_normal_block): Likewise.
8278 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
8279 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
8280 * tree-switch-conversion.c (switch_conversion::collect): Use new
8282 (switch_conversion::check_final_bb): Likewise.
8283 (switch_conversion::gather_default_values): Pass new argument
8285 (switch_conversion::build_constructors): Likewise.
8286 (switch_decision_tree::compute_cases_per_edge): Use new
8287 gimple_switch_edge function.
8288 (switch_decision_tree::analyze_switch_statement): Pass new argument
8290 (switch_decision_tree::try_switch_expansion): Use
8291 gimple_switch_default_edge.
8292 * tree-vrp.c (find_switch_asserts): Pass new argument
8294 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
8295 (vr_values::simplify_switch_using_ranges): Likewise.
8297 2018-08-27 Richard Biener <rguenther@suse.de>
8299 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
8300 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
8302 * tree-ssa-sccvn.h (struct vn_pval): New structure.
8303 (struct vn_nary_op_s): Add unwind_to member. Add
8304 predicated_values flag and put result into a union together
8305 with a linked list of vn_pval.
8306 (struct vn_ssa_aux): Add name member to make maintaining
8307 a map of SSA name to vn_ssa_aux possible. Remove no longer
8308 needed info, dfsnum, low, visited, on_sccstack, use_processed
8309 and range_info_anti_range_p members.
8310 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
8311 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
8313 (vn_valueize): New global.
8314 (vn_context_bb): Likewise.
8315 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
8316 VN_INFO_PTR_INFO): Remove.
8317 * tree-ssa-sccvn.c: ... (rewrite)
8318 (pass_fre::execute): For -O2+ initialize loops and run
8319 RPO VN in optimistic mode (iterating). For -O1 and -Og
8320 run RPO VN in non-optimistic mode.
8321 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
8322 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
8323 * doc/invoke.texi (sccvn-max-scc-size): Remove.
8324 (rpo-vn-max-loop-depth): Document.
8325 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
8326 when valuezing the VUSE signals we walked out of the region.
8327 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
8328 (phi_translate): Set VN context block to use for availability
8330 (compute_avail): Likewise.
8331 (pre_valueize): New function.
8332 (pass_pre::execute): Adjust to the RPO VN API.
8334 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
8335 (propagate_constants_for_unrolling): Remove.
8336 (tree_unroll_loops_completely): Perform value-numbering
8337 on the unrolled bodies loop parent.
8339 2018-08-27 Richard Biener <rguenther@suse.de>
8341 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
8342 for partial antic compute.
8344 2018-08-27 Jakub Jelinek <jakub@redhat.com>
8346 PR rtl-optimization/87065
8347 * combine.c (simplify_if_then_else): Formatting fix.
8348 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
8350 (known_cond): Don't return const_true_rtx for vector modes. Use
8351 CONST0_RTX instead of const0_rtx. Formatting fixes.
8353 2018-08-27 Martin Liska <mliska@suse.cz>
8355 PR gcov-profile/87069
8356 * gcov.c (process_file): Record files already processed
8357 and warn about a file being processed multiple times.
8359 2018-08-27 Martin Liska <mliska@suse.cz>
8362 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8363 Set default values for x_aarch64_*_string strings.
8364 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
8365 prefix. For -mabi do not print '=ABI' in help and use
8366 <option_value> format for -msve-vector-bits and -moverride
8369 2018-08-26 Jeff Law <law@redhat.com>
8371 * config/mips/frame-header-opt.c: Include "backend.h" rather than
8374 2018-08-26 Marek Polacek <polacek@redhat.com>
8376 PR c++/87029, Implement -Wredundant-move.
8377 * doc/invoke.texi: Document -Wredundant-move.
8379 2018-08-25 Martin Sebor <msebor@redhat.com>
8381 PR tree-optimization/87059
8382 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
8383 to the same type as the other.
8384 * fold-const.c (fold_binary_loc): Assert expectation.
8386 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
8388 * config/darwin.c (machopic_legitimize_pic_address): Clean up
8389 extraneous parentheses, dead code section and formatting.
8391 2018-08-24 David Malcolm <dmalcolm@redhat.com>
8394 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
8395 wide enough for jumps in the line-numbering to be visible.
8396 (layout::print_gap_in_line_numbering): New member function.
8397 (layout::calculate_line_spans): When using line numbering, merge
8398 line spans that are only 1 line apart.
8399 (diagnostic_show_locus): When printing line numbers, show gaps in
8400 line numbering directly, rather than printing headers.
8401 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
8402 line-numbering with multiple line spans.
8403 (selftest::test_fixit_insert_containing_newline_2): Add test of
8404 line-numbering, in which the spans are close enough to be merged.
8406 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
8408 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
8409 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
8410 * tree-vrp.c (range_is_nonnull): Remove.
8411 (range_includes_zero_p): Accept value_range instead of min/max.
8412 (extract_range_from_binary_expr_1): Do not early bail on
8414 Use range_includes_zero_p instead of range_is_nonnull.
8415 (extract_range_from_unary_expr): Use range_includes_zero_p instead
8416 of range_is_nonnull.
8417 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
8418 special case VR_ANTI_RANGE.
8419 (vrp_finalize): Same.
8420 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
8422 (range_is_nonnull): Remove.
8423 * vr-values.c (vrp_stmt_computes_nonzero): Use
8424 range_includes_zero_p instead of range_is_nonnull.
8425 (extract_range_basic): Pass value_range to range_includes_zero_p
8426 instead of range_is_nonnull.
8428 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
8430 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
8431 * emit-rtl.h (rtl_data): Remove return_bnd.
8432 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
8433 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
8434 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
8435 (POINTER_BOUNDS_MODE): Remove definition.
8436 (make_pointer_bounds_mode): Remove.
8437 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
8438 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
8439 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
8440 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
8441 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
8442 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
8443 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
8445 * config/i386/i386-modes.def (BND32, BND64): Remove.
8446 * config/i386/i386.c (dbx_register_map): Remove bound registers.
8447 (dbx64_register_map): Ditto.
8448 (svr4_dbx_register_map): Ditto.
8449 (indirect_thunk_bnd_needed): Remove.
8450 (indirect_thunks_bnd_used): Ditto.
8451 (indirect_return_bnd_needed): Ditto.
8452 (indirect_return_via_cx_bnd): Ditto.
8453 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
8454 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
8455 (output_indirect_thunk): Ditto. Remove need_prefix argument.
8456 (output_indirect_thunk_function): Remove handling of
8457 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
8458 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
8459 (ix86_save_reg): Remove handling of crtl->return_bnd.
8460 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
8461 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
8462 and UNSPEC_BNDLX_ADDR.
8463 (ix86_output_indirect_branch_via_reg): Remove handling of
8464 indirect_thunk_prefix_bnd.
8465 (ix86_output_indirect_branch_via_push): Ditto.
8466 (ix86_output_function_return): Ditto.
8467 (ix86_output_indirect_function_return): Ditto.
8468 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
8469 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
8470 (CALL_USED_REGISTERS): Ditto.
8471 (REG_ALLOC_ORDER): Update for removal of bound registers.
8472 (HI_REGISTER_NAMES): Ditto.
8473 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
8474 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
8475 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
8476 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
8477 (FIRST_PSEUDO_REG): Update.
8478 (BND): Remove mode iterator.
8479 * config/i386/predicates.md (bnd_mem_operator): Remove.
8481 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
8483 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
8486 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
8488 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
8489 the case in which the permute needs only a single element and
8490 repeats for every vector of the result. Extend that case to
8491 handle variable-length vectors.
8492 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8494 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
8497 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
8498 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
8500 2018-08-24 Richard Biener <rguenther@suse.de>
8502 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
8503 bb_flags_allocated members.
8504 (auto_flag): New RAII class for allocating flags.
8505 (auto_edge_flag): New RAII class for allocating edge flags.
8506 (auto_bb_flag): New RAII class for allocating bb flags.
8507 * cfgloop.c (verify_loop_structure): Allocate temporary edge
8509 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
8510 in favor of temporarily allocated BB flag.
8511 * hsa-brig.c: Re-order includes.
8512 * hsa-dump.c: Likewise.
8513 * hsa-regalloc.c: Likewise.
8514 * print-rtl.c: Likewise.
8515 * profile-count.c: Likewise.
8517 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
8520 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
8523 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
8526 * wide-int-range.cc (wide_int_range_div): Do not ignore result
8527 from wide_int_range_multiplicative_op.
8529 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8531 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
8534 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
8536 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
8539 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
8541 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
8542 full GENERIC expression used for replacement.
8544 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
8546 * tree-vrp.c (abs_extent_range): Remove.
8547 (extract_range_into_wide_ints): Pass wide ints by reference.
8548 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
8549 Pass wide ints by reference in all calls to
8550 extract_range_into_wide_ints.
8551 * wide-int-range.cc (wide_int_range_div): New.
8552 * wide-int-range.h (wide_int_range_div): New.
8553 (wide_int_range_includes_zero_p): New.
8554 (wide_int_range_zero_p): New.
8556 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
8558 * config/aarch64/aarch64.md (arches): New enum.
8559 (arch): New enum attr.
8560 (arch_enabled): New attr.
8561 (enabled): Now uses arch_enabled only.
8562 (simd, sve, fp16): Removed attribute.
8563 (fp): Attr now defined in terms of 'arch'.
8564 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
8565 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
8566 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
8567 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
8568 attributes into 'arch'.
8569 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
8570 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
8571 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
8572 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
8573 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
8574 'simd' attribute into 'arch'.
8575 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
8576 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
8577 Convert use of 'fp' attribute to 'arch'.
8578 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
8579 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
8581 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
8582 (different modes) Merge 'fp' and 'simd' into 'arch'.
8583 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
8586 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
8588 PR rtl-optimization/87026
8589 * expmed.c (canonicalize_comparison): If we can no longer create
8590 pseudoregisters, don't.
8592 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
8595 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
8597 * config/arm/arm.c (speculation_barrier_libfunc): New static
8599 (arm_init_libfuncs): Initialize it.
8600 (arm_emit_speculation_barrier): New function.
8601 * config/arm/arm.md (speculation_barrier): Call
8602 arm_emit_speculation_barrier for architectures that do not have
8604 (speculation_barrier_insn): Only match on Armv7 or later.
8606 2018-08-23 Richard Biener <rguenther@suse.de>
8609 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
8612 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
8614 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
8615 of single-vector TBLs.
8616 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
8619 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
8622 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
8623 aarch64_evpc_tbl guard.
8625 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
8627 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
8630 2018-08-22 Martin Sebor <msebor@redhat.com>
8633 * tree-pretty-print.c (pretty_print_string): Add argument.
8634 (dump_generic_node): Call to pretty_print_string with string size.
8636 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
8638 PR rtl-optimization/86771
8639 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
8640 of two SETs into those two SETs, one to be placed at i2, if that SETs
8641 destination is modified between i2 and i3.
8643 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8645 PR tree-optimization/86725
8646 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
8648 (vect_analyze_scalar_cycles_1): Check it.
8650 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8652 PR tree-optimization/86725
8653 * tree-vect-loop.c (vect_is_simple_reduction): When treating
8654 an outer loop phi as a double reduction, make sure that the
8655 single user of the phi result is an inner loop phi.
8657 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8659 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
8660 grouped stores with gaps to a strided group.
8662 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8664 * tree-vect-stmts.c (get_group_load_store_type)
8665 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
8666 first statement in a group.
8668 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8670 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
8671 the sequence used in gcc/gcc.c.
8673 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8676 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
8679 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8681 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
8682 Darwin10-specific unwinder-shim.
8683 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
8684 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
8685 New to cater for Darwin10 Rosetta.
8687 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8689 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
8690 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
8692 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8697 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
8698 (dwarf2out_switch_text_section): Generate a local label for the
8699 second function sub-section and apply it as the second FDE start
8701 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
8702 second sub-section start.
8704 2018-08-22 Richard Biener <rguenther@suse.de>
8706 PR tree-optimization/86988
8707 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
8709 2018-08-22 Richard Biener <rguenther@suse.de>
8711 PR tree-optimization/86945
8712 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
8714 2018-08-22 Alexandre Oliva <oliva@adacore.com>
8716 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
8717 a comment about how uses of r2 for .sdata2 come about.
8719 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
8721 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
8723 2018-08-21 Marek Polacek <polacek@redhat.com>
8725 PR c++/86981, Implement -Wpessimizing-move.
8726 * doc/invoke.texi: Document -Wpessimizing-move.
8728 2018-08-21 Jan Hubicka <jh@suse.cz>
8730 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
8731 * tree.h (is_redundant_typedef): Remove.
8732 * dwarf2out.c (is_redundant_typedef): Turn into static function.
8734 2018-08-21 Jan Hubicka <jh@suse.cz>
8736 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
8739 2018-08-21 Tamar Christina <tamar.christina@arm.com>
8741 * expmed.c (extract_low_bits): Reject invalid subregs early.
8743 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
8746 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
8749 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8751 * config/vxworks.h: Guard vxworks_asm_out_constructor and
8752 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
8753 * config/vxworks.c: Likewise.
8755 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8757 * config/vxworks.c: Set targetm.have_ctors_dtors
8758 if HAVE_INITFINI_ARRAY_SUPPORT.
8759 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
8760 if HAVE_INITFINI_ARRAY_SUPPORT.
8762 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8764 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
8765 default search path for VxWorks < 7.
8767 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
8769 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
8770 (get_format_string): Refer to c_getstr.
8772 2018-08-21 Tom de Vries <tdevries@suse.de>
8774 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
8775 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
8776 (debuginfo_early_stop): Declare.
8777 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
8778 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
8779 (debuginfo_early_stop): New function.
8780 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
8781 and debuginfo_early_stop.
8782 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
8783 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
8784 (general_init): Call debuginfo_early_init.
8785 (finalize): Call debuginfo_fini.
8786 (do_compile): Call debuginfo_init.
8787 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
8789 (@item -fdump-debug, @item -fdump-earlydebug): Add.
8791 2018-08-21 Tom de Vries <tdevries@suse.de>
8793 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
8794 flag_dump_noaddr and flag_dump_unnumbered.
8796 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
8798 * wide-int-range.cc (wide_int_range_abs): New.
8799 (wide_int_range_order_set): Rename from wide_int_range_min_max.
8800 * wide-int-range.h (wide_int_range_abs): New.
8801 (wide_int_range_min_max): New.
8802 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
8803 case to call wide_int_range_abs.
8804 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
8805 (extract_range_from_abs_expr): Delete.
8807 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
8810 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
8811 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
8814 2018-08-20 Nathan Sidwell <nathan@acm.org>
8815 Jeff Law <law@redhat.com>
8817 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
8818 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
8820 2018-08-20 David Malcolm <dmalcolm@redhat.com>
8823 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
8824 (decl_attributes): Likewise.
8825 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
8827 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
8828 * diagnostic-core.h (class auto_diagnostic_group): New class.
8829 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
8830 (diagnostic_report_diagnostic): Handle the first diagnostics within
8832 (emit_diagnostic): Add auto_diagnostic_group instance.
8834 (inform_n): Likewise.
8835 (warning): Likewise.
8836 (warning_at): Likewise.
8837 (warning_n): Likewise.
8838 (pedwarn): Likewise.
8839 (permerror): Likewise.
8841 (error_n): Likewise.
8842 (error_at): Likewise.
8844 (fatal_error): Likewise.
8845 (internal_error): Likewise.
8846 (internal_error_no_backtrace): Likewise.
8847 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
8848 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
8849 * diagnostic.h (struct diagnostic_context): Add fields
8850 "diagnostic_group_nesting_depth",
8851 "diagnostic_group_emission_count", "begin_group_cb",
8853 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
8854 Add auto_diagnostic_group instance(s).
8855 (find_explicit_erroneous_behavior): Likewise.
8856 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
8857 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
8858 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
8859 (gimplify_va_arg_expr): Likewise.
8860 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
8861 (HSA_SORRY_AT): Likewise.
8862 * ipa-devirt.c (compare_virtual_tables): Likewise.
8863 (warn_odr): Likewise.
8864 * multiple_target.c (expand_target_clones): Likewise.
8865 * opts-common.c (cmdline_handle_error): Likewise.
8866 * reginfo.c (globalize_reg): Likewise.
8867 * substring-locations.c (format_warning_n_va): Likewise.
8868 * tree-inline.c (expand_call_inline): Likewise.
8869 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
8870 * tree-ssa-loop-niter.c
8871 (do_warn_aggressive_loop_optimizations): Likewise.
8872 * tree-ssa-uninit.c (warn_uninit): Likewise.
8873 * tree.c (warn_deprecated_use): Likewise.
8875 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
8878 * config/i386/i386.md (eh_return): Always update EH return
8879 address in word_mode.
8881 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
8883 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
8884 TARGET_SPLIT_COMPLEX_ARG is defined.
8886 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
8888 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
8890 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
8893 * expr.c (expand_assignment): Assert that bitpos is positive.
8894 (store_field): Likewise
8895 (expand_expr_real_1): Make sure that bitpos is positive.
8896 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
8899 2018-08-20 Nathan Sidwell <nathan@acm.org>
8901 * Makefile.in (CPP_ID_DATA_H): Delete.
8902 (CPP_INTERNAL_H): Don't add it.
8903 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
8904 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
8906 2018-08-20 Richard Biener <rguenther@suse.de>
8908 PR tree-optimization/78655
8909 * tree-vrp.c (extract_range_from_binary_expr_1): Make
8910 pointer + offset nonnull if either operand is nonnull work.
8912 2018-08-20 Tom de Vries <tdevries@suse.de>
8914 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
8915 unless the referenced die describes the added property using
8916 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
8917 Otherwise, add a DW_AT_location to the referenced die.
8919 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
8922 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
8923 register_operand when calling ix86_set_reg_reg_cost.
8924 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
8925 Set *total to 0 for operands that satisfy x86_64_immediate_operand
8926 predicate and to 1 otherwise.
8928 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
8930 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
8931 emit a diagnostic that it is not supported and reset the option.
8932 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
8933 supported and consume the option. (ASM_FINAL_SPEC): New.
8935 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
8937 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
8940 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
8942 C-SKY port: Documentation
8944 * doc/extend.texi (C-SKY Function Attributes): New section.
8945 * doc/invoke.texi (Option Summary): Add C-SKY options.
8946 (C-SKY Options): New section.
8947 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
8949 2018-08-17 Jojo <jijie_rong@c-sky.com>
8950 Huibin Wang <huibin_wang@c-sky.com>
8951 Sandra Loosemore <sandra@codesourcery.com>
8952 Chung-Lin Tang <cltang@codesourcery.com>
8954 C-SKY port: Backend implementation
8956 * config/csky/*: New.
8957 * common/config/csky/*: New.
8959 2018-08-17 Jojo <jijie_rong@c-sky.com>
8960 Huibin Wang <huibin_wang@c-sky.com>
8961 Sandra Loosemore <sandra@codesourcery.com>
8962 Chung-Lin Tang <cltang@codesourcery.com>
8963 Andrew Jenner <andrew@codesourcery.com>
8965 C-SKY port: Configury
8967 * config.gcc (csky-*-*): New.
8968 * configure.ac: Add csky to targets for dwarf2 debug_line support.
8969 * configure: Regenerated.
8971 2018-08-17 David Malcolm <dmalcolm@redhat.com>
8973 * dump-context.h: Include "dumpfile.h".
8974 (dump_context::dump_printf_va): Convert final param from va_list
8975 to va_list *. Convert from ATTRIBUTE_PRINTF to
8976 ATTRIBUTE_GCC_DUMP_PRINTF.
8977 (dump_context::dump_printf_loc_va): Likewise.
8978 * dumpfile.c: Include "stringpool.h".
8979 (make_item_for_dump_printf_va): Delete.
8980 (make_item_for_dump_printf): Delete.
8981 (class dump_pretty_printer): New class.
8982 (dump_pretty_printer::dump_pretty_printer): New ctor.
8983 (dump_pretty_printer::emit_items): New member function.
8984 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
8986 (dump_pretty_printer::emit_item): New member function.
8987 (dump_pretty_printer::stash_item): New member function.
8988 (dump_pretty_printer::format_decoder_cb): New member function.
8989 (dump_pretty_printer::decode_format): New member function.
8990 (dump_context::dump_printf_va): Reimplement in terms of
8991 dump_pretty_printer.
8992 (dump_context::dump_printf_loc_va): Convert final param from va_list
8994 (dump_context::begin_scope): Reimplement call to
8995 make_item_for_dump_printf.
8996 (dump_printf): Update for change to dump_printf_va.
8997 (dump_printf_loc): Likewise.
8998 (selftest::test_capture_of_dump_calls): Convert "stmt" from
8999 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
9000 with %T, %E, and %G.
9001 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
9002 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
9003 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
9004 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
9005 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
9006 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
9007 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
9008 within a dump_printf_loc call to "%wu".
9009 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
9010 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
9011 missing space after "=".
9012 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
9013 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
9014 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
9015 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
9016 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
9017 duplicate "vectorized" from message.
9019 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
9021 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
9022 polyNxK_t element's TYPE_STRING_FLAG.
9024 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
9026 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
9027 (they were unnamed before). Fix comments.
9029 2018-08-17 Nathan Sidwell <nathan@acm.org>
9031 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
9033 2018-08-17 Richard Biener <rguenther@suse.de>
9035 PR tree-optimization/86841
9036 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
9038 2018-08-17 Martin Liska <mliska@suse.cz>
9040 * common.opt: Remove Warn, Init and Report for options with
9041 Ignore/Deprecated flag. Warning is done automatically for
9043 * config/i386/i386.opt: Likewise.
9044 * config/ia64/ia64.opt: Likewise.
9045 * config/rs6000/rs6000.opt: Likewise.
9046 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
9047 Remove usage of flag_check_pointer_bounds.
9048 * lto-wrapper.c (merge_and_complain): Do not handle
9049 OPT_fcheck_pointer_bounds.
9050 (append_compiler_options): Likewise.
9051 * opt-functions.awk: Do not handle Deprecated.
9052 * optc-gen.awk: Check that Var, Report and Init are not
9053 used for an option with Ignore/Deprecated flag.
9054 * opts-common.c (decode_cmdline_option): Do not report
9056 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
9058 * opts.h (struct cl_option): Remove cl_deprecated flag.
9059 (CL_ERR_DEPRECATED): Remove error enum value.
9061 2018-08-17 Richard Biener <rguenther@suse.de>
9064 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
9065 across a va-arg-pack using call adjust its return value accordingly.
9067 2018-08-16 Martin Sebor <msebor@redhat.com>
9069 PR tree-optimization/86853
9070 * gimple-ssa-sprintf.c (struct format_result): Rename member.
9071 (struct fmtresult): Add member and initialize it in ctors.
9072 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
9073 (format_string): Handle %S the same as %ls. Set MAYFAIL.
9074 (format_directive): Set POSUNDER4K when MAYFAIL is set.
9075 (parse_directive): Handle %C same as %c.
9076 (sprintf_dom_walker::compute_format_length): Adjust.
9077 (is_call_safe): Adjust.
9079 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9081 * builtins.c (c_strlen): Add new parameter eltsize. Use it
9082 for determining how to count the elements.
9083 * builtins.h (c_strlen): Adjust prototype.
9084 * expr.c (string_constant): Add new parameter mem_size.
9085 Set *mem_size appropriately.
9086 * expr.h (string_constant): Adjust protoype.
9087 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
9088 * gimple-fold.h (get_range_strlen): Adjust prototype.
9089 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
9090 (format_string): Call get_string_length with eltsize.
9092 2018-08-16 David Malcolm <dmalcolm@redhat.com>
9094 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
9095 to emit the span, rather than setting it as the prefix.
9097 2018-08-16 David Malcolm <dmalcolm@redhat.com>
9099 * diagnostic-show-locus.c (layout::start_annotation_line): Add
9100 "margin_char" parameter, defaulting to space. Use it in place
9101 of pp_space for the initial part of the margin.
9102 (layout::print_leading_fixits): Use '+' when filling the margin
9103 of line-insertion fix-it hints.
9105 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9107 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
9110 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9112 * config/rs6000/altivec.md: Don't set length attribute to the default
9114 * config/rs6000/darwin.md: Ditto.
9115 * config/rs6000/dfp.md: Ditto.
9116 * config/rs6000/htm.md: Ditto.
9117 * config/rs6000/rs6000.md: Ditto.
9118 * config/rs6000/sync.md: Ditto.
9119 * config/rs6000/vsx.md: Ditto.
9121 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9123 * config/rs6000/altivec.md: Don't set length attribute to the default
9124 value, for branch instructions.
9125 * config/rs6000/darwin.md: Ditto.
9126 * config/rs6000/rs6000.md: Ditto.
9128 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9130 * config/rs6000/rs6000.md (length): Always define as const_int 4.
9131 (unnamed conditional branch define_insn): Set length to 4 or 8
9132 depending on offset.
9133 (<bd>_<mode>): Similar, for alternative 0.
9134 (<bd>tf_<mode>): Ditto.
9136 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9138 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
9140 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
9142 * doc/rtl.texi: Replace old RTX class names with new names.
9145 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
9147 * expmed.h (canonicalize_comparison): New declaration.
9148 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
9149 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
9150 * optabs.c (prepare_cmp_insn): Likewise.
9151 * rtl.h (unsigned_condition_p): New function which checks if a
9152 comparison operator is unsigned.
9154 2018-08-16 Nathan Sidwell <nathan@acm.org>
9156 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
9157 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
9159 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9162 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
9163 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
9164 (mov<mov>): ..this and enable unconditionally.
9166 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9168 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
9170 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
9172 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
9173 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
9174 "Common" with "Target".
9176 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
9178 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
9179 * doc/invoke.texi (mmitigate-rop): Remove.
9180 * config/i386/i386.c: Do not include "regrename.h".
9181 (ix86_rop_should_change_byte_p, reg_encoded_number)
9182 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
9184 (ix86_reorg): Remove call to ix86_mitigate_rop.
9185 * config/i386/i386.md (attr "modrm_class"): Remove.
9186 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
9187 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
9188 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
9190 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
9192 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
9193 allow folding of mergeh() and mergel() for the float and double types.
9194 (fold_mergehl_helper): Rework to handle building a permute tree
9197 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
9199 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
9202 2018-08-15 David Malcolm <dmalcolm@redhat.com>
9204 * common.opt (fdiagnostics-show-labels): New option.
9205 * diagnostic-show-locus.c (class layout_range): Add field
9207 (class layout): Add field "m_show_labels_p".
9208 (layout_range::layout_range): Add param "label" and use it to
9210 (make_range): Pass in NULL for new "label" param of layout_range's
9212 (layout::layout): Initialize m_show_labels_p.
9213 (layout::maybe_add_location_range): Pass in loc_range->m_label
9214 when constructing layout_range instances.
9215 (struct line_label): New struct.
9216 (layout::print_any_labels): New member function.
9217 (layout::print_line): Call it if label-printing is enabled.
9218 (selftest::test_one_liner_labels): New test.
9219 (selftest::test_diagnostic_show_locus_one_liner): Call it.
9220 * diagnostic.c (diagnostic_initialize): Initialize
9221 context->show_labels_p.
9222 * diagnostic.h (struct diagnostic_context): Add field
9224 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
9225 -fno-diagnostics-show-labels.
9226 * dwarf2out.c (gen_producer_string): Add
9227 OPT_fdiagnostics_show_labels to the ignored options.
9228 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
9230 (gcc_rich_location::maybe_add_expr): Likewise.
9231 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
9232 label" param, defaulting to NULL.
9233 (gcc_rich_location::add_expr): Add "label" param.
9234 (gcc_rich_location::maybe_add_expr): Likewise.
9235 (class text_range_label): New class.
9236 (class range_label_for_type_mismatch): New class.
9237 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
9238 of format_warning_va.
9239 (fmtwarn_n): Likewise for new params of format_warning_n_va.
9240 * lto-wrapper.c (merge_and_complain): Add
9241 OPT_fdiagnostics_show_labels to the "pick one setting" options.
9242 (append_compiler_options): Likewise to the dropped options.
9243 (append_diag_options): Likewise to the passed-on options.
9244 * opts.c (common_handle_option): Handle the new option.
9245 * selftest-diagnostic.c
9246 (test_diagnostic_context::test_diagnostic_context): Enable
9248 * substring-locations.c: Include "gcc-rich-location.h".
9249 (format_warning_n_va): Add "fmt_label" and "param_label" params
9250 and use them as appropriate.
9251 (format_warning_va): Add "fmt_label" and "param_label" params,
9252 passing them on to format_warning_n_va.
9253 (format_warning_at_substring): Likewise.
9254 (format_warning_at_substring_n): Likewise.
9255 * substring-locations.h (format_warning_va): Add "fmt_label" and
9256 "param_label" params.
9257 (format_warning_n_va): Likewise.
9258 (format_warning_at_substring): Likewise.
9259 (format_warning_at_substring_n): Likewise.
9260 * toplev.c (general_init): Initialize global_dc->show_labels_p.
9262 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
9265 * builtins.c (expand_builtin_memcmp): Do not expand the call
9266 when overflow is detected.
9268 2018-08-15 Martin Sebor <msebor@redhat.com>
9270 PR tree-optimization/71625
9271 * config/aarch64/aarch64-builtins.c
9272 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
9274 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
9276 * config/s390/s390.c (s390_reorg): Remove loop.
9278 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
9281 (darwin_function_switched_text_sections): Delete.
9282 * gcc/config/darwin.h
9283 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
9285 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
9288 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
9289 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
9290 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
9292 2018-08-15 Martin Liska <mliska@suse.cz>
9294 PR tree-optimization/86925
9295 * predict.c (expr_expected_value_1): When taking
9296 later predictor, assign also probability.
9297 Use fold_build2_initializer_loc in order to fold
9298 the expression in -frounding-math.
9300 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
9302 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
9304 (expand_vec_perm_1): Try the new method.
9306 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
9309 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
9310 Check whether lra_live_max_point is 0 before dividing.
9312 2018-08-14 Martin Sebor <msebor@redhat.com>
9314 PR tree-optimization/86650
9315 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
9316 (vrp_prop::check_mem_ref): Same.
9318 2018-08-13 Liu Hao <lh_mouse@126.com>
9320 * pretty-print.c (eat_esc_sequence): Swap the foreground and
9321 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
9322 and clear it thereafter, as it only works for DBCS.
9324 2018-08-13 Liu Hao <lh_mouse@126.com>
9326 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
9327 handle returned by _get_osf_handle().
9329 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9331 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
9332 for folding vec_perm.
9334 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
9337 Add support for gimple-folding of vec_pack() and vec_unpack()
9340 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9342 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
9343 vec_xst variants to the list.
9344 (rs6000_gimple_fold_builtin): Add support for folding unaligned
9345 vector loads and stores.
9347 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
9349 * config.gcc (rs6000-ibm-aix4.x): Delete.
9350 (rs6000-ibm-aix5.1): Delete.
9351 (rs6000-ibm-aix5.2): Delete.
9352 (rs6000-ibm-aix5.3): Delete.
9353 * config/rs6000/aix43.h: Delete.
9354 * config/rs6000/aix51.h: Delete.
9355 * config/rs6000/aix52.h: Delete.
9356 * config/rs6000/t-aix43: Delete.
9358 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
9360 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9362 (s390_decompose_address): Factor out constant pool ref
9365 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9367 * config/nds32/nds32-predicates.c
9368 (nds32_can_use_bclr_p): Change return type as bool.
9369 (nds32_can_use_bset_p): Ditto.
9370 (nds32_can_use_btgl_p): Ditto.
9371 (nds32_can_use_bitci_p): Ditto.
9372 * config/nds32/nds32-protos.h
9373 (nds32_can_use_bclr_p): Change declaration.
9374 (nds32_can_use_bset_p): Ditto.
9375 (nds32_can_use_btgl_p): Ditto.
9376 (nds32_can_use_bitci_p): Ditto.
9378 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9380 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
9381 Support -msched-prolog-epilog option.
9382 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
9384 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9386 * common/config/nds32/nds32-common.c
9387 (nds32_option_optimization_table): Enalbe -malways-align.
9389 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9391 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
9393 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
9394 OPT_misr_secure_ case.
9395 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
9396 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
9397 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
9399 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
9400 (nds32_isr_info): New field security_level.
9401 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
9402 * config/nds32/nds32.md (return_internal): Consider critical attribute.
9403 * config/nds32/nds32.opt (misr-secure): New option.
9404 * config/nds32/nds32_init.inc: New file.
9405 * config/nds32/nds32_isr.h: New file.
9407 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
9409 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
9410 Update comment for atomic instructions.
9411 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
9412 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
9414 (atomic_loaddi): Revise fence expansion to only emit fence prior to
9415 load for __ATOMIC_SEQ_CST model.
9416 (atomic_loaddi_1): Remove float register target.
9417 (atomic_storedi): Handle CONST_INT values.
9418 (atomic_storedi_1): Remove float register source. Add special case
9420 (memory_barrier): New expander and insn.
9422 2018-08-11 Jakub Jelinek <jakub@redhat.com>
9424 PR tree-optimization/86835
9425 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
9426 new_stmt after def_gsi, make sure to insert new_square_stmt after
9427 that stmt, not 2 stmts before it.
9429 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
9432 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
9435 2018-08-10 Martin Liska <mliska@suse.cz>
9438 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
9440 * builtins.c (expand_builtin_expect_with_probability):
9442 (expand_builtin_expect_with_probability): New function.
9443 (build_builtin_expect_predicate): Add new argumnet probability
9444 for BUILT_IN_EXPECT_WITH_PROBABILITY.
9445 (fold_builtin_expect):
9448 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
9449 * builtins.h (fold_builtin_expect): Set new argument.
9450 * doc/extend.texi: Document __builtin_expect_with_probability.
9451 * doc/invoke.texi: Likewise.
9452 * gimple-fold.c (gimple_fold_call): Pass new argument.
9453 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
9454 also BUILT_IN_EXPECT_WITH_PROBABILITY.
9455 * predict.c (get_predictor_value): New function.
9456 (expr_expected_value): Add new argument probability. Assume
9457 that predictor and probability are always non-null.
9458 (expr_expected_value_1): Likewise. For __builtin_expect and
9459 __builtin_expect_with_probability set probability. Handle
9460 combination in binary expressions.
9461 (tree_predict_by_opcode): Simplify code by simply calling
9462 get_predictor_value.
9463 (pass_strip_predict_hints::execute): Add handling of
9464 BUILT_IN_EXPECT_WITH_PROBABILITY.
9465 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
9467 * tree.h (DECL_BUILT_IN_P): New function.
9469 2018-08-10 Martin Liska <mliska@suse.cz>
9471 PR tree-optimization/85799
9472 * passes.def: Add argument for pass_strip_predict_hints.
9473 * predict.c (class pass_strip_predict_hints): Add new argument
9475 (strip_predictor_early): New function.
9476 (pass_strip_predict_hints::execute): Call the function to
9478 (strip_predict_hints): New function.
9479 * predict.def: Fix comment.
9481 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
9483 * Makefile.in: Clarify which tm.texi to copy over to assert the
9484 right to grant a GFDL license for all.
9486 2018-08-09 Jeff Law <law@redhat.com>
9488 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
9491 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
9493 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
9496 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9498 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
9499 reductions for variable-length vectors.
9501 2018-08-09 David Malcolm <dmalcolm@redhat.com>
9504 * common.opt (fdiagnostics-show-line-numbers): New option.
9505 * diagnostic-show-locus.c (class layout): Add fields
9506 "m_show_line_numbers_p" and "m_linenum_width";
9507 (num_digits): New function.
9508 (test_num_digits): New function.
9509 (layout::layout): Initialize new fields. Update m_x_offset
9510 logic to handle any left margin.
9511 (layout::print_source_line): Print line number when requested.
9512 (layout::start_annotation_line): New member function.
9513 (layout::print_annotation_line): Call it.
9514 (layout::print_leading_fixits): Likewise.
9515 (layout::print_trailing_fixits): Likewise. Update calls to
9516 move_to_column for new parameter.
9517 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
9518 it to potentially call start_annotation_line.
9519 (layout::show_ruler): Call start_annotation_line.
9520 (selftest::test_line_numbers_multiline_range): New selftest.
9521 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
9522 and selftest::test_line_numbers_multiline_range.
9523 * diagnostic.c (diagnostic_initialize): Initialize
9524 show_line_numbers_p.
9525 * diagnostic.h (struct diagnostic_context): Add field
9526 "show_line_numbers_p".
9527 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
9528 -fno-diagnostics-show-line-numbers.
9529 * dwarf2out.c (gen_producer_string): Add
9530 OPT_fdiagnostics_show_line_numbers to the ignored options.
9531 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
9532 one setting" options.
9533 (append_compiler_options): Likewise to the dropped options.
9534 (append_diag_options): Likewise to the passed-on options.
9535 * opts.c (common_handle_option): Handle the new option.
9536 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
9538 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
9540 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
9541 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
9542 third argument of type "const signed char" to descriptions of
9543 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
9544 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
9545 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
9546 __builtin_bcdsub_ov functions.
9548 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9550 PR tree-optimization/86858
9551 * tree-vect-loop.c (vect_is_simple_reduction): Restore
9552 flow_bb_inside_loop_p calls.
9554 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9556 PR tree-optimization/86871
9557 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
9558 instead of gimple_assign_lhs.
9560 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
9563 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
9564 register constraint to operand 0.
9565 (add<mode>3_carryinC): Likewise.
9566 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
9568 2018-08-09 Martin Liska <mliska@suse.cz>
9571 * common.opt: Remove extra line.
9573 2018-08-09 Martin Liska <mliska@suse.cz>
9575 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
9576 at the end of a line, make first letter capital and end up
9577 a sentence with a dot.
9578 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
9579 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
9580 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
9581 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
9582 (PARAM_MAX_ISL_OPERATIONS): Likewise.
9583 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
9584 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
9585 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
9586 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
9587 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
9588 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
9589 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
9590 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
9591 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
9592 (PARAM_TREE_REASSOC_WIDTH): Likewise.
9593 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
9594 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
9595 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
9597 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
9600 * config/s390/s390.c (s390_option_override_internal): Reduce the
9601 stack-clash-protection-probe-interval param if it would be too big
9604 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
9607 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
9608 * config/m68k/m68k.c (handle_move_double): Don't call
9609 m68k_final_prescan_insn.
9610 (m68k_adjust_decorated_operand): Renamed from
9611 m68k_final_prescan_insn, remove first and third operand and
9613 (print_operand): Call it.
9614 (print_operand_address): Call it.
9616 2018-08-08 Nathan Sidwell <nathan@acm.org>
9618 * diagnostic.c (diagnostic_report_current_module): Use
9619 linemap_included_from & linemap_included_from_linemap.
9621 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
9623 * config/aarch64/aarch64-cores.def: Add phecda core.
9624 * config/aarch64/aarch64-tune.md: Regenerate.
9625 * doc/invoke.texi: Add phecda core.
9627 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
9630 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
9632 * config/s390/s390.md ("movti"): Add more alternatives for
9633 constant to GPR copies.
9635 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
9637 * config/s390/s390.c: Fix whitespace damage throughout the file.
9638 * config/s390/s390.h: Likewise.
9639 * config/s390/tpf.h: Likewise.
9641 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
9643 * config/s390/s390.c (s390_loadrelative_operand_p):
9644 Remove TARGET_CPU_ZARCH usages.
9645 (s390_rtx_costs): Likewise.
9646 (s390_legitimate_constant_p): Likewise.
9647 (s390_cannot_force_const_mem): Likewise.
9648 (legitimate_reload_constant_p): Likewise.
9649 (s390_preferred_reload_class): Likewise.
9650 (legitimize_pic_address): Likewise.
9651 (legitimize_tls_address): Likewise.
9652 (s390_split_branches): Removed.
9653 (s390_add_execute): Removed.
9654 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
9655 (s390_mainpool_start): Likewise.
9656 (s390_mainpool_finish): Likewise.
9657 (s390_mainpool_cancel): Removed.
9658 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
9659 (s390_chunkify_cancel): Likewise.
9660 (s390_return_addr_rtx): Likewise.
9661 (s390_register_info): Remove split_branches_pending_p uages.
9662 (s390_optimize_register_info): Likewise.
9663 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
9664 split_branches_pending_p usages.
9665 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
9666 (s390_load_got): Likewise.
9667 (s390_expand_split_stack_prologue): Likewise.
9668 (output_asm_nops): Likewise.
9669 (s390_function_profiler): Likewise.
9670 (s390_emit_call): Likewise.
9671 (s390_conditional_register_usage): Likewise.
9672 (s390_optimize_prologue): Likewise.
9673 (s390_reorg): Remove TARGET_CPU_ZARCH and
9674 split_branches_pending_p usages.
9675 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
9677 (s390_output_indirect_thunk_function): Likewise.
9678 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
9679 (TARGET_CPU_ZARCH_P): Removed.
9680 (struct machine_function): Remove split_branches_pending_p.
9681 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
9683 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
9685 * common/config/s390/s390-common.c (processor_flags_table):
9687 * config.gcc: Remove with_arch/with_tune support.
9688 * config/s390/2064.md: Remove cpu attribute comparisons.
9689 * config/s390/driver-native.c (s390_host_detect_local_cpu):
9691 * config/s390/linux.h (ASM_SPEC):
9692 Remove -march support.
9693 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
9694 Use a table to get an arch level.
9695 * config/s390/s390-opts.h (enum processor_type):
9697 * config/s390/s390.c
9698 (processor_table): Remove entries, add arch_level values.
9699 (s390_issue_rate): Remove cases.
9700 (s390_option_override): Adjust
9701 s390_option_override_internal() call.
9702 (s390_option_override_internal): Remove deprecation warning.
9703 (s390_valid_target_attribute_tree): Adjust
9704 s390_option_override_internal() call.
9705 * config/s390/s390.h (struct s390_processor):
9706 Share with s390-c.c, add arch_level field.
9707 * config/s390/s390.md:
9708 Remove occurrences in cpu attribute.
9709 * config/s390/s390.opt: Remove -march/-mtune support.
9710 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
9711 * doc/invoke.texi: Remove deprecation warning.
9713 2018-08-08 Luis Machado <luis.machado@linaro.org>
9715 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
9717 (qdf24xx_tunings): Set vector cost structure to
9718 qdf24xx_vector_cost.
9720 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
9721 <register_sextend>: Set to 3.
9723 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
9726 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
9727 * config/aarch64/aarch64-simd.md
9728 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
9729 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
9730 (aarch64_frecpx<mode>): ...this new pattern.
9731 * config/aarch64/aarch64-simd-builtins.def: Remove comment
9732 about aarch64_frecp<FRECP:frecp_suffix><mode>.
9734 2018-08-07 Martin Liska <mliska@suse.cz>
9737 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
9738 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
9739 * predict.def (PRED_MALLOC_NONNULL): New predictor.
9740 * doc/extend.texi: Document that malloc attribute adds
9743 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
9746 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9747 Define to speculation_safe_value_not_needed.
9749 2018-08-06 Jeff Law <law@redhat.com>
9751 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
9752 the vr_values instance to cprop_into_stmt.
9753 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
9754 (cprop_operand): Also query EVRP to determine if OP is a constant.
9756 2018-08-06 Nathan Sidwell <nathan@acm.org>
9758 * diagnostic.c (diagnostic_report_current_module): Reroll
9759 included-at loop. Translate text.
9761 2018-08-06 David Malcolm <dmalcolm@redhat.com>
9763 * function-tests.c (selftest::test_expansion_to_rtl): Call
9764 free_after_compilation.
9766 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9768 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
9770 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
9772 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
9773 loops with memory block operations from getting unrolled.
9775 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
9778 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9779 Define to speculation_safe_value_not_needed.
9781 2018-08-06 Jeff Law <law@redhat.com>
9783 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
9786 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9789 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
9790 with all enabled __intN types.
9792 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
9794 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9796 * alias.c (record_set): Check for clobber high.
9797 * cfgexpand.c (expand_gimple_stmt): Likewise.
9798 * combine-stack-adj.c (single_set_for_csa): Likewise.
9799 * combine.c (find_single_use_1): Likewise.
9800 (set_nonzero_bits_and_sign_copies): Likewise.
9801 (get_combine_src_dest): Likewise.
9802 (is_parallel_of_n_reg_sets): Likewise.
9803 (try_combine): Likewise.
9804 (record_dead_and_set_regs_1): Likewise.
9805 (reg_dead_at_p_1): Likewise.
9806 (reg_dead_at_p): Likewise.
9807 * dce.c (deletable_insn_p): Likewise.
9808 (mark_nonreg_stores_1): Likewise.
9809 (mark_nonreg_stores_2): Likewise.
9810 * df-scan.c (df_find_hard_reg_defs): Likewise.
9811 (df_uses_record): Likewise.
9812 (df_get_call_refs): Likewise.
9813 * dwarf2out.c (mem_loc_descriptor): Likewise.
9814 * haifa-sched.c (haifa_classify_rtx): Likewise.
9815 * ira-build.c (create_insn_allocnos): Likewise.
9816 * ira-costs.c (scan_one_insn): Likewise.
9817 * ira.c (equiv_init_movable_p): Likewise.
9818 (rtx_moveable_p): Likewise.
9819 (interesting_dest_for_shprep): Likewise.
9820 * jump.c (mark_jump_label_1): Likewise.
9821 * postreload-gcse.c (record_opr_changes): Likewise.
9822 * postreload.c (reload_cse_simplify): Likewise.
9823 (struct reg_use): Add source expr.
9824 (reload_combine): Check for clobber high.
9825 (reload_combine_note_use): Likewise.
9826 (reload_cse_move2add): Likewise.
9827 (move2add_note_store): Likewise.
9828 * print-rtl.c (print_pattern): Likewise.
9829 * recog.c (decode_asm_operands): Likewise.
9830 (store_data_bypass_p): Likewise.
9831 (if_test_bypass_p): Likewise.
9832 * regcprop.c (kill_clobbered_value): Likewise.
9833 (kill_set_value): Likewise.
9834 * reginfo.c (reg_scan_mark_refs): Likewise.
9835 * reload1.c (maybe_fix_stack_asms): Likewise.
9836 (eliminate_regs_1): Likewise.
9837 (elimination_effects): Likewise.
9838 (mark_not_eliminable): Likewise.
9839 (scan_paradoxical_subregs): Likewise.
9840 (forget_old_reloads_1): Likewise.
9841 * reorg.c (find_end_label): Likewise.
9842 (try_merge_delay_insns): Likewise.
9843 (redundant_insn): Likewise.
9844 (own_thread_p): Likewise.
9845 (fill_simple_delay_slots): Likewise.
9846 (fill_slots_from_thread): Likewise.
9847 (dbr_schedule): Likewise.
9848 * resource.c (update_live_status): Likewise.
9849 (mark_referenced_resources): Likewise.
9850 (mark_set_resources): Likewise.
9851 * rtl.c (copy_rtx): Likewise.
9852 * rtlanal.c (reg_referenced_p): Likewise.
9853 (single_set_2): Likewise.
9854 (noop_move_p): Likewise.
9855 (note_stores): Likewise.
9856 * sched-deps.c (sched_analyze_reg): Likewise.
9857 (sched_analyze_insn): Likewise.
9859 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9861 * cse.c (invalidate_reg): New function extracted from...
9862 (invalidate): ...here.
9863 (canonicalize_insn): Check for clobber high.
9864 (invalidate_from_clobbers): invalidate clobber highs.
9865 (invalidate_from_sets_and_clobbers): Likewise.
9866 (count_reg_usage): Check for clobber high.
9867 (insn_live_p): Likewise.
9868 * cselib.c (cselib_expand_value_rtx_1):Likewise.
9869 (cselib_invalidate_regno): Check for clobber in setter.
9870 (cselib_invalidate_rtx): Pass through setter.
9871 (cselib_invalidate_rtx_note_stores):
9872 (cselib_process_insn): Check for clobber high.
9873 * cselib.h (cselib_invalidate_rtx): Add operand.
9875 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9877 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
9878 (mark_not_eliminable): Likewise.
9879 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
9880 * lra-lives.c (process_bb_lives): Check for clobber high.
9881 * lra.c (new_insn_reg): Remember clobber highs.
9882 (collect_non_operand_hard_regs): Check for clobber high.
9883 (lra_set_insn_recog_data): Likewise.
9884 (add_regs_to_insn_regno_info): Likewise.
9885 (lra_update_insn_regno_info): Likewise.
9887 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9889 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
9890 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
9892 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9894 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
9895 (copy_insn_1): Likewise.
9896 (gen_hard_reg_clobber_high): New gen function.
9897 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
9898 * genemit.c (gen_exp): Likewise.
9899 (gen_emit_seq): Pass through info.
9900 (gen_insn): Check for CLOBBER_HIGH.
9901 (gen_expand): Pass through info.
9902 (gen_split): Likewise.
9903 (output_add_clobbers): Likewise.
9904 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
9905 (remove_clobbers): Likewise.
9906 * rtl.h (gen_hard_reg_clobber_high): New declaration.
9908 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9910 * doc/rtl.texi (clobber_high): Add.
9911 (parallel): Add in clobber high
9912 * rtl.c (rtl_check_failed_code3): Add function.
9913 * rtl.def (CLOBBER_HIGH): Add expression.
9914 * rtl.h (RTL_CHECKC3): Add macro.
9915 (rtl_check_failed_code3): Add declaration.
9916 (XC3EXP): Add macro.
9918 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
9921 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
9922 cfun->machine->max_used_stack_alignment if needed.
9924 2018-08-04 Martin Sebor <msebor@redhat.com>
9926 PR tree-optimization/86571
9927 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
9930 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
9933 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9936 2018-08-03 Jeff Law <law@redhat.com>
9939 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9940 Define to speculation_safe_value_not_needed.
9942 2018-08-03 David Malcolm <dmalcolm@redhat.com>
9944 * doc/gcov.texi (-x): Remove duplicate "to".
9945 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
9946 (-Wif-not-aligned): Remove duplicate "is".
9947 (-flto): Remove duplicate "the".
9948 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
9949 duplicate "v5.00.b".
9950 (MSP430 Options): Remove duplicate "and" from the description
9951 of "-mgprel-sec=regexp".
9952 (x86 Options): Remove duplicate copies of "vmldLog102" and
9953 vmlsLog104 from description of "-mveclibabi=type".
9955 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
9957 * internal-fn.h (first_commutative_argument): Declare.
9958 * internal-fn.c (first_commutative_argument): New function.
9959 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
9960 restrictions for pattern statements. Use first_commutative_argument
9961 to look for commutative operands in calls to internal functions.
9963 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
9965 * Makefile.in (wide-int-range.o): New.
9966 * tree-vrp.c: Move all the wide_int_* functions to...
9967 * wide-int-range.cc: ...here.
9968 * tree-vrp.h: Move all the wide_int_* prototypes to...
9969 * wide-int-range.h: ...here.
9971 2018-08-03 Tom de Vries <tdevries@suse.de>
9973 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
9975 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
9976 * except.c (output_function_exception_table): Do early exit if
9977 targetm_common.except_unwind_info (&global_options) == UI_NONE.
9979 2018-08-03 Martin Liska <mliska@suse.cz>
9981 * predict.c (dump_prediction): Change to 2 digits
9984 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
9986 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
9988 (falkor_am_1_gtov_gtov): ... a new insn reservation.
9990 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
9992 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
9993 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
9994 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
9995 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
9996 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
9998 2018-08-02 David Malcolm <dmalcolm@redhat.com>
10000 * diagnostic-show-locus.c (diagnostic_show_locus): Use
10001 pp_take_prefix when saving the existing prefix.
10002 * diagnostic.c (diagnostic_append_note): Likewise.
10003 * langhooks.c (lhd_print_error_function): Likewise.
10004 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
10005 param's type. Free the existing prefix.
10006 (pp_take_prefix): New function.
10007 (pretty_printer::pretty_printer): Drop the prefix parameter.
10008 Rename the length parameter to match the comment.
10009 (pretty_printer::~pretty_printer): Free the prefix.
10010 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
10012 (struct pretty_printer): Drop the "const" from "prefix" field's
10013 type and clarify memory management.
10014 (pp_set_prefix): Drop the "const" from the 2nd param.
10015 (pp_take_prefix): New decl.
10017 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
10019 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
10020 for word_mode_ok here instead of passing as argument.
10021 (expand_block_compare): Change select_block_compare_mode() call.
10022 (expand_strncmp_gpr_sequence): New function.
10023 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
10025 2018-08-02 Jeff Law <law@redhat.com>
10028 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10029 Define to speculation_safe_value_not_needed.
10032 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10033 Define to speculation_safe_value_not_needed.
10035 2018-08-02 Tom de Vries <tdevries@suse.de>
10038 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
10039 function. Return UI_TARGET unconditionally.
10040 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
10041 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
10043 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10045 * genemit.c (print_overload_test): Fix typo.
10047 2018-08-02 Richard Biener <rguenther@suse.de>
10049 PR tree-optimization/86816
10050 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
10051 which checks for value availability before querying it.
10052 (gvn_uses_equal): Use it.
10053 (same_succ_hash): Likewise.
10054 (gimple_equal_p): Likewise.
10056 2018-08-02 Nick Clifton <nickc@redhat.com>
10059 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10060 Define to speculation_safe_value_not_needed.
10063 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10064 Define to speculation_safe_value_not_needed.
10067 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10068 Define to speculation_safe_value_not_needed.
10071 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10072 Define to speculation_safe_value_not_needed.
10075 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10076 Define to speculation_safe_value_not_needed.
10079 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10080 Define to speculation_safe_value_not_needed.
10083 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10084 Define to speculation_safe_value_not_needed.
10087 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10088 Define to speculation_safe_value_not_needed.
10091 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10092 speculation_safe_value_not_needed.
10095 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
10096 to speculation_safe_value_not_needed.
10098 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10100 * doc/md.texi: Expand the documentation of instruction names
10101 to mention port-local uses. Document '@' in pattern names.
10102 * read-md.h (overloaded_instance, overloaded_name): New structs.
10103 (mapping): Declare.
10104 (md_reader::handle_overloaded_name): New member function.
10105 (md_reader::get_overloads): Likewise.
10106 (md_reader::m_first_overload): New member variable.
10107 (md_reader::m_next_overload_ptr): Likewise.
10108 (md_reader::m_overloads_htab): Likewise.
10109 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
10110 m_next_overload_ptr and m_overloads_htab.
10111 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
10112 (get_mode_token, get_code_token, get_int_token): New functions.
10113 (map_attr_string): Add an optional argument that passes back
10114 the associated iterator.
10115 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
10116 (md_reader::handle_overloaded_name, add_overload_instance): New
10118 (apply_iterators): Handle '@' names. Report an error if '@'
10119 is used without iterators.
10120 (initialize_iterators): Initialize the new iterator_group fields.
10121 * genopinit.c (handle_overloaded_code_for)
10122 (handle_overloaded_gen): New functions.
10123 (main): Use them to print declarations of maybe_code_for_* and
10124 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
10125 * genemit.c (print_overload_arguments, print_overload_test)
10126 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
10127 (main): Use it to print definitions of maybe_code_for_* and
10128 maybe_gen_* functions.
10129 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
10130 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
10131 instead of explicit mode checks.
10132 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
10133 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
10134 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
10135 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
10136 (aarch64_expand_compare_and_swap): Likewise
10137 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
10138 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
10139 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
10140 (aarch64_constant_pool_reload_icode): Delete.
10141 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
10142 instead of aarch64_constant_pool_reload_icode. Use
10143 code_for_aarch64_reload_mov instead of explicit mode checks.
10144 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
10145 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
10146 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
10147 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
10148 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
10149 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
10150 (aarch64_atomic_load_op_code): Delete.
10151 (aarch64_emit_atomic_load_op): Likewise.
10152 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
10153 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
10154 instead of aarch64_emit_atomic_load_op.
10155 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
10156 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
10157 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
10158 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
10159 character before the pattern name.
10160 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
10161 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
10162 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
10163 (aarch64_frecps<mode>): Likewise.
10164 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
10165 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
10166 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
10167 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
10168 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
10170 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10172 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
10173 Allow HFmode constants if TARGET_FP_F16INST.
10175 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
10178 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
10179 No longer check last store for clobber of address register.
10181 2018-08-02 Martin Liska <mliska@suse.cz>
10183 PR gcov-profile/86817
10184 * gcov.c (process_all_functions): New function.
10186 (process_file): Move functions processing to
10187 process_all_functions.
10189 2018-08-02 David Malcolm <dmalcolm@redhat.com>
10191 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
10192 "const" to the "gimple *" and "rtx_insn *" parameters.
10193 * dumpfile.h (dump_user_location_t::dump_user_location_t):
10195 (dump_location_t::dump_location_t): Likewise.
10197 2018-08-01 Martin Sebor <msebor@redhat.com>
10199 PR tree-optimization/86650
10200 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
10201 rather than a "gcall *". Directly pass the data of interest
10202 to percent_K_format, rather than building a temporary CALL_EXPR
10204 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
10205 (gimple_fold_builtin_strncat): Adjust.
10206 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
10207 gcall* argument with gimple*.
10208 * gimple-ssa-warn-restrict.c (check_call): Same.
10209 (wrestrict_dom_walker::before_dom_children): Same.
10210 (builtin_access::builtin_access): Same.
10211 (check_bounds_or_overlap): Same
10212 (maybe_diag_overlap): Same.
10213 (maybe_diag_offset_bounds): Same.
10214 * tree-diagnostic.c (default_tree_printer): Move usage of
10215 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
10217 * tree-pretty-print.c (percent_K_format): Add argument.
10218 * tree-pretty-print.h: Add argument.
10219 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
10220 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
10221 (maybe_diag_stxncpy_trunc): Same.
10222 (handle_builtin_stxncpy): Same.
10223 (handle_builtin_strcat): Same.
10225 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10227 * match.pd: Optimise pointer range checks.
10229 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10231 PR tree-optimization/86758
10232 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
10233 to remove pattern statements.
10235 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10237 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
10238 result of dfs_enumerate_from when constructing stmt_vec_infos,
10239 instead of additionally calling get_loop_body.
10241 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10243 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
10245 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
10246 When creating an iv, assert that the step is not known to be zero.
10247 (vect_setup_realignment): Update call accordingly.
10248 * tree-vect-stmts.c (vectorizable_store): Likewise.
10249 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
10251 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10253 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
10254 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
10255 (vectorizable_reduction): Likewise.
10256 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10257 (vect_detect_hybrid_slp_stmts): Likewise.
10258 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
10260 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
10262 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
10263 (wide_int_set_zero_nonzero_bits): ...this.
10264 (zero_nonzero_bits_from_vr): Rename to...
10265 (vrp_set_zero_nonzero_bits): ...this.
10266 (extract_range_from_multiplicative_op_1): Abstract wide int
10268 (wide_int_range_multiplicative_op): ...here.
10269 (extract_range_from_binary_expr_1): Extract wide int binary
10270 operations into their own functions.
10271 (wide_int_range_lshift): New.
10272 (wide_int_range_can_optimize_bit_op): New.
10273 (wide_int_range_shift_undefined_p): New.
10274 (wide_int_range_bit_xor): New.
10275 (wide_int_range_bit_ior): New.
10276 (wide_int_range_bit_and): New.
10277 (wide_int_range_trunc_mod): New.
10278 (extract_range_into_wide_ints): New.
10279 (vrp_shift_undefined_p): New.
10280 (extract_range_from_multiplicative_op): New.
10281 (vrp_can_optimize_bit_op): New.
10282 * tree-vrp.h (value_range::dump): New.
10283 (wide_int_range_multiplicative_op): New.
10284 (wide_int_range_lshift):New.
10285 (wide_int_range_shift_undefined_p): New.
10286 (wide_int_range_bit_xor): New.
10287 (wide_int_range_bit_ior): New.
10288 (wide_int_range_bit_and): New.
10289 (wide_int_range_trunc_mod): New.
10290 (zero_nonzero_bits_from_bounds): Rename to...
10291 (wide_int_set_zero_nonzero_bits): ...this.
10292 (zero_nonzero_bits_from_vr): Rename to...
10293 (vrp_set_zero_nonzero_bits): ...this.
10294 (range_easy_mask_min_max): Rename to...
10295 (wide_int_range_can_optimize_bit_op): this.
10296 * vr-values.c (simplify_bit_ops_using_ranges): Rename
10297 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
10299 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10301 * tree-vectorizer.h (vect_orig_stmt): New function.
10302 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
10303 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
10304 (vect_create_epilog_for_reduction): Likewise.
10305 (vectorizable_live_operation): Likewise.
10306 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
10307 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
10308 * tree-vect-stmts.c (vectorizable_call): Likewise.
10309 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
10311 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10313 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
10315 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
10316 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
10317 (vect_transform_loop): Likewise.
10318 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10320 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10322 * tree-vectorizer.h (vect_schedule_slp): Return void.
10323 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10324 (vect_schedule_slp): Likewise.
10326 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10328 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
10330 (vect_transform_loop): Update calls accordingly. Schedule SLP
10331 instances before the main loop, if any exist.
10333 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10335 PR tree-optimization/86749
10336 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
10337 If the lhs is used in a COND_EXPR, check that it is being used
10338 as the "then" or "else" value.
10340 2018-08-01 Tom de Vries <tdevries@suse.de>
10343 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10344 speculation_safe_value_not_needed.
10346 2018-08-01 Richard Biener <rguenther@suse.de>
10348 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
10349 as base and offset.
10351 2018-08-01 Martin Liska <mliska@suse.cz>
10353 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
10354 format how successful transformation is dumped.
10355 (gimple_mod_pow2_value_transform): Likewise.
10356 (gimple_mod_subtract_transform): Likewise.
10357 (gimple_stringops_transform): Likewise.
10359 2018-08-01 Martin Liska <mliska@suse.cz>
10361 PR value-prof/35543
10362 * value-prof.c (interesting_stringop_to_profile_p):
10363 Simplify the code and add BUILT_IN_MEMMOVE.
10364 (gimple_stringops_transform): Likewise.
10366 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
10368 * config/aarch64/aarch64-simd.md
10369 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
10370 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
10371 use GPI iterator instead of SI mode.
10373 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10375 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
10376 rs6000_speculation_barrier.
10377 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
10378 new barrier pattern name.
10380 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10382 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
10383 (speculation_barrier): New insn.
10385 2018-08-01 Richard Biener <rguenther@suse.de>
10388 * graphite.h: Include isl/id.h and isl/space.h to allow build
10391 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
10394 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
10395 mode for COFF targets.
10396 * defaults.h (TARGET_COFF): Define.
10397 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
10398 TARGET_COFF): Define.
10399 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
10400 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
10401 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
10403 2018-07-31 Alexandre Oliva <oliva@adacore.com>
10405 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
10406 * gimple-streamer-out.c (output_bb): Save it.
10407 * lto-streamer-in.c (input_struct_function_base): Restore
10408 instance discriminator if available. Create map on demand.
10409 * lto-streamer-out.c (output_struct_function_base): Save it if
10411 * final.c (decl_to_instance_map): Document LTO strategy.
10413 2018-07-31 Alexandre Oliva <oliva@adacore.com>
10414 Olivier Hainque <hainque@adacore.com>
10416 * debug.h (decl_to_instance_map_t): New type.
10417 (decl_to_instance_map): Declare.
10418 (maybe_create_decl_to_instance_map): New inline function.
10419 * final.c (bb_discriminator, last_bb_discriminator): New statics,
10420 to track basic block discriminators.
10421 (final_start_function_1): Initialize them.
10422 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
10424 (decl_to_instance_map): New variable.
10425 (map_decl_to_instance, maybe_set_discriminator): New functions.
10426 (notice_source_line): Set discriminator.
10428 2018-07-31 Ian Lance Taylor <iant@golang.org>
10430 * targhooks.c (default_have_speculation_safe_value): Add
10433 2018-07-31 David Malcolm <dmalcolm@redhat.com>
10435 * dump-context.h: Include "pretty-print.h".
10436 (dump_context::refresh_dumps_are_enabled): New decl.
10437 (dump_context::emit_item): New decl.
10438 (class dump_context): Add fields "m_test_pp" and
10440 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
10441 (temp_dump_context::get_dumped_text): New decl.
10442 (class temp_dump_context): Add field "m_pp".
10443 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
10444 (dump_context::refresh_dumps_are_enabled): ...and add a test for
10446 (set_dump_file): Update for above change.
10447 (set_alt_dump_file): Likewise.
10448 (dump_loc): New overload, taking a pretty_printer *.
10449 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
10450 to any test pretty-printer.
10451 (make_item_for_dump_gimple_stmt): New function, adapted from
10452 optinfo::add_gimple_stmt.
10453 (dump_context::dump_gimple_stmt): Call it, and use the result,
10454 eliminating the direct usage of dump_file and alt_dump_file in
10455 favor of indirectly using them via emit_item.
10456 (make_item_for_dump_gimple_expr): New function, adapted from
10457 optinfo::add_gimple_expr.
10458 (dump_context::dump_gimple_expr): Call it, and use the result,
10459 eliminating the direct usage of dump_file and alt_dump_file in
10460 favor of indirectly using them via emit_item.
10461 (make_item_for_dump_generic_expr): New function, adapted from
10463 (dump_context::dump_generic_expr): Call it, and use the result,
10464 eliminating the direct usage of dump_file and alt_dump_file in
10465 favor of indirectly using them via emit_item.
10466 (make_item_for_dump_printf_va): New function, adapted from
10467 optinfo::add_printf_va.
10468 (make_item_for_dump_printf): New function.
10469 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
10470 and use the result, eliminating the direct usage of dump_file and
10471 alt_dump_file in favor of indirectly using them via emit_item.
10472 (make_item_for_dump_dec): New function.
10473 (dump_context::dump_dec): Call it, and use the result,
10474 eliminating the direct usage of dump_file and alt_dump_file in
10475 favor of indirectly using them via emit_item.
10476 (make_item_for_dump_symtab_node): New function, adapted from
10477 optinfo::add_symtab_node.
10478 (dump_context::dump_symtab_node): Call it, and use the result,
10479 eliminating the direct usage of dump_file and alt_dump_file in
10480 favor of indirectly using them via emit_item.
10481 (dump_context::begin_scope): Reimplement, avoiding direct usage
10482 of dump_file and alt_dump_file in favor of indirectly using them
10484 (dump_context::emit_item): New member function.
10485 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
10486 Set up test pretty-printer on the underlying context. Call
10487 refresh_dumps_are_enabled.
10488 (temp_dump_context::~temp_dump_context): Call
10489 refresh_dumps_are_enabled.
10490 (temp_dump_context::get_dumped_text): New member function.
10491 (selftest::verify_dumped_text): New function.
10492 (ASSERT_DUMPED_TEXT_EQ): New macro.
10493 (selftest::test_capture_of_dump_calls): Run all tests twice, with
10494 and then without optinfo enabled. Add uses of
10495 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
10496 * dumpfile.h: Update comment for the dump_* API.
10497 * optinfo-emit-json.cc
10498 (selftest::test_building_json_from_dump_calls): Update for new
10499 param for temp_dump_context ctor.
10500 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
10501 and "m_owned" field.
10502 (optinfo_item::~optinfo_item): Likewise.
10503 (optinfo::add_item): New member function.
10504 (optinfo::emit): Update comment.
10505 (optinfo::add_string): Delete.
10506 (optinfo::add_printf): Delete.
10507 (optinfo::add_printf_va): Delete.
10508 (optinfo::add_gimple_stmt): Delete.
10509 (optinfo::add_gimple_expr): Delete.
10510 (optinfo::add_tree): Delete.
10511 (optinfo::add_symtab_node): Delete.
10512 (optinfo::add_dec): Delete.
10513 * optinfo.h (class dump_context): New forward decl.
10514 (optinfo::add_item): New decl.
10515 (optinfo::add_string): Delete.
10516 (optinfo::add_printf): Delete.
10517 (optinfo::add_printf_va): Delete.
10518 (optinfo::add_gimple_stmt): Delete.
10519 (optinfo::add_gimple_expr): Delete.
10520 (optinfo::add_tree): Delete.
10521 (optinfo::add_symtab_node): Delete.
10522 (optinfo::add_dec): Delete.
10523 (optinfo::add_poly_int): Delete.
10524 (optinfo_item::optinfo_item): Remove "owned" param.
10525 (class optinfo_item): Remove field "m_owned".
10527 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10529 PR middle-end/86705
10530 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
10531 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
10532 requested variable alignment.
10533 (expand_one_ssa_partition): Likewise.
10534 (expand_one_var): Likewise.
10536 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10538 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
10539 to speculation_safe_value_not_needed.
10541 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10543 * targhooks.h (speculation_safe_value_not_needed): New prototype.
10544 * targhooks.c (speculation_safe_value_not_needed): New function.
10545 * target.def (have_speculation_safe_value): Update documentation.
10546 * doc/tm.texi: Regenerated.
10548 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10550 * config/aarch64/iterators.md (ALLI_TI): New iterator.
10551 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
10553 (despeculate_copy<ALLI:mode>_insn): New insn.
10554 (despeculate_copyti_insn): New insn.
10555 (despeculate_simple<ALLI:mode>): New insn
10556 (despeculate_simpleti): New insn.
10557 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
10559 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
10560 aarch64_speculation_safe_value.
10561 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
10563 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10565 * config/aarch64/aarch64-speculation.cc: New file.
10566 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
10567 before pass_reorder_blocks.
10568 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
10570 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
10571 X14 and X15 when tracking speculation.
10572 * config/aarch64/aarch64.md (register name constants): Add
10573 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
10574 (unspec): Add UNSPEC_SPECULATION_TRACKER.
10575 (speculation_barrier): New insn attribute.
10576 (cmp<mode>): Allow SP in comparisons.
10577 (speculation_tracker): New insn.
10578 (speculation_barrier): Add speculation_barrier attribute.
10579 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
10580 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
10581 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
10583 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10585 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
10586 aarch64_track_speculation is true.
10587 (tb<optab><mode>1): Likewise.
10588 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
10589 generate CB[N]Z when tracking speculation.
10590 (aarch64_split_compare_and_swap): Likewise.
10591 (aarch64_split_atomic_op): Likewise.
10593 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10595 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
10597 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10599 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
10600 (speculation_barrier): New insn.
10602 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10604 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
10605 * config/arm/arm.md (speculation_barrier): New expand.
10606 (speculation_barrier_insn): New pattern.
10608 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10610 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
10611 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
10612 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
10613 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
10615 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
10616 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
10617 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
10618 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
10619 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
10620 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
10621 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
10622 * builtins.c (expand_speculation_safe_value): New function.
10623 (expand_builtin): Call it.
10624 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
10625 * doc/extend.texi: Document __builtin_speculation_safe_value.
10626 * doc/md.texi: Document "speculation_barrier" pattern.
10627 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
10628 TARGET_HAVE_SPECULATION_SAFE_VALUE.
10629 * doc/tm.texi: Regenerated.
10630 * target.def (have_speculation_safe_value, speculation_safe_value): New
10632 * targhooks.c (default_have_speculation_safe_value): New function.
10633 (default_speculation_safe_value): New function.
10634 * targhooks.h (default_have_speculation_safe_value): Add prototype.
10635 (default_speculation_safe_value): Add prototype.
10637 2018-07-31 David Malcolm <dmalcolm@redhat.com>
10639 * dump-context.h (dump_context::dump_loc): New decl.
10640 * dumpfile.c (dump_context::dump_loc): New member function.
10641 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
10642 and dump_gimple_stmt.
10643 (dump_context::dump_gimple_expr_loc): Likewise, using
10645 (dump_context::dump_generic_expr_loc): Likewise, using
10647 (dump_context::dump_printf_loc_va): Likewise, using
10649 (dump_context::begin_scope): Explicitly using the global function
10650 "dump_loc", rather than the member function.
10652 2018-07-31 Martin Sebor <msebor@redhat.com>
10654 PR tree-optimization/86741
10655 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
10657 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
10659 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
10660 depend on whether prefetch instructions will be emitted or not.
10661 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
10662 will be emitted or not.
10663 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
10664 (TARGET_SETMEM_PFD): New macros.
10666 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10668 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
10669 (NULL_STMT_VEC_INFO): Delete.
10670 (stmt_vec_info::operator*): Likewise.
10671 (stmt_vec_info::operator gimple *): Likewise.
10672 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
10673 of NULL_STMT_VEC_INFO.
10674 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10675 (vect_reassociating_reduction_p): Likewise.
10676 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
10677 (vectorizable_store): Likewise.
10678 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
10679 (vec_info::free_stmt_vec_infos): Likewise.
10681 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10683 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
10684 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
10686 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10688 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
10689 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
10690 (vec_info::free_stmt_vec_info): New private member functions.
10691 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
10692 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
10693 * tree-parloops.c (gather_scalar_reductions): Remove calls to
10694 set_stmt_vec_info_vec and free_stmt_vec_infos.
10695 * tree-vect-loop.c (_loop_vec_info): Remove call to
10696 set_stmt_vec_info_vec.
10697 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
10698 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
10699 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
10700 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
10701 (vec_info::free_stmt_vec_info): ...these new functions. Remove
10702 assignments in {vec_info::,}new_stmt_vec_info that are redundant
10703 with the clearing in the xcalloc.
10704 (stmt_vec_info_vec): Delete.
10705 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
10706 (vectorize_loops): Likewise.
10707 (vec_info::~vec_info): Remove argument from call to
10708 free_stmt_vec_infos.
10709 (vec_info::add_stmt): Remove vinfo argument from call to
10712 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10714 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
10715 rather than a gimple stmt.
10716 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
10717 information for pattern statements when passed the original
10718 statement; instead wait to be passed the pattern statement itself.
10719 Don't call set_vinfo_for_stmt here.
10720 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
10721 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
10722 stmt_vec_infos here.
10723 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
10724 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
10725 stmt_vec_infos entry.
10727 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10729 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
10730 * tree-vectorizer.c (vec_info::replace_stmt): New function.
10731 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
10732 * tree-vect-stmts.c (vectorizable_call): Likewise.
10733 (vectorizable_simd_clone_call): Likewise.
10735 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10737 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
10738 * tree-vectorizer.c (vec_info::remove_stmt): New function.
10739 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
10740 * tree-vect-loop.c (vect_transform_loop): Likewise.
10741 * tree-vect-slp.c (vect_schedule_slp): Likewise.
10742 * tree-vect-stmts.c (vect_remove_stores): Likewise.
10744 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10746 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
10747 (vect_dr_stmt): Delete.
10748 * tree-vectorizer.c (vec_info::lookup_dr): New function.
10749 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
10751 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
10752 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
10753 (vect_verify_datarefs_alignment, vect_peeling_supportable)
10754 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
10755 (vect_analyze_data_refs): Likewise.
10756 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
10758 (vect_find_same_alignment_drs): Likewise.
10759 (vect_slp_analyze_node_dependences): Update calls accordingly.
10760 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
10761 instead of DR_VECT_AUX.
10762 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
10763 of a vector data references. Use vec_info::lookup_dr instead of
10765 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
10766 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
10767 instead of DR_VECT_AUX.
10769 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10771 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
10773 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
10775 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10776 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10777 (vect_gen_prolog_loop_niters): Likewise.
10779 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10781 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
10782 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
10783 (vect_known_alignment_in_bytes, vect_dr_behavior)
10784 (vect_get_scalar_dr_size): Take references as dr_vec_infos
10785 instead of data_references. Update calls to other routines for
10786 which the same change has been made.
10787 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
10788 dr_vec_infos instead of stmt_vec_infos.
10789 (vect_analyze_data_ref_dependence): Update call accordingly.
10790 (vect_slp_analyze_data_ref_dependence)
10791 (vect_record_base_alignments): Use DR_VECT_AUX.
10792 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
10793 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
10794 (vector_alignment_reachable_p, vect_get_data_access_cost)
10795 (vect_peeling_supportable, vect_analyze_group_access_1)
10796 (vect_analyze_group_access, vect_analyze_data_ref_access)
10797 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
10798 (vect_compile_time_alias, vect_small_gap_p)
10799 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
10800 (vect_supportable_dr_alignment): Take references as dr_vec_infos
10801 instead of data_references. Update calls to other routines for
10802 which the same change has been made.
10803 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
10804 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
10805 (vect_slp_analyze_and_verify_node_alignment)
10806 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
10807 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
10808 (vect_setup_realignment): Use dr_vec_infos. Update calls after
10810 (_vect_peel_info::dr): Replace with...
10811 (_vect_peel_info::dr_info): ...this new field.
10812 (vect_peeling_hash_get_most_frequent)
10813 (vect_peeling_hash_choose_best_peeling): Update accordingly.
10814 (vect_peeling_hash_get_lowest_cost):
10815 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
10816 routines for which the same change has been made.
10817 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
10819 * tree-vect-loop-manip.c (get_misalign_in_elems)
10820 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
10822 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10823 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
10824 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
10825 (get_group_load_store_type, get_negative_load_store_type)
10826 (vect_get_data_ptr_increment, vectorizable_store)
10827 (vectorizable_load): Likewise.
10828 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
10829 Update calls to other routines for which the same change has been made.
10831 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10833 * tree-vectorizer.h (vec_info::move_dr): New member function.
10834 (dataref_aux): Rename to...
10835 (dr_vec_info): ...this and add "dr" and "stmt" fields.
10836 (_stmt_vec_info::dr_aux): Update accordingly.
10837 (_stmt_vec_info::data_ref_info): Delete.
10838 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
10839 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
10840 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
10841 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
10843 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
10844 (STMT_VINFO_DR_INFO): New macro.
10845 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
10846 (set_dr_misalignment): Update after rename of dataref_aux.
10847 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
10848 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
10849 initialization of STMT_VINFO_DATA_REF.
10850 * tree-vectorizer.c (vec_info::move_dr): New function.
10851 * tree-vect-patterns.c (vect_recog_bool_pattern)
10852 (vect_recog_mask_conversion_pattern)
10853 (vect_recog_gather_scatter_pattern): Use it.
10854 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
10855 the "dr" and "stmt" fields of dr_vec_info instead of
10856 STMT_VINFO_DATA_REF.
10858 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10860 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
10861 (is_pattern_stmt_p): Use it.
10862 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
10863 on pattern statements.
10865 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10867 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
10868 original stmt as a stmt_vec_info rather than a gimple stmt.
10869 (vect_pattern_recog_1): Take the statement directly as a
10870 stmt_vec_info, rather than via a gimple_stmt_iterator.
10871 Update call to vect_mark_pattern_stmts.
10872 (vect_pattern_recog): Update calls accordingly.
10874 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10876 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
10877 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
10878 a vect_def_type for the first argument.
10879 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
10880 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
10881 operand if it isn't defined by a vectorized statement.
10882 (vect_build_gather_load_calls): Remove the mask_dt argument and
10883 update calls to vect_get_vec_def_for_stmt_copy.
10884 (vectorizable_bswap): Likewise the dt argument.
10885 (vectorizable_call): Update calls to vectorizable_bswap and
10886 vect_get_vec_def_for_stmt_copy.
10887 (vectorizable_simd_clone_call, vectorizable_assignment)
10888 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
10889 (vectorizable_comparison): Update calls to
10890 vect_get_vec_def_for_stmt_copy.
10891 (vectorizable_store): Likewise. Remove now-unnecessary calls to
10892 vect_is_simple_use.
10893 (vect_get_loop_based_defs): Remove dt argument and update call
10894 to vect_get_vec_def_for_stmt_copy.
10895 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
10896 and vect_get_vec_def_for_stmt_copy.
10897 (vectorizable_load): Update calls to vect_build_gather_load_calls
10898 and vect_get_vec_def_for_stmt_copy.
10899 * tree-vect-loop.c (vect_create_epilog_for_reduction)
10900 (vectorizable_reduction, vectorizable_live_operation): Update calls
10901 to vect_get_vec_def_for_stmt_copy.
10903 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10905 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
10906 and gimple stmt arguments with a stmt_vec_info.
10907 (vect_record_base_alignments): Update calls accordingly.
10908 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
10909 and gimple stmt arguments with a stmt_vec_info.
10910 (vect_build_slp_tree_1): Remove vinfo argument and update call
10911 to vect_record_max_nunits.
10912 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
10913 and vect_record_max_nunits.
10915 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10917 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
10918 file and take a stmt_vec_info instead of a gimple stmt.
10919 (supportable_widening_operation, vect_finish_replace_stmt)
10920 (vect_finish_stmt_generation, vect_get_store_rhs)
10921 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
10922 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
10923 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
10924 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
10925 (vect_create_data_ref_ptr, bump_vector_ptr)
10926 (vect_permute_store_chain, vect_setup_realignment)
10927 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
10928 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
10929 (vectorizable_reduction, vectorizable_induction)
10930 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
10931 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
10932 than gimple stmts as arguments.
10933 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
10934 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
10935 (can_group_stmts_p, vect_check_gather_scatter)
10936 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
10937 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
10938 (vect_permute_load_chain, vect_shift_permute_load_chain)
10939 (vect_transform_grouped_load)
10940 (vect_record_grouped_load_vectors): Likewise.
10941 * tree-vect-loop.c (vect_fixup_reduc_chain)
10942 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
10943 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
10944 (vectorizable_reduction, vectorizable_induction)
10945 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
10946 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
10947 (vect_get_load_store_mask): Likewise.
10948 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
10949 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
10950 * tree-vect-stmts.c (vect_mark_relevant)
10951 (is_simple_and_all_uses_invariant)
10952 (exist_non_indexing_operands_for_use_p, process_use)
10953 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
10954 (vect_get_vec_def_for_operand, vect_get_vec_defs)
10955 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
10956 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
10957 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
10958 (get_negative_load_store_type, get_load_store_type)
10959 (vect_check_load_store_mask, vect_check_store_rhs)
10960 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
10961 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10962 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10963 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10964 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
10965 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
10966 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
10967 (supportable_widening_operation): Likewise.
10969 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10971 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
10972 a stmt_vec_info instead of a gcall.
10973 (vect_check_gather_scatter): Update call accordingly.
10974 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
10976 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
10977 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
10978 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
10979 instead of a gimple stmt.
10980 (vect_transform_loop): Update calls accordingly.
10981 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
10982 stmt_vec_infos instead of gimple stmts.
10983 (vect_analyze_slp_instance): Update use accordingly.
10984 * tree-vect-stmts.c (read_vector_array, write_vector_array)
10985 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
10986 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
10987 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
10988 (vect_gen_widened_results_half, vect_get_loop_based_defs)
10989 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
10990 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
10991 down to subroutines.
10993 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10995 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
10996 of the worklist from a vector of gimple stmts to a vector of
10998 * tree-vect-stmts.c (vect_mark_relevant, process_use)
10999 (vect_mark_stmts_to_be_vectorized): Likewise
11001 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11003 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
11004 statement before passing it to vect_analyze_stmt.
11005 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
11006 the chain of phi vector definitions. Track the exit phi via its
11008 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
11009 STMT_VINFO_REDUC_DEF.
11010 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
11011 stmt_vec_infos to handle the statement chains.
11012 (vect_get_slp_defs): Record the first statement in the node
11013 using a stmt_vec_info.
11014 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
11015 statements here and pass their stmt_vec_info down to subroutines.
11016 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
11017 down to vect_finish_stmt_generation.
11018 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
11019 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
11020 stmt_vec_infos to subroutines.
11021 (vect_remove_stores): Use stmt_vec_infos to handle the statement
11024 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11026 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
11027 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
11028 (vect_permute_store_chain, vect_setup_realignment)
11029 (vect_permute_load_chain, vect_shift_permute_load_chain)
11030 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
11031 stmts internally, and when passing values to other vectorizer routines.
11032 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
11033 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
11034 (vect_analyze_loop_operations, get_initial_def_for_reduction)
11035 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
11036 (vectorizable_reduction, vectorizable_induction)
11037 (vectorizable_live_operation, vect_transform_loop_stmt)
11038 (vect_transform_loop): Likewise.
11039 * tree-vect-patterns.c (vect_reassociating_reduction_p)
11040 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
11041 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
11042 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11043 (vect_slp_analyze_node_operations_1): Likewise.
11044 * tree-vect-stmts.c (vect_mark_relevant, process_use)
11045 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
11046 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
11047 (vect_finish_stmt_generation_1, get_group_load_store_type)
11048 (get_load_store_type, vect_build_gather_load_calls)
11049 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11050 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11051 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11052 (vectorizable_store, vectorizable_load, vectorizable_condition)
11053 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
11054 (supportable_widening_operation): Likewise.
11055 (vect_get_vector_types_for_stmt): Likewise.
11056 * tree-vectorizer.h (vect_dr_behavior): Likewise.
11058 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11060 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11061 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
11062 (vect_permute_store_chain, vect_permute_load_chain)
11063 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
11064 repeated stmt_vec_info lookups.
11065 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
11066 (vect_update_ivs_after_vectorizer): Likewise.
11067 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
11068 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
11069 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
11070 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11071 (vect_bb_slp_scalar_cost): Likewise.
11072 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
11074 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11076 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
11077 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
11078 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
11079 to get gassigns and gcalls, rather than operating on generc gimple
11081 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
11082 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
11083 (vectorizable_load, vect_analyze_stmt): Likewise.
11084 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
11086 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11088 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
11089 return stmt_vec_infos rather than gimple stmts. Do not accept
11091 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
11093 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
11094 Update use of get_later_stmt.
11095 (vect_get_constant_vectors): Update call accordingly.
11096 (vect_schedule_slp_instance): Likewise
11097 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
11098 (vect_slp_analyze_instance_dependence): Likewise.
11099 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
11101 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11103 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
11104 (stmt_info_for_cost::stmt_info): ...this new field.
11105 (add_stmt_costs): Update accordingly.
11106 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11107 (vect_get_known_peeling_cost): Likewise.
11108 (vect_estimate_min_profitable_iters): Likewise.
11109 * tree-vect-stmts.c (record_stmt_cost): Likewise.
11111 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11113 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
11114 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11115 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
11117 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
11119 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11121 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
11122 a gimple stmt to a stmt_vec_info.
11123 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
11125 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11127 * tree-vectorizer.h (vec_info::grouped_stores): Change from
11128 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11129 (_loop_vec_info::reduction_chains): Likewise.
11130 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
11132 * tree-vect-slp.c (vect_analyze_slp): Likewise.
11134 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11136 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
11137 a gimple stmt to a stmt_vec_info.
11138 (_stmt_vec_info::next_element): Likewise.
11139 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
11140 (vect_slp_analyze_and_verify_node_alignment)
11141 (vect_analyze_group_access_1, vect_analyze_group_access)
11142 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
11143 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
11144 (vect_supportable_dr_alignment): Update accordingly.
11145 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
11146 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
11147 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
11148 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11149 * tree-vect-slp.c (vect_build_slp_tree_1)
11150 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
11151 (vect_split_slp_store_group, vect_analyze_slp_instance)
11152 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
11153 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
11154 (get_group_load_store_type, get_load_store_type)
11155 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
11156 (vect_transform_stmt, vect_remove_stores): Likewise.
11158 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11160 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
11161 than a gimple stmt.
11162 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11163 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
11164 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
11165 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
11166 (vector_alignment_reachable_p, vect_get_data_access_cost)
11167 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
11168 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
11169 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
11170 (vect_analyze_group_access_1, vect_analyze_group_access)
11171 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
11172 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
11173 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
11174 result of vect_dr_stmt and use the stmt_vec_info instead of
11175 the associated gimple stmt.
11176 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11177 (vect_gen_prolog_loop_niters): Likewise.
11178 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11180 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11182 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
11183 to a vec<stmt_vec_info>.
11184 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
11185 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
11186 vec<stmt_vec_info>.
11187 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
11188 to a vec<stmt_vec_info>.
11189 (bst_traits::value_type, bst_traits::value_type): Likewise.
11190 (bst_traits::hash): Update accordingly.
11191 (vect_get_and_check_slp_defs): Change the stmts parameter from
11192 a vec<gimple *> to a vec<stmt_vec_info>.
11193 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
11194 (vect_build_slp_tree): Likewise.
11195 (vect_build_slp_tree_2): Likewise. Update uses of
11196 SLP_TREE_SCALAR_STMTS.
11197 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
11198 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
11199 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
11200 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
11201 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
11202 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
11203 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
11204 (vect_get_constant_vectors, vect_get_slp_defs)
11205 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
11206 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
11207 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
11208 instead of gimple stmts.
11209 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
11210 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
11211 (vect_slp_analyze_instance_dependence): Update uses of
11212 SLP_TREE_SCALAR_STMTS.
11213 (vect_slp_analyze_and_verify_node_alignment): Likewise.
11214 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
11215 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
11216 (get_initial_defs_for_reduction): Likewise.
11217 (vect_create_epilog_for_reduction): Likewise.
11218 (vectorize_fold_left_reduction): Likewise.
11219 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
11220 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
11221 (can_vectorize_live_stmts): Likewise.
11223 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11225 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
11226 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11227 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
11229 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
11231 (gather_scalar_reductions): Update after above interface changes.
11232 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
11233 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
11235 (vect_force_simple_reduction): Likewise.
11236 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
11237 LOOP_VINFO_REDUCTIONS.
11238 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11240 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11242 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
11243 a gimple stmt to a stmt_vec_info.
11244 * tree-vect-loop.c (vect_active_double_reduction_p)
11245 (vect_force_simple_reduction, vectorizable_reduction): Update
11248 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11250 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
11251 vec<gimple *> to a vec<stmt_vec_info>.
11252 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
11253 the reduction_phis argument from a vec<gimple *> to a
11254 vec<stmt_vec_info>.
11255 (vectorizable_reduction): Likewise the phis local variable that
11256 is passed to vect_create_epilog_for_reduction. Update for new type
11257 of SLP_TREE_VEC_STMTS.
11258 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
11259 (vectorizable_live_operation): Likewise.
11260 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
11261 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
11263 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11265 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
11266 a gimple stmt to a stmt_vec_info.
11267 (vectorizable_condition, vectorizable_live_operation)
11268 (vectorizable_reduction, vectorizable_induction): Pass back the
11269 vectorized statement as a stmt_vec_info.
11270 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
11271 use of STMT_VINFO_VEC_STMT.
11272 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
11273 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
11274 as stmt_vec_infos rather than gimple stmts.
11275 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
11276 to a stmt_vec_info.
11277 (vectorizable_live_operation): Likewise.
11278 (vectorizable_reduction, vectorizable_induction): Likewise,
11279 updating use of STMT_VINFO_VEC_STMT.
11280 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
11281 of STMT_VINFO_VEC_STMT.
11282 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
11283 (vectorizable_simd_clone_call, vectorizable_conversion)
11284 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11285 (vectorizable_store, vectorizable_load, vectorizable_condition)
11286 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
11287 from a gimple stmt to a stmt_vec_info.
11288 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
11289 pointer to a stmt_vec_info to the vectorizable_* routines.
11291 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11293 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
11294 a gimple stmt to a stmt_vec_info.
11295 (is_pattern_stmt_p): Update accordingly.
11296 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
11297 (vect_record_grouped_load_vectors): Likewise.
11298 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
11299 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
11300 (vect_model_reduction_cost): Likewise.
11301 (vect_create_epilog_for_reduction): Likewise.
11302 (vectorizable_reduction, vectorizable_induction): Likewise.
11303 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11304 Return the stmt_vec_info for the pattern statement.
11305 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
11306 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
11307 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
11308 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
11309 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
11310 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
11311 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
11312 (free_stmt_vec_info, vect_is_simple_use): Likewise.
11314 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11316 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
11317 (vect_finish_stmt_generation): Likewise.
11318 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
11319 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
11320 (vect_build_gather_load_calls): Use the return value of the above
11321 functions instead of a separate call to vinfo_for_stmt. Use narrow
11322 scopes for the input gimple stmt and wider scopes for the associated
11323 stmt_vec_info. Use vec_info::lookup_def when setting these
11324 stmt_vec_infos from an SSA_NAME definition.
11325 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11326 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11327 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11328 (vectorizable_store, vectorizable_load, vectorizable_condition)
11329 (vectorizable_comparison): Likewise.
11330 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
11331 (vectorizable_reduction): Likewise.
11333 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11335 * tree-vectorizer.h (vect_is_simple_use): Add an optional
11336 stmt_vec_info * parameter before the optional gimple **.
11337 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11338 (process_use, vect_get_vec_def_for_operand_1): Update callers.
11339 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
11340 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11341 (vectorizable_live_operation): Likewise.
11342 * tree-vect-patterns.c (type_conversion_p): Likewise.
11343 (vect_look_through_possible_promotion): Likewise.
11344 (vect_recog_rotate_pattern): Likewise.
11345 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11347 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11349 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
11350 a typedef to a wrapper class.
11351 (NULL_STMT_VEC_INFO): New macro.
11352 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
11353 (stmt_vec_info::operator*): New function.
11354 (stmt_vec_info::operator gimple *): Likewise.
11355 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
11356 (add_stmt_costs): Likewise.
11357 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
11358 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11359 (vect_get_known_peeling_cost): Likewise.
11360 (vect_estimate_min_profitable_iters): Likewise.
11361 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11362 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
11363 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
11364 (vectorizable_store, free_stmt_vec_infos): Likewise.
11365 (new_stmt_vec_info): Change return type of xcalloc to
11368 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11370 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
11371 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
11372 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
11373 a single_imm_use-based sequence.
11374 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
11376 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11378 * tree-vectorizer.h (vec_info::lookup_def): Declare.
11379 * tree-vectorizer.c (vec_info::lookup_def): New function.
11380 * tree-vect-patterns.c (vect_get_internal_def): Use it.
11381 (vect_widened_op_tree): Likewise.
11382 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11383 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
11384 (vectorizable_reduction): Likewise.
11385 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
11387 (vect_is_slp_reduction): Update calls accordingly. Use
11388 vec_info::lookup_def.
11389 (vect_is_simple_reduction): Likewise
11390 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
11392 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11394 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
11395 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
11396 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
11398 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
11399 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
11400 (vect_update_vf_for_slp, vect_analyze_loop_operations)
11401 (vect_is_slp_reduction, vectorizable_induction)
11402 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
11403 * tree-vect-patterns.c (vect_init_pattern_stmt):
11404 (vect_determine_min_output_precision_1, vect_determine_precisions)
11405 (vect_pattern_recog): Likewise.
11406 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
11407 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
11408 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
11409 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
11410 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
11411 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
11412 info field from a loop to a loop_vec_info.
11414 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11416 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
11417 (vec_info::add_stmt): Declare.
11418 * tree-vectorizer.c (vec_info::add_stmt): New function.
11419 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
11420 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
11421 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
11422 (vectorizable_induction): Likewise.
11423 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
11424 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
11425 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
11426 (vectorizable_load): Likewise.
11427 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11428 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
11429 (vect_recog_gather_scatter_pattern): Likewise.
11430 (append_pattern_def_seq): Likewise. Remove a check that is
11431 performed by add_stmt itself.
11433 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11435 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
11436 which make_ssa_name was called with new_stmt before new_stmt
11439 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11441 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
11443 (vect_is_slp_reduction): ...here...
11444 (vect_is_simple_reduction): ...and here. Remove repetition of tests
11445 that are already known to be false.
11447 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11449 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
11450 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
11451 STMT_VINFO_NUM_SLP_USES when it's true.
11452 (vect_free_slp_instance): Add a final_p parameter and pass it to
11453 vect_free_slp_tree.
11454 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
11455 (vect_analyze_slp_instance): Likewise.
11456 (vect_slp_analyze_operations): Likewise.
11457 (vect_slp_analyze_bb_1): Likewise.
11458 * tree-vectorizer.c (vec_info): Likewise.
11459 * tree-vect-loop.c (vect_transform_loop): Likewise.
11461 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11463 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11464 function is not called for second and subsequent members of
11467 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11469 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
11470 cases for nested loops from here to ...
11471 (vect_create_epilog_for_reduction): ...here. Only call
11472 vect_is_simple_use for inner-loop reductions.
11474 2018-07-31 Martin Liska <mliska@suse.cz>
11476 PR gcov-profile/85338
11477 PR gcov-profile/85350
11478 PR gcov-profile/85372
11479 * profile.c (struct location_triplet): New.
11480 (struct location_triplet_hash): Likewise.
11481 (output_location): Do not output a BB that
11482 is already recorded for a line.
11483 (branch_prob): Use streamed_locations.
11485 2018-07-31 Martin Liska <mliska@suse.cz>
11487 PR gcov-profile/85370
11488 * coverage.c (coverage_begin_function): Do not mark target
11489 clones as artificial functions.
11491 2018-07-31 Martin Liska <mliska@suse.cz>
11493 PR gcov-profile/83813
11494 PR gcov-profile/84758
11495 PR gcov-profile/85217
11496 PR gcov-profile/85332
11497 * profile.c (branch_prob): Do not record GOTO expressions
11498 for GIMPLE statements which locations are already streamed.
11500 2018-07-31 Olivier Hainque <hainque@adacore.com>
11502 * gcc.c (handle_spec_function): Accept a soft_matched_part
11503 argument, as do_spec_1. Pass it down to ...
11504 (eval_spec_function): Accept a soft_matched_part argument,
11505 and pass it down to ...
11506 (do_spec_2): Accept a soft_matched_part argument, and pass
11507 it down to do_spec_1.
11508 (do_spec_1): Pass soft_matched_part to handle_spec_function.
11509 (handle_braces): Update call to handle_spec_function.
11510 (driver::set_up_specs): Update calls to do_spec_2.
11511 (compare_debug_dump_opt_spec_function): Likewise.
11512 (compare_debug_self_opt_spec_function): Likewise.
11514 2018-07-31 Olivier Hainque <hainque@adacore.com>
11516 * common.opt (nolibc): New option.
11517 * doc/invoke.texi (Link Options): Document it.
11518 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
11519 * config/alpha/linux.h: Likewise.
11520 * config/arc/elf.h: Likewise.
11521 * config/arm/uclinux-elf.h: Likewise.
11522 * config/arm/unknown-elf.h: Likewise.
11523 * config/avr/avrlibc.h: Likewise.
11524 * config/bfin/bfin.h: Likewise.
11525 * config/bfin/linux.h: Likewise.
11526 * config/bfin/uclinux.h: Likewise.
11527 * config/darwin.h: Likewise.
11528 * config/darwin10.h: Likewise.
11529 * config/darwin12.h: Likewise.
11530 * config/gnu-user.h: Likewise.
11531 * config/lm32/uclinux-elf.h: Likewise.
11532 * config/pa/pa-hpux11.h: Likewise.
11533 * config/pa/pa64-hpux.h: Likewise.
11534 * config/sparc/sparc.h: Likewise.
11536 2018-07-31 Olivier Hainque <hainque@adacore.com>
11538 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
11539 undefined variables.
11541 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
11544 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
11545 instead of GEN_INT.
11547 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11549 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
11550 terminated string literal.
11552 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
11554 PR rtl-optimization/85160
11555 * combine.c (is_just_move): New function.
11556 (try_combine): Allow combining two instructions into two if neither of
11557 the original instructions was a move.
11559 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
11562 * doc/extend.texi (Global Register Variables): Discourage use of type
11564 (Local Register Variables): Likewise.
11566 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
11568 PR tree-optimization/86506
11569 * hwint.h (ceil_log2): Resync with hwint.c implementation.
11571 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
11574 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
11575 hard_regno, make sure no insn between `from` and `to` clobbers it.
11577 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
11578 Tom de Vries <tdevries@suse.de>
11580 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
11581 (PTX_DEFAULT_RUNTIME_DIM): ... this.
11582 (nvptx_goacc_validate_dims): Set default worker and gang dims to
11583 PTX_DEFAULT_RUNTIME_DIM.
11584 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
11586 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
11588 * config/pa/pa.c (pa_output_addr_vec): Align address table.
11589 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
11590 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
11592 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
11594 * config/rs6000/constraints.md (wG constraint): Delete, no longer
11596 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
11597 predicate to reflect toc fusion has been deleted.
11598 (toc_fusion_mem_raw): Delete, no longer used.
11599 (toc_fusion_mem_wrapped): Likewise.
11600 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
11602 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
11603 Delete, no longer used.
11604 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
11605 meant to be used for toc fusion.
11606 (rs6000_debug_print_mode): Delete toc fusion debugging.
11607 (rs6000_debug_reg_global): Likewise.
11608 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
11609 fusion and secondary reload support that were never used.
11610 (rs6000_option_override_internal): Delete TOC fusion, that was only
11611 partially defined, and it did not work unless you also used the
11613 (rs6000_legitimate_address_p): Delete TOC fusion support.
11614 (rs6000_opt_masks): Likewise.
11615 (fusion_wrap_memory_address): Delete function, no longer used.
11616 (fusion_split_address); Delete TOC fusion support.
11617 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
11618 longer used with toc fusion being deleted.
11619 (TARGET_TOC_FUSION_FP): Likewise.
11620 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
11622 (toc fusion spliter): Delete TOC fusion support.
11623 (toc_fusionload_<mode>): Likewise.
11624 (toc_fusionload_di): Likewise.
11625 (fusion_gpr_load_<mode>): Delete generator function, this insn no
11626 longer needs to be named. Rename predicate to delete TOC fusion.
11627 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
11628 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
11629 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
11630 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
11631 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
11633 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11635 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
11636 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
11637 __int128 in built-in function prototypes.
11638 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
11639 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
11641 2018-07-27 Martin Sebor <msebor@redhat.com>
11643 PR tree-optimization/86696
11644 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
11645 types, including enums.
11646 (handle_char_store): Be prepared for the above function to fail.
11648 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
11650 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
11651 when optimization level is lower than 2 or optimize for size.
11653 2018-07-26 Martin Sebor <msebor@redhat.com>
11655 PR tree-optimization/86043
11656 PR tree-optimization/86042
11657 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
11658 (get_string_cst_length): Rename...
11659 (get_min_string_length): ...to this. Add argument.
11660 (handle_char_store): Extend to handle multi-character stores by
11662 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
11663 * tree.h (initializer_zerop): Add argument.
11665 2018-07-26 Jakub Jelinek <jakub@redhat.com>
11667 PR middle-end/86660
11668 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
11669 declare target to variables if they have always,{to,from,tofrom} map
11672 2018-07-26 Martin Liska <mliska@suse.cz>
11675 * lto-wrapper.c: Add linker_output as prefix
11676 for ltrans_output_file.
11678 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
11680 PR rtl-optimization/85805
11681 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
11682 value for hard registers if that was written in the same mode.
11684 2018-07-26 Martin Liska <mliska@suse.cz>
11686 PR gcov-profile/86536
11687 * gcov.c (format_gcov): Use printf format %.*f directly
11688 and do not handle special values.
11690 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11692 * common/config/arc/arc-common.c (arc_option_optimization_table):
11693 Update default optimizations for size.
11695 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11697 * config/arc/arc.md (movsf_insn): Add short instruction selection.
11698 * config/arc/constraints.md (CfZ): New constraint.
11699 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
11700 (subsf3_fpu): Likewise.
11701 (cmpsf_fpu): Likewise.
11702 (cmpsf_fpu_uneq): Likewise.
11704 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11706 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
11708 (arc_is_uncached_mem_p): uncached applies to both the variable and
11711 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11713 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
11716 2018-07-25 David Malcolm <dmalcolm@redhat.com>
11718 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
11719 field "m_scopes" from vec to auto_vec.
11721 2018-07-25 Martin Liska <mliska@suse.cz>
11723 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
11726 2018-07-25 Richard Biener <rguenther@suse.de>
11729 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
11730 special wrt context_die late.
11731 (gen_subprogram_die): Re-use DIEs in local scope.
11733 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
11735 PR tree-optimization/86644
11736 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
11738 2018-07-25 Martin Liska <mliska@suse.cz>
11740 PR middle-end/86645
11741 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
11742 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
11744 2018-07-25 Martin Liska <mliska@suse.cz>
11747 * params.def: Explain ASan abbreviation and provide
11748 a documentation link.
11750 2018-07-24 Martin Sebor <msebor@redhat.com>
11752 PR tree-optimization/86622
11753 PR tree-optimization/86532
11754 * builtins.h (string_length): Declare.
11755 * builtins.c (c_strlen): Correct handling of non-constant offsets.
11756 (check_access): Be prepared for non-constant length ranges.
11757 (string_length): Make extern.
11758 * expr.c (string_constant): Only handle the minor non-constant
11759 array index. Use string_constant to compute the length of
11760 a generic string constant.
11762 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
11764 PR tree-optimization/86618
11765 * tree-vect-stmts.c (vectorizable_call): Don't take the address
11766 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
11768 2018-07-24 David Malcolm <dmalcolm@redhat.com>
11770 PR tree-optimization/86636
11771 * json.cc (json::object::set): Fix comment. Add assertions.
11772 (json::array::append): Move here from json.h. Add comment and an
11774 (json::string::string): Likewise.
11775 * json.h (json::array::append): Move to json.cc.
11776 (json::string::string): Likewise.
11777 * optinfo-emit-json.cc
11778 (optrecord_json_writer::impl_location_to_json): Assert that we
11779 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
11780 wrapper around it. Expand the location once, rather than three
11782 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
11783 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
11785 (optrecord_json_writer::optinfo_to_json): Likewise, in four
11786 places. Fix some overlong lines.
11788 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
11790 * config/aarch64/aarch64-simd.md
11791 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
11792 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
11793 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
11794 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
11795 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
11796 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
11797 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
11798 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
11799 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
11801 2018-07-24 Jakub Jelinek <jakub@redhat.com>
11803 PR middle-end/86627
11804 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
11805 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
11806 and abs_d == d, do the power of two handling if profitable.
11808 2018-07-24 Richard Biener <rguenther@suse.de>
11810 * match.pd: Add BIT_FIELD_REF canonicalizations.
11812 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11815 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
11817 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11819 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
11820 terminated STRING_CST object.
11822 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11824 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
11826 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
11828 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
11829 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
11830 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
11831 the elements into a register.
11832 (rs6000_split_v4si_init_di_reg): Delete.
11833 (rs6000_split_v4si_init): Delete.
11834 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
11835 (vsx_init_v4si): Rewrite as a define_expand.
11837 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
11839 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
11840 zero_extend argument from memory): New.
11842 2018-07-22 Martin Sebor <msebor@redhat.com>
11845 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
11846 diagnosing calls with unknown arguments unless -Walloca-larger-than
11847 is restricted to less than PTRDIFF_MAX bytes.
11849 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
11851 * doc/gcov.texi (Invoking Gcov): Editorial changes.
11853 2018-07-20 David Malcolm <dmalcolm@redhat.com>
11855 * pretty-print.c (text_info::set_location): Remove redundant
11856 "line_table" parameter from call to rich_location::set_range.
11858 2018-07-20 Martin Sebor <msebor@redhat.com>
11860 PR middle-end/82063
11861 * builtins.c (expand_builtin_alloca): Adjust.
11862 * calls.c (alloc_max_size): Simplify.
11863 * cgraphunit.c (cgraph_node::expand): Adjust.
11864 * common.opt (larger_than_size, warn_frame_larger_than): Remove
11866 (frame_larger_than_size): Same.
11867 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
11868 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
11870 * doc/invoke.texi (GCC Command Options): Document option arguments.
11871 Explain byte-size arguments and suffixes.
11872 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
11873 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
11874 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
11875 * doc/options.texi (UInteger): Expand.
11876 (Host_Wide_Int, ByteSize): Document new properties.
11877 * final.c (final_start_function_1): Include sizes in an error message.
11878 * function.c (frame_offset_overflow): Same.
11879 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
11880 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
11881 Diagnose unbounded alloca calls only for limits of less than
11883 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
11884 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
11886 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
11887 only for limits of less than PTRDIFF_MAX.
11888 * langhooks-def.h (lhd_handle_option): Change function argument
11890 * langhooks.c (lhd_handle_option): Same.
11891 * langhooks.h (handle_option): Same.
11892 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
11894 (var_type, var_type_struct): Same.
11895 (var_set): Handle ByteSize flag.
11896 * optc-gen.awk: Add comments to output to ease debugging. Make
11897 use of HOST_WIDE_INT where appropriate.
11898 * opts-gen-save.awk: Use %lx to format unsigned long.
11899 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
11900 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
11901 arguments. Parse bytes-size suffixes.
11902 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
11903 (enum_value_to_arg): Same.
11904 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
11905 (handle_option): Adjust.
11906 (generate_option): Change function argument to HOST_WIDE_INT.
11907 (cmdline_handle_error): Adjust.
11908 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
11909 (set_option): Change function argument to HOST_WIDE_INT.
11910 (option_enabled): Handle cl_host_wide_int.
11911 (get_option_state): Handle CLVC_SIZE.
11912 (control_warning_option): Same.
11913 * opts.c (common_handle_option): Change function argument to
11914 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
11915 OPT_Wvla_larger_than_.
11916 * opts.h (enum cl_var_type): Add an enumerator.
11917 * stor-layout.c (layout_decl): Print a more meaningful warning.
11918 * toplev.c (output_stack_usage): Adjust.
11920 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
11922 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
11923 call to inline_expand_builtin_string_cmp.
11924 (expand_builtin_strcmp): Likewise.
11925 (expand_builtin_strncmp): Likewise.
11926 (inline_string_cmp): Delete the last parameter, change char_type_node
11927 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
11929 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
11930 the inlining expansion on target where the type of the call has same or
11931 narrower precision than unsigned char.
11933 2018-07-20 David Malcolm <dmalcolm@redhat.com>
11935 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
11936 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
11937 * common.opt (fsave-optimization-record): New option.
11938 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
11939 * doc/invoke.texi (-fsave-optimization-record): New option.
11940 * dumpfile.c: Include "optinfo-emit-json.h".
11941 (struct kv_pair): Move to coretypes.h.
11942 (optgroup_options): Make non-static.
11943 (dump_context::end_scope): Call
11944 optimization_records_maybe_pop_dump_scope.
11945 * dumpfile.h (optgroup_options): New decl.
11946 * json.cc: New file.
11947 * json.h: New file.
11948 * optinfo-emit-json.cc: New file.
11949 * optinfo-emit-json.h: New file.
11950 * optinfo.cc: Include "optinfo-emit-json.h".
11951 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
11952 (optinfo_enabled_p): Check optimization_records_enabled_p.
11953 (optinfo_wants_inlining_info_p): Likewise.
11954 * optinfo.h: Update comment.
11955 * profile-count.c (profile_quality_as_string): New function.
11956 * profile-count.h (profile_quality_as_string): New decl.
11957 (profile_count::quality): New accessor.
11958 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
11959 and optinfo_emit_json_cc_tests.
11960 * selftest.h (selftest::json_cc_tests): New decl.
11961 (selftest::optinfo_emit_json_cc_tests): New decl.
11962 * toplev.c: Include "optinfo-emit-json.h".
11963 (compile_file): Call optimization_records_finish.
11964 (do_compile): Call optimization_records_start.
11965 * tree-ssa-live.c: Include optinfo.h.
11966 (remove_unused_scope_block_p): Retain inlining information if
11967 optinfo_wants_inlining_info_p returns true.
11969 2018-07-20 Richard Biener <rguenther@suse.de>
11972 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
11973 to cover -flto-partition=none.
11975 2018-07-20 Martin Liska <mliska@suse.cz>
11977 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
11978 (get_decl_source_range): Remove unused function.
11980 2018-07-20 Richard Biener <rguenther@suse.de>
11982 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
11983 (struct vn_phi_s): Likewise.
11984 (struct vn_reference_s): Likewise.
11985 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
11986 for searching the slot of an entry known to be in the hash itself.
11987 (vn_phi_hasher::equal): Likewise.
11988 (vn_reference_hasher::equal): Likewise.
11989 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
11991 (optimistic_info, current_info): Remove, keeping only valid_info.
11992 (vn_reference_lookup_1): Remove fallback lookup.
11993 (vn_reference_lookup_2): Likewise.
11994 (vn_nary_op_lookup_1): Likewise.
11995 (vn_phi_lookup): Likewise.
11996 (vn_nary_build_or_lookup_1): Make sure to not chain the built
11998 (vn_reference_insert): Adjust, chain the inserted hash element
11999 at last_inserted_ref.
12000 (vn_reference_insert_pieces): Likewise.
12001 (visit_reference_op_call): Likewise.
12002 (vn_nary_op_insert_into): Chain the inserted hash element at
12003 last_inserted_nary.
12004 (vn_nary_op_insert_pieces): Adjust.
12005 (vn_nary_op_insert): Likewise.
12006 (vn_nary_op_insert_stmt): Likewise.
12007 (vn_phi_insert): Adjust, chain the inserted hash element at
12009 (process_scc): Remove clearing and copying the optimistic
12010 table. Instead remove elements inserted during an optimistic
12011 iteration from the single table we maintain.
12012 (init_scc_vn): Adjust.
12013 (free_scc_vn): Likewise.
12014 (sccvn_dom_walker::record_cond): Likewise.
12015 (sccvn_dom_walker::after_dom_children): Likewise.
12017 2018-07-19 Martin Sebor <msebor@redhat.com>
12019 PR tree-optimization/84047
12020 PR tree-optimization/83776
12021 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
12022 (check_array_bounds): Call it.
12024 2018-07-19 Martin Sebor <msebor@redhat.com>
12026 * align.h (align_flags): Use member initialization.
12028 2018-07-19 David Malcolm <dmalcolm@redhat.com>
12030 * Makefile.in (OBJS): Add optinfo.o.
12031 * coretypes.h (class symtab_node): New forward decl.
12032 (struct cgraph_node): New forward decl.
12033 (class varpool_node): New forward decl.
12034 * dump-context.h: New file.
12035 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
12037 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
12038 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
12039 (set_alt_dump_file): Likewise.
12040 (dump_context::~dump_context): New dtor.
12041 (dump_gimple_stmt): Move implementation to...
12042 (dump_context::dump_gimple_stmt): ...this new member function.
12043 Add the stmt to any pending optinfo, creating one if need be.
12044 (dump_gimple_stmt_loc): Move implementation to...
12045 (dump_context::dump_gimple_stmt_loc): ...this new member function.
12046 Start a new optinfo and add the stmt to it.
12047 (dump_gimple_expr): Move implementation to...
12048 (dump_context::dump_gimple_expr): ...this new member function.
12049 Add the stmt to any pending optinfo, creating one if need be.
12050 (dump_gimple_expr_loc): Move implementation to...
12051 (dump_context::dump_gimple_expr_loc): ...this new member function.
12052 Start a new optinfo and add the stmt to it.
12053 (dump_generic_expr): Move implementation to...
12054 (dump_context::dump_generic_expr): ...this new member function.
12055 Add the tree to any pending optinfo, creating one if need be.
12056 (dump_generic_expr_loc): Move implementation to...
12057 (dump_context::dump_generic_expr_loc): ...this new member
12058 function. Add the tree to any pending optinfo, creating one if
12060 (dump_printf): Move implementation to...
12061 (dump_context::dump_printf_va): ...this new member function. Add
12062 the text to any pending optinfo, creating one if need be.
12063 (dump_printf_loc): Move implementation to...
12064 (dump_context::dump_printf_loc_va): ...this new member function.
12065 Start a new optinfo and add the stmt to it.
12066 (dump_dec): Move implementation to...
12067 (dump_context::dump_dec): ...this new member function. Add the
12068 value to any pending optinfo, creating one if need be.
12069 (dump_context::dump_symtab_node): New member function.
12070 (dump_context::get_scope_depth): New member function.
12071 (dump_context::begin_scope): New member function.
12072 (dump_context::end_scope): New member function.
12073 (dump_context::ensure_pending_optinfo): New member function.
12074 (dump_context::begin_next_optinfo): New member function.
12075 (dump_context::end_any_optinfo): New member function.
12076 (dump_context::s_current): New global.
12077 (dump_context::s_default): New global.
12078 (dump_scope_depth): Delete global.
12079 (dumpfile_ensure_any_optinfo_are_flushed): New function.
12080 (dump_symtab_node): New function.
12081 (get_dump_scope_depth): Reimplement in terms of dump_context.
12082 (dump_begin_scope): Likewise.
12083 (dump_end_scope): Likewise.
12084 (selftest::temp_dump_context::temp_dump_context): New ctor.
12085 (selftest::temp_dump_context::~temp_dump_context): New dtor.
12086 (selftest::verify_item): New function.
12087 (ASSERT_IS_TEXT): New macro.
12088 (ASSERT_IS_TREE): New macro.
12089 (ASSERT_IS_GIMPLE): New macro.
12090 (selftest::test_capture_of_dump_calls): New test.
12091 (selftest::dumpfile_c_tests): Call it.
12092 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
12093 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
12094 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
12095 descriptive comment.
12096 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
12097 (dump_node, dump_bb): Move these unrelated decls.
12098 (class dump_manager): Add leading comment.
12099 * optinfo.cc: New file.
12100 * optinfo.h: New file.
12102 2018-07-19 Michael Collison <michael.collison@arm.com>
12103 Richard Henderson <rth@redhat.com>
12105 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
12106 (subti): Handle op1 zero.
12107 (subvti4, usub4ti4): New.
12108 (*sub<GPI>3_compare1_imm): New.
12109 (sub<GPI>3_carryinCV): New.
12110 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
12111 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
12113 2018-07-19 Michael Collison <michael.collison@arm.com>
12114 Richard Henderson <rth@redhat.com>
12116 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
12117 (addti3): Create simpler code if low part is already known to be 0.
12118 (addvti4, uaddvti4): New.
12119 (*add<GPI>3_compareC_cconly_imm): New.
12120 (*add<GPI>3_compareC_cconly): New.
12121 (*add<GPI>3_compareC_imm): New.
12122 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
12123 handle constants within this pattern..
12124 (*add<GPI>3_compareV_cconly_imm): New.
12125 (*add<GPI>3_compareV_cconly): New.
12126 (*add<GPI>3_compareV_imm): New.
12127 (add<GPI>3_compareV): New.
12128 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
12129 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
12130 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
12131 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
12132 with 'comparison' operator.
12133 (*add<GPI>3_compareV_cconly_imm): Ditto.
12134 (*add<GPI>3_compareV_cconly): Ditto.
12135 (*add<GPI>3_compareV_imm): Ditto.
12136 (add<GPI>3_compareV): Ditto.
12137 (add<mode>3_carryinC): Ditto.
12138 (*add<mode>3_carryinC_zero): Ditto.
12139 (*add<mode>3_carryinC): Ditto.
12140 (add<mode>3_carryinV): Ditto.
12141 (*add<mode>3_carryinV_zero): Ditto.
12142 (*add<mode>3_carryinV): Ditto.
12144 2018-07-19 Michael Collison <michael.collison@arm.com>
12145 Richard Henderson <rth@redhat.com>
12147 * config/aarch64/aarch64-modes.def (CC_V): New.
12148 * config/aarch64/aarch64-protos.h
12149 (aarch64_addti_scratch_regs): Declare
12150 (aarch64_subvti_scratch_regs): Declare.
12151 (aarch64_expand_subvti): Declare.
12152 (aarch64_gen_unlikely_cbranch): Declare
12153 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
12154 for signed overflow using CC_Vmode.
12155 (aarch64_get_condition_code_1): Handle CC_Vmode.
12156 (aarch64_gen_unlikely_cbranch): New function.
12157 (aarch64_addti_scratch_regs): New function.
12158 (aarch64_subvti_scratch_regs): New function.
12159 (aarch64_expand_subvti): New function.
12161 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12163 * config/aarch64/aarch64-option-extensions.def: New entry for profile
12165 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
12166 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
12169 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12172 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
12173 address check not strict.
12175 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12177 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
12179 (store_pair_lanes<mode>): Likewise.
12180 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
12181 enum value 'ADDR_QUERY_LDP_STP_N'.
12182 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
12183 (aarch64_print_address_internal): Add declaration.
12184 (aarch64_print_ldpstp_address): Remove.
12185 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
12186 (aarch64_print_operand): Change printing of 'y'.
12187 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
12188 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
12189 'true' rather than '1'.
12190 * config/aarch64/constraints.md (Uml): Likewise.
12191 (Uml): Rename to Umn.
12194 2018-07-19 Richard Biener <rguenther@suse.de>
12196 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
12198 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
12199 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
12200 (vn_reference_hasher): Likewise.
12201 (struct vn_tables_s): Remove obstack and alloc-pool members.
12202 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
12203 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
12204 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
12205 (vn_reference_insert_pieces): Likewise.
12206 (alloc_vn_nary_op_noinit): Adjust.
12207 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
12208 (vn_phi_eq): Adjust.
12209 (shared_lookup_phiargs): Remove.
12210 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
12211 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
12212 (visit_reference_op_call): Likewise.
12213 (copy_nary, copy_phi, copy_reference): Remove.
12214 (process_scc): Rewind the obstack when iterating. Do not
12215 copy the elements to valid_info but just move them from one
12216 hashtable to the other.
12217 (allocate_vn_table): Adjust.
12218 (free_vn_table): Likewise.
12219 (init_scc_vn): Likewise.
12220 (free_scc_vn): Likewise.
12222 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
12225 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
12226 indirect_return as function type attribute.
12227 (ix86_attribute_table): Change indirect_return to function
12229 * doc/extend.texi: Update indirect_return attribute.
12231 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
12233 * wide-int.h (widest2_int): New.
12234 * gimple-fold.c (arith_overflowed_p): Use it.
12235 * tree.h (widest2_int_cst): New.
12236 * tree-vrp.c (wide_int_binop_overflow): Rename from
12237 vrp_int_const_binop.
12238 Rewrite to work on trees.
12239 (extract_range_from_multiplicative_op_1): Abstract code to...
12240 (wide_int_range_min_max): ...here.
12241 (wide_int_range_cross_product): ...and here.
12242 (extract_range_from_binary_expr_1): Abstract overflow code to...
12243 (wide_int_range_mult_wrapping): ...here.
12244 * tree-vrp.h (wide_int_range_cross_product): New.
12245 (wide_int_range_mult_wrapping): New.
12247 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
12248 Julia Koval <julia.koval@intel.com>
12250 * config/i386/x86-tune-costs.h (skylake_memcpy,
12251 skylake_memset): Replace rep_prefix with unrolling for size 512.
12253 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
12255 PR middle-end/86544
12256 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
12257 comparision with EQ_EXPR in last stmt.
12259 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
12261 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
12262 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
12263 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
12264 previously known as "PowerPC AltiVec Built-in Functions". Move
12265 some material to new subsubsections "PowerPC AltiVec Built-in
12266 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
12268 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
12269 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
12270 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
12271 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
12273 2018-07-18 Richard Biener <rguenther@suse.de>
12275 PR tree-optimization/86557
12276 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
12279 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
12281 * config/s390/s390.c (s390_function_profiler): Generate CFI.
12283 2018-07-17 Jeff Law <law@redhat.com>
12285 * config/arm/arm.c (get_label_padding): Update for recent
12286 changes to label_to_alignment.
12288 PR tree-optimization/86010
12289 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
12291 * config/mips/mips.c (vr4130_align_insns): Update for recent
12292 changes to label_to_alignment.
12294 * config/frv/frv.c (frv_label_align): Update for recent changes
12295 to label_to_alignment.
12297 * config/nios2/nios2.c (nios2_label_align): Update for recent
12298 changes which dropped ALIGN_LABELS_LOG.
12300 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
12302 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
12303 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
12305 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12307 * config/arc/arc.c (arc_label_align): Use align_labels instead of
12308 deprecated align_labels_log.
12310 2018-07-17 Richard Biener <rguenther@suse.de>
12313 * dwarf2out.c (init_sections_and_labels): Always generate
12314 a debug_line_str_section for early LTO debug.
12315 (dwarf2out_finish): Reset debug_line_str_hash output early.
12316 Bump counter for extra dwarf5 .debug_loc labels to not conflict
12318 (dwarf2out_early_finish): Output debug_line_str.
12320 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
12322 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
12323 index register on z196 or later.
12325 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
12327 * config/s390/s390.c (s390_default_align): Set default function
12329 (s390_override_options_after_change): Call s390_default align.
12330 (s390_option_override_internal): Call s390_default_align.
12331 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
12333 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12335 PR middle-end/86542
12336 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
12337 to _looptemp_ clauses, other than the first two.
12339 2018-07-17 Martin Liska <mliska@suse.cz>
12341 * opts.c: Do not enable OPT_falign_* for -Os.
12343 2018-07-17 Martin Liska <mliska@suse.cz>
12345 * align.h (MAX_CODE_ALIGN): New.
12346 (MAX_CODE_ALIGN_VALUE): New.
12347 * common/config/i386/i386-common.c (ix86_handle_option):
12348 (MAX_CODE_ALIGN): Moved to align.h.
12349 * final.c (MAX_CODE_ALIGN): Likewise.
12350 * opts.c (parse_and_check_align_values):
12351 (MAX_CODE_ALIGN): Likewise.
12352 (MAX_CODE_ALIGN_VALUE): Likewise.
12354 2018-07-17 Martin Liska <mliska@suse.cz>
12356 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
12357 in order to fulfil coding style.
12358 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
12359 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12360 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
12361 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
12362 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
12363 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
12364 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
12365 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
12366 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12368 2018-07-17 Martin Liska <mliska@suse.cz>
12370 * align.h: New file.
12371 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
12373 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
12374 align_flags of label_to_alignment.
12375 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
12377 * config/m68k/m68k.c: Do not use removed align_labels_value and
12379 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
12380 (LOOP_ALIGN): Likewise.
12381 (LABEL_ALIGN): Likewise.
12382 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12383 Remove not used macro.
12384 (rs6000_loop_align): Change return type to align_flags.
12385 (rs6000_loop_align_max_skip): Remove.
12386 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
12387 Change return type to align_flags.
12388 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12389 Remove not used macro.
12390 (rs6000_loop_align): Change return type to align_flags.
12391 (rs6000_loop_align_max_skip): Remove.
12392 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
12393 * config/rx/rx-protos.h (rx_align_for_label): Make it
12395 * config/rx/rx.c (rx_align_for_label): Change return type
12397 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
12399 into align_flags class.
12400 (LABEL_ALIGN): Likewise.
12401 (LOOP_ALIGN): Likewise.
12402 * config/s390/s390.c (s390_label_align): Use align_flags
12404 (s390_asm_output_function_label): Likewise.
12405 * config/sh/sh.c (sh_override_options_after_change):
12406 Use align_flags class directly without macros.
12407 (find_barrier): Likewise.
12408 (barrier_align): Likewise.
12409 (sh_loop_align): Likewise.
12410 * config/spu/spu.c (spu_option_override):
12411 Use align_flags_tuple::get_value instead of removed macros.
12412 (spu_sched_init): Likewise.
12413 * config/spu/spu.h (GTY): Likewise.
12414 * config/visium/visium.c (visium_option_override):
12415 Set "8" as default secondary alignment.
12416 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
12417 in order to guarantee secondary alignment of 8.
12418 * coretypes.h: Include align.h header file.
12419 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
12420 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
12421 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
12422 * doc/tm.texi.in: Likewise.
12423 * final.c (struct label_alignment): Remove not used structure.
12424 (LABEL_ALIGN): Change type to align_flags.
12425 (LOOP_ALIGN): Likewise.
12426 (JUMP_ALIGN): Likewise.
12427 (default_loop_align_max_skip): Remove.
12428 (default_label_align_max_skip): Likewise.
12429 (default_jump_align_max_skip): Likewise.
12430 (default_label_align_after_barrier_max_skip):
12431 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
12432 (LABEL_TO_MAX_SKIP): Remove.
12433 (label_to_alignment): Return align_flags type instead of integer.
12434 (label_to_max_skip): Remove.
12435 (align_fuzz): Use align_flags type.
12436 (compute_alignments): Use align_flags type and use align_flags::max
12437 to combine multiple alignments.
12438 (grow_label_align): Grow vec instead of C array.
12439 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
12440 (shorten_branches): Use align_flags type and use align_flags::max
12441 to combine multiple alignments.
12442 (final_scan_insn_1): Remove usage of secondary alignment that comes
12443 from label alignment, but instead use proper secondary alignment
12444 which is computed in grow_label_align.
12445 * flags.h (struct align_flags_tuple): Move to align.h.
12446 (struct align_flags): Likewise.
12447 (state_align_loops): Rename to align_loops.
12448 (state_align_jumps): Rename to align_jumps.
12449 (state_align_labels): Rename to align_labels.
12450 (state_align_functions): Rename to align_functions.
12451 (align_loops_log): Remove.
12452 (align_jumps_log): Remove.
12453 (align_labels_log): Remove.
12454 (align_functions_log): Remove.
12455 (align_loops_max_skip): Remove.
12456 (align_jumps_max_skip): Remove.
12457 (align_labels_max_skip): Remove.
12458 (align_functions_max_skip): Remove.
12459 (align_loops_value): Remove.
12460 (align_jumps_value): Remove.
12461 (align_labels_value): Remove.
12462 (align_functions_value): Remove.
12463 * output.h (label_to_alignment): Change return type to align_flags.
12464 (label_to_max_skip): Remove.
12465 * target.def: Remove loop_align_max_skip, label_align_max_skip,
12466 jump_align_max_skip macros.
12467 * targhooks.h (default_loop_align_max_skip): Remove.
12468 (default_label_align_max_skip): Likewise.
12469 (default_jump_align_max_skip): Likewise.
12470 (default_label_align_after_barrier_max_skip): Remove.
12471 * toplev.c (read_log_maxskip): Use ::normalize function.
12472 (parse_N_M): Remove not used argument and also call ::normalize.
12473 (parse_alignment_opts): Do not pass unused arguments.
12474 * varasm.c (assemble_start_function): Use directly align_functions
12475 instead of removed macros.
12476 * system.h: Do not poison removed macros.
12478 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12480 PR middle-end/86539
12481 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
12482 and cond temporaries don't have reference type if iterator has
12483 pointer type. For init use &for_pre_body instead of pre_p if
12484 for_pre_body is non-empty.
12486 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12488 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
12489 double-double modes to SFmode directly directly.
12490 (trunc<mode>sf2_fprs): Delete.
12492 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12494 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
12495 for conversions between IFmode and the decimal floating point modes.
12496 (init_float128_ieee): Use the correct names for conversions between
12497 KFmode and the decimal floating point modes.
12499 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12501 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
12502 for the conversions between TDmode and IFmode.
12503 (init_float128_ieee): Use more correct names for the conversions
12504 between TDmode and KFmode.
12506 2018-07-16 Jakub Jelinek <jakub@redhat.com>
12508 PR tree-optimization/86526
12509 * builtins.c (expand_builtin_memcmp): Formatting fixes.
12510 (inline_expand_builtin_string_cmp): Likewise.
12511 (inline_string_cmp): Likewise. Use c_readstr instead of
12512 builtin_memcpy_read_str. Add unit_mode temporary.
12514 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
12516 PR middle-end/86528
12517 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
12518 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
12520 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
12522 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
12523 Alphabetize prototypes of built-in functions, separating out
12524 built-in functions that are listed in this section but should be
12525 described elsewhere.
12527 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
12530 * expmed.c (emit_store_flag): Do not emit setcc followed by a
12531 conditional move when trapping comparison was split to a
12532 non-trapping one (and vice versa).
12534 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12536 * config/s390/s390.c (s390_function_profiler): Generate nops
12537 instead of profiler call sequences.
12538 * config/s390/s390.opt: Add the new option.
12540 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12542 * config/s390/s390.c (s390_function_profiler): Generate
12543 __mcount_loc section.
12544 * config/s390/s390.opt: Add the new option.
12546 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12548 * common.opt: Add the new warning.
12549 * config/s390/s390.c (s390_function_profiler): Emit "brasl
12550 %r0,__fentry__" when -mfentry is specified.
12551 (s390_option_override_internal): Disallow -mfentry for 31-bit
12553 * config/s390/s390.opt: Add the new option.
12555 2018-07-16 Richard Biener <rguenther@suse.de>
12558 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
12559 for function-local FUNCTION_DECL and RESULT_DECL immediately.
12561 2018-07-16 Martin Liska <mliska@suse.cz>
12564 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
12567 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
12569 * config/arc/arcHS.md: Update ARCHS scheduling rules.
12571 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
12573 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
12575 * config/arc/arc-cpus.def (hs4x): New cpu.
12577 * config/arc/arc-tables.opt: Regenerate.
12578 * config/arc/arc.c (arc_sched_issue_rate): New function.
12579 (TARGET_SCHED_ISSUE_RATE): Define.
12580 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
12581 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
12583 (attr tune): Add ARCHS4x tune values.
12584 (attr tune_dspmpy): Define.
12585 (*tst): Correct instruction type.
12586 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
12587 * config/arc/arcHS4x.md: New file.
12588 * config/arc/fpu.md: Update instruction type attributes.
12589 * config/arc/t-multilib: Regenerate.
12591 2018-07-16 Tom de Vries <tdevries@suse.de>
12594 * var-tracking.c (vt_initialize): Fix pre_dec handling.
12596 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12598 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
12601 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
12603 PR tree-optimization/86514
12604 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
12605 conversion to a boolean type from a type with greater precision.
12607 2018-07-16 Tom de Vries <tdevries@suse.de>
12609 * var-tracking.c (vt_initialize): Print adjusted insn slim if
12610 dump_flags request TDF_SLIM.
12612 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
12614 * fold-const.c (int_const_binop_1): Abstract...
12615 (wide_int_binop): ...wide int code here.
12616 (poly_int_binop): ...poly int code here.
12617 Abstract the rest of int_const_binop_1 into int_const_binop.
12618 * fold-const.h (wide_int_binop): New.
12619 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
12620 Remove useless PLUS/MINUS_EXPR case.
12621 (zero_nonzero_bits_from_vr): Move wide int code...
12622 (zero_nonzero_bits_from_bounds): ...here.
12623 (extract_range_from_binary_expr_1): Move mask optimization code...
12624 (range_easy_mask_min_max): ...here.
12625 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
12626 (range_easy_mask_min_max): New.
12628 2018-07-15 Jeff Law <law@redhat.com>
12631 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
12634 2018-07-14 Jim Wilson <jimw@sifive.com>
12636 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
12638 2018-07-14 Paul Koning <ni1d@arrl.net>
12640 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
12642 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
12644 * lto-streamer-out.c (copy_function_or_variable): Dump info about
12647 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
12648 Steve Munroe <munroesj52@gmail.com>
12650 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
12651 (_mm_andnot_si128): Likewise.
12652 (_mm_or_si128): Likewise.
12653 (_mm_xor_si128): Likewise.
12655 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
12657 PR middle-end/78809
12658 * builtins.c (expand_builtin_memcmp): Inline the calls first
12659 when result_eq is false.
12660 (expand_builtin_strcmp): Inline the calls first.
12661 (expand_builtin_strncmp): Likewise.
12662 (inline_string_cmp): New routine. Expand a string compare
12663 call by using a sequence of char comparison.
12664 (inline_expand_builtin_string_cmp): New routine. Inline expansion
12665 a call to str(n)cmp/memcmp.
12666 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
12668 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
12670 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
12672 * config/arm/driver-arm.c: Include arm-native.h.
12673 (host_detect_local_cpu): Use auto-generated data tables.
12674 (vendors, arm_cpu_table): Delete. Move part information to ...
12675 * config/arm/arm-cpus.in: ... here.
12676 * config/arm/parsecpu.awk (gen_native): New function.
12677 (vendor, part): New CPU fields.
12678 (END): Add support for building the native CPU detection tables.
12679 * config/arm/t-arm (arm-native.h): Add build rule.
12680 (driver-arm.o): Add dependency on arm-native.h.
12682 2018-07-13 Richard Biener <rguenther@suse.de>
12684 PR middle-end/85974
12685 * match.pd (addr1 - addr2): Allow either of the operand to
12688 2018-07-13 Tom de Vries <tdevries@suse.de>
12690 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
12693 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
12695 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
12696 arrays instead of numbered variables.
12698 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
12700 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
12701 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
12703 2018-07-13 Richard Biener <rguenther@suse.de>
12706 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
12707 instead of get_context_die.
12709 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12710 Richard Biener <rguenther@suse.de>
12712 PR middle-end/86489
12713 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
12714 that the loop latch destination where phi is defined.
12716 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
12718 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
12719 (riscv_expand_epilogue): Add assertion to check interrupt mode.
12720 (riscv_set_current_function): Extract getting interrupt type to new
12722 (riscv_get_interrupt_type): New function.
12723 (riscv_merge_decl_attributes): New function, checking interrupt type is
12725 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
12727 2018-07-12 Paul Koning <ni1d@arrl.net>
12729 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
12732 2018-07-12 Paul Koning <ni1d@arrl.net>
12734 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
12735 zero reference, add doloop_end instead.
12736 * doc/md.texi (decrement_and_branch_until_zero): Remove.
12737 (Looping patterns): Remove decrement_and_branch_until_zero. Add
12738 detail for doloop_end.
12740 2018-07-12 Martin Sebor <msebor@redhat.com>
12743 * attribs.c (decl_attributes): Reject conflicting attributes before
12744 calling attribute handlers.
12746 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
12748 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
12750 (gcc::dump_manager::get_dump_file_name): likewise.
12751 (dump_begin): Likewise.
12752 * dumpfile.h (dump_begin): Update prototype.
12753 (gcc::dump_manager::get_dump_file_name,
12754 gcc::dump_manager::get_dump_file_name): Update prototype.
12756 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12758 * internal-fn.h (vectorizable_internal_fn_p): New function.
12759 * tree-vect-slp.c (compatible_calls_p): Likewise.
12760 (vect_build_slp_tree_1): Remove nops argument. Handle calls
12761 to internal functions.
12762 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
12764 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12766 * fold-const.h (inverse_conditions_p): Declare.
12767 * fold-const.c (inverse_conditions_p): New function.
12768 * match.pd: Use inverse_conditions_p. Add folds of view_converts
12769 that test the inverse condition of a conditional internal function.
12770 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
12771 * internal-fn.c (internal_fn_mask_index): Handle conditional
12772 internal functions.
12773 (vectorized_internal_fn_supported_p): New function.
12774 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
12775 (any_pred_load_store): Replace with...
12776 (need_to_predicate): ...this new variable.
12777 (redundant_ssa_names): New variable.
12778 (ifcvt_can_use_mask_load_store): Move initial checks to...
12779 (ifcvt_can_predicate): ...this new function. Handle tree codes
12780 for which a conditional internal function exists.
12781 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
12782 instead of ifcvt_can_use_mask_load_store. Update after variable
12784 (predicate_load_or_store): New function, split out from
12785 predicate_mem_writes.
12786 (check_redundant_cond_expr): New function.
12787 (value_available_p): Likewise.
12788 (predicate_rhs_code): Likewise.
12789 (predicate_mem_writes): Rename to...
12790 (predicate_statements): ...this. Use predicate_load_or_store
12791 and predicate_rhs_code.
12792 (combine_blocks, tree_if_conversion): Update after above name changes.
12793 (ifcvt_local_dce): Handle redundant_ssa_names.
12794 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
12795 general conditional functions.
12796 * tree-vect-stmts.c (vectorizable_call): Likewise.
12798 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12799 Alan Hayward <alan.hayward@arm.com>
12800 David Sherwood <david.sherwood@arm.com>
12802 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
12803 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
12804 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
12805 plus and minus and convert them into IFN_COND_FMA-based sequences.
12806 (convert_mult_to_fma): Handle conditional plus and minus.
12808 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12810 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
12811 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
12812 (cond_fnms_optab): New optabs.
12813 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
12814 internal functions.
12815 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
12816 * internal-fn.h (get_conditional_internal_fn): Declare.
12817 (get_unconditional_internal_fn): Likewise.
12818 * internal-fn.c (cond_ternary_direct): New macro.
12819 (expand_cond_ternary_optab_fn): Likewise.
12820 (direct_cond_ternary_optab_supported_p): Likewise.
12821 (FOR_EACH_COND_FN_PAIR): Likewise.
12822 (get_conditional_internal_fn): New function.
12823 (get_unconditional_internal_fn): Likewise.
12824 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
12825 (gimple_match_op::gimple_match_op): Add a new overload for 5
12827 (gimple_match_op::set_op): Likewise.
12828 (gimple_resimplify5): Declare.
12829 * genmatch.c (decision_tree::gen): Generate simplifications for
12831 * gimple-match-head.c (gimple_simplify): Define an overload for
12832 5 operands. Handle calls with 5 arguments in the top-level overload.
12833 (convert_conditional_op): Handle conversions from unconditional
12834 internal functions to conditional ones.
12835 (gimple_resimplify5): New function.
12836 (build_call_internal): Pass a fifth operand.
12837 (maybe_push_res_to_seq): Likewise.
12838 (try_conditional_simplification): Try converting conditional
12839 internal functions to unconditional internal functions.
12840 Handle 3-operand unconditional forms.
12841 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
12842 Define ternary equivalents of the current rules for binary conditional
12843 internal functions.
12844 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
12845 ternary operations.
12846 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
12847 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
12848 (optab): Handle them.
12849 (SVE_COND_FP_TERNARY): New int iterator.
12850 (sve_fmla_op, sve_fmad_op): New int attributes.
12851 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
12852 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
12853 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
12855 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12857 * target.def (preferred_else_value): New target hook.
12858 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
12859 * doc/tm.texi: Regenerate.
12860 * targhooks.h (default_preferred_else_value): Declare.
12861 * targhooks.c (default_preferred_else_value): New function.
12862 * internal-fn.h (conditional_internal_fn_code): Declare.
12863 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
12864 (get_conditional_internal_fn): Use it.
12865 (conditional_internal_fn_code): New function.
12866 * gimple-match.h (gimple_match_cond): New struct.
12867 (gimple_match_op): Add a cond member function.
12868 (gimple_match_op::gimple_match_op): Update all forms to take a
12870 * genmatch.c (expr::gen_transform): Use the same condition as res_op
12871 for the suboperation, but don't specify a particular else_value.
12872 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
12873 (visit_nary_op, visit_reference_op_load): Pass
12874 gimple_match_cond::UNCOND to the gimple_match_op constructor.
12875 * gimple-match-head.c: Include tree-eh.h
12876 (convert_conditional_op): New function.
12877 (maybe_resimplify_conditional_op): Likewise.
12878 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
12879 (gimple_resimplify2): Likewise.
12880 (gimple_resimplify3): Likewise.
12881 (gimple_resimplify4): Likewise.
12882 (maybe_push_res_to_seq): Return null for conditional operations.
12883 (try_conditional_simplification): New function.
12884 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
12886 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
12888 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
12890 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
12892 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
12894 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12896 (hash_tree): Do not hash DECL_FCONTEXT
12897 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
12898 Do not stream DECL_FCONTEXT.
12899 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
12900 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
12902 2018-07-12 Richard Biener <rguenther@suse.de>
12905 * dwarf2out.c (gen_block_die): Only output blocks when they have
12906 at least one !DECL_IGNORED_P variable.
12908 2018-07-12 Richard Biener <rguenther@suse.de>
12911 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
12912 Remove -mieee-fp handling.
12914 2018-07-12 Richard Biener <rguenther@suse.de>
12916 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
12917 left-over from last patch.
12919 2018-07-12 Jakub Jelinek <jakub@redhat.com>
12921 PR tree-optimization/86492
12922 * gimple-ssa-store-merging.c
12923 (imm_store_chain_info::coalesce_immediate_stores): Call
12924 check_no_overlap even for the merge_overlapping case. Formatting fix.
12926 2018-07-12 Richard Biener <rguenther@suse.de>
12928 PR middle-end/86479
12929 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
12930 move possibly trapping operations into the conditional.
12932 2018-07-12 Richard Biener <rguenther@suse.de>
12934 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
12935 (vn_lookup_simplify_result): Remove recursion limit applied
12937 (vn_nary_build_or_lookup_1): Adjust.
12938 (try_to_simplify): Likewise.
12939 * gimple-match-head.c (gimple_resimplify1): Instead apply one
12941 (gimple_resimplify2): Likewise.
12942 (gimple_resimplify3): Likewise.
12943 (gimple_resimplify4): Likewise.
12945 2018-07-11 Jakub Jelinek <jakub@redhat.com>
12947 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
12948 Use __mmask64 type instead of __mmask8 for __M argument.
12949 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
12950 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
12952 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
12953 __mmask16 for __M argument.
12954 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
12955 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
12956 to __mmask16 instead of __mmask8.
12957 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
12958 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
12959 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
12960 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
12961 instead of __mmask16 for __U argument.
12962 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
12963 __mmask16 instead of __mmask8 for __U argument.
12964 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
12966 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
12968 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
12970 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
12972 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
12974 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
12976 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
12978 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
12980 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
12982 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
12984 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
12985 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
12986 return type as well as __M argument type and all casts from __mmask8
12988 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
12989 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
12990 return type as well as __M argument type and all casts from __mmask8
12992 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
12993 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
12994 return type as well as __M argument type and all casts from __mmask8
12996 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
12997 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
12998 return type as well as __M argument type and all casts from __mmask8
13000 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
13001 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
13004 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
13006 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
13007 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
13010 2018-07-11 Paul Koning <ni1d@arrl.net>
13012 * doc/md.texi (define_subst): Document how multiple occurrences of
13013 the same argument in the replacement pattern are handled.
13015 2018-07-11 Paul Koning <ni1d@arrl.net>
13017 * doc/extend.texi (Common Variable Attributes): Move "mode" into
13018 alphabetical order.
13019 (Common Type Attributes): Add "mode" attribute.
13021 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
13023 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
13024 stream DECL_ORIGINAL_TYPE.
13025 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
13026 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
13027 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
13028 Do not walk original type.
13029 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
13031 (write_ts_decl_non_common_tree_pointers): Do not stream
13033 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
13034 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
13036 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
13038 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
13039 thread twice from the same starting edge.
13041 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
13043 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
13044 * gimple.c (gimple_call_nonnull_result_p): ...here...
13045 (gimple_call_nonnull_arg): ...and here.
13046 * gimple.h (gimple_call_nonnull_result_p): New.
13047 (gimple_call_nonnull_arg): New.
13049 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
13051 * config/arm/arm-cpus.in: Move information from fpu field of each
13052 cpu definition to the isa field.
13053 * config/arm/parsecpu.awk (fpu): Delete match rule.
13054 (gen_comm_data): Don't add bits from the CPU's FPU entry.
13056 2018-07-11 Richard Biener <rguenther@suse.de>
13059 * dwarf2out.c (init_sections_and_labels): Use
13060 output_asm_line_debug_info consistently.
13061 (dwarf2out_early_finish): Likewise.
13062 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
13065 2018-07-11 Richard Biener <rguenther@suse.de>
13067 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
13068 Rework father_bb setting in a way to avoid propagating constants
13069 multiple times on a loop body.
13071 2018-07-10 Mark Wielaard <mark@klomp.org>
13074 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
13076 2018-07-10 Richard Biener <rguenther@suse.de>
13078 * hash-map.h (hash_map::iterator::operator*): Return
13079 references to key and value.
13081 2018-07-10 Jakub Jelinek <jakub@redhat.com>
13084 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
13085 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
13086 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
13087 (gimplify_omp_for): For composite loops, move outer
13088 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
13089 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
13090 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
13091 TREE_LIST for both the original class iterator and the "last" helper
13092 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
13093 loop, remember has_decl_expr from outer composite loops for the
13094 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
13096 2018-07-09 Martin Sebor <msebor@redhat.com>
13098 PR middle-end/77357
13099 PR middle-end/86428
13100 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
13101 accessing implicitly initialized array elements.
13102 * expr.c (string_constant): Handle string initializers of
13103 character arrays within aggregates.
13104 * gimple-fold.c (fold_array_ctor_reference): Add argument.
13105 Store element offset. As a special case, handle zero size.
13106 (fold_nonarray_ctor_reference): Same.
13107 (fold_ctor_reference): Add argument. Store subobject offset.
13108 * gimple-fold.h (fold_ctor_reference): Add argument.
13110 2018-07-09 Paul Koning <ni1d@arrl.net>
13112 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
13113 (pdp11_insn_cost): New function.
13114 (pdp11_md_asm_adjust): New function.
13115 (TARGET_INVALID_WITHIN_DOLOOP): Define.
13116 (pdp11_rtx_costs): Update to match machine better.
13117 (output_addr_const_pdp11): Correct format mismatch warnings.
13118 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
13119 * config/pdp11/pdp11.md: General change to add base_cost and/or
13120 length attributes for use by new pdp11_insn_cost function.
13121 (MIN_BRANCH): Correct definition.
13123 (doloop_end): Use standard pattern name for looping pattern.
13124 (doloop_end_nocc): New.
13125 (movsf): Add another constraint alternative.
13126 (zero_extendqihi2): Add constraint alternatives for not in place
13128 (zero_extendhisi2): Remove.
13129 (shift patterns): Add CC handling variants.
13133 (define_peephole2): New peephole to recognize mov that sets CC for
13136 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13138 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
13139 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
13142 2018-07-09 Jakub Jelinek <jakub@redhat.com>
13145 * real.c (real_nextafter): Return true if result is denormal.
13147 2018-07-09 Martin Liska <mliska@suse.cz>
13149 * common.opt: Add back wrongly removed attribute.
13151 2018-07-09 Richard Biener <rguenther@suse.de>
13154 * dwarf2out.c (gen_block_die): For an early generated DIE
13155 always output high/low PC attributes.
13157 2018-07-09 Tom de Vries <tdevries@suse.de>
13159 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
13160 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
13161 onto VAR_DECL with abstract origin.
13163 2018-07-07 Jim Wilson <jimw@sifive.com>
13165 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
13167 2018-07-07 Tom de Vries <tdevries@suse.de>
13169 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
13171 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
13173 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
13175 (combine_bound): Use wide-int overflow calculation instead of
13177 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
13179 * fold-const.c (int_const_binop_2): Same.
13180 (extract_muldiv_1): Same.
13181 (fold_div_compare): Same.
13182 (fold_abs_const): Same.
13184 * poly-int.h (add): Same.
13188 * predict.c (predict_iv_comparison): Same.
13189 * profile-count.c (slow_safe_scale_64bit): Same.
13190 * simplify-rtx.c (simplify_const_binary_operation): Same.
13191 * tree-chrec.c (tree_fold_binomial): Same.
13192 * tree-data-ref.c (split_constant_offset_1): Same.
13193 * tree-if-conv.c (idx_within_array_bound): Same.
13194 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
13195 * tree-ssa-phiopt.c (minmax_replacement): Same.
13196 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
13197 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
13198 * vr-values.c (vr_values::adjust_range_with_scev): Same.
13199 * wide-int.cc (wi::add_large): Same.
13200 (wi::mul_internal): Same.
13201 (wi::sub_large): Same.
13202 (wi::divmod_internal): Same.
13203 * wide-int.h: Change overflow type to overflow_type for neg, add,
13204 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
13205 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
13206 mul_internal, divmod_internal.
13207 (overflow_type): New enum.
13208 (accumulate_overflow): New.
13210 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
13212 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
13213 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
13215 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
13217 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
13218 argument is checked for zero before entering loop, avoid checking again.
13220 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
13222 * gimplify.h (generic_expr_could_trap_p): Set as global function.
13223 * gimplify.h (generic_expr_could_trap_p): Likwise.
13224 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
13226 2018-07-06 Jakub Jelinek <jakub@redhat.com>
13228 PR tree-optimization/86401
13229 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
13230 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
13231 (fold_bit_and_mask): ... here. New helper function for match.pd.
13232 * fold-const.h (fold_bit_and_mask): Declare.
13233 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
13235 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
13238 * target.def (translate_mode_attribute): New hook.
13239 * targhooks.h (default_translate_mode_attribute): Declare.
13240 * targhooks.c (default_translate_mode_attribute): New function.
13241 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
13242 * doc/tm.texi: Regenerate.
13243 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
13244 (rs6000_translate_mode_attribute): New function.
13246 2018-07-06 Paul Koning <ni1d@arrl.net>
13248 * doc/md.texi (define_split): Document DONE and FAIL.
13249 (define_peephole2): Ditto.
13251 2018-07-05 Jeff Law <law@redhat.com>
13253 PR tree-optimization/86010
13254 * tree-ssa-dse.c (compute_trims): More aggressively trim at
13255 both the head and tail of mem* and str* calls.
13257 2018-07-05 Jim Wilson <jimw@sifive.com>
13259 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
13261 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
13263 * config/aarch64/aarch64-simd.md: correct flags text for
13264 MIN_EXPR replacement.
13266 2018-07-05 James Clarke <jrtc27@jrtc27.com>
13268 * configure: Regenerated.
13270 2018-07-05 Carl Love <cel@us.ibm.com>
13272 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
13273 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
13274 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
13275 VSX_BUILTIN_DOUBLEL_V4SF.
13277 2018-07-05 Martin Sebor <msebor@redhat.com>
13280 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
13281 than its domain to compute its the upper bound of a char array.
13283 2018-07-05 Nathan Sidwell <nathan@acm.org>
13285 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
13286 * doc/cpp.texi: Update comment.
13287 * doc/tm.texi: Rebuilt.
13288 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
13289 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
13290 * doc/extend.texi (Backwards Compatibility): Clarify it is system
13291 headers affected by extern "C".
13292 * system.h: Poison NO_IMPLICIT_EXTERN_C.
13293 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
13294 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
13295 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
13296 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
13297 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
13298 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
13299 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
13300 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
13301 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
13302 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
13303 config/sparc/sp64-elf.h, config/spu/spu.h,
13304 config/stormy16/stormy16.h, config/v850/v850.h,
13305 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
13306 define NO_IMPLICIT_EXTERN_C.
13307 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
13309 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13312 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
13313 instead of GET_MODE_SIZE when comparing Units.
13315 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13318 * rtlanal.c (set_noop_p): Constrain on mode change,
13319 include hard-reg-set.h
13321 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13323 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
13325 2018-07-05 Jakub Jelinek <jakub@redhat.com>
13328 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
13331 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
13333 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13335 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
13338 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
13340 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13342 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
13344 PR middle-end/86380
13345 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
13347 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
13349 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
13350 neg_*_op* variables.
13352 2018-07-04 Martin Liska <mliska@suse.cz>
13354 * tree-switch-conversion.c: Define
13355 max_ratio_for_speed and max_ratio_for_size constants.
13357 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13358 Martin Liska <mliska@suse.cz>
13360 PR middle-end/66240
13363 * common.opt: Rename align options with 'str_' prefix.
13364 * common/config/i386/i386-common.c (set_malign_value): New
13366 (ix86_handle_option): Use it to set -falign-* options/
13367 * config/aarch64/aarch64-protos.h (struct tune_params): Change
13368 type from int to string.
13369 * config/aarch64/aarch64.c: Update default values from int
13371 * config/alpha/alpha.c (alpha_override_options_after_change):
13373 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
13374 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13375 max skip conditionally.
13376 * config/i386/freebsd.h (SUBALIGN_LOG): New.
13377 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13378 max skip conditionally.
13379 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13380 max skip conditionally.
13381 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
13382 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13383 max skip conditionally.
13384 * config/i386/i386.c (struct ptt): Change type from int to
13386 (ix86_default_align): Set default values.
13387 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
13388 max skip conditionally.
13389 * config/i386/iamcu.h (SUBALIGN_LOG): New.
13390 (ASM_OUTPUT_MAX_SKIP_ALIGN):
13391 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
13392 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13393 max skip conditionally.
13394 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
13395 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
13396 * config/i386/x86-64.h (SUBALIGN_LOG): New.
13397 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13398 max skip conditionally.
13399 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13400 * config/ia64/ia64.c (ia64_option_override): Set default values
13401 for alignment options.
13402 * config/m68k/m68k.c: Handle new str_align_* options.
13403 * config/mips/mips.c (mips_set_compression_mode): Change
13405 (mips_option_override): Set default values for options.
13406 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
13408 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13410 * config/rx/rx.c (rx_option_override): Likewise.
13411 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
13412 (LABEL_ALIGN): Use align_labels_log.
13413 (LOOP_ALIGN): Use align_loops_align.
13414 * config/s390/s390.c (s390_asm_output_function_label): Use new
13416 * config/sh/sh.c (sh_override_options_after_change):
13417 Change type of constants.
13418 * config/spu/spu.c (spu_sched_init): Likewise.
13419 * config/sparc/sparc.c (sparc_option_override): Set default
13420 values for options.
13421 * config/visium/visium.c (visium_option_override): Likewise.
13422 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
13423 emit p2align format with last argument if it's not needed.
13424 * doc/invoke.texi: Document extended format of -falign-*.
13425 * final.c: Use align_labels alignment.
13426 * flags.h (struct target_flag_state): Change type to use
13428 (struct align_flags_tuple): New.
13429 (struct align_flags): Likewise.
13430 (align_loops_log): Redefine macro to use new types.
13431 (align_loops_max_skip): Redefine macro to use new types.
13432 (align_jumps_log): Redefine macro to use new types.
13433 (align_jumps_max_skip): Redefine macro to use new types.
13434 (align_labels_log): Redefine macro to use new types.
13435 (align_labels_max_skip): Redefine macro to use new types.
13436 (align_functions_log): Redefine macro to use new types.
13437 (align_loops): Redefine macro to use new types.
13438 (align_jumps): Redefine macro to use new types.
13439 (align_labels): Redefine macro to use new types.
13440 (align_functions): Redefine macro to use new types.
13441 (align_functions_max_skip): Redefine macro to use new types.
13442 (align_loops_value): New macro.
13443 (align_jumps_value): New macro.
13444 (align_labels_value): New macro.
13445 (align_functions_value): New macro.
13446 * function.c (invoke_set_current_function_hook): Propagate
13447 alignment values from flags to global variables default in
13449 * ipa-icf.c (sem_function::equals_wpa): Use
13450 cl_optimization_option_eq instead of memcmp.
13451 * lto-streamer.h (cl_optimization_stream_out): Support streaming
13453 (cl_optimization_stream_in): Likewise.
13454 * optc-save-gen.awk: Support strings in cl_optimization.
13455 * opth-gen.awk: Likewise.
13456 * opts.c (finish_options): Remove error checking of invalid
13458 (MAX_CODE_ALIGN): Remove.
13459 (MAX_CODE_ALIGN_VALUE): Likewise.
13460 (parse_and_check_align_values): New function.
13461 (check_alignment_argument): Likewise.
13462 (common_handle_option): Use check_alignment_argument.
13463 * opts.h (parse_and_check_align_values): Declare.
13464 * toplev.c (init_alignments): Remove.
13465 (read_log_maxskip): New.
13466 (parse_N_M): Likewise.
13467 (parse_alignment_opts): Likewise.
13468 (backend_init_target): Remove usage of init_alignments.
13469 * toplev.h (parse_alignment_opts): Declare.
13470 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
13472 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
13473 * tree.c (cl_option_hasher::equal): New.
13474 * varasm.c: Use new global macros.
13476 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13478 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
13479 Use a simpler align directive also if MAXSKIP = ALIGN-1.
13480 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13481 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13482 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13483 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13484 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
13485 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
13486 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
13487 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13488 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13489 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13490 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13492 2018-07-04 Martin Liska <mliska@suse.cz>
13493 Jonathan Wakely <jwakely@redhat.com>
13495 * coverage.c: Use correct type.
13496 * doc/invoke.texi: Language correction.
13498 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
13501 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
13502 ENDBRANCH for non-tail call which may return via indirect branch.
13503 * doc/extend.texi: Document indirect_return attribute.
13505 2018-07-03 Martin Sebor <msebor@redhat.com>
13507 PR tree-optimization/86274
13508 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
13510 (format_floating): Correct handling of infinities and NaNs.
13512 2018-07-03 Martin Sebor <msebor@redhat.com>
13514 * print-tree.c (print_real_cst): New function.
13515 (print_node_brief): Call it.
13516 (print_node): Ditto.
13518 2018-07-03 Jeff Law <law@redhat.com>
13520 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
13521 into a single pattern.
13523 * config/h8300/h8300.md (ors code_iterator): New.
13524 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
13525 a single pattern and single splitter.
13526 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
13527 (iorqi3_1, xorqi3_1): Likewise.
13528 (iorqi3, xorqi3 expanders): Similarly.
13530 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
13531 (movmd_internal) into a single pattern using the P mode iterator.
13532 (movmd splitters): Similarly.
13533 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
13534 (movsd splitters): Similarly.
13536 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
13537 ADDB, ADDW and ADDL into a single ADD attribute which selects the
13538 right table based on the size of the operand.
13539 * config/h8300/h8300.md (length_table): Corresponding changes. All
13540 references to "addb", "addw" and "addl" changed to "add".
13541 (btst patterns): Merge two variants into a single pattern.
13542 (tstqi, tsthi): Likewise.
13543 (addhi3_incdec, addsi3_incdec): Likewise.
13544 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
13545 (mulhi3, mulsi3): Likewise.
13546 (udivhi3, udivsi3): Likewise.
13547 (divhi3, divsi3): Likewise.
13548 (andorqi3, andorhi3, andorsi3): Likewise.
13550 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
13553 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
13554 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
13556 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13558 PR tree-optimization/85694
13559 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
13560 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
13562 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
13563 (<u>avg<mode>3_ceil): New patterns.
13565 2018-07-03 David Malcolm <dmalcolm@redhat.com>
13567 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
13568 scan-tree-dump directive.
13569 * gcc.dg/vect/slp-perm-2.c: Likewise.
13570 * gcc.dg/vect/slp-perm-3.c: Likewise.
13571 * gcc.dg/vect/slp-perm-5.c: Likewise.
13572 * gcc.dg/vect/slp-perm-6.c: Likewise.
13573 * gcc.dg/vect/slp-perm-7.c: Likewise.
13574 * gcc.dg/vect/slp-perm-8.c: Likewise.
13576 2018-07-03 Marek Polacek <polacek@redhat.com>
13578 PR middle-end/86202
13579 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
13581 2018-07-03 Richard Biener <rguenther@suse.de>
13584 * tree-ssa-structalias.c (find_func_clobbers): Properly
13585 handle indirect calls.
13587 2018-07-03 Jeff Law <law@redhat.com>
13589 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
13590 (shifts): New code iterator.
13591 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
13592 expander. Fix HImode handling on H8/SX.
13593 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
13594 (subqi3, subhi3, subsi3 expanders): Likewise.
13595 (andqi3, andhi3, andsi3 expanders): Likewise.
13596 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
13597 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
13598 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
13599 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
13600 (zero_extendqihi2, zero_extendqisi2): Likewise.
13601 (extendqihi2, extendqisi2): Likewise.
13602 (rotlqi3, rotlhi3, rotlsi3): Likewise.
13603 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
13604 (rotlqi3_1, rotlhi3_1): Likewise.
13605 (logicalhi3_sn, logicalsi3_sn): Likewise.
13606 (logicalhi3, logicalsi3): Likewise.
13608 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13610 * tree-vect-patterns.c (vect_recog_rotate_pattern)
13611 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
13612 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
13613 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
13614 type to append_pattern_def_seq instead of creating a stmt_vec_info
13616 (build_mask_conversion): Likewise. Remove vinfo argument.
13617 (vect_add_conversion_to_patterm): Likewise, renaming to...
13618 (vect_add_conversion_to_pattern): ...this.
13619 (vect_recog_mask_conversion_pattern): Update call to
13620 build_mask_conversion. Pass the vector type to
13621 append_pattern_def_seq here too.
13622 (vect_recog_gather_scatter_pattern): Update call to
13623 vect_add_conversion_to_pattern.
13625 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13627 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
13628 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13629 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
13630 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
13631 STMT_VINFO_PATTERN_DEF_SEQ to null here.
13632 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
13633 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
13634 append_pattern_def_seq instead of new_pattern_def_seq.
13635 (vect_recog_divmod_pattern): Do both of the above.
13636 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
13639 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13641 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
13642 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
13643 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
13644 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
13645 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
13646 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
13647 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
13648 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
13649 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
13650 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
13651 parameter with a single stmt_vec_info.
13652 (vect_recog_func_ptr): Likewise.
13653 (vect_recog_gather_scatter_pattern): Likewise, folding in...
13654 (vect_try_gather_scatter_pattern): ...this.
13655 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
13656 the stmt_vec_info of the statement to be matched. Don't clear
13657 STMT_VINFO_RELATED_STMT.
13658 (vect_pattern_recog): Update call accordingly.
13660 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13662 PR tree-optimization/85694
13663 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
13664 (uavgM3_ceil): Document new optabs.
13665 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
13666 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
13668 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
13669 (savg_ceil_optab): New optabs.
13670 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
13671 (vect_vect_recog_func_ptrs): Add it.
13672 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
13673 constant directly from the associated lhs.
13675 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13677 * tree-vect-patterns.c (vect_split_statement): New function.
13678 (vect_convert_input): Use it to try to split an existing cast.
13680 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13682 * poly-int.h (print_hex): New function.
13683 * dumpfile.h (dump_dec, dump_hex): Declare.
13684 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
13685 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
13686 min_input_precision, operation_precision and operation_sign.
13687 * tree-vect-patterns.c (vect_get_range_info): New function.
13688 (vect_same_loop_or_bb_p, vect_single_imm_use)
13689 (vect_operation_fits_smaller_type): Delete.
13690 (vect_look_through_possible_promotion): Add an optional
13691 single_use_p parameter.
13692 (vect_recog_over_widening_pattern): Rewrite to use new
13693 stmt_vec_info infomration. Handle one operation at a time.
13694 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
13695 (vect_truncatable_operation_p, vect_set_operation_type)
13696 (vect_set_min_input_precision): New functions.
13697 (vect_determine_min_output_precision_1): Likewise.
13698 (vect_determine_min_output_precision): Likewise.
13699 (vect_determine_precisions_from_range): Likewise.
13700 (vect_determine_precisions_from_users): Likewise.
13701 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
13702 (vect_vect_recog_func_ptrs): Put over_widening first.
13704 (vect_pattern_recog): Call vect_determine_precisions.
13706 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13708 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
13709 statements that have been replaced by further pattern statements.
13710 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
13712 2018-07-03 Richard Biener <rguenther@suse.de>
13714 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
13715 always set *dt. Dump vectype in vectype overload.
13716 * dumpfile.h (dump_gimple_expr): New function.
13717 (dump_gimple_expr_loc): Likewise.
13718 * dumpfile.c (dump_gimple_expr): New function.
13719 (dump_gimple_expr_loc): Likewise.
13721 2018-07-02 Jeff Law <law@redhat.com>
13723 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
13724 the H8/300, H8/300H and H8/S variants into a single pattern.
13725 (movhi_h8300, movqi_h8300hs): Similarly.
13726 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
13727 (QHI mode iterator): New.
13729 * config/h8300/h8300.md: Remove trailing whitespace.
13731 2018-07-02 Jim Wilson <jimw@sifive.com>
13733 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
13734 instead of emit_insn for interrupt returns.
13735 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
13736 (riscv_sret, riscv_uret): Likewise.
13738 2018-07-02 David Malcolm <dmalcolm@redhat.com>
13740 * pretty-print.c (selftest::test_pp_format): Move save and restore
13741 of quotes to class auto_fix_quotes, and add an instance.
13742 * selftest.c: Include "intl.h".
13743 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
13744 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
13745 * selftest.h (selftest::auto_fix_quotes): New class.
13747 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
13749 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
13750 (aarch64_sve_prepare_conditional_op): Remove.
13751 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
13752 Allow aarch64_simd_reg_or_zero as select operand; remove
13753 the aarch64_sve_prepare_conditional_op call.
13754 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
13755 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
13756 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
13757 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
13758 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
13759 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
13760 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
13761 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
13762 and a splitters to match all of the *_any patterns.
13763 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
13765 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
13766 (SVE_COND_FP_BINARY_REV): Remove.
13767 (sve_int_op_rev, sve_fp_op_rev): New.
13768 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
13769 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
13770 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
13771 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
13772 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
13773 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
13774 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
13775 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
13776 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
13778 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
13779 Remove match_dup 1 from the inner unspec.
13780 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
13782 * config/aarch64/aarch64.md (movprfx): New attr.
13783 (length): Default movprfx to 8.
13784 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
13785 (*madd<SVE_I>, *msub<SVE_I): Likewise.
13786 (*<su>mul<SVE_I>3_highpart): Likewise.
13787 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
13788 (*v<ASHIFT><SVE_I>3): Likewise.
13789 (*<su><MAXMIN><SVE_I>3): Likewise.
13790 (*<su><MAXMIN><SVE_F>3): Likewise.
13791 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
13792 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
13793 (*div<SVE_F>4): Likewise.
13795 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
13797 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
13800 2018-07-02 Richard Biener <rguenther@suse.de>
13802 PR tree-optimization/86363
13803 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
13804 memset argument refers to a non-variable address.
13806 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
13808 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
13809 {PLUS,MINUS}_EXPR code to...
13810 (adjust_symbolic_bound): ...here,
13811 (combine_bound): ...here,
13812 (set_value_range_with_overflow): ...and here.
13814 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
13816 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
13818 (extract_range_from_abs_expr): ...here.
13820 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
13822 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
13823 -fno-omit-frame-pointer when not optimizing.
13825 2018-07-02 Martin Liska <mliska@suse.cz>
13828 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
13829 (propagate_nothrow): Likewise.
13831 2018-07-02 Martin Liska <mliska@suse.cz>
13834 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
13836 2018-07-02 David Malcolm <dmalcolm@redhat.com>
13838 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
13839 function in r262149, changing "loc" param from source_location to
13840 const dump_location_t &.
13841 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
13842 declaration, as above.
13844 2018-07-01 Paul Koning <ni1d@arrl.net>
13846 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
13847 -munit-asm, -mgnu-asm, -mdec-asm.
13848 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
13849 (pdp11_output_labelref): New.
13850 (pdp11_output_def): New.
13851 (pdp11_output_addr_vec_elt): New.
13852 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
13853 %# and %@ format codes.
13854 (pdp11_option_override): New.
13855 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
13856 (pdp11_output_ident): New.
13857 (pdp11_asm_named_section): New.
13858 (pdp11_asm_init_sections): New.
13859 (pdp11_file_start): New.
13860 (pdp11_file_end): New.
13861 (output_ascii): Use .ascii/.asciz for -mdec-asm.
13862 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
13863 %o, like %c but octal.
13864 (pdp11_option_override): New.
13865 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
13867 (DATA_SECTION_ASM_OP): Ditto.
13868 (READONLY_DATA_SECTION_ASM_OP): New.
13869 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
13870 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
13871 (ASM_OUTPUT_LABELREF): Ditto.
13872 (ASM_OUTPUT_DEF): Ditto.
13873 (ASM_OUTPUT_EXTERNAL): New.
13874 (ASM_OUTPUT_SOURCE_FILENAME): New.
13875 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
13876 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
13877 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
13878 %# and %@ format codes.
13879 * config/pdp11/pdp11.opt (mgnu-asm): New.
13880 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
13881 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
13882 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
13884 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
13886 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
13887 dereferencing path[] beyond its length.
13889 (debug_all_paths): New.
13890 (rewire_first_differing_edge): New.
13891 (adjust_paths_after_duplication): New.
13892 (duplicate_thread_path): Call adjust_paths_after_duplication.
13894 (thread_through_all_blocks): Add new argument to
13895 duplicate_thread_path.
13897 2018-06-30 Jim Wilson <jimw@sifive.com>
13899 * config/riscv/predicates.md (p2m1_shift_operand): New.
13900 (high_mask_shift_operand): New.
13901 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
13902 pattern using p2m1_shift_operand.
13903 (lshsi3_zero_extend_3+2): New combiner pattern using
13904 high_mask_shift_operand.
13906 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13908 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
13910 (vect_recog_rotate_pattern): ...here.
13911 (vect_convert_input): Try to insert casts of invariants in the
13913 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
13914 preheader to be empty.
13916 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13918 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
13919 vector type. If given, install it in the new statement's
13920 STMT_VINFO_VECTYPE.
13921 (vect_element_precision): New function.
13922 (vect_unpromoted_value): New struct.
13923 (vect_unpromoted_value::vect_unpromoted_value): New function.
13924 (vect_unpromoted_value::set_op): Likewise.
13925 (vect_look_through_possible_promotion): Likewise.
13926 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
13927 (vect_widened_op_tree, vect_convert_input): Likewise.
13928 (vect_convert_inputs, vect_convert_output): Likewise.
13929 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
13930 to handle the optional cast of the multiplication result and
13931 vect_widened_op_tree to detect the widened multiplication itself.
13932 Do not require the input and output of promotion casts to have
13933 the same sign, but base the signedness of the operation on the
13934 input rather than the result. If the pattern includes two
13935 promotions, check that those promotions have the same sign.
13936 Do not restrict the MULT_EXPR handling to a double-width result;
13937 handle quadruple-width results and wider. Use vect_convert_inputs
13938 to convert the inputs to the common type.
13939 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
13940 to handle the optional cast of the ABS result. Also allow a sign
13941 change or a sign extension between the ABS and MINUS.
13942 Use vect_widened_op_tree to detect the widened subtraction and use
13943 vect_convert_inputs to convert the inputs to the common type.
13944 (vect_handle_widen_op_by_const): Delete.
13945 (vect_recog_widen_op_pattern): New function.
13946 (vect_recog_widen_mult_pattern): Use it.
13947 (vect_recog_widen_shift_pattern): Likewise.
13948 (vect_recog_widen_sum_pattern): Use
13949 vect_look_through_possible_promotion to handle the promoted
13952 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13954 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
13955 the containing gimple_seq *.
13956 * gimple-iterator.h (gsi_for_stmt): Declare it.
13957 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
13958 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
13959 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
13960 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
13961 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
13963 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
13965 (vect_mark_pattern_stmts): ...here. Handle cases in which the
13966 statement being replaced is part of an existing pattern
13967 definition sequence, inserting the new pattern statements before
13969 (vect_pattern_recog_1): Don't return a bool. If the statement
13970 is already part of a pattern, instead apply pattern matching
13971 to the pattern definition statements. Don't clear the
13972 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
13973 (vect_pattern_recog): Don't break after the first match;
13974 continue processing the pattern definition statements instead.
13975 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
13977 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13979 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
13980 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13981 (vect_recog_widen_sum_pattern): Use it.
13983 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13985 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13986 phi is not a pattern statement and has not been replaced by
13987 a pattern statement.
13988 * tree-vect-patterns.c (type_conversion_p): Don't check
13989 STMT_VINFO_IN_PATTERN_P.
13990 (vect_recog_vector_vector_shift_pattern): Likewise.
13991 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
13992 the pattern statement rather than the original statement; check
13993 directly for a WIDEN_MULT_EXPR here.
13994 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
13995 vect_is_simple_use to return the pattern statement rather
13996 than the original statement; use is_pattern_stmt_p to check
13997 for such a pattern statement.
13998 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
13999 to return the pattern statement rather than the original statement;
14000 don't do the same transformation here.
14001 (vect_is_simple_use): If the defining statement has been replaced
14002 by a pattern statement, return the pattern statement instead.
14003 Remove the corresponding (local) transformation from the vectype
14006 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14008 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
14009 end and default to null.
14010 * tree-vect-loop.c (vect_create_epilog_for_reduction)
14011 (vectorizable_reduction): Update calls accordingly, dropping the
14012 gimple ** argument if the passed-back statement isn't needed.
14013 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
14014 (vect_recog_rotate_pattern): Likewise.
14015 (vect_recog_mask_conversion_pattern): Likewise.
14016 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
14017 (vect_mask_constant_operand_p): Likewise.
14018 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
14019 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
14020 (get_group_load_store_type, get_load_store_type): Likewise.
14021 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
14022 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
14023 (vectorizable_conversion, vectorizable_assignment): Likewise.
14024 (vectorizable_shift, vectorizable_operation): Likewise.
14025 (vectorizable_store, vect_is_simple_cond): Likewise.
14026 (vectorizable_condition, vectorizable_comparison): Likewise.
14027 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
14028 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
14029 and move it to the end. Cope with null def_stmt_outs.
14031 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
14033 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
14035 2018-06-29 Jeff Law <law@redhat.com>
14037 * config/v850/v850.c (v850_legitimate_address_p): Handle large
14038 displacements for TARGET_V850E2V3 and newer.
14039 (TARGET_LRA_P): Remove. Defaults to LRA now.
14040 * config/v850/v850.md (sign23byte_load): Remove.
14041 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
14042 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
14044 2018-06-29 Martin Liska <mliska@suse.cz>
14047 * coverage.c (coverage_init): Mangle full path name.
14048 * doc/invoke.texi: Document the change.
14049 * gcov-io.c (mangle_path): New.
14050 * gcov-io.h (mangle_path): Likewise.
14051 * gcov.c (mangle_name): Use mangle_path for path mangling.
14053 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14055 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
14056 if starting source register is not even.
14058 2018-06-29 Martin Liska <mliska@suse.cz>
14060 PR tree-optimization/86263
14061 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
14062 Make edge redirection.
14064 2018-06-29 David Malcolm <dmalcolm@redhat.com>
14066 * dumpfile.c (dump_loc): Add indentation based on scope depth.
14067 (dump_scope_depth): New variable.
14068 (get_dump_scope_depth): New function.
14069 (dump_begin_scope): New function.
14070 (dump_end_scope): New function.
14071 * dumpfile.h (get_dump_scope_depth): New declaration.
14072 (dump_begin_scope): New declaration.
14073 (dump_end_scope): New declaration.
14074 (class auto_dump_scope): New class.
14075 (AUTO_DUMP_SCOPE): New macro.
14076 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
14079 2018-06-29 Richard Biener <rguenther@suse.de>
14081 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
14082 compute_all_dependences succeeds.
14083 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
14084 exceed --param loop-max-datarefs-for-datadeps.
14086 2018-06-29 Jakub Jelinek <jakub@redhat.com>
14088 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
14090 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
14093 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
14094 alternative 0 in preferred_for_speed attribute.
14096 2018-06-28 Paul Koning <ni1d@arrl.net>
14098 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
14099 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
14100 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
14101 * config/pdp11/pdp11.md: Correct "length" attribute calculation
14102 for shift insn patterns.
14104 2018-06-28 David Malcolm <dmalcolm@redhat.com>
14106 * cgraph.c (cgraph_node::get_body): Replace assignments to
14107 "dump_file" with calls to set_dump_file.
14108 * dumpfile.c (alt_dump_file): Make static, and group with...
14109 (alt_flags): ...this definition.
14110 (dumps_are_enabled): New variable.
14111 (refresh_dumps_are_enabled): New function.
14112 (set_dump_file): New function.
14113 (set_alt_dump_file): New function.
14114 (gcc::dump_manager::dump_start): Replace assignments to
14115 "dump_file" and "alt_dump_file" with calls to set_dump_file and
14117 (gcc::dump_manager::dump_finish): Likewise.
14118 * dumpfile.h (alt_dump_file): Delete decl.
14119 (dumps_are_enabled): New variable decl.
14120 (set_dump_file): New function decl.
14121 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
14123 * tree-nested.c (lower_nested_functions): Replace assignments to
14124 "dump_file" with calls to set_dump_file.
14126 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
14128 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
14129 goto_locus of each outgoing edge of each basic block.
14131 2018-06-28 Richard Biener <rguenther@suse.de>
14133 * dwarf2out.c (decl_scope_table): Remove.
14134 (push_decl_scope): Likewise.
14135 (pop_decl_scope): Likewise.
14136 (gen_type_die_for_member): Do not call push/pop_decl_scope.
14137 (gen_struct_or_union_type_die): Likewise.
14138 (gen_tagged_type_die): Likewise.
14139 (dwarf2out_init): Do not initialize decl_scope_table.
14140 (dwarf2out_c_finalize): Do not free it.
14142 2018-06-28 Richard Biener <rguenther@suse.de>
14144 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
14145 deciding whether to not re-use a DIE.
14147 2018-06-28 Richard Biener <rguenther@suse.de>
14149 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
14150 DW_AT_abstract_origin attribute.
14152 2018-06-28 Martin Liska <mliska@suse.cz>
14154 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
14155 Use newly introduced constants.
14156 * tree-switch-conversion.h (struct jump_table_cluster):
14157 Define max_ratio_for_size and max_ratio_for_speed.
14159 2018-06-28 Martin Liska <mliska@suse.cz>
14161 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14162 Add new checking assert to catch invalid state.
14163 (jump_table_cluster::can_be_handled): Handle single case
14165 (jump_table_cluster::is_beneficial): Bail out for such case.
14166 (bit_test_cluster::find_bit_tests):
14167 Add new checking assert to catch invalid state.
14168 (bit_test_cluster::can_be_handled): Handle single case
14170 (bit_test_cluster::is_beneficial): Bail out for such case.
14171 (switch_decision_tree::analyze_switch_statement):
14174 2018-06-28 Martin Liska <mliska@suse.cz>
14176 * common.opt: Introduce -completion option.
14177 * gcc.c (driver_handle_option): Handle it.
14178 (driver::main): Print completions if completion
14180 * opt-suggestions.c (option_proposer::get_completions):
14182 (option_proposer::suggest_completion): Likewise.
14183 (option_proposer::find_param_completions): Likewise.
14184 (verify_autocompletions): Likewise.
14185 (test_completion_valid_options): Likewise.
14186 (test_completion_valid_params): Likewise.
14187 (in_completion_p): Likewise.
14188 (empty_completion_p): Likewise.
14189 (test_completion_partial_match): Likewise.
14190 (test_completion_garbage): Likewise.
14191 (opt_proposer_c_tests): Likewise.
14192 * opt-suggestions.h: Declare new functions.
14193 * opts.c (common_handle_option): Handle OPT__completion_.
14194 * selftest-run-tests.c (selftest::run_tests): Add
14195 opt_proposer_c_tests.
14196 * selftest.c (assert_str_startswith): New.
14197 * selftest.h (assert_str_startswith): Likewise.
14198 (opt_proposer_c_tests): New.
14199 (ASSERT_STR_STARTSWITH): Likewise.
14201 2018-06-28 Martin Liska <mliska@suse.cz>
14203 * Makefile.in: Add opt-suggestions.o.
14204 * gcc-main.c: Include opt-suggestions.h.
14205 * gcc.c (driver::driver): Likewise.
14206 (driver::~driver): Remove m_option_suggestions.
14207 (driver::build_option_suggestions): Moved to option_proposer.
14208 (driver::suggest_option): Likewise.
14209 (driver::handle_unrecognized_options): Use option_proposer.
14210 * gcc.h (class driver): Add new memver m_option_proposer.
14211 * opt-suggestions.c: New file.
14212 * opt-suggestions.h: New file.
14214 2018-06-28 Martin Liska <mliska@suse.cz>
14216 * vec.h (class auto_string_vec): New (moved from auto_argvec).
14217 (auto_string_vec::~auto_string_vec): Likewise.
14219 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
14221 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
14222 prevent_decl_creation_for_types fields up and add reset_location field.
14223 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
14224 statement if id->reset_location is true.
14225 (copy_edges_for_bb): Do not set goto_locus on the new edges if
14226 id->reset_location is true.
14227 (copy_phis_for_bb): Force input_location on the arguments if
14228 id->reset_location is true.
14229 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
14230 is set on the function to be inlined.
14232 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
14234 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
14236 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
14238 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
14239 registers for Pmode.
14240 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
14241 hard registers for the clobbered pseudo.
14243 2018-06-27 Paul Koning <ni1d@arrl.net>
14245 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
14246 mutually exclusive options.
14247 * config/pdp11/constraints.md (h): New constraint.
14248 (O): Update definition to match shift code generation.
14249 (D): New constraint.
14250 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
14252 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
14254 (output_jump): Change arguments.
14255 (pdp11_fixed_cc_regs): New function.
14256 (pdp11_cc_mode): Ditto.
14257 (pdp11_expand_shift): Ditto.
14258 (pdp11_assemble_shift): Ditto.
14259 (pdp11_small_shift): Ditto.
14260 (pdp11_branch_cost): Remove.
14261 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
14263 (pdp11_register_move_cost): Update for CC registers.
14264 (pdp11_rtx_costs): Add case for LSHIFTRT.
14265 (pdp11_output_jump): Add CCNZ mode conditional branches.
14266 (notice_update_cc_on_set): Remove.
14267 (pdp11_cc_mode): New function.
14268 (simple_memory_operand): Correct pre/post decrement case.
14269 (no_side_effect_operand): New function.
14270 (pdp11_regno_reg_class): Add CC_REGS class.
14271 (pdp11_fixed_cc_regs): New function.
14272 (pdp11_small_shift): New function.
14273 (pdp11_expand_shift): New function to expand shift insns.
14274 (pdp11_assemble_shift): New function to output shifts.
14275 (pdp11_branch_cost): Remove.
14276 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
14277 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
14278 (WCHAR_TYPE): Ditto.
14279 (PTRDIFF_TYPE): Ditto.
14280 (ADJUST_INSN_LENGTH): New macro.
14281 (FIXED_REGISTERS): Add CC registers.
14282 (CALL_USED_REGISTERS): Ditto.
14283 (reg_class): Ditto.
14284 (REG_CLASS_NAMES): Ditto.
14285 (REG_CLASS_CONTENTS): Ditto.
14286 (SELECT_CC_MODE): Use new function.
14287 (TARGET_FLAGS_REGNUM): New macro.
14288 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
14289 (cc0_reg_rtx): Remove.
14290 (CC_STATUS_MDEP): Remove.
14291 (CC_STATUS_MDEFP_INIT): Remove.
14292 (CC_IN_FPU): Remove.
14293 (NOTICE_UPDATE_CC): Remove.
14294 (REGISTER_NAMES): Add CC registers.
14295 (BRANCH_COST): Change to constant 1.
14296 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
14298 * config/pdp11/pdp11.opt (mbcopy): Remove.
14299 (mbcopy-builtin): Remove.
14300 (mbranch-cheap): Remove.
14301 (mbranch-expensive): Remove.
14302 * config/pdp11/predicates.md (expand_shift_operand): Update to
14303 match shift code generation.
14304 (ccnz_operator): New predicate.
14305 * doc/invoke.texi (PDP-11 Options): Remove deleted options
14306 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
14307 Remove non-existent option -mabshi, -mno-abshi. Document mutually
14309 * doc/md.texi (PDP-11): Document new D and h constraints. Update
14310 description of O constraint.
14312 2018-06-27 Jeff Law <law@redhat.com>
14313 Austin Law <austinklaw@gmail.com>
14315 * config/v850/v850.md (addsi3_set_flags): New pattern.
14316 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
14317 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
14318 (zero_extendhisi2_v850_set_flags): Likewise.
14319 (zero_extendqisi2_v850_set_flags): Likewise.
14320 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
14321 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
14322 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
14324 * config/v850/v850-protos.h (notice_update_cc): Remove.
14325 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
14326 (v850_print_operand): Handle 'D' and "d".
14327 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
14328 Add handling of arithmetic/logical operations compared against zero.
14329 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
14330 Do not look at v850_compare_op, instead get mode from last argument.
14331 (v850_gen_compare): Remove
14332 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
14333 after reload for prologue insns.
14334 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
14336 (construct_save_jarl): Likewise.
14337 (TARGET_FLAGS_REGNUM): Define.
14338 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
14339 (NOTICE_UPDATE_CC): Remove.
14340 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
14341 than cc0. Conditionalize on reload_completed.
14342 (cmpsi_insn, setfcc_insn): Likewise.
14343 (tst1 splitter): Turn into define_and_split which sets the flags
14345 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
14346 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
14347 (cstoresf4, cstoredf4): Clobber the flags.
14348 (cmpsi, cmpsf, cmpdf): Remove expanders.
14349 (setf_insn): Remove pattern.
14350 (addsi3): Turn into define_and_split which clobbers the flags after
14351 reload and a suitable pattern (addsi3_clobber_flags) for use after
14353 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
14354 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
14355 (ashrsi3, ashrsi3_v850e2): Likewise.
14356 (bins): Clobber the flags.
14357 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
14358 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
14359 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
14360 (call_value_internal_short, call_value_internal_long): Likewise.
14361 (callt_save_interrupt, callt_return_interrupt): Likewise.
14362 (save_interrupt, return_interrupt): Likewise.
14363 (callt_save_all_interrupt, save_all_interrupt): Likewise.
14364 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
14365 (restore_all_interrupt, _restore_all_interrupt): Likewise.
14366 (All FP comparisons): Only allow after reload has completed.
14368 (divh, divhu): Tweak output template.
14369 (branch_z_normal, branch_z_invert): Remove
14370 (branch_nz_normal, branch_nz_invert): Likewise.
14371 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
14373 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
14374 * config/v850/v850.c (notice_update_cc): Remove.
14375 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
14376 (CC_NO_CARRY): Likewise.
14377 (NOTICE_UPDATE_CC): Define to nothing.
14378 * config/v850/v850.md: Remove block comment on cc0 handling
14379 Remove "cc" attribute from all patterns. Remove cc_status handling
14380 from all patterns. Minor formatting fixes.
14382 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14384 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
14385 (cortex-a76.cortex-a55): Likewise.
14386 * config/aarch64/aarch64-tune.md: Regenerate.
14387 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
14388 cortex-a76.cortex-a55.
14390 2018-06-27 Jeff Law <law@redhat.com>
14392 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
14393 (MULTILIB_DIRNAMES): Similarly.
14395 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
14397 * gimple.h (gimple_return_retbnd): Delete.
14398 (gimple_return_set_retbnd): Likewise.
14399 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
14400 gimple_return_set_retbnd.
14401 * gimple-pretty-print.c (dump_gimple_return): Remove call to
14402 gimple_return_retbnd and adjust.
14403 * tree-inline.h (struct copy_body_data): Remove retbnd field.
14404 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
14405 Explicitly return NULL in a couple more cases. Move assertion
14406 on debug statements and remove unreachable code.
14407 (reset_debug_binding): Do not test id->retbnd.
14408 (expand_call_inline): Do not set it.
14410 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
14412 * configure.ac: Add --disable-gcov option.
14413 * configure: Regenerate.
14414 * Makefile.in: Honour @enable_gcov@.
14415 * doc/install.texi: Document --disable-gcov.
14417 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14419 * config/arm/arm-cpus.in (cortex-a76): New entry.
14420 (cortex-a76.cortex-a55): Likewise.
14421 * config/arm/arm-tables.opt: Regenerate.
14422 * config/arm/arm-tune.md: Likewise.
14423 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
14424 * doc/invoke.texi (ARM Options): Document cortex-a76 and
14425 cortex-a76.cortex-a55.
14427 2018-06-27 Tamar Christina <tamar.christina@arm.com>
14430 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
14432 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
14434 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
14436 (EPILOGUE_USES): Likewise.
14438 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
14440 * tree-inline.c (remap_location): New function extracted from...
14441 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
14442 (copy_phis_for_bb): ...here. Call remap_location.
14443 (copy_cfg_body): Adjust call to copy_edges_for_bb.
14445 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
14447 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
14448 unaligned vsx for 16B memset.
14450 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
14453 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
14454 ieee128_float_type_node to long_double_type_node unless
14455 TARGET_LONG_DOUBLE_128 is set.
14457 2018-06-26 David Malcolm <dmalcolm@redhat.com>
14459 * cfgloop.c (get_loop_location): Convert return type from
14460 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
14461 by implicit construction from rtx_insn *, and using
14462 dump_user_location_t::from_function_decl for the fallback case.
14463 * cfgloop.h (get_loop_location): Convert return type from
14464 location_t to dump_user_location_t.
14465 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
14466 dump_printf_loc to pass in a dump_location_t rather than a
14467 location_t, via the gimple stmt.
14468 * coverage.c (get_coverage_counts): Update calls to
14469 dump_printf_loc to pass in dump_location_t rather than a
14471 * doc/optinfo.texi (Dump types): Convert example of
14472 dump_printf_loc from taking "locus" to taking "insn". Update
14473 description of the "_loc" calls to cover dump_location_t.
14474 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
14476 (dump_user_location_t::dump_user_location_t): New constructors,
14477 from gimple *stmt and rtx_insn *.
14478 (dump_user_location_t::from_function_decl): New function.
14479 (dump_loc): Make static.
14480 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
14481 const dump_location_t &.
14482 (dump_generic_expr_loc): Delete.
14483 (dump_printf_loc): Convert param "loc" from location_t to
14484 const dump_location_t &.
14485 (selftest::test_impl_location): New function.
14486 (selftest::dumpfile_c_tests): New function.
14487 * dumpfile.h: Include "profile-count.h".
14488 (class dump_user_location_t): New class.
14489 (struct dump_impl_location_t): New struct.
14490 (class dump_location_t): New class.
14491 (dump_printf_loc): Convert 2nd param from source_location to
14492 const dump_location_t &.
14493 (dump_generic_expr_loc): Delete.
14494 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
14495 const dump_location_t &.
14496 * gimple-fold.c (fold_gimple_assign): Update call to
14497 dump_printf_loc to pass in a dump_location_t rather than a
14498 location_t, via the gimple stmt.
14499 (gimple_fold_call): Likewise.
14500 * gimple-loop-interchange.cc
14501 (loop_cand::analyze_iloop_reduction_var): Update for change to
14502 check_reduction_path.
14503 (tree_loop_interchange::interchange): Update for change to
14504 find_loop_location.
14505 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
14506 change in return-type of find_loop_location.
14507 (graphite_regenerate_ast_isl): Likewise.
14508 * graphite-optimize-isl.c (optimize_isl): Likewise.
14509 * graphite.c (graphite_transform_loops): Likewise.
14510 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
14511 pass in a dump_location_t rather than a location_t, via the
14513 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14514 * ipa.c (walk_polymorphic_call_targets): Likewise.
14515 * loop-unroll.c (report_unroll): Convert "locus" param from
14516 location_t to dump_location_t.
14517 (decide_unrolling): Update for change to get_loop_location's
14519 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
14520 location_t to dump_user_location_t.
14521 (grid_find_single_omp_among_assignments_1): Updates calls to
14522 dump_printf_loc to pass in a dump_location_t rather than a
14523 location_t, via the gimple stmt.
14524 (grid_parallel_clauses_gridifiable): Convert "tloc" from
14525 location_t to dump_location_t. Updates calls to dump_printf_loc
14526 to pass in a dump_location_t rather than a location_t, via the
14528 (grid_inner_loop_gridifiable_p): Likewise.
14529 (grid_dist_follows_simple_pattern): Likewise.
14530 (grid_gfor_follows_tiling_pattern): Likewise.
14531 (grid_target_follows_gridifiable_pattern): Likewise.
14532 (grid_attempt_target_gridification): Convert initialization
14533 of local "grid" from memset to zero-initialization; FIXME: does
14534 this require C++11? Update call to dump_printf_loc to pass in a
14535 optinfo_location rather than a location_t, via the gimple stmt.
14536 * profile.c (read_profile_edge_counts): Updates call to
14537 dump_printf_loc to pass in a dump_location_t rather than a
14539 (compute_branch_probabilities): Likewise.
14540 * selftest-run-tests.c (selftest::run_tests): Call
14542 * selftest.h (dumpfile_c_tests): New decl.
14543 * tree-loop-distribution.c (pass_loop_distribution::execute):
14544 Update for change in return type of find_loop_location.
14545 * tree-parloops.c (parallelize_loops): Likewise.
14546 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
14547 "locus" from location_t to dump_user_location_t.
14548 (canonicalize_loop_induction_variables): Likewise.
14549 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
14550 for change in return type of find_loop_location.
14551 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
14552 to dump_printf_loc to pass in a dump_location_t rather than a
14553 location_t, via the stmt.
14554 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
14556 * tree-vect-loop-manip.c (find_loop_location): Convert return
14557 type from source_location to dump_user_location_t.
14558 (vect_do_peeling): Update for above change.
14559 (vect_loop_versioning): Update for change in type of
14561 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
14562 from location_t to dump_user_location_t.
14563 (vect_estimate_min_profitable_iters): Update for change in type
14565 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
14566 location_t to dump_location_t.
14567 (vect_slp_bb): Update for change in type of vect_location.
14568 * tree-vectorizer.c (vect_location): Convert from source_location
14569 to dump_user_location_t.
14570 (try_vectorize_loop_1): Update for change in vect_location's type.
14571 (vectorize_loops): Likewise.
14572 (increase_alignment): Likewise.
14573 * tree-vectorizer.h (vect_location): Convert from source_location
14574 to dump_user_location_t.
14575 (find_loop_location): Convert return type from source_location to
14576 dump_user_location_t.
14577 (check_reduction_path): Convert 1st param from location_t to
14578 dump_user_location_t.
14579 * value-prof.c (check_counter): Update call to dump_printf_loc to
14580 pass in a dump_user_location_t rather than a location_t; update
14581 call to error_at for change in type of "locus".
14582 (check_ic_target): Update call to dump_printf_loc to
14583 pass in a dump_user_location_t rather than a location_t, via the
14586 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
14588 * config/s390/s390.h (enum processor_flags): Do not use
14589 default tune parameter when -march was specified.
14591 2018-06-26 Jakub Jelinek <jakub@redhat.com>
14594 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
14595 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
14598 2018-06-26 Richard Biener <rguenther@suse.de>
14600 PR tree-optimization/86287
14602 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
14603 (vect_analyze_loop): Initialize n_stmts.
14605 2018-06-26 Richard Biener <rguenther@suse.de>
14607 PR middle-end/86271
14608 * fold-const.c (fold_convertible_p): Pointer extension
14611 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
14614 * dwarf2out.c (loc_list_has_views): Adjust comments.
14615 (dw_loc_list): Split single cross-partition range with
14618 2018-06-25 Jeff Law <law@redhat.com>
14620 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
14621 on -mbig-switch by default.
14623 * config/v850/predicates.md (const_float_1_operand): Fix match_code
14625 (const_float_0_operand): Remove unused predicate.
14626 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
14627 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
14628 (recipsf2): New expander. Original pattern now called
14630 (recipdf2, recipdf2_insn): Similarly.
14631 (rsqrtsf2, rsqrtsf2_insn): Similarly
14632 (rsqrtdf2, rsqrtdf2_insn): Similarly
14634 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
14636 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
14637 Simplify logic for FreeBSD (twice).
14639 2018-06-25 Martin Sebor <msebor@redhat.com>
14641 PR tree-optimization/86204
14642 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
14643 a strnlen result if it's less than the length of the string.
14645 2018-06-25 Martin Sebor <msebor@redhat.com>
14647 PR tree-optimization/85700
14648 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
14649 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
14650 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
14652 2018-06-25 Martin Sebor <msebor@redhat.com>
14654 * doc/extend.texi (Zero-length arrays): Update and clarify.
14656 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
14658 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
14659 added IEEE/IBM long double multilib support on PowerPC little
14660 endian Linux systems.
14661 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
14662 (MULTILIB_DEFAULTS): Likewise.
14663 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14665 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
14666 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
14667 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
14669 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
14671 PR middle-end/86311
14672 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
14673 (REORDER_45): Likewise.
14675 2018-06-25 Jeff Law <law@redhat.com>
14677 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
14678 dividend to 32 bits. Adjust length.
14679 (udivmodhi4): Cleanup output template. Fix length.
14681 2018-06-25 Carl Love <cel@us.ibm.com>
14683 * config/rs6000/vsx.md: Change word selector to prefered location.
14685 2018-06-25 Richard Biener <rguenther@suse.de>
14687 PR tree-optimization/86304
14688 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
14689 epilogue-if-converted loops as well.
14691 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
14693 * lto-section-out.c (lto_begin_section): Do not print section
14694 name for noaddr and unnumbered dumps.
14696 2018-06-25 Richard Biener <rguenther@suse.de>
14698 * tree-vectorizer.h (struct vec_info_shared): New structure
14699 with parts split out from struct vec_info and loop_nest from
14700 struct _loop_vec_info.
14701 (struct vec_info): Adjust accordingly.
14702 (struct _loop_vec_info): Likewise.
14703 (LOOP_VINFO_LOOP_NEST): Adjust.
14704 (LOOP_VINFO_DATAREFS): Likewise.
14705 (LOOP_VINFO_DDRS): Likewise.
14706 (struct _bb_vec_info): Likewise.
14707 (BB_VINFO_DATAREFS): Likewise.
14708 (BB_VINFO_DDRS): Likewise.
14709 (struct _stmt_vec_info): Add dr_aux member.
14710 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
14711 (DR_MISALIGNMENT_UNINITIALIZED): New.
14712 (set_dr_misalignment): Adjust.
14713 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
14714 (vect_analyze_loop): Adjust prototype.
14715 (vect_analyze_loop_form): Likewise.
14716 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14717 Compute dependences lazily.
14718 (vect_record_base_alignments): Use shared datarefs/ddrs.
14719 (vect_verify_datarefs_alignment): Likewise.
14720 (vect_analyze_data_refs_alignment): Likewise.
14721 (vect_analyze_data_ref_accesses): Likewise.
14722 (vect_analyze_data_refs): Likewise.
14723 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
14724 constructor parameter for shared part.
14725 (vect_analyze_loop_form): Pass in shared part and adjust.
14726 (vect_analyze_loop_2): Pass in storage for the number of
14727 stmts. Move loop nest finding to the caller. Compute
14729 (vect_analyze_loop): Pass in shared part.
14730 (vect_transform_loop): Verify shared datarefs are unchanged.
14731 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
14732 constructor parameter for shared part.
14733 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
14734 (vect_slp_bb): Verify shared datarefs are unchanged before
14736 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
14738 (new_stmt_vec_info): Initialize DR_AUX misalignment to
14739 DR_MISALIGNMENT_UNINITIALIZED.
14740 * tree-vectorizer.c (vec_info::vec_info): Add constructor
14741 parameter for shared part.
14742 (vec_info::~vec_info): Adjust.
14743 (vec_info_shared::vec_info_shared): New.
14744 (vec_info_shared::~vec_info_shared): Likewise.
14745 (vec_info_shared::save_datarefs): Likewise.
14746 (vec_info_shared::check_datarefs): Likewise.
14747 (try_vectorize_loop_1): Construct shared part live for analyses
14748 of a single loop for multiple vector sizes.
14749 * tree-parloops.c (gather_scalar_reductions): Adjust.
14751 2018-06-25 Richard Biener <rguenther@suse.de>
14753 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
14754 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
14755 (vect_analyze_data_refs): Remove similar code from here and
14756 simplify accordingly.
14758 2018-06-25 Richard Biener <rguenther@suse.de>
14760 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
14761 for reverse storage order accesses rather than asserting
14762 they cannot happen here.
14764 2018-06-25 Tom de Vries <tdevries@suse.de>
14767 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
14768 Use data16 instead of .byte for insn prefix.
14770 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
14773 * parser.c (make_char_string_pack): Pass this literal chars
14774 through cpp_interpret_string.
14775 (cp_parser_userdef_numeric_literal): Check the result of
14776 make_char_string_pack.
14778 2018-06-24 Maya Rashish <coypu@sdf.org>
14780 * ginclude/stddef.h: Simplify conditions around avoiding
14781 re-definition of __size_t.
14783 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14785 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
14786 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
14788 2018-06-22 Maya Rashish <coypu@sdf.org>
14790 * doc/invoke.texi (mno-fancy-math-387): Update for changes
14791 made to OpenBSD and NetBSD through the years.
14793 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
14795 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
14796 behavior of vec_pack (vector double, vector double) to match
14797 behavior of vec_float2 (vector double, vector double).
14799 2018-06-22 Olivier Hainque <hainque@adacore.com>
14801 * gimplify.c (gimplify_function_tree): Prevent creation
14802 of a trampoline for the address of the current function
14803 passed to entry/exit instrumentation hooks.
14805 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
14808 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
14811 2018-06-22 Martin Liska <mliska@suse.cz>
14813 PR tree-optimization/86263
14814 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14815 Bail out if is_enabled is false.
14816 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
14818 (jump_table_cluster::is_enabled): New function.
14820 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14822 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
14823 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
14824 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
14825 (lto_input_ts_binfo_tree_pointers): Likewise.
14826 * tree-streamer-out.c (streamer_write_tree_bitfields,
14827 write_ts_binfo_tree_pointers): Likewise.
14828 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
14830 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14832 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
14834 2018-06-22 Martin Liska <mliska@suse.cz>
14836 * symbol-summary.h (get): Make it pure and inline move
14837 functionality from ::get function.
14838 (get): Remove and inline into ::get and ::get_create.
14839 (get_create): Move code from ::get function.
14841 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14844 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
14845 -x assembler-with-cpp.
14847 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14849 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
14850 _FILE_OFFSET_BITS=64 for C++.
14852 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
14854 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
14855 conversion insn that shows up when pr85657-3.c is compiled using
14856 IEEE 128-bit long double.
14857 (neg<mode>2_internal): Use the correct mode to check whether the
14858 mode is IBM extended.
14859 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
14860 multiply and divide external functions from being created more
14863 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14865 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
14867 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
14868 the edge can be forwarded.
14869 (cfg_layout_merge_blocks): Likewise.
14871 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14873 * except.c (finish_eh_generation): Commit edge insertions only after
14874 the EH edges have been redirected from post-landing to landing pads.
14876 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14878 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
14879 create_tmp_var_for to create the FRAME decl.
14880 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
14882 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14884 * tree-inline.c (copy_edges_for_bb): Minor tweak.
14885 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
14886 debug statement when resetting its value.
14887 (expand_call_inline): Copy the locus of the call onto the assignment
14888 of the return value, if any. Use local variable in more cases.
14890 2018-06-21 Martin Liska <mliska@suse.cz>
14892 * ipa-pure-const.c (propagate_nothrow): Use
14893 funct_state_summaries->get.
14894 (dump_malloc_lattice): Likewise.
14895 (propagate_malloc): Likewise.
14897 2018-06-21 Richard Biener <rguenther@suse.de>
14899 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
14900 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
14901 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
14903 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
14904 BLOCK_ABSTRACT_ORIGIN unconditionally.
14906 2018-06-21 David Malcolm <dmalcolm@redhat.com>
14908 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
14910 * ipa-reference.c (ipa_reference_c_finalize): Delete
14911 ipa_ref_opt_sum_summaries and set it to NULL.
14913 2018-06-21 Tom de Vries <tdevries@suse.de>
14915 PR tree-optimization/85859
14916 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
14917 test with comment from bb_no_side_effects_p.
14919 2018-06-21 Richard Biener <rguenther@suse.de>
14921 PR tree-optimization/86232
14922 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
14923 max for constant niter.
14925 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
14927 * config/aarch64/aarch64-simd.md
14928 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
14930 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
14932 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
14933 Make opernads of the unspec commutative.
14935 2018-06-21 Richard Biener <rguenther@suse.de>
14937 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
14938 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14939 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
14940 (vect_analyze_data_ref_dependence): Re-order checks to deal with
14942 (vect_record_base_alignments): Do not record base alignment
14943 for gathers or scatters.
14944 (vect_compute_data_ref_alignment): Drop return value that is always
14945 true. Bail out early for gathers or scatters.
14946 (vect_enhance_data_refs_alignment): Bail out early for gathers
14948 (vect_find_same_alignment_drs): Likewise.
14949 (vect_analyze_data_refs_alignment): Remove dead code.
14950 (vect_slp_analyze_and_verify_node_alignment): Likewise.
14951 (vect_analyze_data_refs): For possible gathers or scatters do
14952 not create an alternate DR, just check their possible validity
14953 and mark them. Adjust DECL_NONALIASED handling to not rely
14954 on DR_BASE_ADDRESS.
14955 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
14956 update inits of gathers or scatters.
14957 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
14958 Also copy gather/scatter flag to pattern vinfo.
14960 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
14962 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
14963 behavior of vec_packsu (vector unsigned long long, vector unsigned
14964 long long) to match behavior of vec_packs with same signature.
14966 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
14967 Thomas Schwinge <thomas@codesourcery.com>
14968 Cesar Philippidis <cesar@codesourcery.com>
14970 * gimplify.c (gimplify_scan_omp_clauses): Add support for
14971 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
14972 (gimplify_adjust_omp_clauses): Likewise.
14973 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
14974 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
14975 (gimplify_omp_target_update): Update handling of acc update and
14977 * omp-low.c (install_var_field): Remove unused parameter
14978 base_pointers_restrict.
14979 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
14980 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
14982 (omp_target_base_pointers_restrict_p): Delete.
14983 (scan_omp_target): Update call to scan_sharing_clauses.
14984 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
14986 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
14987 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
14988 (convert_local_omp_clauses): Likewise.
14989 * tree-pretty-print.c (dump_omp_clause): Likewise.
14990 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
14992 (omp_clause_code_name): Likewise.
14994 2018-06-20 Jakub Jelinek <jakub@redhat.com>
14997 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
15000 PR tree-optimization/86231
15001 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
15002 anti-range don't overwrite *vr0min before using it to compute *vr0max.
15004 2018-06-20 Tom de Vries <tdevries@suse.de>
15006 PR tree-optimization/86097
15007 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
15008 iv type if signedness of iv type is not the same as that of *nit.
15010 2018-06-20 Jakub Jelinek <jakub@redhat.com>
15012 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
15013 EDGE_EH edges, verify they are all EDGE_EH.
15015 2018-06-20 Maya Rashish <coypu@sdf.org>
15017 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
15019 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15021 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
15022 * config/aarch64/aarch64.c (xgene1_tunings): Add
15023 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
15024 (aarch64_mode_valid_for_sched_fusion_p):
15025 Allow 16-byte modes.
15026 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
15027 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
15029 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
15031 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
15032 * config/aarch64/iterators.md (VQ2): New mode iterator.
15034 2018-06-20 Martin Liska <mliska@suse.cz>
15036 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
15037 Change default ratio from 10 to 8.
15039 2018-06-20 Martin Liska <mliska@suse.cz>
15041 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15043 (bit_test_cluster::find_bit_tests): Likewise.
15044 (switch_decision_tree::analyze_switch_statement): Find clusters.
15045 * tree-switch-conversion.h (struct jump_table_cluster): Document
15048 2018-06-20 Martin Liska <mliska@suse.cz>
15050 * tree-switch-conversion.c (switch_conversion::collect):
15051 Record m_uniq property.
15052 (switch_conversion::expand): Bail out for special conditions.
15053 (group_cluster::~group_cluster): New.
15054 (group_cluster::group_cluster): Likewise.
15055 (group_cluster::dump): Likewise.
15056 (jump_table_cluster::emit): New.
15057 (switch_decision_tree::fix_phi_operands_for_edges): New.
15058 (struct case_node): Remove struct.
15059 (jump_table_cluster::can_be_handled): New.
15060 (case_values_threshold): Moved to header.
15061 (reset_out_edges_aux): Likewise.
15062 (jump_table_cluster::is_beneficial): New.
15063 (bit_test_cluster::can_be_handled): Likewise.
15064 (add_case_node): Remove.
15065 (bit_test_cluster::is_beneficial): New.
15066 (case_bit_test::cmp): New.
15067 (bit_test_cluster::emit): New.
15068 (expand_switch_as_decision_tree_p): Remove.
15069 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
15070 (fix_phi_operands_for_edge): Likewise.
15071 (switch_decision_tree::analyze_switch_statement): New.
15072 (compute_cases_per_edge): Move ...
15073 (switch_decision_tree::compute_cases_per_edge): ... here.
15074 (try_switch_expansion): Likewise.
15075 (switch_decision_tree::try_switch_expansion): Likewise.
15076 (record_phi_operand_mapping): Likewise.
15077 (switch_decision_tree::record_phi_operand_mapping): Likewise.
15078 (emit_case_decision_tree): Likewise.
15079 (switch_decision_tree::emit): Likewise.
15080 (balance_case_nodes): Likewise.
15081 (switch_decision_tree::balance_case_nodes): Likewise.
15082 (dump_case_nodes): Likewise.
15083 (switch_decision_tree::dump_case_nodes): Likewise.
15084 (emit_jump): Likewise.
15085 (switch_decision_tree::emit_jump): Likewise.
15086 (emit_cmp_and_jump_insns): Likewise.
15087 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
15088 (emit_case_nodes): Likewise.
15089 (switch_decision_tree::emit_case_nodes): Likewise.
15090 (conditional_probability): Remove.
15091 * tree-switch-conversion.h (enum cluster_type): New.
15093 (struct cluster): Likewise.
15094 (cluster::cluster): Likewise.
15095 (struct simple_cluster): Likewise.
15096 (simple_cluster::simple_cluster): Likewise.
15097 (struct group_cluster): Likewise.
15098 (struct jump_table_cluster): Likewise.
15099 (struct bit_test_cluster): Likewise.
15100 (struct min_cluster_item): Likewise.
15101 (struct case_tree_node): Likewise.
15102 (case_tree_node::case_tree_node): Likewise.
15103 (jump_table_cluster::case_values_threshold): Likewise.
15104 (struct case_bit_test): Likewise.
15105 (struct switch_decision_tree): Likewise.
15106 (struct switch_conversion): Likewise.
15107 (switch_decision_tree::reset_out_edges_aux): Likewise.
15109 2018-06-20 Martin Liska <mliska@suse.cz>
15111 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
15112 (hoist_edge_and_branch_if_true): Likewise.
15113 (expand_switch_using_bit_tests_p): Likewise.
15114 (struct case_bit_test): Likewise.
15115 (case_bit_test_cmp): Likewise.
15116 (emit_case_bit_tests): Likewise.
15117 (switch_conversion::switch_conversion): New class.
15118 (struct switch_conv_info): Remove old struct.
15119 (collect_switch_conv_info): More to ...
15120 (switch_conversion::collect): ... this.
15121 (check_range): Likewise.
15122 (switch_conversion::check_range): Likewise.
15123 (check_all_empty_except_final): Likewise.
15124 (switch_conversion::check_all_empty_except_final): Likewise.
15125 (check_final_bb): Likewise.
15126 (switch_conversion::check_final_bb): Likewise.
15127 (create_temp_arrays): Likewise.
15128 (switch_conversion::create_temp_arrays): Likewise.
15129 (free_temp_arrays): Likewise.
15130 (gather_default_values): Likewise.
15131 (switch_conversion::gather_default_values): Likewise.
15132 (build_constructors): Likewise.
15133 (switch_conversion::build_constructors): Likewise.
15134 (constructor_contains_same_values_p): Likewise.
15135 (switch_conversion::contains_same_values_p): Likewise.
15136 (array_value_type): Likewise.
15137 (switch_conversion::array_value_type): Likewise.
15138 (build_one_array): Likewise.
15139 (switch_conversion::build_one_array): Likewise.
15140 (build_arrays): Likewise.
15141 (switch_conversion::build_arrays): Likewise.
15142 (gen_def_assigns): Likewise.
15143 (switch_conversion::gen_def_assigns): Likewise.
15144 (prune_bbs): Likewise.
15145 (switch_conversion::prune_bbs): Likewise.
15146 (fix_phi_nodes): Likewise.
15147 (switch_conversion::fix_phi_nodes): Likewise.
15148 (gen_inbound_check): Likewise.
15149 (switch_conversion::gen_inbound_check): Likewise.
15150 (process_switch): Use the newly created class.
15151 (switch_conversion::expand): New.
15152 (switch_conversion::~switch_conversion): New.
15153 * tree-switch-conversion.h: New file.
15155 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15157 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
15158 tree-vect-patterns.c.
15159 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
15160 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
15161 (vect_recog_sad_pattern): Likewise.
15162 (vect_recog_widen_sum_pattern): Likewise.
15163 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
15164 (vect_recog_widen_shift_pattern): Remove the type_in argument.
15165 (vect_recog_rotate_pattern): Likewise.
15166 (vect_recog_mult_pattern): Likewise.
15167 (vect_recog_vector_vector_shift_pattern): Likewise.
15168 (vect_recog_divmod_pattern): Likewise.
15169 (vect_recog_mixed_size_cond_pattern): Likewise.
15170 (vect_recog_bool_pattern): Likewise.
15171 (vect_recog_mask_conversion_pattern): Likewise.
15172 (vect_try_gather_scatter_pattern): Likewise.
15173 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
15174 (vect_recog_over_widening_pattern): Likewise.
15175 (vect_recog_gather_scatter_pattern): Likewise.
15176 (vect_recog_func_ptr): Move from tree-vectorizer.h
15177 (vect_vect_recog_func_ptrs): Move further down the file.
15178 (vect_recog_func): Likewise. Remove the third argument.
15179 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
15180 (vect_pattern_recog_1): Expect the pattern function to do any
15181 necessary target tests. Also expect it to provide a vector type.
15182 Remove the type_in handling.
15184 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15186 * tree-vect-patterns.c (vect_pattern_detected): New function.
15187 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
15188 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
15189 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
15190 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
15191 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
15192 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
15193 (vect_recog_mask_conversion_pattern)
15194 (vect_try_gather_scatter_pattern): Likewise.
15196 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15198 * tree-vect-patterns.c (vect_get_internal_def): New function.
15199 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15200 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
15201 (search_type_for_mask_1): Use it.
15203 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15205 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
15206 redundant WIDEN_SUM_EXPR handling.
15207 (vect_recog_sad_pattern): Likewise.
15209 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15211 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
15212 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
15213 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
15214 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
15215 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
15217 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15219 * tree-vect-stmts.c (vectorizable_call): Make sure that we
15220 use the stmt_vec_info of the original bb statement for the
15221 new zero assignment, even if the call is part of a pattern.
15223 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15225 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
15226 that the sequence is attached to the original statement rather
15227 than the pattern statement.
15228 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
15229 PATTERN_DEF_SEQ from the original statement rather than
15230 the main pattern statement.
15231 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
15232 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
15233 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
15235 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15237 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
15238 definition statements before the early exit for statements that aren't
15240 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
15242 (vect_transform_loop): ...here. Process pattern definition
15243 statements without first checking whether the main pattern
15244 statement is live or relevant.
15246 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15248 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
15249 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
15251 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
15253 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
15254 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
15255 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
15256 (expand_block_compare): Change select_block_compare_mode call.
15257 (expand_strncmp_align_check): Use new functions, fix comment.
15258 (emit_final_str_compare_gpr): New function.
15259 (expand_strn_compare): Refactor and clean up code.
15260 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
15262 2018-06-19 Tony Reix <tony.reix@atos.com>
15263 Damien Bergamini <damien.bergamini@atos.com>
15264 David Edelsohn <dje.gcc@gmail.com>
15266 * collect2.c (static_obj): New variable.
15267 (static_libs): New variable.
15268 (is_in_list): Uncomment declaration.
15269 (main): Track AIX libraries linked statically.
15270 (is_in_list): Uncomment definition.
15271 (scan_prog_file): Don't add AIX shared libraries initializer
15272 to constructor list if linking statically.
15274 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
15276 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
15278 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
15280 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
15282 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
15285 2018-06-19 Martin Liska <mliska@suse.cz>
15287 * config/i386/i386.c (ix86_can_inline_p): Do not use
15288 ipa_fn_summaries::get_create.
15289 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
15291 (devirtualization_time_bonus): Likewise.
15292 (ipcp_propagate_stage): Likewise.
15293 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
15294 (edge_set_predicate): Likewise.
15295 (evaluate_conditions_for_known_args): Likewise.
15296 (evaluate_properties_for_edge): Likewise.
15297 (ipa_call_summary::reset): Tranform to ...
15298 (ipa_call_summary::~ipa_call_summary): ... this.
15299 (ipa_fn_summary::reset): Transform to ...
15300 (ipa_fn_summary::~ipa_fn_summary): ... this.
15301 (ipa_fn_summary_t::remove): Rename to ...
15302 (ipa_fn_summary_t::remove_callees): ... this.
15303 (ipa_fn_summary_t::duplicate): Use placement new
15304 instead of memory copy.
15305 (ipa_call_summary_t::duplicate): Likewise.
15306 (ipa_call_summary_t::remove): Remove.
15307 (dump_ipa_call_summary): Change get_create to get.
15308 (ipa_dump_fn_summary): Dump only when summary exists.
15309 (analyze_function_body): Use symbol_summary::get instead
15311 (compute_fn_summary): Likewise.
15312 (estimate_edge_devirt_benefit): Likewise.
15313 (estimate_edge_size_and_time): Likewise.
15314 (inline_update_callee_summaries): Likewise.
15315 (remap_edge_change_prob): Likewise.
15316 (remap_edge_summaries): Likewise.
15317 (ipa_merge_fn_summary_after_inlining): Likewise.
15318 (write_ipa_call_summary): Likewise.
15319 (ipa_fn_summary_write): Likewise.
15320 (ipa_free_fn_summary): Likewise.
15321 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
15322 (struct ipa_call_summary): Likewise.
15323 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
15325 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
15326 (estimate_size_after_inlining): Likewise.
15327 (estimate_growth): Likewise.
15328 (growth_likely_positive): Likewise.
15329 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
15330 (inline_call): Likewise.
15331 * ipa-inline.c (caller_growth_limits): Likewise.
15332 (can_inline_edge_p): Likewise.
15333 (can_inline_edge_by_limits_p): Likewise.
15334 (compute_uninlined_call_time): Likewise.
15335 (compute_inlined_call_time): Likewise.
15336 (want_inline_small_function_p): Likewise.
15337 (edge_badness): Likewise.
15338 (update_caller_keys): Likewise.
15339 (update_callee_keys): Likewise.
15340 (inline_small_functions): Likewise.
15341 (inline_to_all_callers_1): Likewise.
15342 (dump_overall_stats): Likewise.
15343 (early_inline_small_functions): Likewise.
15344 (early_inliner): Likewise.
15345 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15346 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15347 * ipa-pure-const.c (malloc_candidate_p): Likewise.
15348 * ipa-split.c (execute_split_functions): Likewise.
15349 * symbol-summary.h: Likewise.
15350 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
15352 2018-06-19 Richard Biener <rguenther@suse.de>
15354 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
15355 (vectorize_loops): ... here. Fix dbgcnt handling.
15356 (try_vectorize_loop): Wrap try_vectorize_loop_1.
15358 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
15361 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
15362 ieee128 argument takes up only one (vector) register, not two (floating
15365 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15367 * gimplify.c (gimplify_init_constructor): Really never clear for an
15368 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
15370 2018-06-19 Richard Biener <rguenther@suse.de>
15372 PR tree-optimization/86179
15373 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
15374 after failed recognition.
15376 2018-06-18 Martin Sebor <msebor@redhat.com>
15378 PR middle-end/85602
15379 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
15380 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
15381 Handle integer subtraction.
15382 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
15383 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
15385 2018-06-18 David Malcolm <dmalcolm@redhat.com>
15387 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
15388 param from rtx to rtx_insn *.
15389 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
15391 (frv_ifcvt_modify_insn): Likwise.
15392 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
15393 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
15394 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
15395 as_a <rtx_insn *> cast to local "unprotected_region" once
15396 it's been established that it's not NULL or pc_rtx.
15397 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
15398 param "sethi" from rtx to rtx_insn *.
15399 (nds32_group_float_insns): Likewise for param "insn".
15400 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
15402 (vax_output_int_subtract): Likewise.
15403 * config/vax/vax.c (vax_output_int_add): Likewise for param
15405 (vax_output_int_subtract): Likewise.
15406 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
15407 (emit_pattern_after): Likewise for param "after".
15408 (emit_insn_after): Likewise.
15409 (emit_jump_insn_after): Likewise.
15410 (emit_call_insn_after): Likewise.
15411 (emit_debug_insn_after): Likewise.
15412 (emit_pattern_before): Likewise for param "before".
15413 (emit_insn_before): Likewise.
15414 (emit_jump_insn_before): Likewise.
15415 * final.c (get_insn_template): Likewise for param "insn", removing
15417 * output.h (get_insn_template): Likewise for 2nd param.
15418 * rtl.h (emit_insn_before): Likewise.
15419 (emit_jump_insn_before): Likewise.
15420 (emit_debug_insn_before_noloc): Likewise.
15421 (emit_insn_after): Likewise.
15422 (emit_jump_insn_after): Likewise.
15423 (emit_call_insn_after): Likewise.
15424 (emit_debug_insn_after): Likewise.
15425 (set_insn_deleted): Likewise for param.
15427 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
15430 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
15431 floating point modes, so that IFmode is numerically greater than
15432 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
15433 to declare the ordering. This prevents IFmode from being
15434 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
15435 machine. Include rs6000-modes.h to share the fractional values
15436 between genmodes* and the rest of the compiler.
15437 (IFmode): Likewise.
15438 (KFmode): Likewise.
15439 (TFmode): Likewise.
15440 * config/rs6000/rs6000-modes.h: New file.
15441 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
15442 meaning of rs6000_long_double_size so that 126..128 selects an
15443 appropriate 128-bit floating point type.
15444 (rs6000_option_override_internal): Likewise.
15445 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
15446 (TARGET_LONG_DOUBLE_128): Change the meaning of
15447 rs6000_long_double_size so that 126..128 selects an appropriate
15448 128-bit floating point type.
15449 (LONG_DOUBLE_TYPE_SIZE): Update comment.
15450 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
15451 source and destination to match the standard usage.
15452 (truncifkf2): Likewise.
15453 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
15454 ISA 2.07 to use an explicit clobber, instead of passing in a
15456 (copysign<mode>3_soft): Likewise.
15458 2018-06-18 David Malcolm <dmalcolm@redhat.com>
15460 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15461 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
15462 (vect_slp_analyze_instance_dependence): Likewise.
15463 (vect_enhance_data_refs_alignment): Likewise.
15464 (vect_analyze_data_refs_alignment): Likewise.
15465 (vect_slp_analyze_and_verify_instance_alignment
15466 (vect_analyze_data_ref_accesses): Likewise.
15467 (vect_prune_runtime_alias_test_list): Likewise.
15468 (vect_analyze_data_refs): Likewise.
15469 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
15470 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15471 (vect_analyze_scalar_cycles_1): Likewise.
15472 (vect_get_loop_niters): Likewise.
15473 (vect_analyze_loop_form_1): Likewise.
15474 (vect_update_vf_for_slp): Likewise.
15475 (vect_analyze_loop_operations): Likewise.
15476 (vect_analyze_loop): Likewise.
15477 (vectorizable_induction): Likewise.
15478 (vect_transform_loop): Likewise.
15479 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
15480 * tree-vect-slp.c (vect_analyze_slp): Likewise.
15481 (vect_make_slp_decision): Likewise.
15482 (vect_detect_hybrid_slp): Likewise.
15483 (vect_slp_analyze_operations): Likewise.
15484 (vect_slp_bb): Likewise.
15485 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
15486 (vectorizable_bswap): Likewise.
15487 (vectorizable_call): Likewise.
15488 (vectorizable_simd_clone_call): Likewise.
15489 (vectorizable_conversion): Likewise.
15490 (vectorizable_assignment): Likewise.
15491 (vectorizable_shift): Likewise.
15492 (vectorizable_operation): Likewise.
15493 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
15495 2018-06-18 Martin Sebor <msebor@redhat.com>
15497 PR tree-optimization/81384
15498 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
15499 * builtins.c (expand_builtin_strnlen): New function.
15500 (expand_builtin): Call it.
15501 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
15502 * builtins.def (BUILT_IN_STRNLEN): New.
15503 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
15504 Warn for bounds in excess of maximum object size.
15505 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
15506 single-value ranges. Handle strnlen.
15507 (handle_builtin_strlen): Handle strnlen.
15508 (strlen_check_and_optimize_stmt): Same.
15509 * doc/extend.texi (Other Builtins): Document strnlen.
15511 2018-06-18 Maya Rashish <coypu@sdf.org>
15513 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
15514 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
15515 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
15517 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
15519 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
15521 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15523 * tree.c (escaped_string::escape): Replace cast to char * by
15524 const_cast<char *> (unescaped).
15526 2018-06-18 Nick Clifton <nickc@redhat.com>
15529 * tree.c (escaped_string): New class. Converts an unescaped
15530 string into its escaped equivalent.
15531 (warn_deprecated_use): Use the new class to convert the
15532 deprecation message, if present.
15533 (test_escaped_strings): New self test.
15534 (test_c_tests): Add test_escaped_strings.
15535 * doc/extend.texi (deprecated): Add a note that the
15536 deprecation message is affected by the -fmessage-length
15537 option, and that control characters will be escaped.
15538 (#pragma GCC error): Document this pragma.
15539 (#pragma GCC warning): Likewise.
15540 * doc/invoke.texi (-fmessage-length): Document this option's
15541 effect on the #warning and #error preprocessor directives and
15542 the deprecated attribute.
15544 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
15546 * tree.c (decl_value_expr_lookup): Revert latest change.
15547 (decl_value_expr_insert): Likewise.
15549 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
15551 * gimplify.c (nonlocal_vlas): Delete.
15552 (nonlocal_vla_vars): Likewise.
15553 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
15555 (gimplify_body): Do not create and destroy nonlocal_vlas.
15556 * tree-nested.c: Include diagnostic.h.
15557 (use_pointer_in_frame): Tweak.
15558 (lookup_field_for_decl): Add assertion and declare the transformation.
15559 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
15560 internal error when the reference is in a wrong context. Do not
15561 create a debug decl by default.
15562 (note_nonlocal_block_vlas): Delete.
15563 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
15564 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
15565 create a debug decl by default.
15566 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
15567 call is in a wrong context.
15568 (fixup_vla_decls): New function.
15569 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
15570 debug variables were created.
15571 * tree.c (decl_value_expr_lookup): Add checking assertion.
15572 (decl_value_expr_insert): Likewise.
15574 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
15576 PR middle-end/82479
15577 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
15578 * tree-scalar-evolution.c (interpret_expr): Likewise.
15579 (expression_expensive_p): Likewise.
15580 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
15581 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
15582 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
15583 (ssa_defined_by_minus_one_stmt_p): New.
15585 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
15587 PR middle-end/64946
15588 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
15589 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
15590 * dojump.c (do_jump): Likewise.
15591 * expr.c (expand_expr_real_2): Check operand type's sign.
15592 * fold-const.c (const_unop): Handle ABSU_EXPR.
15593 (fold_abs_const): Likewise.
15594 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
15595 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
15596 (strip_sign_op_1): Likesise.
15597 * match.pd: Add new pattern to generate ABSU_EXPR.
15598 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
15599 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15600 * tree-eh.c (operation_could_trap_helper_p): Likewise.
15601 * tree-inline.c (estimate_operator_cost): Likewise.
15602 * tree-pretty-print.c (dump_generic_node): Likewise.
15603 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
15604 * tree.def (ABSU_EXPR): New.
15606 2018-06-16 Jakub Jelinek <jakub@redhat.com>
15608 PR middle-end/86095
15609 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
15610 documented as preserved for backward compatibility only.
15611 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
15613 PR rtl-optimization/86108
15614 * bb-reorder.c (create_forwarder_block): Renamed to ...
15615 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
15616 jump from new landing pad to the second part.
15617 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
15620 2018-06-15 Jakub Jelinek <jakub@redhat.com>
15622 PR middle-end/85878
15623 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
15624 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
15625 Only call store_expr for halves if the mode is the same.
15627 PR middle-end/86123
15628 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
15629 Fix up comment formatting.
15631 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15633 * typed-splay-tree.h (typed_splay_tree::remove): New function.
15634 (typed_splay_tree::closure,
15635 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
15636 (typed_splay_tree::typed_splay_tree,
15637 typed_splay_tree::operator =): Declared private.
15638 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
15639 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
15640 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
15641 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
15642 typed_splay_tree::splay_tree_splay,
15643 typed_splay_tree::splay_tree_foreach_helper,
15644 typed_splay_tree::splay_tree_insert,
15645 typed_splay_tree::splay_tree_remove,
15646 typed_splay_tree::splay_tree_lookup,
15647 typed_splay_tree::splay_tree_predecessor,
15648 typed_splay_tree::splay_tree_successor,
15649 typed_splay_tree::splay_tree_min,
15650 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
15651 (typed_splay_tree::root, typed_splay_tree::comp,
15652 typed_splay_tree::delete_key,
15653 typed_splay_tree::delete_value): New data members.
15654 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
15655 typed_splay_tree::remove.
15657 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
15659 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
15660 -mginv and -mno-ginv to the assembler.
15661 * config/mips/mips.opt (-mcrc): New option.
15662 (-mginv): Likewise.
15663 * doc/invoke.text (-mcrc): Document.
15664 (-mginv): Likewise.
15666 2018-06-15 Nick Clifton <nickc@redhat.com>
15669 * tree.c (escaped_string): New class. Converts an unescaped
15670 string into its escaped equivalent.
15671 (warn_deprecated_use): Use the new class to convert the
15672 deprecation message, if present.
15673 (test_escaped_strings): New self test.
15674 (test_c_tests): Add test_escaped_strings.
15675 * doc/extend.texi (deprecated): Add a note that the
15676 deprecation message is affected by the -fmessage-length
15677 option, and that control characters will be escaped.
15678 (#pragma GCC error): Document this pragma.
15679 (#pragma GCC warning): Likewise.
15680 * doc/invoke.texi (-fmessage-length): Document this option's
15681 effect on the #warning and #error preprocessor directives and
15682 the deprecated attribute.
15684 2018-06-15 Richard Biener <rguenther@suse.de>
15686 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
15687 here, also noting vector size used.
15688 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
15689 size used in MSG_OPTIMIZED_LOCATIONS dump.
15690 (pass_slp_vectorize::execute): Adjust.
15692 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
15695 * config/arc/arc.c (arc_return_address_register): Fix
15698 2018-06-15 Richard Biener <rguenther@suse.de>
15700 PR middle-end/86159
15701 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
15702 leave useless conversion stripping to force_gimple_operand_gsi.
15703 (gimplify_build2): Likewise.
15704 (gimplify_build1): Likewise.
15706 2018-06-15 Richard Biener <rguenther@suse.de>
15708 PR middle-end/86076
15709 * tree-cfg.c (move_stmt_op): unshare invariant addresses
15710 before adjusting their block.
15712 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
15714 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
15715 multilibs for *-*-rtems*.
15716 * config/riscv/t-rtems: New file.
15718 2018-06-14 Jakub Jelinek <jakub@redhat.com>
15720 PR middle-end/86122
15721 * match.pd ((A +- CST1) +- CST2): Punt if last resort
15722 unsigned_type_for returns NULL.
15725 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
15726 subregs of multi-word pseudos unless the float mode has word size.
15728 2018-06-14 Richard Biener <rguenther@suse.de>
15730 PR middle-end/86139
15731 * tree-vect-generic.c (build_word_mode_vector_type): Remove
15732 duplicate and harmful type_hash_canon.
15733 * tree.c (type_hash_canon): Assert we didn't find ourselves.
15735 2018-06-14 Richard Biener <rguenther@suse.de>
15738 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
15741 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
15743 * config/rtems.h (STDINT_LONG32): Define.
15745 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
15746 Prachi Godbole <prachi.godbole@imgtec.com>
15748 * config/mips/mips-cpus.def: Define P6600.
15749 * config/mips/mips-tables.opt: Regenerate.
15750 * config/mips/mips.c (mips_ucbranch_type): New enum.
15751 (mips_rtx_cost_data): Add support for P6600.
15752 (mips_issue_rate): Likewise.
15753 (mips_multipass_dfa_lookahead): Likewise.
15754 (mips_avoid_hazard): Likewise.
15755 (mips_reorg_process_insns): Likewise.
15756 (mips_classify_branch_p6600): New function.
15757 * config/mips/mips.h (TUNE_P6600): New define.
15758 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
15759 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
15760 * config/mips/mips.md: Include p6600.md.
15761 (processor): Add p6600.
15762 * config/mips/p6600.md: New file.
15763 * doc/invoke.texi: Add p6600 to supported architectures.
15765 2018-06-13 Martin Sebor <msebor@redhat.com>
15767 PR tree-optimization/86114
15768 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
15770 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
15772 2018-06-13 Richard Biener <rguenther@suse.de>
15774 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
15775 Properly set vector type of the intermediate stmt.
15776 * tree-vect-stmts.c (vectorizable_operation): The destination
15777 var always has vectype_out type.
15779 2018-06-13 Jeff Law <law@redhat.com>
15781 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
15782 integer 0 for argument to print_rtl_with_bb.
15783 (rl78_reorg): Likewise.
15785 2018-06-13 David Malcolm <dmalcolm@redhat.com>
15787 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
15788 from rtx to rtx_insn *.
15789 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
15791 (add_sched_insns_for_speculation): Likewise for local "target",
15792 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
15793 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
15794 from rtx_insn ** to rtx_code_label **.
15795 (reorg_emit_nops): Likewise.
15796 (c6x_reorg): Likewise for local "call_labels".
15797 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
15799 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
15800 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
15801 the loops over LABEL_REFS.
15802 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
15804 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
15805 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
15806 (split_branches): Strengthen local "olabel" from rtx to
15807 rtx_insn *, adding a safe_as_a cast.
15808 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
15810 (add_insn_after): Likewise for first two params.
15811 (add_insn_before): Likewise.
15812 (remove_insn): Likewise for param.
15813 (emit_pattern_before_noloc): Likewise for second and third params.
15814 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
15815 (emit_call_insn_before_noloc): Likewise.
15816 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
15818 (emit_barrier_before): Likewise.
15819 (emit_label_before): Strengthen "label" param from "rtx" to
15820 "rtx_code_label *". Strengthen "before" param from "rtx" to
15822 (emit_insn_after_1): Strengthen "after" param from "rtx" to
15824 (emit_pattern_after_noloc): Likewise.
15825 (emit_insn_after_noloc): Likewise.
15826 (emit_jump_insn_after_noloc): Likewise.
15827 (emit_call_insn_after_noloc): Likewise.
15828 (emit_debug_insn_after_noloc): Likewise.
15829 (emit_barrier_after): Likewise.
15830 (emit_label_after): Likewise for both params.
15831 (emit_pattern_after_setloc): Likewise for "after" param. Convert
15832 "loc" param from "int" to "location_t".
15833 (emit_insn_after_setloc): Likewise.
15834 (emit_jump_insn_after_setloc): Likewise.
15835 (emit_call_insn_after_setloc): Likewise.
15836 (emit_debug_insn_after_setloc): Likewise.
15837 (emit_pattern_before_setloc): Likewise for "before" param. Convert
15838 "loc" param from "int" to "location_t".
15839 (emit_pattern_before): Convert NULL_RTX to NULL.
15840 (emit_insn_before_setloc): Convert "loc" param from "int" to
15842 (emit_jump_insn_before_setloc): Likewise.
15843 (emit_call_insn_before_setloc): Likewise.
15844 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
15845 rtx_insn *. Convert "loc" param from "int" to "location_t".
15846 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
15847 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
15848 Convert 3rd param from "int" to "location_t".
15849 (emit_barrier_before, emit_barrier_after, next_real_insn):
15850 Strengthen param from rtx to rtx_insn *.
15851 (emit_label_before): Strengthen 1st param from "rtx" to
15852 "rtx_code_label *". Strengthen 2nd param from "rtx" to
15854 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
15855 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
15856 Strengthen 2nd param from "rtx" to "rtx_insn *".
15857 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
15858 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
15859 Likewise. Convert 3rd param from "int" to "location_t".
15860 (emit_label_after): Strengthen 1st param from "rtx" to
15861 "rtx_code_label *".
15862 (next_real_insn, remove_insn): Strengthen param from "rtx" to
15864 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
15865 from "rtx" to "rtx_insn *".
15867 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
15869 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
15870 bodies streamed in with -Q.
15871 * dumpfile.c (dump_files): Add lto-stream-out dump file.
15872 * dumpfile.h (tree_dump_index): Add lto_stream_out.
15873 * gimple-streamer-out.c: Include gimple-pretty-print.h
15874 (output_bb): Dump stmts streamed.
15875 * lto-section-out.c: Include print-tree.h
15876 (lto_begin_section): Dump sections created.
15877 (lto_output_decl_index): Dump decl encoded.
15878 * lto-streamer-out.c: Include print-tree.h
15879 (create_output_block): Dump output block created.
15880 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
15881 (output_function): Dump function output.
15882 (output_constructor): Dump constructor streamed.
15883 (write_global_stream): Output indexes encoded.
15884 (produce_asm_for_decls): Dump streams encoded.
15885 * lto-streamer.c (streamer_dump_file): New global var.
15886 * lto-streamer.h (streamer_dump_file): Declare.
15887 * passes.c (ipa_write_summaries): Initialize streamer dump.
15888 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
15891 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
15894 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
15895 offsets for register save directives. Emit a second batch of save
15896 directives, if need be, when the function accesses prior frames.
15898 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15900 * config/arc/fpu.md (fmasf4): Force operand to register.
15901 (fnmasf4): Likewise.
15903 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15905 * config/arc/arc-protos.h (arc_pad_return): Remove.
15906 * config/arc/arc.c (machine_function): Remove force_short_suffix
15908 (arc_print_operand): Adjust printing of '&'.
15909 (arc_verify_short): Remove conditional printing of short suffix.
15910 (arc_final_prescan_insn): Remove reference to size_reason.
15911 (pad_return): New function.
15912 (arc_reorg): Call pad_return.
15913 (arc_pad_return): Remove.
15914 (arc_init_machine_status): Remove reference to force_short_suffix.
15915 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
15916 (attr length): When attribute iscompact is true force to 2
15917 regardless; in the case of maybe check if we want to force the
15918 instruction to have 4 bytes length.
15919 (nopv): Change it to generate 4 byte long nop as well.
15920 (blockage): New pattern.
15921 (simple_return): Remove call to arc_pad_return.
15922 (p_return_i): Likewise.
15924 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15926 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
15928 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15930 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
15933 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15935 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
15936 for ARC700 and ARCv2.
15938 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
15941 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
15942 operands[2] instead of operands[1].
15945 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15947 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
15948 case, check whether the outer register overlaps an unallocatable
15949 register, not just whether it fits the required class.
15951 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15953 * poly-int.h (can_div_trunc_p): Add new overload in which all values
15955 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
15956 (memrefs_conflict_p): Likewise.
15957 (init_alias_analysis): Likewise.
15958 * cfgexpand.c (expand_debug_expr): Likewise.
15959 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
15960 * cse.c (fold_rtx): Likewise.
15961 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
15962 * expr.c (emit_block_move_hints): Likewise.
15963 (clear_storage_hints, push_block, emit_push_insn): Likewise.
15964 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
15965 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
15966 (emit_group_store): Likewise.
15967 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
15968 to read the PRE/POST_MODIFY increment.
15969 * calls.c (store_one_arg): Use strip_offset.
15970 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
15972 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
15974 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
15975 (simplify_binary_operation_1): Extend CONST_INT handling to
15977 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
15978 than a HOST_WIDE_INT.
15979 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
15981 (adjust_mems, add_stores): Update accodingly.
15982 (vt_canonicalize_addr): Track polynomial offsets.
15983 (emit_note_insn_var_location): Likewise.
15984 (vt_add_function_parameter): Likewise.
15985 (vt_initialize): Likewise.
15987 2018-06-12 Jeff Law <law@redhat.com>
15989 * config.gcc (alpha*-*-freebsd*): Remove.
15990 * config/alpha/freebsd.h: Remove.
15992 2018-06-12 David Malcolm <dmalcolm@redhat.com>
15995 * spellcheck-tree.c (levenshtein_distance): Rename to...
15996 (get_edit_distance): ...this, and update for underlying renaming.
15997 * spellcheck-tree.h (levenshtein_distance): Rename to...
15998 (get_edit_distance): ...this.
15999 * spellcheck.c (levenshtein_distance): Rename to...
16000 (get_edit_distance): ...this. Convert from Levenshtein distance
16001 to Damerau-Levenshtein distance by supporting transpositions of
16002 adjacent characters. Rename "v1" to "v_next" and "v0" to
16004 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
16005 (selftest::test_edit_distance_unit_test_oneway): ...this, and
16006 update for underlying renaming.
16007 (selftest::levenshtein_distance_unit_test): Rename to...
16008 (selftest::test_get_edit_distance_unit): ...this, and update for
16009 underlying renaming.
16010 (selftest::test_find_closest_string): Add example from PR 69968
16011 where transposition helps
16012 (selftest::test_metric_conditions): Update for renaming.
16013 (selftest::test_metric_conditions): Likewise.
16014 (selftest::spellcheck_c_tests): Likewise.
16015 * spellcheck.h (levenshtein_distance): Rename both overloads to...
16016 (get_edit_distance): ...this.
16017 (best_match::consider): Update for renaming.
16019 2018-06-12 Martin Sebor <msebor@redhat.com>
16021 PR tree-optimization/85259
16022 * builtins.c (compute_objsize): Handle constant offsets.
16023 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
16024 true iff a warning has been issued.
16025 * gimple.h (gimple_nonartificial_location): New function.
16026 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
16027 gimple_nonartificial_location and handle -Wno-system-headers.
16028 (handle_builtin_stxncpy): Same.
16030 2018-06-12 Martin Sebor <msebor@redhat.com>
16033 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
16035 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16037 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16038 BUILTIN_VEC_XST entries for pointer to double and long long.
16040 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
16043 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
16045 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
16048 2018-06-12 Martin Liska <mliska@suse.cz>
16050 * doc/options.texi: Document IntegerRange.
16052 2018-06-12 Martin Liska <mliska@suse.cz>
16054 * config/i386/i386.opt: Make MPX-related options as Deprecated.
16055 * opt-functions.awk: Handle Deprecated flag.
16056 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
16058 (read_cmdline_option): Report warning for a deprecated option.
16059 * opts.h (struct cl_option): Add new field cl_deprecated.
16060 (CL_ERR_DEPRECATED): New.
16062 2018-06-12 Martin Liska <mliska@suse.cz>
16064 * doc/options.texi: Document Deprecated option flag.
16066 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16068 * config/arc/arc-arch.h (arc_extras): New enum.
16069 (arc_cpu_t):Add field extra.
16070 (arc_cpu_types): Consider the extras.
16071 * config/arc/arc-cpus.def: Add extras info.
16072 * config/arc/arc-opts.h (processor_type): Consider extra field.
16073 * config/arc/arc.c (arc_override_options): Handle extra field.
16075 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16077 * config/arc/arc-arch.h: Update ARC_OPTX macro.
16078 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
16080 * config/arc/arc.c (arc_init): Update pic warning.
16081 (irq_range): Update irq range parsing warnings.
16082 (arc_override_options): Update various warning messages.
16083 (arc_handle_aux_attribute): Likewise.
16085 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
16087 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
16089 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16091 * doc/sourcebuild.texi: Document usage of line number 0 in verify
16092 compiler messages directives.
16094 2018-06-12 Matthew Fortune <mfortune@gmail.com>
16096 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
16097 * config/mips/mips-tables.opt: Regenerate.
16098 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
16100 * doc/invoke.texi: Document -march=i6500.
16102 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
16104 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
16105 (i6400_gpmul): Add cpu_unit.
16106 (i6400_gpdiv): Likewise.
16107 (i6400_msa_add_d): Update reservations.
16108 (i6400_msa_int_add) Likewise.
16109 (i6400_msa_short_logic3) Likewise.
16110 (i6400_msa_short_logic2) Likewise.
16111 (i6400_msa_short_logic) Likewise.
16112 (i6400_msa_move) Likewise.
16113 (i6400_msa_cmp) Likewise.
16114 (i6400_msa_short_float2) Likewise.
16115 (i6400_msa_div_d) Likewise.
16116 (i6400_msa_long_logic1) Likewise.
16117 (i6400_msa_long_logic2) Likewise.
16118 (i6400_msa_mult) Likewise.
16119 (i6400_msa_long_float2) Likewise.
16120 (i6400_msa_long_float4) Likewise.
16121 (i6400_msa_long_float5) Likewise.
16122 (i6400_msa_long_float8) Likewise.
16123 (i6400_fpu_fadd): Include frint type.
16124 (i6400_fpu_store): New define_insn_reservation.
16125 (i6400_fpu_load): Likewise.
16126 (i6400_fpu_move): Likewise.
16127 (i6400_fpu_fcmp): Likewise.
16128 (i6400_fpu_fmadd): Likewise.
16129 (i6400_int_mult): Include imul3nc type and update reservation.
16130 (i6400_int_div): Include idiv3 type and update reservation.
16131 (i6400_int_load): Update to check type not move_type.
16132 (i6400_int_store): Likewise.
16133 (i6400_int_prefetch): Set zero latency.
16135 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
16137 * gcc.c: Document new %@{...} sequence.
16138 (LINK_COMMAND_SPEC): Use it for the -L switches.
16139 (cpp_unique_options): Use it for the -I switches.
16140 (at_file_argbuf): New global variable.
16141 (in_at_file): Likewise.
16142 (alloc_args): Create at_file_argbuf.
16143 (clear_args): Truncate at_file_argbuf.
16144 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
16145 (open_at_file): New function.
16146 (close_at_file): Likewise.
16147 (create_at_file): Delete.
16148 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
16151 (validate_switches_from_spec): Deal with %@{...} sequence.
16152 (validate_switches): Likewise.
16153 (driver::finalize): Call clear_args.
16155 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
16157 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
16159 2018-06-11 Martin Sebor <msebor@redhat.com>
16161 * doc/invoke.texi (-Wall): List -Wc++17-compat.
16162 (Wno-class-memaccess): Add @opindex.
16163 (Wno-templates, Wno-multiple-inheritance): Same.
16164 (Wno-virtual-inheritance, Wno-namespaces): Same.
16165 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
16166 (Wno-format-overflow, Wno-format-truncation): Same.
16167 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
16168 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
16169 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
16170 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
16171 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
16172 (Wno-misspelled-isr): Same.
16174 2018-06-11 Martin Sebor <msebor@redhat.com>
16176 * PR tree-optimization/86083
16177 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
16179 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
16181 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
16183 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
16186 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
16187 on the correct operand.
16188 (*movdi_internal64): Ditto.
16190 2018-06-11 Martin Liska <mliska@suse.cz>
16192 PR tree-optimization/86089
16193 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
16195 2018-06-11 Julia Koval <julia.koval@intel.com>
16197 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
16198 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
16199 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
16201 2018-06-11 Olivier Hainque <hainque@adacore.com>
16203 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
16204 for Ada with strict dwarf2.
16206 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
16209 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
16212 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
16214 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
16216 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
16218 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
16221 2018-06-08 Martin Liska <mliska@suse.cz>
16223 * tree-cfg.h (debug_function): Fix argument type to match
16226 2018-06-08 Martin Liska <mliska@suse.cz>
16228 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
16229 Remove usage of MPX-related (and removed) fields.
16230 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
16232 2018-06-08 David Malcolm <dmalcolm@redhat.com>
16234 * cfg.c (debug): Use TDF_NONE rather than 0.
16235 * cfghooks.c (debug): Likewise.
16236 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
16237 (struct dump_option_value_info): Convert to...
16238 (struct kv_pair): ...this template type.
16239 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
16241 (optinfo_verbosity_options): Likewise.
16242 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
16244 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
16245 than int for "optgroup_flags" param.
16246 (dump_generic_expr_loc): Use dump_flags_t rather than int for
16248 (dump_dec): Likewise.
16249 (dump_finish): Use TDF_NONE rather than 0.
16250 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
16251 rather than int for "optgroup_flags" param. Use TDF_NONE rather
16252 than 0. Update for change to option_ptr.
16253 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
16254 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
16255 0. Update for changes to optinfo_verbosity_options and
16257 (opt_info_switch_p): Convert optgroup_flags from int to
16259 (dump_basic_block): Use dump_flags_t rather than int
16260 for "dump_kind" param.
16261 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
16262 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
16263 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
16264 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
16265 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
16266 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
16267 TDF_NONE): Convert from macros to...
16268 (enum dump_flag): ...this new enum.
16269 (dump_flags_t): Update to use enum.
16270 (operator|, operator&, operator~, operator|=, operator&=):
16271 Implement for dump_flags_t.
16272 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
16273 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
16274 Convert from macros to...
16275 (enum optgroup_flag): ...this new enum.
16276 (optgroup_flags_t): New typedef.
16277 (operator|, operator|=): Implement for optgroup_flags_t.
16278 (struct dump_file_info): Convert field "alt_flags" to
16279 dump_flags_t. Convert field "optgroup_flags" to
16281 (dump_basic_block): Use dump_flags_t rather than int for param.
16282 (dump_generic_expr_loc): Likewise.
16283 (dump_dec): Likewise.
16284 (dump_register): Convert param "optgroup_flags" to
16286 (opt_info_enable_passes): Likewise.
16287 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
16289 * gimple-pretty-print.c (debug): Likewise.
16290 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
16291 (merged_store_group::apply_stores): Likewise.
16292 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
16293 * gimple.c (verify_gimple_pp): Likewise.
16294 * graphite-poly.c (print_pbb_body): Likewise.
16295 * passes.c (pass_manager::register_one_dump_file): Convert
16296 local "optgroup_flags" to optgroup_flags_t.
16297 * print-tree.c (print_node): Use TDF_NONE rather than 0.
16299 (debug_body): Likewise.
16300 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
16301 to optgroup_flags_t.
16302 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
16304 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
16305 (convert_mult_to_fma): Likewise.
16306 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
16307 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
16308 * tree-vect-data-refs.c (dump_lower_bound): Convert param
16309 "dump_kind" to dump_flags_t.
16311 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
16313 * config/rs6000/rs6000.c (min, max): Delete.
16315 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
16317 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
16318 -mabi=spe and -mabi=no-spe.
16320 2018-06-08 Martin Liska <mliska@suse.cz>
16322 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
16323 where we expect an existing summary.
16325 2018-06-08 Martin Liska <mliska@suse.cz>
16327 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
16328 * ipa-inline.h (estimate_edge_growth): Likewise.
16330 2018-06-08 Martin Liska <mliska@suse.cz>
16332 * cgraph.c (function_version_hasher::hash): Use
16333 cgraph_node::get_uid ().
16334 (function_version_hasher::equal):
16335 * cgraph.h (cgraph_node::get_uid): New method.
16336 * ipa-inline.c (update_caller_keys): Use
16337 cgraph_node::get_uid ().
16338 (update_callee_keys): Likewise.
16339 * ipa-utils.c (searchc): Likewise.
16340 (ipa_reduced_postorder): Likewise.
16341 * lto-cgraph.c (input_node): Likewise.
16342 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
16343 * symbol-summary.h (symtab_insertion): Likewise.
16344 (symtab_removal): Likewise.
16345 (symtab_duplication): Likewise.
16346 * tree-pretty-print.c (dump_function_header): Likewise.
16347 * tree-sra.c (convert_callers_for_node): Likewise.
16349 2018-06-08 Martin Liska <mliska@suse.cz>
16351 * cgraph.c (symbol_table::create_edge): Always assign a new
16353 (symbol_table::free_edge): Do not recycle numbers.
16354 * cgraph.h (cgraph_edge::get): New method.
16355 * symbol-summary.h (symtab_removal): Use it.
16356 (symtab_duplication): Likewise.
16357 (call_summary::hashable_uid): Remove.
16359 2018-06-08 Martin Liska <mliska@suse.cz>
16361 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
16362 (initialize_growth_caches): Remove.
16363 (free_growth_caches): Likewise.
16364 (do_estimate_edge_time): Use edge_growth_cache.
16365 (do_estimate_edge_size): Likewise.
16366 (do_estimate_edge_hints): Likewise.
16367 * ipa-inline.c (reset_edge_caches): Likewise.
16368 (recursive_inlining): Likewise.
16369 (inline_small_functions): Likewise.
16370 * ipa-inline.h (initialize_growth_caches): Remove.
16371 (estimate_edge_size): Likewise.
16372 (estimate_edge_time): Likewise.
16373 (estimate_edge_hints): Likewise.
16374 (reset_edge_growth_cache): Likewise.
16375 * symbol-summary.h (call_summary::remove): New method.
16377 2018-06-08 Martin Liska <mliska@suse.cz>
16379 * ipa-cp.c (class edge_clone_summary): New summary.
16380 (grow_edge_clone_vectors): Remove.
16381 (ipcp_edge_duplication_hook): Remove.
16382 (class edge_clone_summary_t): New call_summary class.
16383 (ipcp_edge_removal_hook): Remove.
16384 (edge_clone_summary_t::duplicate): New function.
16385 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
16386 (create_specialized_node): Likewise.
16387 (ipcp_driver): Initialize edge_clone_summaries and do not
16390 2018-06-08 Martin Liska <mliska@suse.cz>
16392 * symbol-summary.h (get): New function.
16393 (call_summary::m_initialize_when_cloning): New class member.
16395 2018-06-08 Martin Liska <mliska@suse.cz>
16397 * cgraph.c (cgraph_node::remove): Do not recycle uid.
16398 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
16399 (symbol_table::allocate_cgraph_symbol): Do not set uid.
16400 * passes.c (uid_hash_t): Record removed_nodes by their uids.
16401 (remove_cgraph_node_from_order): Use the removed_nodes set.
16402 (do_per_function_toporder): Likwise.
16403 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
16404 instead of summary_uid.
16405 (symtab_removal): Likewise.
16406 (symtab_duplication): Likewise.
16408 2018-06-08 Martin Liska <mliska@suse.cz>
16410 * ipa-cp.c (ipcp_store_bits_results): Use
16411 ipcp_transformation_sum.
16412 (ipcp_store_vr_results): Likewise.
16413 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
16415 (ipcp_transformation_initialize): ... this.
16416 (ipa_set_node_agg_value_chain):
16417 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
16418 (write_ipcp_transformation_info): Likewise.
16419 (read_ipcp_transformation_info): Likewise.
16420 (ipcp_update_bits): Likewise.
16421 (ipcp_update_vr): Likewise.
16422 (ipcp_transform_function): Likewise.
16423 * ipa-prop.h: Rename ipcp_transformation_summary to
16424 ipcp_transformation.
16425 (class ipcp_transformation_t): New function summary.
16426 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
16427 (ipa_get_agg_replacements_for_node): Likewise.
16429 2018-06-08 Martin Liska <mliska@suse.cz>
16431 * ipa-pure-const.c (struct funct_state_d): Do it class instead
16433 (class funct_state_summary_t): New function_summary class.
16434 (has_function_state): Remove.
16435 (get_function_state): Likewise.
16436 (set_function_state): Likewise.
16437 (add_new_function): Likewise.
16438 (funct_state_summary_t::insert): New function.
16439 (duplicate_node_data): Remove.
16440 (remove_node_data): Remove.
16441 (funct_state_summary_t::duplicate): New function.
16442 (register_hooks): Create new funct_state_summaries.
16443 (pure_const_generate_summary): Use it.
16444 (pure_const_write_summary): Likewise.
16445 (pure_const_read_summary): Likewise.
16446 (propagate_pure_const): Likewise.
16447 (propagate_nothrow): Likewise.
16448 (dump_malloc_lattice): Likewise.
16449 (propagate_malloc): Likewise.
16450 (execute): Do not register hooks, just remove summary
16452 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
16455 2018-06-08 Martin Liska <mliska@suse.cz>
16457 * ipa-reference.c (remove_node_data): Remove.
16458 (duplicate_node_data): Likewise.
16459 (class ipa_ref_var_info_summary_t): New class.
16460 (class ipa_ref_opt_summary_t): Likewise.
16461 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
16462 (get_reference_optimization_summary): Use
16463 ipa_ref_opt_sum_summaries.
16464 (set_reference_vars_info): Remove.
16465 (set_reference_optimization_summary): Likewise.
16466 (ipa_init): Create summaries.
16467 (init_function_info): Use function summary.
16468 (ipa_ref_opt_summary_t::duplicate): New function.
16469 (ipa_ref_opt_summary_t::remove): New function.
16470 (get_read_write_all_from_node): Fix GNU coding style.
16471 (propagate): Use function summary.
16472 (write_node_summary_p): Fix GNU coding style.
16473 (stream_out_bitmap): Likewise.
16474 (ipa_reference_read_optimization_summary): Use function summary.
16475 (ipa_reference_c_finalize): Do not release hooks.
16477 2018-06-08 Martin Liska <mliska@suse.cz>
16479 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
16480 (analyze_function_body): Extract multiple calls of get_create.
16481 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
16482 * ipa-inline.c (recursive_inlining): Use ::get method.
16483 * ipa-inline.h (estimate_edge_growth): Likewise.
16485 2018-06-08 Martin Liska <mliska@suse.cz>
16487 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
16489 (hsa_function_summary::hsa_function_summary): Use the new enum
16491 (hsa_gpu_implementation_p): Use hsa_summaries::get.
16492 * hsa-gen.c (hsa_get_host_function): Likewise.
16493 (get_brig_function_name): Likewise.
16494 * ipa-hsa.c (process_hsa_functions): Likewise.
16495 (ipa_hsa_write_summary): Likewise.
16496 * symbol-summary.h (symtab_duplication): Use ::get function/
16497 (get): New function.
16499 2018-06-08 Martin Liska <mliska@suse.cz>
16501 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
16503 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
16504 (hsa_register_kernel): Likewise.
16505 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
16506 * hsa-gen.c (hsa_get_host_function): Likewise.
16507 (get_brig_function_name): Likewise.
16508 (generate_hsa): Likewise.
16509 (pass_gen_hsail::execute): Likewise.
16510 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
16511 (devirtualization_time_bonus): Likewise.
16512 (ipcp_propagate_stage): Likewise.
16513 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
16514 (edge_set_predicate): Likewise.
16515 (evaluate_conditions_for_known_args): Likewise.
16516 (evaluate_properties_for_edge): Likewise.
16517 (ipa_fn_summary::reset): Likewise.
16518 (ipa_fn_summary_t::duplicate): Likewise.
16519 (dump_ipa_call_summary): Likewise.
16520 (ipa_dump_fn_summary): Likewise.
16521 (analyze_function_body): Likewise.
16522 (compute_fn_summary): Likewise.
16523 (estimate_edge_devirt_benefit): Likewise.
16524 (estimate_edge_size_and_time): Likewise.
16525 (estimate_calls_size_and_time): Likewise.
16526 (estimate_node_size_and_time): Likewise.
16527 (inline_update_callee_summaries): Likewise.
16528 (remap_edge_change_prob): Likewise.
16529 (remap_edge_summaries): Likewise.
16530 (ipa_merge_fn_summary_after_inlining): Likewise.
16531 (ipa_update_overall_fn_summary): Likewise.
16532 (read_ipa_call_summary): Likewise.
16533 (inline_read_section): Likewise.
16534 (write_ipa_call_summary): Likewise.
16535 (ipa_fn_summary_write): Likewise.
16536 (ipa_free_fn_summary): Likewise.
16537 * ipa-hsa.c (process_hsa_functions): Likewise.
16538 (ipa_hsa_write_summary): Likewise.
16539 (ipa_hsa_read_section): Likewise.
16540 * ipa-icf.c (sem_function::merge): Likewise.
16541 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
16542 (do_estimate_edge_time): Likewise.
16543 (estimate_size_after_inlining): Likewise.
16544 (estimate_growth): Likewise.
16545 (growth_likely_positive): Likewise.
16546 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
16547 (inline_call): Likewise.
16548 * ipa-inline.c (caller_growth_limits): Likewise.
16549 (can_inline_edge_p): Likewise.
16550 (can_inline_edge_by_limits_p): Likewise.
16551 (compute_uninlined_call_time): Likewise.
16552 (compute_inlined_call_time): Likewise.
16553 (want_inline_small_function_p): Likewise.
16554 (edge_badness): Likewise.
16555 (update_caller_keys): Likewise.
16556 (update_callee_keys): Likewise.
16557 (recursive_inlining): Likewise.
16558 (inline_small_functions): Likewise.
16559 (inline_to_all_callers_1): Likewise.
16560 (dump_overall_stats): Likewise.
16561 (early_inline_small_functions): Likewise.
16562 (early_inliner): Likewise.
16563 * ipa-inline.h (estimate_edge_growth): Likewise.
16564 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16565 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16566 * ipa-prop.h (IPA_NODE_REF): Likewise.
16567 (IPA_EDGE_REF): Likewise.
16568 * ipa-pure-const.c (malloc_candidate_p): Likewise.
16569 (propagate_malloc): Likewise.
16570 * ipa-split.c (execute_split_functions): Likewise.
16571 * symbol-summary.h: Rename get to get_create.
16573 (get_create): Likewise.
16574 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
16576 2018-06-08 Martin Liska <mliska@suse.cz>
16578 * symbol-summary.h (release): Move definition out of class
16580 (symtab_removal): Likewise.
16581 (symtab_duplication): Likewise.
16583 2018-06-08 Martin Liska <mliska@suse.cz>
16585 * symbol-summary.h (function_summary): Move constructor
16586 implementation out of class declaration.
16587 (release): Likewise.
16588 (symtab_insertion): Likewise.
16589 (symtab_removal): Likewise.
16590 (symtab_duplication): Likewise.
16593 2018-06-08 Martin Liska <mliska@suse.cz>
16595 * Makefile.in: Remove support for MPX (macros, related functions,
16596 fields in cgraph_node, ...).
16597 * builtin-types.def (BT_BND): Likewise.
16598 (BT_FN_BND_CONST_PTR): Likewise.
16599 (BT_FN_CONST_PTR_BND): Likewise.
16600 (BT_FN_VOID_PTR_BND): Likewise.
16601 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
16602 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
16603 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
16604 (expand_builtin_mempcpy_with_bounds): Likewise.
16605 (expand_builtin_memset_with_bounds): Likewise.
16606 (expand_builtin_memset_args): Likewise.
16607 (std_expand_builtin_va_start): Likewise.
16608 (expand_builtin): Likewise.
16609 (expand_builtin_with_bounds): Likewise.
16610 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
16611 (DEF_LIB_BUILTIN_CHKP): Likewise.
16612 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
16613 (DEF_CHKP_BUILTIN): Likewise.
16614 (BUILT_IN_MEMCPY): Likewise.
16615 (BUILT_IN_MEMMOVE): Likewise.
16616 (BUILT_IN_MEMPCPY): Likewise.
16617 (BUILT_IN_MEMSET): Likewise.
16618 (BUILT_IN_STPCPY): Likewise.
16619 (BUILT_IN_STRCAT): Likewise.
16620 (BUILT_IN_STRCHR): Likewise.
16621 (BUILT_IN_STRCPY): Likewise.
16622 (BUILT_IN_STRLEN): Likewise.
16623 (BUILT_IN_MEMCPY_CHK): Likewise.
16624 (BUILT_IN_MEMMOVE_CHK): Likewise.
16625 (BUILT_IN_MEMPCPY_CHK): Likewise.
16626 (BUILT_IN_MEMSET_CHK): Likewise.
16627 (BUILT_IN_STPCPY_CHK): Likewise.
16628 (BUILT_IN_STRCAT_CHK): Likewise.
16629 (BUILT_IN_STRCPY_CHK): Likewise.
16630 * calls.c (store_bounds): Likewise.
16631 (emit_call_1): Likewise.
16632 (special_function_p): Likewise.
16633 (maybe_warn_nonstring_arg): Likewise.
16634 (initialize_argument_information): Likewise.
16635 (finalize_must_preallocate): Likewise.
16636 (compute_argument_addresses): Likewise.
16637 (expand_call): Likewise.
16638 * cfgexpand.c (expand_call_stmt): Likewise.
16639 (expand_return): Likewise.
16640 (expand_gimple_stmt_1): Likewise.
16641 (pass_expand::execute): Likewise.
16642 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16643 (cgraph_node::remove): Likewise.
16644 (cgraph_node::dump): Likewise.
16645 (cgraph_node::verify_node): Likewise.
16646 * cgraph.h (chkp_function_instrumented_p): Likewise.
16647 (symtab_node::get_alias_target): Likewise.
16648 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
16649 (cgraph_local_p): Likewise.
16650 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
16651 (cgraph_edge::rebuild_references): Likewise.
16652 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
16653 (walk_polymorphic_call_targets): Likewise.
16654 (cgraph_node::expand_thunk): Likewise.
16655 (symbol_table::output_weakrefs): Likewise.
16656 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
16657 (ix86_handle_option): Likewise.
16658 * config/i386/constraints.md: Likewise.
16659 * config/i386/i386-builtin-types.def (BND): Likewise.
16663 * config/i386/i386-builtin.def (BDESC_END): Likewise.
16664 (BDESC_FIRST): Likewise.
16666 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
16667 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
16668 * config/i386/i386.c (enum reg_class): Likewise.
16669 (ix86_target_string): Likewise.
16670 (ix86_option_override_internal): Likewise.
16671 (ix86_conditional_register_usage): Likewise.
16672 (ix86_valid_target_attribute_inner_p): Likewise.
16673 (ix86_set_indirect_branch_type): Likewise.
16674 (ix86_set_current_function): Likewise.
16675 (ix86_function_arg_regno_p): Likewise.
16676 (init_cumulative_args): Likewise.
16677 (ix86_function_arg_advance): Likewise.
16678 (ix86_function_arg): Likewise.
16679 (ix86_pass_by_reference): Likewise.
16680 (ix86_function_value_regno_p): Likewise.
16681 (ix86_function_value_1): Likewise.
16682 (ix86_function_value_bounds): Likewise.
16683 (ix86_return_in_memory): Likewise.
16684 (ix86_setup_incoming_vararg_bounds): Likewise.
16685 (ix86_va_start): Likewise.
16686 (indirect_thunk_need_prefix): Likewise.
16687 (print_reg): Likewise.
16688 (ix86_print_operand): Likewise.
16689 (ix86_expand_call): Likewise.
16690 (ix86_output_function_return): Likewise.
16691 (reg_encoded_number): Likewise.
16692 (BDESC_VERIFYS): Likewise.
16693 (ix86_init_mpx_builtins): Likewise.
16694 (ix86_init_builtins): Likewise.
16695 (ix86_emit_cmove): Likewise.
16696 (ix86_emit_move_max): Likewise.
16697 (ix86_expand_builtin): Likewise.
16698 (ix86_builtin_mpx_function): Likewise.
16699 (ix86_get_arg_address_for_bt): Likewise.
16700 (ix86_load_bounds): Likewise.
16701 (ix86_store_bounds): Likewise.
16702 (ix86_load_returned_bounds): Likewise.
16703 (ix86_store_returned_bounds): Likewise.
16704 (ix86_class_likely_spilled_p): Likewise.
16705 (ix86_hard_regno_mode_ok): Likewise.
16706 (x86_order_regs_for_local_alloc): Likewise.
16707 (ix86_mitigate_rop): Likewise.
16708 (ix86_bnd_prefixed_insn_p): Likewise.
16709 (ix86_mpx_bound_mode): Likewise.
16710 (ix86_make_bounds_constant): Likewise.
16711 (ix86_initialize_bounds): Likewise.
16712 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
16713 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
16714 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
16715 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
16716 (TARGET_CHKP_BOUND_MODE): Likewise.
16717 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
16718 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
16719 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
16720 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
16721 * config/i386/i386.h (TARGET_MPX): Likewise.
16722 (TARGET_MPX_P): Likewise.
16723 (VALID_BND_REG_MODE): Likewise.
16724 (FIRST_BND_REG): Likewise.
16725 (LAST_BND_REG): Likewise.
16726 (enum reg_class): Likewise.
16727 (BND_REG_P): Likewise.
16728 (BND_REGNO_P): Likewise.
16729 (BNDmode): Likewise.
16730 (ADJUST_INSN_LENGTH): Likewise.
16731 * config/i386/i386.md: Likewise.
16732 * config/i386/i386.opt: Likewise.
16733 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
16734 (defined): Likewise.
16735 (LINK_MPX): Likewise.
16736 (MPX_SPEC): Likewise.
16737 (LIBMPX_SPEC): Likewise.
16738 (LIBMPXWRAPPERS_SPEC): Likewise.
16739 (CHKP_SPEC): Likewise.
16740 * config/i386/predicates.md: Likewise.
16741 * dbxout.c (dbxout_type): Likewise.
16742 * doc/extend.texi: Likewise.
16743 * doc/invoke.texi: Likewise.
16744 * doc/md.texi: Likewise.
16745 * doc/tm.texi: Likewise.
16746 * doc/tm.texi.in: Likewise.
16747 * dwarf2out.c (is_base_type): Likewise.
16748 (gen_formal_types_die): Likewise.
16749 (gen_subprogram_die): Likewise.
16750 (gen_type_die_with_usage): Likewise.
16751 (gen_decl_die): Likewise.
16752 (dwarf2out_late_global_decl): Likewise.
16753 * expr.c (expand_assignment): Likewise.
16754 (emit_storent_insn): Likewise.
16755 (store_expr_with_bounds): Likewise.
16756 (store_expr): Likewise.
16757 (expand_expr_real_1): Likewise.
16758 * expr.h (store_expr_with_bounds): Likewise.
16759 * function.c (use_register_for_decl): Likewise.
16760 (struct bounds_parm_data): Likewise.
16761 (assign_parms_augmented_arg_list): Likewise.
16762 (assign_parm_find_entry_rtl): Likewise.
16763 (assign_parm_is_stack_parm): Likewise.
16764 (assign_parm_load_bounds): Likewise.
16765 (assign_bounds): Likewise.
16766 (assign_parms): Likewise.
16767 (expand_function_start): Likewise.
16768 * gcc.c (CHKP_SPEC): Likewise.
16769 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16770 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
16771 (wrestrict_dom_walker::check_call): Likewise.
16772 * gimple.c (gimple_build_call_from_tree): Likewise.
16773 * gimple.h (enum gf_mask): Likewise.
16774 (gimple_call_with_bounds_p): Likewise.
16775 (gimple_call_set_with_bounds): Likewise.
16776 * gimplify.c (gimplify_init_constructor): Likewise.
16777 * ipa-cp.c (initialize_node_lattices): Likewise.
16778 (propagate_constants_across_call): Likewise.
16779 (find_more_scalar_values_for_callers_subset): Likewise.
16780 * ipa-hsa.c (process_hsa_functions): Likewise.
16781 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
16782 * ipa-icf.c (sem_function::merge): Likewise.
16783 * ipa-inline.c (early_inliner): Likewise.
16784 * ipa-pure-const.c (warn_function_noreturn): Likewise.
16785 (warn_function_cold): Likewise.
16786 (propagate_pure_const): Likewise.
16787 * ipa-ref.h (enum GTY): Likewise.
16788 * ipa-split.c (find_retbnd): Likewise.
16789 (consider_split): Likewise.
16790 (split_function): Likewise.
16791 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
16792 * ipa.c (walk_polymorphic_call_targets): Likewise.
16793 (symbol_table::remove_unreachable_nodes): Likewise.
16794 (process_references): Likewise.
16795 (cgraph_build_static_cdtor_1): Likewise.
16796 * lto-cgraph.c (lto_output_node): Likewise.
16797 (output_refs): Likewise.
16798 (compute_ltrans_boundary): Likewise.
16799 (input_overwrite_node): Likewise.
16800 (input_node): Likewise.
16801 (input_cgraph_1): Likewise.
16802 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
16803 * passes.c (pass_manager::execute_early_local_passes): Likewise.
16804 (class pass_chkp_instrumentation_passes): Likewise.
16805 (make_pass_chkp_instrumentation_passes): Likewise.
16806 * passes.def: Likewise.
16807 * rtl.h (struct GTY): Likewise.
16808 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
16809 * stor-layout.c (layout_type): Likewise.
16810 * symtab.c: Likewise.
16811 * target.def: Likewise.
16812 * targhooks.c (default_chkp_bound_type): Likewise.
16813 (default_chkp_bound_mode): Likewise.
16814 (default_builtin_chkp_function): Likewise.
16815 (default_chkp_function_value_bounds): Likewise.
16816 (default_chkp_make_bounds_constant): Likewise.
16817 (default_chkp_initialize_bounds): Likewise.
16818 * targhooks.h (default_chkp_bound_type): Likewise.
16819 (default_chkp_bound_mode): Likewise.
16820 (default_builtin_chkp_function): Likewise.
16821 (default_chkp_function_value_bounds): Likewise.
16822 (default_chkp_make_bounds_constant): Likewise.
16823 (default_chkp_initialize_bounds): Likewise.
16824 * toplev.c (compile_file): Likewise.
16825 (process_options): Likewise.
16826 * tree-core.h (DEF_BUILTIN): Likewise.
16827 (DEF_BUILTIN_CHKP): Likewise.
16828 * tree-inline.c (declare_return_variable): Likewise.
16829 (remap_gimple_stmt): Likewise.
16830 (copy_bb): Likewise.
16831 (initialize_inlined_parameters): Likewise.
16832 (expand_call_inline): Likewise.
16833 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
16834 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
16835 (make_pass_ipa_chkp_produce_thunks): Likewise.
16836 (make_pass_chkp): Likewise.
16837 (make_pass_chkp_opt): Likewise.
16838 (make_pass_chkp_instrumentation_passes): Likewise.
16839 * tree-pretty-print.c (dump_generic_node): Likewise.
16840 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
16841 * tree-ssa-dce.c (propagate_necessity): Likewise.
16842 (eliminate_unnecessary_stmts): Likewise.
16843 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
16844 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
16845 * tree-ssa-sccvn.h: Likewise.
16846 * tree-ssa-strlen.c (get_string_length): Likewise.
16847 (valid_builtin_call): Likewise.
16848 (adjust_last_stmt): Likewise.
16849 (handle_builtin_strchr): Likewise.
16850 (handle_builtin_strcpy): Likewise.
16851 (handle_builtin_stxncpy): Likewise.
16852 (handle_builtin_memcpy): Likewise.
16853 (handle_builtin_strcat): Likewise.
16854 (strlen_check_and_optimize_stmt): Likewise.
16855 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
16856 * tree-streamer-in.c: Likewise.
16857 * tree-streamer.c (record_common_node): Likewise.
16858 * tree.c (tree_code_size): Likewise.
16859 (wide_int_to_tree_1): Likewise.
16860 (type_contains_placeholder_1): Likewise.
16861 (build_common_tree_nodes): Likewise.
16862 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
16863 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
16864 (POINTER_BOUNDS_P): Likewise.
16865 (BOUNDED_TYPE_P): Likewise.
16866 (BOUNDED_P): Likewise.
16867 (CALL_WITH_BOUNDS_P): Likewise.
16868 (pointer_bounds_type_node): Likewise.
16869 * value-prof.c (gimple_ic): Likewise.
16870 * var-tracking.c (vt_add_function_parameters): Likewise.
16871 * varasm.c (make_decl_rtl): Likewise.
16872 (assemble_start_function): Likewise.
16873 (output_constant): Likewise.
16874 (maybe_assemble_visibility): Likewise.
16875 * varpool.c (ctor_for_folding): Likewise.
16876 * chkp-builtins.def: Remove.
16877 * ipa-chkp.c: Remove.
16878 * ipa-chkp.h: Remove.
16879 * rtl-chkp.c: Remove.
16880 * rtl-chkp.h: Remove.
16881 * tree-chkp-opt.c: Remove.
16882 * tree-chkp.c: Remove.
16883 * tree-chkp.h: Remove.
16885 2018-06-07 Carl Love <cel@us.ibm.com>
16887 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16888 vextract_fp_from_shortl): Add BE support.
16890 2018-06-07 Paul Koning <ni1d@arrl.net>
16892 * compare-elim.c (try_merge_compare): Don't merge compare if
16893 address contains a side effect.
16894 (try_eliminate_compare): Likewise.
16896 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
16898 * config.gcc: Support "tremont".
16899 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
16900 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16902 * config/i386/i386.c (m_TREMONT): Define.
16903 (processor_target_table): Add "tremont".
16904 (PTA_TREMONT): Define.
16905 (ix86_lea_outperforms): Add TARGET_TREMONT.
16906 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
16907 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
16908 and M_INTEL_GOLDMONT_PLUS.
16909 (fold_builtin_cpu): Add "tremont".
16910 (ix86_add_stmt_cost): Add TARGET_TREMONT.
16911 (ix86_option_override_internal): Add "tremont".
16912 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
16913 (processor_type): Add PROCESSOR_TREMONT.
16914 * config/i386/x86-tune.def: Add m_TREMONT.
16915 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
16917 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16919 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
16920 symbol defined for msp430i* devices to be lower case.
16922 2018-06-07 Richard Biener <rguenther@suse.de>
16924 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
16925 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
16926 Properly wrap signed arithmetic if overflow wraps.
16928 2018-06-07 Jakub Jelinek <jakub@redhat.com>
16930 PR tree-optimization/69615
16931 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
16932 of a cast from a same precision integral SSA_NAME in a bb dominated
16933 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
16934 cast to utype if rhs2 has already a compatible type.
16936 2018-06-07 Richard Biener <rguenther@suse.de>
16938 PR tree-optimization/85935
16939 * graphite-scop-detection.c (find_params_in_bb): Analyze
16940 condition operands with respect to the correct loop. Assert
16941 the analysis doesn't fail.
16943 2018-06-04 Carl Love <cel@us.ibm.com>
16945 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
16946 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
16947 as it is slightly cheaper.
16948 (first_match_or_eos_index_<mode>):
16949 Calculate index using natural element order.
16950 (first_match_index_<mode>):
16951 Calculate index using natural element order.
16952 (first_match_or_eos_index_<mode>):
16953 Calculate index using natural order.
16954 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
16955 for BE and LE modes.
16956 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
16957 P9V_BUILTIN_VCLZLSBB_V16QI.
16958 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
16961 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
16963 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16964 indentation and line wrap for many prototypes. Add missing
16965 @smallexample directives around block of prototypes for vec_xl and
16968 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
16970 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
16971 track if we pass or return IEEE 128-bit floating point.
16972 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
16973 C++ mangling that is compatible with GCC 8.1.
16974 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
16975 (init_cumulative_args): Note if we pass or return IEEE 128-bit
16976 floating point types.
16977 (rs6000_function_arg_advance_1): Likewise.
16978 (rs6000_mangle_type): Optionally generate mangled names that match
16979 what GCC 8.1 generated for IEEE 128-bit floating point types.
16980 (rs6000_globalize_decl_name): If we have an external function that
16981 passes or returns IEEE 128-bit types, generate a weak reference
16982 from the mangled name used in GCC 8.1 to the current mangled
16984 (rs6000_init_builtins): Make __ibm128 use the long double type if
16985 long double is IBM extended double. Make __float128 use the long
16986 double type if long double is IEEE 128-bit.
16989 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
16990 macro for __ibm128 built-in functions.
16991 (PACK_IF): Add __ibm128 pack/unpack functions.
16992 (UNPACK_IF): Likewise.
16993 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
16994 enable long double built-in functions if long double is IEEE
16995 128-bit floating point.
16996 (rs6000_invalid_builtin): Update long double built-in function
16998 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
16999 functions, adjust the built-in function to use the long double
17000 built-in function if __ibm128 and long double are the same type.
17001 * doc/extend.texi (PowerPC builtins): Update documention for
17002 __builtin_{,un}pack_longdouble. Add documentation for
17003 __builtin_{,un}pack_ibm128.
17005 2018-06-06 Jim Wilson <jimw@sifive.com>
17007 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
17008 (struct machine_function): New field interrupt_mode.
17009 (riscv_handle_type_attribute): New function. Add forward declaration.
17010 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
17011 (riscv_expand_epilogue): Check interrupt_mode field.
17012 (riscv_set_current_function): Check interrupt attribute args and
17013 set interrupt_mode field.
17014 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
17015 (riscv_sret, riscv_uret): New.
17016 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
17017 new arguments to interrupt attribute.
17019 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
17022 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
17023 Don't handle -mcpu=power8 if -mpower9-vector is also used.
17025 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
17027 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
17028 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
17029 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
17030 several redundant entries.
17032 2018-06-06 David Malcolm <dmalcolm@redhat.com>
17034 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
17035 type from "rtx" to "rtx_insn *".
17036 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
17037 for local "call_insn", removing cast.
17038 (ix86_expand_call): Likewise, introducing a "call_insn" local.
17040 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
17042 PR tree-optimization/86066
17043 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
17044 for BIT_INSERT_EXPR stores.
17046 2018-06-06 Richard Biener <rguenther@suse.de>
17048 PR tree-optimization/86062
17049 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
17050 component refs ontop
17051 of to be offsetted base.
17053 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17055 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
17056 to be static and remove check on interrupt attribute name.
17058 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
17060 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
17061 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
17063 2018-06-05 Steve Ellcey <sellcey@cavium.com>
17066 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
17068 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
17069 Remove second argument, change how error is called.
17070 (aarch64_layout_arg): Remove second argument from
17071 aarch64_err_no_fpadvsimd call.
17072 (aarch64_init_cumulative_args): Ditto.
17073 (aarch64_gimplify_va_arg_expr): Ditto.
17074 * config/aarch64/aarch64.md (mov<mode>): Ditto.
17076 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
17078 * config/i386/i386.md (simple_return_indirect_internal): New expander.
17079 (*simple_return_indirect_internal<mode>): Rename from
17080 simple_return_indirect_internal. Use W mode iterator.
17081 (rstorssp): New expander.
17082 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
17083 (clrssbsy): New expander.
17084 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
17086 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17088 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
17090 (cmse_check_pointed_object): Likewise.
17092 2018-06-05 Martin Liska <mliska@suse.cz>
17094 PR gcov-profile/47618
17095 * doc/invoke.texi: Document how -fprofile-dir format
17098 2018-06-05 Richard Biener <rguenther@suse.de>
17100 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
17101 removal pretend DOM info isn't available so we do not update
17102 it and only remove edges, not dominated blocks. Actually free
17103 DOM info in case we removed something. Remove unreachable blocks.
17104 (mfb_keep_latches): Work with either DOM info or marked backedges.
17105 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
17106 first. Mark backedges if DOM info isn't available.
17107 (Re-)compute DOM info after cleanup_control_flow_pre.
17109 2018-06-05 Richard Biener <rguenther@suse.de>
17111 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
17112 (locus_discrim_hasher::hash): Adjust.
17113 (locus_discrim_hasher::equal): Likewise.
17114 (next_discriminator_for_locus): Work on line directly.
17115 (same_line_p): Pass in expanded locus1 as well.
17116 (assign_discriminators): Avoid redundant location expansions.
17118 2018-06-05 Richard Biener <rguenther@suse.de>
17120 PR tree-optimization/86046
17121 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
17122 if required after clearing TREE_ADDRESSABLE.
17124 2018-06-05 Richard Biener <rguenther@suse.de>
17126 PR tree-optimization/86047
17127 * tree-ssa-loop.c (for_each_index): Glob handling of all
17128 decls and constants and really handle all of them.
17130 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17133 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
17134 qualifier_void_pointer and qualifier_const_void_pointer.
17135 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
17136 (arm_init_builtins): Handle the above.
17137 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
17138 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
17139 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
17142 2018-06-05 Martin Liska <mliska@suse.cz>
17144 * auto-profile.c (read_autofdo_file): Do not use
17145 gcov_ctr_summary struct.
17146 (afdo_callsite_hot_enough_for_early_inline): Likewise.
17147 * coverage.c (struct counts_entry): Likewise.
17148 (read_counts_file): Read just single summary entry.
17149 (get_coverage_counts): Use gcov_summary struct.
17150 * coverage.h (get_coverage_counts): Likewise.
17151 * gcov-dump.c (dump_working_sets): Likewise.
17152 (tag_summary): Dump just single summary.
17153 * gcov-io.c (gcov_write_summary): Write just histogram
17155 (gcov_read_summary): Read just single summary.
17156 (compute_working_sets): Use gcov_summary struct.
17157 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
17158 of GCOV_COUNTERS_SUMMABLE.
17159 (GCOV_COUNTERS_SUMMABLE): Remove.
17160 (GCOV_FIRST_VALUE_COUNTER): Replace with
17161 GCOV_COUNTER_V_INTERVAL.
17162 (struct gcov_ctr_summary): Remove.
17163 (struct gcov_summary): Directly use fields of former
17165 (compute_working_sets): Use gcov_summary struct.
17166 * gcov.c (read_count_file): Do not use ctrs fields.
17167 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
17169 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
17171 * profile.c: Likewise.
17172 * profile.h: Likewise.
17174 2018-06-05 Martin Liska <mliska@suse.cz>
17176 PR gcov-profile/84846
17177 * gcov.c (output_lines): Print working directory only
17178 in intermediate format.
17180 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
17182 * config/s390/s390-builtin-types.def: Add void function type.
17183 * config/s390/s390-builtins.def: Use the function type for the
17186 2018-06-04 Jim Wilson <jimw@sifive.com>
17188 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
17190 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
17191 and maybe_eh_return. Change regno to unsigned int. Use new args to
17192 handle EH_RETURN_DATA_REGNO registers properly.
17193 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
17194 (riscv_expand_epilogue): Update comment. Change argument name and
17195 type. Update code to use new name and type. Pass new args to
17196 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
17198 * config/riscv/riscv.md (NORMAL_RETURN): New.
17199 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
17200 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
17201 (eh_return): Call gen_eh_return_internal and emit barrier.
17202 (eh_return_internal): Call riscv_expand_epilogue.
17204 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
17206 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
17207 bit_insertion field and declare can_be_merged_into method.
17208 (merged_store_group::can_be_merged_into): New method.
17209 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
17210 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
17211 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
17213 2018-06-04 Richard Biener <rguenther@suse.de>
17215 PR tree-optimization/85955
17216 * builtins.c (fold_builtin_sincos): Convert pointers to
17217 destination to appropriate type before dereferencing.
17219 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
17221 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
17223 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
17225 * expr.c (expand_expr_real_1): Force the operand into memory if
17226 its TYPE_MODE is BLKmode and if there is no integer mode for
17227 the number of bits being extracted.
17229 2018-06-04 Jakub Jelinek <jakub@redhat.com>
17233 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
17234 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
17236 2018-06-04 Richard Biener <rguenther@suse.de>
17238 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
17239 (cleanup_tree_cfg_noloop): ... single caller. Do
17240 start_recording_case_labels later.
17242 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
17244 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
17245 to _IMMINTRIN_H_INCLUDED.
17246 * config/i386/pconfigintrin.h: Ditto.
17247 * config/i386/waitpkgintrin.h: Ditto.
17248 * config/i386/immintrin.h: Add includes for sgxintrin.h,
17249 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
17250 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
17251 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
17252 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
17253 waitpkgintrin.h and cldemoteintrin.h.
17255 2018-06-04 Richard Biener <rguenther@suse.de>
17257 PR tree-optimization/86038
17258 * tracer.c (find_best_successor): Check probability for
17259 being initialized, bail out if not.
17261 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
17264 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
17265 of bits to ignore when comparing architectures.
17267 2018-06-04 Jakub Jelinek <jakub@redhat.com>
17269 PR tree-optimization/69615
17270 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
17271 maximum or minimum of the type, try to merge it also as if
17272 range1 is + [-, x - 1] or + [x + 1, -].
17275 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
17277 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
17279 PR tree-optimization/86034
17280 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
17281 the unsigned bitfield type in a bit insertion sequence if it does not
17282 have a larger precision than the bitfield size.
17283 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
17285 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
17287 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
17289 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
17291 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
17292 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
17293 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
17294 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
17296 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
17298 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
17299 Disable -fdelete-null-pointer-checks for ELF toolchain.
17301 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
17302 Kito Cheng <kito.cheng@gmail.com>
17304 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
17305 (nds32le-*-*, nds32be-*-*): Integrate checking process.
17306 (nds32*-*-*): Add glibc and uclibc conditions.
17307 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
17308 (TARGET_EXCEPT_UNWIND_INFO): Define.
17309 * config/nds32/elf.h: New file.
17310 * config/nds32/linux.h: New file.
17311 * config/nds32/nds32-elf.opt: New file.
17312 * config/nds32/nds32-linux.opt: New file.
17313 * config/nds32/nds32-fp-as-gp.c
17314 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
17315 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
17317 (nds32_asm_file_end): Ditto.
17318 (nds32_print_operand): Ditto.
17319 (nds32_insert_attributes): Ditto.
17320 (nds32_init_libfuncs): New function.
17321 (TARGET_HAVE_TLS): Define.
17322 (TARGET_INIT_LIBFUNCS): Define.
17323 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
17325 (TARGET_ELF): Apply different mcmodel setting.
17326 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
17327 been migrated into elf.h and linux.h files.
17328 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
17329 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
17330 (mcmodel): The content has been migrated into nds32-elf.opt and
17331 nds32-linux.opt files.
17332 * config/nds32/t-elf: New file.
17333 * config/nds32/t-linux: New file.
17335 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
17336 Shiva Chen <shiva0217@gmail.com>
17338 * config/nds32/constants.md (unspec_volatile_element): Add
17339 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
17340 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
17342 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
17343 (make_pass_nds32_fp_as_gp): Declare.
17344 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
17346 (nds32_asm_function_end_prologue): Remove unused asm output.
17347 (nds32_asm_function_begin_epilogue): Remove unused asm output.
17348 (nds32_asm_file_start): Output necessary fp_as_gp information.
17349 (nds32_option_override): Adjust register usage.
17350 (nds32_expand_prologue): Consider fp_as_gp situation.
17351 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
17352 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
17355 (simple_return): Ditto.
17356 (omit_fp_begin): Output special directive for fp_as_gp.
17357 (omit_fp_end): Output special directive for fp_as_gp.
17358 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
17359 mforbid-fp-as-gp): New options.
17361 2018-06-01 Mark Wielaard <mark@klomp.org>
17363 * dwarf2out.c (dwarf2out_finish): Remove generation of
17364 DW_AT_loclists_base.
17366 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
17368 * gimple-ssa-store-merging.c: Include gimple-fold.h.
17369 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
17370 (struct merged_store_group): Add bit_insertion field.
17371 (dump_char_array): Use standard hexadecimal format.
17372 (merged_store_group::merged_store_group): Set bit_insertion to false.
17373 (merged_store_group::apply_stores): Use optimal buffer size. Deal
17374 with BIT_INSERT_EXPR stores. Move up code updating the mask and
17375 also print the mask in the dump file.
17376 (pass_store_merging::gate): Minor tweak.
17377 (imm_store_chain_info::coalesce_immediate): Fix wrong association
17378 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
17379 stores with INTEGER_CST stores.
17380 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
17381 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
17382 and use it throughout. Generate bit insertion sequences if need be.
17383 (pass_store_merging::process_store): Remove redundant condition.
17384 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
17386 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
17388 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
17389 the 128-bit floating point types. Fix function comment.
17391 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17393 * config/aarch64/aarch64-simd.md
17394 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
17396 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
17399 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
17401 PR tree-optimization/85989
17402 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
17404 (backprop::intersect_uses): Check it when deciding whether this
17405 is a backedge reference.
17406 (backprop::process_block): Add each phi to m_visited_phis
17407 after visiting it, then clear it at the end.
17409 2018-06-01 Richard Biener <rguenther@suse.de>
17411 * tree-vectorizer.h (vect_dr_stmt): New function.
17412 (vect_get_load_cost): Adjust.
17413 (vect_get_store_cost): Likewise.
17414 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
17415 Use vect_dr_stmt instead of DR_SMTT.
17416 (vect_record_base_alignments): Likewise.
17417 (vect_calculate_target_alignment): Likewise.
17418 (vect_compute_data_ref_alignment): Likewise and make static.
17419 (vect_update_misalignment_for_peel): Likewise.
17420 (vect_verify_datarefs_alignment): Likewise.
17421 (vector_alignment_reachable_p): Likewise.
17422 (vect_get_data_access_cost): Likewise. Pass down
17423 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
17424 (vect_get_peeling_costs_all_drs): Likewise.
17425 (vect_peeling_hash_get_lowest_cost): Likewise.
17426 (vect_enhance_data_refs_alignment): Likewise.
17427 (vect_find_same_alignment_drs): Likewise.
17428 (vect_analyze_data_refs_alignment): Likewise.
17429 (vect_analyze_group_access_1): Likewise.
17430 (vect_analyze_group_access): Likewise.
17431 (vect_analyze_data_ref_access): Likewise.
17432 (vect_analyze_data_ref_accesses): Likewise.
17433 (vect_vfa_segment_size): Likewise.
17434 (vect_small_gap_p): Likewise.
17435 (vectorizable_with_step_bound_p): Likewise.
17436 (vect_prune_runtime_alias_test_list): Likewise.
17437 (vect_analyze_data_refs): Likewise.
17438 (vect_supportable_dr_alignment): Likewise.
17439 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
17440 (vect_gen_prolog_loop_niters): Likewise.
17441 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17442 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
17444 (vect_recog_mask_conversion_pattern): Likewise.
17445 (vect_try_gather_scatter_pattern): Likewise.
17446 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
17447 to vect_get_store_cost.
17448 (vect_get_store_cost): Get stmt_info instead of DR.
17449 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
17450 (vect_get_load_cost): Get stmt_info instead of DR.
17452 2018-06-01 Richard Biener <rguenther@suse.de>
17454 PR middle-end/86017
17455 * gimple-fold.c (var_decl_component_p): Also allow offsetted
17456 vars wrapped in MEM_REFs.
17458 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
17460 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17461 Fix subreg tests so that we only return a choice between
17462 GENERAL_REGS and FP_REGS if the original classes included both.
17464 2018-06-01 Richard Biener <rguenther@suse.de>
17467 * tree-ssa-structalias.c (get_function_part_constraint):
17468 Handle NULL fi->decl.
17469 (find_func_aliases_for_call): Properly handle indirect
17470 fi from direct call.
17471 (find_func_clobbers): Likewise.
17472 (ipa_pta_execute): Likewise.
17473 (create_variable_info_for): For functions that are ifunc_resolver
17474 resolve to a varinfo that contains the result of the resolver call.
17475 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
17478 2018-05-31 Michael Collison <michael.collison@arm.com>
17480 * config/aarch64/aarch64.md:
17481 (*fix_to_zero_extenddfdi2): New pattern.
17482 * gcc.target/aarch64/fix_extend1.c: New testcase.
17484 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
17486 PR middle-end/78809
17487 PR middle-end/83026
17488 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
17489 and BUILT_IN_STRNCMP_EQ.
17490 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
17491 BUILT_IN_STRNCMP_EQ.
17492 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
17493 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
17494 (gimple_fold_builtin): Likewise.
17495 * tree-ssa-strlen.c (compute_string_length): New function.
17496 (determine_min_obsize): New function.
17497 (handle_builtin_string_cmp): New function to handle calls to
17498 string compare functions.
17499 (strlen_optimize_stmt): Add handling to builtin string compare
17501 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17502 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
17503 * tree.c (build_common_builtin_nodes): Add new defines of
17504 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
17506 2018-05-31 Jakub Jelinek <jakub@redhat.com>
17509 * bb-reorder.c (pass_partition_blocks::gate): Return false for
17510 functions with naked attribute.
17512 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
17514 * config/i386/sse.md (avx_vec_concat<mode>):
17515 Substitute concat_tg_mode mode attribute with xtg_mode.
17516 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
17517 (concat_tg_mode): Remove mode attribute.
17519 2018-05-31 Martin Sebor <msebor@redhat.com>
17522 * calls.c (alloc_max_size): Correct a logic error/typo.
17523 Treat excessive arguments as infinite. Warn for invalid arguments.
17524 * doc/invoke.texi (-Walloc-size-larger-than): Update.
17526 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
17529 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
17530 and movx for Haswell.
17532 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
17533 Cesar Philippidis <cesar@codesourcery.com>
17535 PR middle-end/85879
17536 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
17537 when emitting error on private/firstprivate reductions.
17538 * omp-low.c (lower_omp_target): Avoid reference-type processing
17539 on pointers for firstprivate clause.
17541 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
17543 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
17546 * config/aarch64/aarch64-simd.md
17547 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
17548 (aarch64_ld1_x3_<mode>): Likewise
17549 (aarch64_st1x2<VALLDIF:mode>): Likewise
17550 (aarch64_st1_x2_<mode>): Likewise
17551 (aarch64_st1x3<VALLDIF:mode>): Likewise
17552 (aarch64_st1_x3_<mode>): Likewise
17553 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
17554 (vld1_s8_x3): Likewise.
17555 (vld1_u16_x3): Likewise.
17556 (vld1_s16_x3): Likewise.
17557 (vld1_u32_x3): Likewise.
17558 (vld1_s32_x3): Likewise.
17559 (vld1_u64_x3): Likewise.
17560 (vld1_s64_x3): Likewise.
17561 (vld1_f16_x3): Likewise.
17562 (vld1_f32_x3): Likewise.
17563 (vld1_f64_x3): Likewise.
17564 (vld1_p8_x3): Likewise.
17565 (vld1_p16_x3): Likewise.
17566 (vld1_p64_x3): Likewise.
17567 (vld1q_u8_x3): Likewise.
17568 (vld1q_s8_x3): Likewise.
17569 (vld1q_u16_x3): Likewise.
17570 (vld1q_s16_x3): Likewise.
17571 (vld1q_u32_x3): Likewise.
17572 (vld1q_s32_x3): Likewise.
17573 (vld1q_u64_x3): Likewise.
17574 (vld1q_s64_x3): Likewise.
17575 (vld1q_f16_x3): Likewise.
17576 (vld1q_f32_x3): Likewise.
17577 (vld1q_f64_x3): Likewise.
17578 (vld1q_p8_x3): Likewise.
17579 (vld1q_p16_x3): Likewise.
17580 (vld1q_p64_x3): Likewise.
17581 (vst1_s64_x2): Likewise.
17582 (vst1_u64_x2): Likewise.
17583 (vst1_f64_x2): Likewise.
17584 (vst1_s8_x2): Likewise.
17585 (vst1_p8_x2): Likewise.
17586 (vst1_s16_x2): Likewise.
17587 (vst1_p16_x2): Likewise.
17588 (vst1_s32_x2): Likewise.
17589 (vst1_u8_x2): Likewise.
17590 (vst1_u16_x2): Likewise.
17591 (vst1_u32_x2): Likewise.
17592 (vst1_f16_x2): Likewise.
17593 (vst1_f32_x2): Likewise.
17594 (vst1_p64_x2): Likewise.
17595 (vst1q_s8_x2): Likewise.
17596 (vst1q_p8_x2): Likewise.
17597 (vst1q_s16_x2): Likewise.
17598 (vst1q_p16_x2): Likewise.
17599 (vst1q_s32_x2): Likewise.
17600 (vst1q_s64_x2): Likewise.
17601 (vst1q_u8_x2): Likewise.
17602 (vst1q_u16_x2): Likewise.
17603 (vst1q_u32_x2): Likewise.
17604 (vst1q_u64_x2): Likewise.
17605 (vst1q_f16_x2): Likewise.
17606 (vst1q_f32_x2): Likewise.
17607 (vst1q_f64_x2): Likewise.
17608 (vst1q_p64_x2): Likewise.
17609 (vst1_s64_x3): Likewise.
17610 (vst1_u64_x3): Likewise.
17611 (vst1_f64_x3): Likewise.
17612 (vst1_s8_x3): Likewise.
17613 (vst1_p8_x3): Likewise.
17614 (vst1_s16_x3): Likewise.
17615 (vst1_p16_x3): Likewise.
17616 (vst1_s32_x3): Likewise.
17617 (vst1_u8_x3): Likewise.
17618 (vst1_u16_x3): Likewise.
17619 (vst1_u32_x3): Likewise.
17620 (vst1_f16_x3): Likewise.
17621 (vst1_f32_x3): Likewise.
17622 (vst1_p64_x3): Likewise.
17623 (vst1q_s8_x3): Likewise.
17624 (vst1q_p8_x3): Likewise.
17625 (vst1q_s16_x3): Likewise.
17626 (vst1q_p16_x3): Likewise.
17627 (vst1q_s32_x3): Likewise.
17628 (vst1q_s64_x3): Likewise.
17629 (vst1q_u8_x3): Likewise.
17630 (vst1q_u16_x3): Likewise.
17631 (vst1q_u32_x3): Likewise.
17632 (vst1q_u64_x3): Likewise.
17633 (vst1q_f16_x3): Likewise.
17634 (vst1q_f32_x3): Likewise.
17635 (vst1q_f64_x3): Likewise.
17636 (vst1q_p64_x3): Likewise.
17638 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17640 * config/msp430/msp430.c (msp430_output_labelref): Prepend
17641 user_label_prefix to name.
17643 * tree-core.h: Update comment about the format of NAME string
17644 passed to handler in attribute_spec.
17646 * config/msp430/msp430.md: Remove erroneous subreg expression from
17647 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
17648 zero_extend{q,h}isi2.
17650 2018-05-30 Borislav Petkov <bp@suse.de>
17652 * doc/extend.texi: Document some architecture specific
17653 constraints and sort entries.
17655 2018-05-30 Martin Sebor <msebor@redhat.com>
17657 PR middle-end/85369
17658 * builtins.c (expand_builtin_stpcpy_1): New function.
17659 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
17660 only if the former succeeds.
17662 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
17664 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
17667 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17669 * doc/invoke.texi (-flinker-output): Document
17671 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17673 * passes.c (ipa_write_summaries): Only modify statements if body
17675 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
17676 incrementally linking.
17677 (ipa_passes): Likewise.
17678 * lto-cgraph.c (lto_output_node): When incrementally linking do not
17679 pass down resolution info.
17680 * common.opt (flag_incremental_link): Update info.
17681 * gcc.c (plugin specs): Turn flinker-output=* to
17682 -plugin-opt=-linker-output-known
17683 * toplev.c (compile_file): Also cut compilation when doing incremental
17685 * flag-types. (enum lto_partition_model): Add
17686 LTO_LINKER_OUTPUT_NOLTOREL.
17687 (invoke.texi): Add -flinker-output docs.
17688 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
17689 link same way as WPA; do not stream in dead initializers.
17691 * dwarf2out.c (dwarf2out_die_ref_for_decl,
17692 darf2out_register_external_decl): Support incremental link.
17694 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17696 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
17698 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17700 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
17701 it down to simple_object_copy_lto_debug_sections.
17702 (run_gcc): Determine incremental LTO link time and configure
17703 lto1 into non-wpa mode, disable renaming of debug sections.
17705 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
17707 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
17708 descriptions of various incorrectly documented functions.
17710 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
17713 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
17714 address check not strict.
17716 2018-05-30 Richard Biener <rguenther@suse.de>
17718 PR tree-optimization/85964
17719 * tracer.c (better_p): Drop initialized count check, we only
17720 call the function with initialized counts now.
17721 (find_best_successor): Do find a best edge if one
17722 has uninitialized count.
17723 (find_best_predecessor): Likewise. Do BB frequency check only
17724 if count is initialized.
17726 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
17728 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
17729 (aarch64_ldrstr_offset_compare): New.
17730 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
17731 load/store orderings.
17732 (aarch64_gen_adjusted_ldpstp): Likewise.
17734 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17736 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17737 Check for subset of GENERAL_REGS and FP_REGS.
17738 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
17741 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
17743 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
17744 and wi::to_poly_offset. Add the current offset and then check
17745 whether the sum fits, rather than using an unchecked addition of
17746 a checked term. Check for a shwi rather than a uhwi.
17747 * expr.c (get_bit_range): Use tree_to_poly_uint64.
17748 (store_constructor): Use poly_int_tree_p.
17749 (expand_expr_real_1): Likewise.
17750 * function.c (assign_temp): Likewise.
17751 * fold-const.c (const_binop): Use poly_int_tree_p and
17752 wi::to_poly_offset.
17753 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
17755 * ipa-icf-gimple.c (func_checker::compare_operand): Use
17756 to_poly_offset for MEM offsets.
17757 * ipa-icf.c (sem_variable::equals): Likewise.
17758 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
17759 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
17760 wi::to_poly_offset for BIT_FIELD_REF offsets.
17761 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
17762 wi::to_poly_offset.
17763 * var-tracking.c (emit_note_insn_var_location): Use
17764 tree_to_poly_uint64.
17766 2018-05-29 Jim Wilson <jimw@sifive.com>
17768 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
17770 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
17773 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
17774 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
17776 (sse4_1_round<mode>2): Use nonimmediate_operand
17777 for operand 1 predicate.
17779 2018-05-29 Martin Sebor <msebor@redhat.com>
17780 Richard Biener <rguenther@suse.de>
17783 * calls.c (get_size_range): Call determine_value_range instead
17784 of get_value_range..
17785 * tree-vrp.h (determine_value_range): Declared new function.
17786 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
17788 2018-05-29 Richard Biener <rguenther@suse.de>
17790 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
17791 sure to use non-pattern stmts for get_earlier_stmt arguments.
17792 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
17793 called on pattern stmts.
17794 (get_later_stmt): Likewise.
17796 2018-05-29 Martin Liska <mliska@suse.cz>
17798 PR gcov-profile/85759
17799 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
17802 2018-05-29 Jakub Jelinek <jakub@redhat.com>
17804 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
17806 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
17810 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
17811 VEC_PACK_FLOAT_EXPR): New tree codes.
17812 * tree-pretty-print.c (op_code_prio): Handle
17813 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
17814 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
17815 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
17816 * tree-inline.c (estimate_operator_cost): Likewise.
17817 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
17818 * fold-const.c (const_binop): Likewise.
17819 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
17820 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
17821 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17822 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
17823 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
17824 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
17825 * expr.c (expand_expr_real_2): Likewise.
17826 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
17827 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
17828 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
17830 * optabs.c (expand_widen_pattern_expr): For
17831 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
17832 sign from result type rather than operand's type.
17833 (expand_binop_directly): For vec_packu_float_optab and
17834 vec_packs_float_optab allow result type to be different from operand's
17836 * optabs-tree.c (optab_for_tree_code): Handle
17837 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
17838 VEC_PACK_FLOAT_EXPR. Formatting fixes.
17839 * tree-vect-generic.c (expand_vector_operations_1): Handle
17840 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
17841 VEC_PACK_FLOAT_EXPR.
17842 * tree-vect-stmts.c (supportable_widening_operation): Handle
17844 (supportable_narrowing_operation): Handle FLOAT_EXPR.
17845 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
17846 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
17847 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
17848 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
17850 (vec_pack<floatprefix>_float_<mode>): New expander.
17851 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
17853 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
17854 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
17855 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
17856 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
17857 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
17859 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
17860 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
17861 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
17862 VEC_PACK_FLOAT_EXPR): Document.
17864 2018-05-29 Richard Biener <rguenther@suse.de>
17866 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
17868 (stmt_vec_info_vec): Make pointer.
17869 (init_stmt_vec_info_vec): Remove.
17870 (free_stmt_vec_info_vec): Likewise.
17871 (set_stmt_vec_info_vec): New function.
17872 (free_stmt_vec_infos): Likewise.
17873 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
17874 (set_vinfo_for_stmt): Likewise.
17875 (get_earlier_stmt): Likewise.
17876 (get_later_stmt): Likewise.
17877 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
17878 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
17879 (vec_info::~vec_info): Free stmt_vec_infos.
17880 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
17881 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
17882 (pass_slp_vectorize::execute): Likewise.
17883 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
17884 (free_stmt_vec_info_vec): Likewise.
17885 (set_stmt_vec_info_vec): New function.
17886 (free_stmt_vec_infos): Likewise.
17887 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
17888 the global stmt_vec_info_vec.
17889 * tree-parloops.c (gather_scalar_reductions): Use
17890 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
17893 2018-05-29 Richard Biener <rguenther@suse.de>
17895 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
17897 2018-05-29 Martin Liska <mliska@suse.cz>
17898 David Malcolm <dmalcolm@redhat.com>
17900 * vec.c (test_reverse): New.
17901 (vec_c_tests): Add new test.
17902 * vec.h (vl_ptr>::reverse): New function.
17904 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
17906 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
17908 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
17911 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17913 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
17915 2018-05-28 Richard Biener <rguenther@suse.de>
17917 PR tree-optimization/85933
17918 * tree-vect-data-refs.c (vect_record_base_alignments): Only
17919 look at stmts marked as vectorizable.
17921 2018-05-28 Richard Biener <rguenther@suse.de>
17923 PR tree-optimization/85934
17924 * tree-vect-generic.c (expand_vector_operations_1): Hoist
17925 vector boolean check before scalar optimization.
17927 2018-05-28 Jakub Jelinek <jakub@redhat.com>
17929 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
17932 2018-05-28 Mark Wielaard <mark@klomp.org>
17934 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
17935 if it is an expression containing a minus sign.
17937 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
17939 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
17941 2018-05-27 Paul Koning <ni1d@arrl.net>
17943 * config/pdp11/pdp11.md (truncsihi2): Remove.
17945 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
17946 Chung-Ju Wu <jasonwucj@gmail.com>
17948 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
17950 (unaligned_store_dw): Ditto.
17951 * config/nds32/nds32-memory-manipulation.c
17952 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
17953 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
17954 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
17955 (emit_setmem_word_loop): Rename to ...
17956 (emit_setmem_doubleword_loop): ... this.
17957 (nds32_gen_dup_4_byte_to_word_value): New function.
17958 (nds32_gen_dup_8_byte_to_double_word_value): New function.
17959 (nds32_expand_setmem_loop): Refine implementation.
17960 (nds32_expand_setmem_loop_v3m): Ditto.
17961 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
17964 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
17966 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
17968 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
17970 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
17971 (nds32_init_machine_status): Initialize machine->attr_naked_p and
17972 machine->attr_no_prologue_p.
17973 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
17974 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
17975 (nds32_expand_epilogue): Consider attr_naked_p.
17976 (nds32_expand_epilogue_v3pop): Likewise.
17977 (nds32_can_use_return_insn): Likewise.
17978 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
17979 attr_no_prologue_p fields.
17980 * config/nds32/nds32.opt (mret-in-naked-func): New option.
17982 2018-05-27 Jakub Jelinek <jakub@redhat.com>
17985 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
17987 * config/i386/sse.md
17988 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
17990 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
17992 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
17994 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
17996 (*<floatsuffix>floatv2div2sf2): Rename to ...
17997 (*float<floatunssuffix>v2div2sf2): ... this.
17998 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
17999 (float<floatunssuffix>v2div2sf2_mask): ... this.
18000 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
18001 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
18002 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
18004 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
18006 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
18008 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
18010 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
18012 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
18014 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
18015 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
18016 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
18017 gen_ufix_truncv8dfv8si2.
18018 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
18019 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
18020 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
18021 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
18022 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
18023 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
18024 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
18025 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
18027 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
18031 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
18033 2018-05-25 Jim Wilson <jimw@sifive.com>
18035 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
18036 * config/riscv/riscv.c (struct machine_function): Add
18037 interrupt_handler_p and attribute_checked_p fields.
18038 (riscv_attribute_table): Add interrupt.
18039 (riscv_interrupt_type_p): New.
18040 (riscv_save_reg_p): Save extra regs for interrupt handler.
18041 (riscv_use_save_libcall): Return false for interrupt handler.
18042 (riscv_first_stack_step): Add forward declaration.
18043 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
18044 for interrupt handler with large frame. Use it for saved reg list.
18045 (riscv_expand_prologue): Move flag_stack_usage_info support to
18046 eliminate duplication.
18047 (riscv_expand_epilogue): Generate mret for interrupt handler.
18048 (riscv_epilogue_uses): New.
18049 (riscv_can_use_return_insn): Return false for interrupt handler.
18050 (riscv_function_ok_for_sibcall): Likewise.
18051 (riscv_set_current_function): Add interrupt handler support.
18052 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
18053 * config/riscv/riscv.md (UNSPECV_MRET): New.
18055 (riscv_frflags, riscv_fsflags): Use tab after opcode.
18057 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
18059 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
18061 PR tree-optimization/85712
18062 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
18063 this candidate has already been replaced in-situ by a copy.
18065 2018-05-25 Jason Merrill <jason@redhat.com>
18067 PR c++/80485 - inline function non-zero address.
18068 * symtab.c (nonzero_address): Check DECL_COMDAT.
18070 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
18073 * config/alpha/alpha.md (ashlsi3): New insn pattern.
18074 (*ashlsi_se): Rename from *ashldi_se. Define as sign
18075 extension of SImode operation. Use const123_operand predicate.
18076 (*saddsi_1): Remove.
18077 (*saddl_se_1): Ditto.
18078 (*ssubsi_1): Ditto.
18079 (*ssubl_se_1): Ditto.
18080 * config/alpha/predicates.md (const123_operand): New predicate.
18081 * config/alpha/constraints.md (P): Use IN_RANGE.
18083 2018-05-25 Richard Biener <rguenther@suse.de>
18085 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
18087 (ref_maybe_used_by_stmt_p): Likewise.
18088 (stmt_may_clobber_ref_p): Likewise.
18089 (stmt_may_clobber_ref_p_1): Likewise.
18090 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
18092 (ref_maybe_used_by_stmt_p): Likewise.
18093 (stmt_may_clobber_ref_p): Likewise.
18094 (stmt_may_clobber_ref_p_1): Likewise.
18095 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
18096 the alias oracle to disambiguate DRs with stmts DR analysis
18098 (vect_analyze_data_refs): Do not give up on not analyzable
18099 DRs for BB vectorization. Remove code truncating the dataref
18102 2018-05-25 Jakub Jelinek <jakub@redhat.com>
18105 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
18106 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
18107 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
18109 2018-05-25 Richard Biener <rguenther@suse.de>
18111 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
18112 function, combining stmt data ref gathering and fatal analysis
18114 (vect_analyze_data_refs): Remove now redudnant code and simplify.
18115 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
18116 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
18117 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
18118 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
18120 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18122 PR tree-optimization/85720
18123 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
18124 SCC if all partitions are builtins.
18125 (version_loop_by_alias_check): New parameter. Generate cancelable
18126 runtime alias check if all partitions are builtins.
18127 (distribute_loop): Update call to above function.
18129 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18131 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
18132 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
18133 (parm_default_def_partition_arg): Ditto.
18134 (set_parm_default_def_partition): Ditto.
18135 (get_parm_default_def_partitions): Ditto and make it static.
18136 (get_undefined_value_partitions): Ditto and make it static.
18137 (remove_ssa_form): Refactor call to init_var_map here.
18138 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
18139 computation for loop region.
18140 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
18141 (register_default_def): Delete.
18142 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
18143 (parm_default_def_partition_arg): Ditto.
18144 (set_parm_default_def_partition): Ditto.
18145 (get_parm_default_def_partitions): Ditto and make it static.
18146 (get_undefined_value_partitions): Ditto and make it static.
18147 (coalesce_with_default, coalesce_with_default): Update comment.
18148 (create_coalesce_list_for_region): New func factored out from
18149 create_outofssa_var_map.
18150 (populate_coalesce_list_for_outofssa): New func factored out from
18151 create_outofssa_var_map and coalesce_ssa_name.
18152 (create_outofssa_var_map): Delete.
18153 (coalesce_ssa_name): Refactor to support live range computation.
18154 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
18155 (get_parm_default_def_partitions): Delete.
18156 (get_undefined_value_partitions): Ditto.
18157 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
18158 computation for loop region.
18159 (new_tree_live_info, loe_visit_block): Ditto.
18160 (live_worklist, set_var_live_on_entry): Ditto.
18161 (calculate_live_on_exit, verify_live_on_entry): Ditto.
18162 * tree-ssa-live.h (struct _var_map): New fields.
18163 (init_var_map): Change decl.
18164 (region_contains_p): New.
18166 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18168 * tree-ssa-live.h (live_merge_and_clear): Delete.
18170 2018-05-25 Richard Biener <rguenther@suse.de>
18173 * tree-dump.c (dequeue_and_dump): Remove access to removed
18174 operand 2 of a SWITCH_EXPR.
18176 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18178 * doc/sourcebuild.texi (vect_double_cond_arith): Include
18179 multiplication and division.
18180 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
18181 (cond_udiv@var{m}, cond_umod@var{m}): Document.
18182 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
18183 (cond_udiv_optab, cond_umod_optab): New optabs.
18184 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
18185 (IFN_COND_RDIV): New internal functions.
18186 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
18187 TRUNC_MOD_EXPR and RDIV_EXPR.
18188 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
18189 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
18191 (SVE_INT_BINARY): Include mult.
18192 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
18193 (optab, sve_int_op): Handle mult.
18194 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
18196 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
18197 for SVE_INT_BINARY_SD.
18199 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18201 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
18202 (optab, sve_int_op): Handle div and udiv.
18203 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
18204 for SVE_INT_BINARY_SD.
18205 (*<optab><mode>3): New insn for the same.
18207 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18209 * tree-vect-patterns.c: Include predict.h.
18210 (vect_recog_divmod_pattern): Restrict check for division support
18211 to when optimizing for size.
18213 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18215 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
18216 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
18217 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
18218 (gimple_match_op::set_op): Likewise.
18219 (gimple_resimplify4): Declare.
18220 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
18221 (expr::gen_transform): Likewise.
18222 (decision_tree::gen): Generate a simplification routine for 4 operands.
18223 * gimple-match-head.c (gimple_simplify): Add an overload for
18224 4 operands. In the top-level function, handle up to 4 call
18225 arguments and call gimple_resimplify4.
18226 (gimple_resimplify4): New function.
18227 (build_call_internal): Pass a fourth operand.
18228 (maybe_push_to_seq): Likewise.
18229 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
18230 Fold VEC_COND_EXPRs of an operation and a default value into
18231 an IFN_COND_* function if possible.
18232 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
18234 (SVE_COND_FP_BINARY): Include them.
18235 (optab, sve_fp_op): Handle them.
18236 (SVE_INT_BINARY_REV): New code iterator.
18237 (SVE_COND_FP_BINARY_REV): New int iterator.
18238 (commutative): New int attribute.
18239 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
18241 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
18243 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
18244 (*cond_<optab><mode>): New patterns for reversed operands.
18246 2018-05-25 Richard Biener <rguenther@suse.de>
18248 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
18249 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
18250 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
18251 (STMT_VINFO_GROUPED_ACCESS): Adjust.
18252 * tree-vect-data-refs.c (everywhere): Adjust users.
18253 * tree-vect-loop.c (everywhere): Likewise.
18254 * tree-vect-slp.c (everywhere): Likewise.
18255 * tree-vect-stmts.c (everywhere): Likewise.
18256 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
18258 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18260 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
18262 (gcc_cv_as_section_exclude): ... this.
18263 Try Solaris as #exclude syntax.
18264 * configure: Regenerate.
18265 * config.in: Regenerate.
18266 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
18268 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
18269 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
18271 * varasm.c (default_elf_asm_named_section): Don't check if
18272 HAVE_GAS_SECTION_EXCLUDE is defined.
18274 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18276 * doc/md.texi: Update the documentation of the cond_* optabs
18277 to mention the new final operand. Fix GET_MODE_NUNITS call.
18278 Describe the scalar case too.
18279 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
18280 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
18282 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
18283 (get_conditional_internal_fn): Update comment.
18284 * tree-vect-loop.c (vectorizable_reduction): Pass the original
18285 accumulator value as a final argument to conditional functions.
18286 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
18287 a define_expand and add an "else" operand. Assert for now that
18288 the else operand is equal to operand 2. Use SVE_INT_BINARY and
18289 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
18290 (*cond_<optab><mode>): New patterns.
18291 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
18292 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
18293 (UNSPEC_COND_EOR): Delete.
18294 (optab): Remove associated mappings.
18295 (SVE_INT_BINARY): New code iterator.
18296 (sve_int_op): Remove int attribute and add "minus" to the code
18298 (SVE_COND_INT_OP): Delete.
18299 (SVE_COND_FP_OP): Rename to...
18300 (SVE_COND_FP_BINARY): ...this.
18302 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18304 * optabs.c (can_reuse_operands_p): New function.
18305 (maybe_legitimize_operands): Try to reuse the results for
18308 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
18310 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
18311 Add {q} suffix to insn mnemonic.
18313 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18315 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
18316 (msp430_warn_func_return): New.
18318 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
18320 * fold-const.c (tree_nonzero_bits): New function.
18321 * fold-const.h (tree_nonzero_bits): Likewise.
18322 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
18323 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
18325 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
18329 * varasm.c (assemble_alias): Check ifunc_resolver only on
18332 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
18335 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
18336 when memory input operand is handled.
18338 2018-05-24 Luis Machado <luis.machado@linaro.org>
18340 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
18342 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
18344 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18346 * match.pd: Delay FMA folds until after vectorization.
18348 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
18351 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
18352 address check not strict.
18354 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18356 * gimple-match.h (gimple_match_op): New class.
18357 (mprts_hook): Replace parameters with a gimple_match_op *.
18358 (maybe_build_generic_op): Likewise.
18359 (gimple_simplified_result_is_gimple_val): Replace parameters with
18360 a const gimple_match_op *.
18361 (gimple_simplify): Replace code_helper * and tree * parameters with
18362 a gimple_match_op * parameter.
18363 (gimple_resimplify1): Replace code_helper *, tree and tree *
18364 parameters with a gimple_match_op * parameter.
18365 (gimple_resimplify2): Likewise.
18366 (gimple_resimplify3): Likewise.
18367 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
18368 parameters with a gimple_match_op * parameter.
18369 * gimple-match-head.c (gimple_simplify): Change prototypes of
18370 auto-generated functions to take a gimple_match_op * instead of
18371 separate code_helper * and tree * parameters. Make the same
18372 change in the top-level overload and update calls to the
18373 gimple_resimplify routines. Update calls to the auto-generated
18374 functions and to maybe_push_res_to_seq in the publicly-facing
18375 operation-specific gimple_simplify overloads.
18376 (gimple_match_op::MAX_NUM_OPS): Define.
18377 (gimple_resimplify1): Replace rcode and ops with a single res_op
18378 parameter. Update call to gimple_simplify.
18379 (gimple_resimplify2): Likewise.
18380 (gimple_resimplify3): Likewise.
18381 (mprts_hook): Replace parameters with a gimple_match_op *.
18382 (maybe_build_generic_op): Likewise.
18383 (build_call_internal): Replace type, nargs and ops with
18384 a gimple_match_op *.
18385 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
18386 with a single gimple_match_op *. Update calls to mprts_hook,
18387 build_call_internal and gimple_simplified_result_is_gimple_val.
18388 Factor out code that is common to the tree_code and combined_fn cases.
18389 * genmatch.c (expr::gen_transform): Replace tem_code and
18390 tem_ops with a gimple_match_op called tem_op. Update calls
18391 to the gimple_resimplify functions and maybe_push_res_to_seq.
18392 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
18393 res_ops. Update call to the gimple_resimplify functions.
18394 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
18395 (decision_tree::gen): Make the functions take a gimple_match_op *
18396 called res_op instead of separate res_code and res_ops parameters.
18397 Update call accordingly.
18398 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
18399 and ops with a single res_op parameter. Update calls to
18400 maybe_build_generic_op and maybe_push_res_to_seq.
18401 (fold_stmt_1): Update calls to gimple_simplify and
18402 replace_stmt_with_simplification.
18403 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
18404 and gimple_simplified_result_is_gimple_val.
18405 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
18407 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
18408 with a gimple_match_op *.
18409 (vn_nary_build_or_lookup): Likewise. Update call to
18410 vn_nary_build_or_lookup_1.
18411 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
18412 gimple_match_op *. Update calls to the gimple_resimplify routines
18413 and to gimple_simplified_result_is_gimple_val.
18414 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
18415 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
18416 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
18417 (visit_nary_op): Likewise.
18418 (visit_reference_op_load): Likewise.
18420 2018-05-23 Luis Machado <luis.machado@linaro.org>
18422 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
18423 modifier for printing the step amount.
18425 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18428 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
18431 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
18433 * doc/sourcebuild.texi (Endianness): New subsubsection.
18435 2018-05-23 Luis Machado <luis.machado@linaro.org>
18437 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18438 <prefetch_dynamic_strides>: New const bool field.
18439 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18440 prefetch_dynamic_strides.
18441 (exynosm1_prefetch_tune): Likewise.
18442 (thunderxt88_prefetch_tune): Likewise.
18443 (thunderx_prefetch_tune): Likewise.
18444 (thunderx2t99_prefetch_tune): Likewise.
18445 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
18447 (aarch64_override_options_internal): Update to set
18448 PARAM_PREFETCH_DYNAMIC_STRIDES.
18449 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18450 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18451 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18452 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18453 prefetch-dynamic-strides setting.
18455 2018-05-23 Luis Machado <luis.machado@linaro.org>
18457 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18458 <minimum_stride>: New const int field.
18459 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18460 minimum_stride field defaulting to -1.
18461 (exynosm1_prefetch_tune): Likewise.
18462 (thunderxt88_prefetch_tune): Likewise.
18463 (thunderx_prefetch_tune): Likewise.
18464 (thunderx2t99_prefetch_tune): Likewise.
18465 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
18466 <default_opt_level>: Set to 3.
18467 (aarch64_override_options_internal): Update to set
18468 PARAM_PREFETCH_MINIMUM_STRIDE.
18469 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18470 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18471 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18472 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18473 stride is constant and is below the minimum stride threshold.
18475 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18477 * config/arm/arm-cpus.in (mode26): Delete.
18478 (armv4): Delete mode26 reference.
18479 * config/arm/arm.c (arm_configure_build_target): Delete use of
18482 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
18484 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
18486 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
18487 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
18489 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
18490 (floatunsdidf2): Ditto.
18492 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
18494 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
18495 (fixuns_trunc<mode>si2_avx512f): Ditto.
18496 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
18497 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
18498 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
18500 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
18502 PR rtl-optimization/79985
18503 * df-scan.c (df_insn_refs_collect): Remove special case for
18504 global registers and asm statements.
18506 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
18508 * extend.texi (Global Register Variables): Rewrite the bullet list.
18509 Note that the register is available for allocation. Note that access
18510 via inline asm must use constraints. Add note about async-signal
18511 handlers. Remove paragraph about automagic register selection.
18513 2018-05-23 Richard Biener <rguenther@suse.de>
18515 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
18516 of fixed offset from memset VN.
18518 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
18520 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
18521 first_interp field.
18522 (alloc_cand_and_find_basis): Initialize first_interp field.
18523 (slsr_process_mul): Modify first_interp field.
18524 (slsr_process_add): Likewise.
18525 (slsr_process_cast): Modify first_interp field for each new
18527 (slsr_process_copy): Likewise.
18528 (dump_candidate): Dump first_interp field.
18529 (replace_mult_candidate): Process all interpretations, not just
18531 (replace_rhs_if_not_dup): Likewise.
18532 (replace_one_candidate): Likewise.
18534 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
18536 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
18538 (aarch64_needs_frame_chain): New function.
18539 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
18541 2018-05-23 Sudakshina Das <sudi.das@arm.com>
18544 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
18545 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
18546 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
18547 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
18548 as true for strict-align.
18549 (aarch64_can_inline_p): Perform checks even when callee has no
18550 attributes to check for strict alignment.
18551 * doc/extend.texi (AArch64 Function Attributes): Document
18553 * doc/invoke.texi: (AArch64 Options): Likewise.
18555 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
18557 PR tree-optimization/85853
18558 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
18559 the handling of the root of the node to...
18560 (vect_slp_analyze_node_operations_1): ...this new function,
18561 and run the whole thing with the child nodes' def types
18562 set according to their SLP node's def type.
18564 2018-05-23 Richard Biener <rguenther@suse.de>
18566 PR middle-end/85874
18567 * tree-data-ref.c (create_runtime_alias_checks): Defer
18568 and ignore overflow warnings.
18570 2018-05-23 Yury Gribov <tetra2005@gmail.com>
18572 PR tree-optimization/85822
18573 * tree-vrp.c (is_masked_range_test): Fix handling of negative
18576 2018-05-23 Richard Biener <rguenther@suse.de>
18578 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
18579 memset constants via native_interpret_expr.
18581 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
18584 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
18586 (cgraph_node::create_alias): Likewise.
18587 (cgraph_node::get_availability): Check ifunc_resolver instead
18588 of looking up ifunc attribute.
18589 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
18590 * varasm.c (do_assemble_alias): Likewise.
18591 (assemble_alias): Likewise.
18592 (default_binds_local_p_3): Likewise.
18593 * cgraph.h (cgraph_node): Add ifunc_resolver.
18594 (cgraph_node::only_called_directly_or_aliased_p): Return false
18595 for IFUNC resolver.
18596 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
18598 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
18599 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
18600 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
18601 instead of looking up ifunc attribute.
18603 2018-05-22 Luis Machado <luis.machado@linaro.org>
18605 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
18607 2018-05-22 Martin Sebor <msebor@redhat.com>
18609 PR middle-end/85359
18610 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
18611 only when expasion succeeds.
18612 (expand_builtin_strcmp): Same.
18613 (expand_builtin_strncmp): Same.
18615 2018-05-22 Martin Sebor <msebor@redhat.com>
18617 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
18619 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
18620 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18622 * config/aarch64/aarch64-ldpstp.md: Replace uses of
18623 aarch64_mem_pair_operand with memory_operand and delete operand swapping
18625 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
18626 Add check for legitimate_address.
18627 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
18628 (aarch64_swap_ldrstr_operands): New.
18629 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
18632 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
18633 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18635 * config/aarch64/aarch64.md: New patterns to generate stp
18637 (store_pair_sw, store_pair_dw): New patterns to generate stp for
18638 single words and double words.
18639 (load_pair_sw, load_pair_dw): Likewise.
18640 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
18642 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
18644 * config/aarch64/aarch64-ldpstp.md: Modify peephole
18645 for different mode ldpstp and add peephole for merged zero stores.
18646 Likewise for loads.
18647 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
18649 (aarch64_gen_store_pair): Rename calls to match new patterns.
18650 (aarch64_gen_load_pair): Rename calls to match new patterns.
18651 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
18652 (load_pair<DREG:mode><DREG2:mode>): ... This.
18653 (store_pair<mode>): Rename to...
18654 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
18655 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
18656 New mode iterators.
18657 (V_INT_EQUIV): Handle SImode.
18658 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
18661 2018-05-22 Martin Sebor <msebor@redhat.com>
18664 * calls.c (maybe_warn_nonstring_arg): Use string length to set
18665 or ajust the presumed bound on an operation to avoid unnecessary
18668 2018-05-22 Martin Sebor <msebor@redhat.com>
18670 PR tree-optimization/85826
18671 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
18672 assuming that a DECL necesarily has a constant size.
18674 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
18676 PR middle-end/85862
18677 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
18679 2018-05-22 Richard Biener <rguenther@suse.de>
18681 PR tree-optimization/85834
18682 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
18683 non-constant and non-zero memset arguments.
18685 2018-05-22 Martin Liska <mliska@suse.cz>
18688 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
18690 2018-05-22 Richard Biener <rguenther@suse.de>
18692 PR tree-optimization/85863
18693 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
18694 comparisons when vectype is specified.
18695 (vectorizable_condition): Do not specify vectype for
18696 vect_is_simple_cond when SLP vectorizing.
18698 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
18701 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
18702 define __ibm128 as long double.
18703 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
18704 as a distinct type when IEEE 128-bit support is enabled.
18705 (init_float128_ieee): Fix up conversions between IFmode and IEEE
18706 128-bit types to use the correct functions.
18707 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
18708 convert between 128-bit floating point types that have different
18709 modes but the same representation, instead of using gen_lowpart to
18711 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
18713 (IFKF_reg): New attributes to give the register constraints for
18715 (extend<mode>tf2_internal): New insns to mark an explicit
18716 conversion between 128-bit floating point types that have a
18717 different mode but share the same representation.
18719 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
18721 PR tree-optimization/85814
18722 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
18723 a null return from get_strinfo when unsharing the next
18724 strinfo in the chain.
18726 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
18729 * varasm.c (weak_finish): Clean up weak_decls.
18731 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18733 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
18734 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
18735 UNSPEC_UADALP values.
18736 * config/aarch64/iterators.md (ABAL): New int iterator.
18739 (sur): Add mappings for the above.
18740 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
18742 (aarch64_<sur>abal<mode>_4): Likewise.
18743 (aarch64_<sur>adalp<mode>_3): Likewise.
18744 (<sur>sadv16qi): New define_expand.
18746 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
18748 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
18749 (*movdf_internal): Ditto.
18750 (*rcpsf2_sse): Ditto.
18751 (*rsqrtsf2_sse): Ditto.
18752 (*sqrt<mode>2_sse): Ditto.
18754 2018-05-21 Tamar Christina <tamar.christina@arm.com>
18756 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
18758 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
18759 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
18760 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
18761 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
18763 * config/aarch64/arm_neon.h: Likewise.
18764 * config/aarch64/iterators.md (VQ_I): New.
18766 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
18768 * config.gcc: Add arc/t-multilib-linux to tmake_file for
18770 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
18773 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
18775 * config/nds32/constraints.md (S): New constraint.
18776 * config/nds32/nds32.md (call_internal): Use constraint S.
18777 (call_value_internal): Likewise.
18778 (sibcall_internal): Likewise.
18779 (sibcall_value_internal): Likewise.
18781 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
18782 Chung-Ju Wu <jasonwucj@gmail.com>
18784 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
18785 into consideration.
18787 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
18788 Chung-Ju Wu <jasonwucj@gmail.com>
18790 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
18791 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
18792 (nds32_rtx_costs_impl): Simplify.
18793 (nds32_address_cost_impl): Simplify.
18794 (nds32_init_rtx_costs): New function.
18795 (nds32_rtx_costs_speed_prefer): Likewise.
18796 (nds32_rtx_costs_size_prefer): Likewise.
18797 (nds32_address_cost_speed_prefer): Likewise.
18798 (nds32_address_cost_speed_fwprop): Likewise.
18799 (nds32_address_cost_size_prefer): Likewise.
18800 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
18801 * config/nds32/nds32.c (nds32_option_override): Use
18802 nds32_init_rtx_costs function.
18804 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
18806 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
18807 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
18808 (TARGET_PIPELINE_N8): Likewise.
18809 (TARGET_PIPELINE_N10): Likewise.
18810 (TARGET_PIPELINE_N13): Likewise.
18811 (TARGET_PIPELINE_GRAYWOLF): Likewise.
18813 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
18815 * config/nds32/nds32-fpu.md: Update copyright year.
18817 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18819 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
18821 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18823 * config/nds32/nds32.c
18824 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
18825 * config/nds32/nds32.opt (minline-asm-r15): New option.
18827 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18829 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
18831 * config/nds32/nds32.md (abssi2): New pattern.
18833 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
18835 * config/i386/i386.md (rex64namesuffix): New mode attribute.
18836 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
18837 Merge insn pattern from sse_cvtsi2ss<round_name> and
18838 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
18839 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
18840 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
18841 using SWI48 mode iterator.
18842 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
18843 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
18844 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
18845 pattern from sse_cvttss2si<round_saeonly_name>
18846 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
18847 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
18848 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
18849 using SWI48 mode iterator.
18850 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
18851 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
18852 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
18853 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
18854 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
18855 using SWI48 mode iterator.
18856 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
18857 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
18858 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
18859 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
18860 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
18861 SWI48 mode iterator.
18862 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
18863 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
18864 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
18865 pattern from sse_cvttsd2si<round_saeonly_name>
18866 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
18868 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18870 * config/nds32/nds32-md-auxiliary.c
18871 (nds32_valid_smw_lwm_base_p): Refine.
18872 (nds32_output_smw_single_word): Refine.
18873 (nds32_output_smw_double_word): New.
18874 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
18876 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18878 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
18879 (nds32_output_stack_pop): Refine.
18880 (nds32_expand_unaligned_load): Refine.
18881 (nds32_expand_unaligned_store): Refine.
18883 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18884 Chung-Ju Wu <jasonwucj@gmail.com>
18886 * config/nds32/constants.md: Add TP_REGNUM constant.
18887 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
18888 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
18890 * config/nds32/nds32-doubleword.md: Consider flag_pic.
18891 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
18892 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
18893 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
18894 and PIC code generation.
18895 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
18897 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
18899 * config/nds32/nds32.md: Support TLS and PIC.
18900 * config/nds32/nds32.c: Support TLS and PIC.
18901 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
18902 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
18905 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18907 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
18908 mode with E_ prefix.
18910 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18911 Chung-Ju Wu <jasonwucj@gmail.com>
18913 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
18914 * config/nds32/nds32-md-auxiliary.c
18915 (symbolic_reference_mentioned_p): New.
18916 (nds32_legitimize_ict_address): New.
18917 (nds32_expand_ict_move): New.
18918 (nds32_indirect_call_referenced_p): New.
18919 (nds32_symbol_binds_local_p): Delete.
18920 (nds32_long_call_p): Modify.
18921 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
18922 * config/nds32/nds32-protos.h
18923 (symbolic_reference_mentioned_p): Declare.
18924 (nds32_legitimize_ict_address): Declare.
18925 (nds32_expand_ict_move): Declare.
18926 (nds32_indirect_call_referenced_p): Declare.
18927 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
18928 (nds32_relax_group): Use nds32_ict_const_p as condition.
18929 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
18930 (nds32_asm_file_start): Output ict_model directive in asm code.
18931 (nds32_legitimate_address_p): Consider indirect call.
18932 (nds32_print_operand): Consider indirect call.
18933 (nds32_print_operand_address): Consider indirect call.
18934 (nds32_insert_attributes): Handle "indirect_call" attribute.
18935 (TARGET_LEGITIMATE_ADDRESS_P): Define.
18936 (TARGET_LEGITIMATE_CONSTANT_P): Define.
18937 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18938 (TARGET_DELEGITIMIZE_ADDRESS): Define.
18939 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
18940 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
18941 (TARGET_ICT_MODEL_SMALL): Define.
18942 (TARGET_ICT_MODEL_LARGE): Define.
18943 * config/nds32/nds32.md (movsi): Consider ict model.
18944 (call, call_value): Consider ict model.
18945 (sibcall, sibcall_value): Consider ict model.
18946 * config/nds32/nds32.opt (mict-model): New option.
18947 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
18950 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
18951 Monk Chiang <sh.chiang04@gmail.com>
18952 Jim Wilson <jimw@sifive.com>
18954 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
18955 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
18956 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
18957 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
18958 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
18959 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
18960 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
18961 compute save_libcall_adjustment properly.
18962 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
18963 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
18964 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
18965 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
18966 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
18967 (ABI_SPEC): Handle mabi=ilp32e.
18968 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
18969 (RVE): Add RVE mask.
18970 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
18971 <-march>: Add rv32e as an example.
18973 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
18976 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
18977 (intra_create_variable_infos): Handle C++ constructors.
18979 2018-05-18 Martin Liska <mliska@suse.cz>
18981 * passes.def: Remove a redundant pass.
18983 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
18986 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
18988 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18990 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
18992 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
18994 (armv2, armv2a, armv3, armv3m): Delete architectures.
18995 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
18996 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
18997 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
18999 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
19000 (*mulsidi3adddi): Likewise.
19001 (mulsidi3): Likewise.
19002 (*mulsidi3_nov6): Likewise.
19003 (umulsidi3): Likewise.
19004 (umulsidi3_nov6): Likewise.
19005 (umaddsidi4): Likewise.
19006 (*umulsidi3adddi): Likewise.
19007 (smulsi3_highpart): Likewise.
19008 (*smulsi3_highpart_nov6): Likewise.
19009 (umulsi3_highpart): Likewise.
19010 (*umulsi3_highpart_nov6): Likewise.
19011 * config/arm/arm.h (arm_arch3m): Delete.
19012 * config/arm/arm.c (arm_arch3m): Delete.
19013 (arm_option_override_internal): Update armv3-related comment.
19014 (arm_configure_build_target): Delete use of isa_bit_mode32.
19015 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
19016 (arm_rtx_costs_internal): Delete check of arm_arch3m.
19017 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
19018 (mulsa3): Likewise.
19019 (mulusa3): Likewise.
19020 * config/arm/arm-protos.h (arm_arch3m): Delete.
19021 * config/arm/arm-tables.opt: Regenerate.
19022 * config/arm/arm-tune.md: Likewise.
19023 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
19024 deleted architectures.
19026 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19028 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
19029 (armv5t, armv5te): New features.
19030 (ARMv5, ARMv5e): Delete fgroups.
19031 (ARMv5t, ARMv5te): Adjust for above changes.
19032 (ARMv6m): Likewise.
19033 (armv5, armv5e): Delete arches.
19034 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
19036 (*call_reg_arm): Likewise.
19037 (*call_value_reg_armv5): Likewise.
19038 (*call_value_reg_arm): Likewise.
19039 (*call_symbol): Likewise.
19040 (*call_value_symbol): Likewise.
19041 (*sibcall_insn): Likewise.
19042 (*sibcall_value_insn): Likewise.
19043 (clzsi2): Likewise.
19044 (prefetch): Likewise.
19045 (define_split and define_peephole2 dependent on arm_arch5):
19047 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
19049 (TARGET_ARM_QBIT): Likewise.
19050 (TARGET_DSP_MULTIPLY): Likewise.
19051 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
19052 (arm_arch5, arm_arch5e): Delete.
19053 (arm_arch5t, arm_arch5te): Declare.
19054 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
19055 (arm_arch5t): Declare.
19056 (arm_option_reconfigure_globals): Update for the above.
19057 (arm_options_perform_arch_sanity_checks): Update comment, replace
19058 use of arm_arch5 with arm_arch5t.
19059 (use_return_insn): Likewise.
19060 (arm_emit_call_insn): Likewise.
19061 (output_return_instruction): Likewise.
19062 (arm_final_prescan_insn): Likewise.
19063 (arm_coproc_builtin_available): Likewise.
19064 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
19065 arm_arch5e with arm_arch5t and arm_arch5te.
19066 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
19067 (arm_arch5t, arm_arch5te): Declare.
19068 * config/arm/arm-tables.opt: Regenerate.
19069 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
19070 * config/arm/t-multilib: Likewise.
19071 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
19072 instead of arm_arch5.
19073 (*call_reg_thumb1): Likewise.
19074 (*call_value_reg_thumb1_v5): Likewise.
19075 (*call_value_reg_thumb1): Likewise.
19076 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
19078 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
19080 2018-05-18 Martin Liska <mliska@suse.cz>
19082 PR gcov-profile/84846
19083 * doc/gcov.texi: Document -t option of gcov tool.
19085 2018-05-18 Martin Liska <mliska@suse.cz>
19087 PR gcov-profile/84846
19088 * gcov.c (print_usage): Add new -t option.
19089 (process_args): Handle the option.
19090 (generate_results): Use stdout as output when requested by
19093 2018-05-18 Martin Liska <mliska@suse.cz>
19095 PR gcov-profile/84846
19096 * coverage.c (coverage_init): Write PWD to .gcno file.
19097 * doc/gcov.texi: Document how working directory is printed.
19098 * gcov-dump.c (dump_gcov_file): Print PWD.
19099 * gcov.c (output_intermediate_file): Likewise.
19100 (read_graph_file): Read PWD string.
19101 (output_lines): Print PWD.
19103 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19105 PR middle-end/85817
19106 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
19107 for retval and return false if all args to phi are zero.
19109 2018-05-18 Richard Biener <rguenther@suse.de>
19111 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
19113 (evrp_dom_walker::before_dom_children): Call it.
19115 2018-05-18 Richard Biener <rguenther@suse.de>
19117 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
19118 results when processing array refs with variable index.
19120 2018-05-18 Toon Moene <toon@moene.org>
19122 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
19123 directly after that of -floop-interchange. Indicate that both
19124 options are enabled by default when specifying -O3.
19126 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19128 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
19129 iterator. Delete separate integer-mode vec_set<mode> expander.
19130 (aarch64_simd_vec_setv2di): Delete.
19131 (vec_setv2di): Delete.
19132 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
19133 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
19134 the "w, r" alternative.
19136 2018-05-18 Martin Liska <mliska@suse.cz>
19138 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
19139 * tree-pass.h (make_pass_lower_switch_O0): New function.
19140 * tree-switch-conversion.c (node_has_low_bound): Remove.
19141 (node_has_high_bound): Likewise.
19142 (node_is_bounded): Likewise.
19143 (class pass_lower_switch): Make it a template type and create
19145 (pass_lower_switch::execute): Add template argument.
19146 (make_pass_lower_switch): New function.
19147 (make_pass_lower_switch_O0): New function.
19148 (do_jump_if_equal): Remove.
19149 (emit_case_nodes): Simplify to just handle all 3 cases and leave
19150 all the hard work to tree optimization passes.
19152 2018-05-18 Martin Liska <mliska@suse.cz>
19154 * dbgcnt.c (limit_low): Renamed from limit.
19155 (limit_high): New variable.
19156 (dbg_cnt_is_enabled): Check for upper limit.
19157 (dbg_cnt): Adjust dumping.
19158 (dbg_cnt_set_limit_by_index): Add new argument for high
19160 (dbg_cnt_set_limit_by_name): Likewise.
19161 (dbg_cnt_process_single_pair): Parse new format.
19162 (dbg_cnt_process_opt): Use strtok.
19163 (dbg_cnt_list_all_counters): Remove 'value' and add
19165 * doc/invoke.texi: Document changes.
19167 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
19169 * doc/sourcebuild.texi (scalar_all_fma): Document.
19170 * tree.def (FMA_EXPR): Delete.
19171 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
19172 * internal-fn.c (ternary_direct): New macro.
19173 (expand_ternary_optab_fn): Likewise.
19174 (direct_ternary_optab_supported_p): Likewise.
19175 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
19176 * builtins.c (fold_builtin_fma): Delete.
19177 (fold_builtin_3): Don't call it.
19178 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
19179 * expr.c (expand_expr_real_2): Likewise.
19180 * fold-const.c (operand_equal_p): Likewise.
19181 (fold_ternary_loc): Likewise.
19182 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
19183 * gimple.c (DEFTREECODE): Likewise.
19184 * gimplify.c (gimplify_expr): Likewise.
19185 * optabs-tree.c (optab_for_tree_code): Likewise.
19186 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19187 * tree-eh.c (operation_could_trap_p): Likewise.
19188 (stmt_could_throw_1_p): Likewise.
19189 * tree-inline.c (estimate_operator_cost): Likewise.
19190 * tree-pretty-print.c (dump_generic_node): Likewise.
19191 (op_code_prio): Likewise.
19192 * tree-ssa-loop-im.c (stmt_cost): Likewise.
19193 * tree-ssa-operands.c (get_expr_operands): Likewise.
19194 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
19195 * fold-const-call.h (fold_fma): Delete.
19196 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
19197 CFN_FNMA and CFN_FNMS.
19198 (fold_fma): Delete.
19199 * genmatch.c (combined_fn): New enum.
19200 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
19201 (commutative_op): New function.
19202 (commutate): Use it. Handle more than 2 operands.
19203 (dt_operand::gen_gimple_expr): Use commutative_op.
19204 (parser::parse_expr): Allow :c to be used with non-binary
19205 operators if the commutative operand is known.
19206 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
19207 CFN_FMS, CFN_FNMA and CFN_FNMS.
19208 (backprop::process_assign_use): Remove FMA_EXPR handling.
19209 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
19210 (gen_hsa_fma): New function.
19211 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
19212 IFN_FNMA and IFN_FNMS.
19213 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
19214 * gimple-fold.h (follow_all_ssa_edges): Declare.
19215 * gimple-fold.c (follow_all_ssa_edges): New function.
19216 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
19217 gimple_build interface and use follow_all_ssa_edges to fold the result.
19218 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
19219 instead of checking for optabs directly.
19220 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
19221 rather than FMA_EXPRs.
19222 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
19223 call to IFN_FMA instead of an FMA_EXPR.
19225 2018-05-17 Jim Wilson <jimw@sifive.com>
19227 * expr.c (do_tablejump): When converting index to Pmode, if we have a
19228 sign extended promoted subreg, and the range does not have the sign bit
19229 set, then do a sign extend.
19231 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
19232 test, check for sign extended subreg and/or constant operands, and
19233 do a sign extend in that case.
19235 2018-05-17 Steve Ellcey <sellcey@cavium.com>
19237 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
19238 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
19240 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
19241 Change logics_shift_reg to logics_shift_imm.
19242 (thunderx2t99_fp_loadpair_basic): Delete.
19243 (thunderx2t99_fp_storepair_basic): Delete.
19244 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
19245 (thunderx2t99_asimd_polynomial): Delete.
19246 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
19247 and neon_fp_mul_d_scalar_q.
19248 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
19249 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
19250 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
19251 (thunderx2t99_asimd_lut): Add missing tbl types.
19252 (thunderx2t99_asimd_ext): Delete.
19253 (thunderx2t99_asimd_load1_1_mult): Delete.
19254 (thunderx2t99_asimd_load1_2_mult): Delete.
19255 (thunderx2t99_asimd_load1_ldp): New.
19256 (thunderx2t99_asimd_load1): New.
19257 (thunderx2t99_asimd_load2): Add missing *load2* types.
19258 (thunderx2t99_asimd_load3): New.
19259 (thunderx2t99_asimd_load4): New.
19260 (thunderx2t99_asimd_store1_1_mult): Delete.
19261 (thunderx2t99_asimd_store1_2_mult): Delete.
19262 (thunderx2t99_asimd_store2_mult): Delete.
19263 (thunderx2t99_asimd_store2_onelane): Delete.
19264 (thunderx2t99_asimd_store_stp): New.
19265 (thunderx2t99_asimd_store1): New.
19266 (thunderx2t99_asimd_store2): New.
19267 (thunderx2t99_asimd_store3): New.
19268 (thunderx2t99_asimd_store4): New.
19270 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
19272 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
19273 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
19275 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
19276 Segher Boessenkool <segher@kernel.crashing.org>
19279 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
19282 2018-05-17 Richard Biener <rguenther@suse.de>
19284 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
19285 for pruning loop and prune defs feeding only already visited PHIs.
19287 2018-05-17 Richard Biener <rguenther@suse.de>
19289 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
19291 2018-05-17 Bin Cheng <bin.cheng@arm.com>
19292 Richard Biener <rguenther@suse.de>
19294 PR tree-optimization/85793
19295 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
19296 for VMAT_ELEMENTWISE.
19298 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
19300 * internal-fn.h (lookup_internal_fn): Declare
19301 * internal-fn.c (lookup_internal_fn): New function.
19302 * gimple.c (gimple_build_call_from_tree): Handle calls to
19303 internal functions.
19304 * gimple-pretty-print.c (dump_gimple_call): Print "." before
19305 internal function names.
19306 * tree-pretty-print.c (dump_generic_node): Likewise.
19307 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
19309 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
19311 * gimple-fold.h (gimple_build): Make the function forms take
19312 combined_fn rather than built_in_function.
19313 (gimple_simplify): Likewise.
19314 * gimple-match-head.c (gimple_simplify): Likewise.
19315 * gimple-fold.c (gimple_build): Likewise.
19316 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
19317 rather than gimple_build_call_internal.
19318 (get_initial_defs_for_reduction): Likewise.
19319 (vect_create_epilog_for_reduction): Likewise.
19320 (vectorizable_live_operation): Likewise.
19322 2018-05-17 Martin Liska <mliska@suse.cz>
19324 * gimple-ssa-sprintf.c (format_directive): Do not use
19325 space in between 'G_' and '('.
19327 2018-05-17 Jakub Jelinek <jakub@redhat.com>
19330 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
19331 even if the mask is not all ones.
19334 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
19336 (ix86_gimple_fold_builtin): Likewise.
19339 * config/i386/i386.c: Include tree-vector-builder.h.
19340 (ix86_vector_shift_count): New function.
19341 (ix86_fold_builtin): Fold shift builtins by scalar count.
19342 (ix86_gimple_fold_builtin): Likewise.
19344 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
19345 _mm512_setzero): New intrinsics.
19347 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
19348 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19350 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
19351 code generation for cases where splatting a value is not useful.
19352 * simplify-rtx.c (simplify_ternary_operation): Simplify
19353 vec_merge across a vec_duplicate and a paradoxical subreg forming
19354 a vector mode to a vec_concat.
19356 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
19358 * config.gcc: Support "goldmont-plus".
19359 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19361 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19362 PROCESSOR_GOLDMONT_PLUS.
19363 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
19364 (processor_target_table): Add "goldmont-plus".
19365 (PTA_GOLDMONT_PLUS): Define.
19366 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
19367 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
19368 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
19369 (fold_builtin_cpu): Add "goldmont-plus".
19370 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
19371 (ix86_option_override_internal): Add "goldmont-plus".
19372 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
19373 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
19374 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
19375 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
19377 2018-05-17 Richard Biener <rguenther@suse.de>
19379 PR tree-optimization/85757
19380 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
19381 remove defs that only feed that PHI from further processing.
19383 2018-05-16 Jim Wilson <jimw@sifive.com>
19385 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
19387 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
19389 2018-05-16 Mark Wielaard <mark@klomp.org>
19391 * dwarf2out.c (count_index_strings): New function.
19392 (output_indirect_strings): Call count_index_strings and generate
19393 header for dwarf_version >= 5.
19395 2018-05-16 Mark Wielaard <mark@klomp.org>
19397 * dwarf2out.c (dwarf_FORM): New function.
19398 (set_indirect_string): Use dwarf_FORM.
19399 (reset_indirect_string): Likewise.
19400 (size_of_die): Likewise.
19401 (value_format): Likewise.
19402 (output_die): Likewise.
19403 (add_skeleton_AT_string): Likewise.
19404 (output_macinfo_op): Likewise.
19405 (index_string): Likewise.
19406 (output_index_string_offset): Likewise.
19407 (output_index_string): Likewise.
19408 (count_index_strings): Likewise.
19410 2018-05-16 Carl Love <cel@us.ibm.com>
19412 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
19413 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
19415 2018-05-16 Martin Jambor <mjambor@suse.cz>
19417 * ipa-prop.c (ipa_free_all_edge_args): Remove.
19418 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
19420 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
19422 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
19423 (fnma<mode>4): Likewise.
19424 (fms<mode>4): Likewise.
19425 (fnms<mode>4): Likewise.
19426 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
19427 (aarch64_fnma<mode>4): Likewise.
19428 (aarch64_fms<mode>4): Likewise.
19429 (aarch64_fnms<mode>4): Likewise.
19430 (aarch64_fnmadd<mode>4): Likewise.
19432 2018-05-16 Jason Merrill <jason@redhat.com>
19434 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
19436 2018-05-16 Richard Biener <rguenther@suse.de>
19438 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
19439 (dump_stmt_cost): Declare.
19440 (add_stmt_cost): Dump cost we add.
19441 (add_stmt_costs): New function.
19442 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
19443 No longer exported.
19444 (vect_analyze_stmt): Adjust prototype.
19445 (vectorizable_condition): Likewise.
19446 (vectorizable_live_operation): Likewise.
19447 (vectorizable_reduction): Likewise.
19448 (vectorizable_induction): Likewise.
19449 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
19450 cost vector to pass to vectorizable_ and record afterwards.
19451 (vect_model_reduction_cost): Take cost vector argument and adjust.
19452 (vect_model_induction_cost): Likewise.
19453 (vectorizable_reduction): Likewise.
19454 (vectorizable_induction): Likewise.
19455 (vectorizable_live_operation): Likewise.
19456 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
19457 SLP_TREE_NUMBER_OF_VEC_STMTS.
19458 (vect_analyze_slp_cost_1): Remove.
19459 (vect_analyze_slp_cost): Likewise.
19460 (vect_slp_analyze_node_operations): Take visited args and
19461 a target cost vector. Avoid processing already visited stmt sets.
19462 (vect_slp_analyze_operations): Use a local cost vector to gather
19463 costs and register those of non-discarded instances.
19464 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
19465 (vect_schedule_slp_instance): Remove copying of
19466 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
19468 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
19469 adding cost. Record cost entry location.
19470 (vect_prologue_cost_for_slp_op): Function to compute cost of
19471 a constant or invariant generated for SLP vect in the prologue,
19472 split out from vect_analyze_slp_cost_1.
19473 (vect_model_simple_cost): Make static. Adjust for SLP costing.
19474 (vect_model_promotion_demotion_cost): Likewise.
19475 (vect_model_store_cost): Likewise, make static.
19476 (vect_model_load_cost): Likewise.
19477 (vectorizable_bswap): Add cost vector arg and adjust.
19478 (vectorizable_call): Likewise.
19479 (vectorizable_simd_clone_call): Likewise.
19480 (vectorizable_conversion): Likewise.
19481 (vectorizable_assignment): Likewise.
19482 (vectorizable_shift): Likewise.
19483 (vectorizable_operation): Likewise.
19484 (vectorizable_store): Likewise.
19485 (vectorizable_load): Likewise.
19486 (vectorizable_condition): Likewise.
19487 (vectorizable_comparison): Likewise.
19488 (can_vectorize_live_stmts): Likewise.
19489 (vect_analyze_stmt): Likewise.
19490 (vect_transform_stmt): Adjust calls to vectorizable_*.
19491 * tree-vectorizer.c: Include gimple-pretty-print.h.
19492 (dump_stmt_cost): New function.
19494 2018-05-16 Richard Biener <rguenther@suse.de>
19496 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
19497 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
19498 * tree-ssa-dse.c: Include tree-ssa-loop.h.
19499 (check_name): New callback.
19500 (dse_classify_store): Track cycles via a visited bitmap of PHI
19501 defs and simplify handling of in-loop and across loop dead stores
19502 and properly fail for loop-variant refs. Handle byte-tracking with
19503 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
19506 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
19508 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
19509 (vect_get_mask_type_for_stmt): Likewise.
19510 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
19512 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
19513 to determine the statement's vector type and the vector type that
19514 should be used for calculating nunits. Deal with cases in which
19515 the type has to be deferred.
19516 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
19517 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
19518 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
19519 (vect_determine_vf_for_stmt): New functions, split out from...
19520 (vect_determine_vectorization_factor): ...here.
19521 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
19522 (vect_get_mask_type_for_stmt): New functions, split out from
19523 vect_determine_vectorization_factor.
19525 2018-05-16 Richard Biener <rguenther@suse.de>
19527 * tree-cfg.c (verify_gimple_assign_ternary): Properly
19528 verify the [VEC_]COND_EXPR embedded comparison.
19530 2018-05-15 Martin Sebor <msebor@redhat.com>
19532 PR tree-optimization/85753
19533 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
19534 RECORD_TYPE in addition to ARRAY_TYPE.
19536 2018-05-15 Martin Sebor <msebor@redhat.com>
19538 PR middle-end/85643
19539 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
19541 2018-05-15 Richard Biener <rguenther@suse.de>
19543 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
19544 add by_clobber_p one. Change algorithm to collect all defs
19545 representing uses we need to walk and try reducing them to
19546 a single one before failing.
19547 (dse_dom_walker::dse_optimize_stmt): Adjust.
19549 2018-05-13 Mark Wielaard <mark@klomp.org>
19551 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
19552 (size_of_loc_descr): Likewise.
19553 (output_loc_operands): Likewise.
19554 (output_loc_operands_raw): Likewise.
19555 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
19556 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
19557 (hash_loc_operands): Likewise.
19558 (compare_loc_operands): Likewise.
19560 2018-05-14 Mark Wielaard <mark@klomp.org>
19562 * dwarf2out.c (count_index_addrs): New function.
19563 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
19565 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19567 PR tree-optimization/83648
19568 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
19569 return value as malloc candidate.
19571 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19574 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
19575 param as true in call to suggest_attribute.
19577 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
19579 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
19580 -mreadonly-in-sdata.
19582 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19584 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
19586 (aarch64_crypto_aesd_fused): Likewise.
19588 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
19590 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
19591 (movsi_aarch64): Likewise.
19592 (load_pairsi): Likewise.
19593 (load_pairdi): Likewise.
19594 (store_pairsi): Likewise.
19595 (store_pairdi): Likewise.
19596 (load_pairsf): Likewise.
19597 (load_pairdf): Likewise.
19598 (store_pairsf): Likewise.
19599 (store_pairdf): Likewise.
19600 (zero_extend): Likewise.
19601 (trunc): Swap alternatives.
19602 (fcvt_target): Add '?' to prefer w over r.
19604 2018-05-14 Jakub Jelinek <jakub@redhat.com>
19607 * config/i386/i386.md: Disallow non-commutative arithmetics in
19608 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
19609 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
19610 in the peephole2 before it.
19612 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
19614 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
19615 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
19616 (ix86_handle_option): Handle -mcldemote.
19617 * config.gcc: New header.
19618 * config/i386/cldemoteintrin.h: New file.
19619 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
19620 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19622 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19623 OPTION_MASK_ISA_CLDEMOTE.
19624 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
19625 (ix86_valid_target_attribute_inner_p): Ditto.
19626 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
19627 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
19628 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
19629 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
19630 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
19632 * config/i386/i386.opt: Add -mcldemote.
19633 * config/i386/x86intrin.h: New header.
19634 * doc/invoke.texi: Add -mcldemote.
19636 2018-05-14 Richard Biener <rguenther@suse.de>
19638 * doc/match-and-simplify.texi: Adjust :s documentation.
19640 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
19642 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
19643 intended memcpy size.
19644 (REORDER_45): Likewise.
19646 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
19648 * sort.cc: New file.
19649 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
19650 * vec.c (qsort_chk): Use gcc_qsort.
19651 * Makefile.in (OBJS-libcommon): Add sort.o.
19652 (build/sort.o): New target. Use it...
19653 (BUILD_RTL): ... here, and...
19654 (build/gencfn-macros): ... here, and...
19655 (build/genmatch): ... here.
19657 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19658 Chung-Ju Wu <jasonwucj@gmail.com>
19660 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
19661 * config/nds32/nds32-graywolf.md: New file.
19662 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
19663 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
19665 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
19666 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
19667 * config/nds32/nds32.md (pipeline_model): Add graywolf.
19668 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
19669 * config/nds32/pipelines.md: Include n15 settings.
19671 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19672 Chung-Ju Wu <jasonwucj@gmail.com>
19674 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
19675 * config/nds32/nds32-n13.md: New file.
19676 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
19677 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
19679 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
19680 * config/nds32/nds32.md (pipeline_model): Add n13.
19681 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
19682 * config/nds32/pipelines.md: Include n13 settings.
19684 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19685 Chung-Ju Wu <jasonwucj@gmail.com>
19687 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
19688 * config/nds32/nds32-n10.md: New file.
19689 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
19690 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
19692 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
19693 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
19694 * config/nds32/nds32.md (pipeline_model): Add n10.
19695 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
19696 * config/nds32/pipelines.md: Include n10 settings.
19698 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
19699 Kito Cheng <kito.cheng@gmail.com>
19700 Chung-Ju Wu <jasonwucj@gmail.com>
19702 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
19703 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19704 Add enum values for DSP extension instructions.
19705 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
19707 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
19708 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
19709 New code iterators.
19710 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
19711 * config/nds32/nds32-dspext.md: New file for DSP implementation.
19712 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
19713 * config/nds32/nds32-intrinsic.md: Likewise.
19714 * config/nds32/nds32_intrinsic.h: Likewise.
19715 * config/nds32/nds32-md-auxiliary.c: Likewise.
19716 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
19717 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
19718 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
19719 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
19720 * config/nds32/nds32-protos.h: New declarations for DSP extension.
19721 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
19722 TYPE_DMAC in switch statement.
19723 * config/nds32/nds32.c: New checking and implementation for DSP
19724 extension instructions.
19725 * config/nds32/nds32.h: Likewise.
19726 * config/nds32/nds32.md: Likewise.
19727 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
19728 * config/nds32/predicates.md: Implement new predicates for DSP
19731 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19733 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
19734 Reformat alternatives and attributes so it is easier to identify
19735 which constraints/attributes go with which instruction.
19736 (mov<mode>_hardfloat32, FMOVE64): Likewise.
19737 (mov<mode>_softfloat32, FMOVE64): Likewise.
19738 (mov<mode>_hardfloat64, FMOVE64): Likewise.
19739 (mov<mode>_softfloat64, FMOVE64): Likewise.
19741 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19743 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19745 (Basic PowerPC Built-in Functions): The new name of the
19746 subsection previously known as "PowerPC Built-in Functions".
19747 (Basic PowerPC Built-in Functions Available on all Configurations):
19749 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
19750 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
19751 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
19752 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
19754 2018-05-11 Martin Jambor <mjambor@suse.cz>
19757 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
19760 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
19763 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
19765 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
19767 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
19768 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
19769 (ix86_handle_option): Handle -mwaitpkg.
19770 * config.gcc: New header.
19771 * config/i386/cpuid.h (bit_WAITPKG): New bit.
19772 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
19773 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
19775 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19776 OPTION_MASK_ISA_WAITPKG.
19777 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
19778 (ix86_option_override_internal): Add PTA_WAITPKG.
19779 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
19780 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
19781 IX86_BUILTIN_TPAUSE.
19782 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
19783 __builtin_ia32_umwait and __builtin_ia32_tpause.
19784 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
19785 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
19786 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
19787 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
19788 UNSPECV_TPAUSE): New.
19789 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
19790 * config/i386/i386.opt: Add -mwaitpkg.
19791 * config/i386/waitpkgintrin.h: New file.
19792 * config/i386/x86intrin.h: New header.
19793 * doc/invoke.texi: Add -mwaitpkg.
19795 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
19798 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
19800 (cortex-m0): Use armv6s-m isa.
19801 (cortex-m0plus): Likewise.
19802 (cortex-m1): Likewise.
19803 (cortex-m0.small-multiply): Likewise.
19804 (cortex-m0plus.small-multiply): Likewise.
19805 (cortex-m1.small-multiply): Likewise.
19807 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
19808 Jakub Jelinek <jakub@redhat.com>
19810 PR tree-optimization/85692
19811 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
19812 source permute as well.
19814 2018-05-11 Martin Liska <mliska@suse.cz>
19817 * doc/extend.texi: Document LLVM style format for no_sanitize
19820 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
19822 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
19823 mode_supports_vsx_dform_quad to mode_supports_dq_form.
19824 (mode_supports_vsx_dform_quad): Likewise.
19825 (mode_supports_vmx_dform): Move these functions to be next to the
19826 other mode_supports functions.
19827 (mode_supports_dq_form): Likewise.
19828 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
19829 mode_supports_dq_form.
19830 (reg_offset_addressing_ok_p): Likewise.
19831 (offsettable_ok_by_alignment): Likewise.
19832 (rs6000_legitimate_offset_address_p): Likewise.
19833 (legitimate_lo_sum_address_p): Likewise.
19834 (rs6000_legitimize_address): Likewise.
19835 (rs6000_legitimize_reload_address): Likewise.
19836 (rs6000_secondary_reload_inner): Likewise.
19837 (rs6000_preferred_reload_class): Likewise.
19838 (rs6000_output_move_128bit): Likewise.
19840 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
19842 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
19843 Generate SImode target register for null target.
19844 <case IX86_BUILTIN_XGETBV>: Ditto.
19845 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
19846 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
19848 2018-05-10 Carl Love <cel@us.ibm.com>
19850 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
19851 dcbtt and dcbtstt if operands[2] is 0.
19853 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
19856 * config/i386/sse.md (usadv64qi): New expander.
19858 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
19860 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
19861 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
19862 -maltivec=be support.
19863 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
19864 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
19865 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19866 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
19867 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
19868 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
19869 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
19870 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
19871 altivec_vsumsws): Adjust.
19872 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
19873 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
19874 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
19876 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
19877 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
19878 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
19879 (altivec_lve<VI_char>x): Delete expand.
19880 (*altivec_lve<VI_char>x_internal): Rename to...
19881 (altivec_lve<VI_char>x): ... this.
19882 (altivec_lvxl_<mode>): Delete expand.
19883 (*altivec_lvxl_<mode>_internal): Rename to ...
19884 (altivec_lvxl_<mode>): ... this.
19885 (altivec_stvxl_<mode>): Delete expand.
19886 (*altivec_stvxl_<mode>_internal): Rename to ...
19887 (altivec_stvxl_<mode>): ... this.
19888 (altivec_stve<VI_char>x): Delete expand.
19889 (*altivec_stve<VI_char>x_internal): Rename to ...
19890 (altivec_stve<VI_char>x): ... this.
19891 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
19892 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
19893 reduc_plus_scal_<mode>): Adjust.
19894 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
19896 (rs6000_cpu_cpp_builtins): Adjust.
19897 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
19898 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
19899 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
19900 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
19901 -maltivec=be support.
19902 (rs6000_split_vec_extract_var): Adjust.
19903 (rs6000_split_v4si_init): Adjust.
19904 (swap_selector_for_mode): Delete.
19905 (altivec_expand_lvx_be, altivec_expand_stvx_be,
19906 altivec_expand_stvex_be): Delete.
19907 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
19908 -maltivec=be support.
19909 (rs6000_gimple_fold_builtin): Ditto.
19910 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
19912 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
19913 (TARGET_DIRECT_MOVE_64BIT): Adjust.
19914 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
19915 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
19916 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
19917 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
19918 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
19919 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
19920 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
19921 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
19922 anonymous split): Adjust.
19923 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
19924 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
19926 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
19928 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
19929 when --with-gxx-include-dir is also specified.
19930 * configure: Regenerate.
19932 2018-05-09 Jim Wilson <jimw@sifive.com>
19935 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
19936 * config/riscv/t-withmultilib: New.
19937 * config/riscv/withmultilib.h: New.
19938 * doc/install.texi: Document RISC-V --with-multilib-list support.
19940 2018-05-09 Richard Biener <rguenther@suse.de>
19942 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
19944 (vect_bb_vectorization_profitable_p): Adjust. Compute
19945 actual scalar cost using the cost vector and the add_stmt_cost
19948 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19950 PR rtl-optimization/85645
19951 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
19952 in the REG_CFA_REGISTER note for LR, don't leave it empty.
19954 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19956 PR rtl-optimization/85645
19957 * shrink-wrap.c (spread_components): Return a boolean saying if
19958 anything was changed.
19959 (try_shrink_wrapping_separate): Iterate spread_components until
19960 nothing changes anymore.
19962 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19964 PR rtl-optimization/85645
19965 * regrename.c (build_def_use): Also kill the chains that include the
19966 destination of a REG_CFA_REGISTER note.
19968 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19970 PR rtl-optimization/85645
19971 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
19972 insn that has a REG_CFA_REGISTER note.
19974 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
19976 * cfgexpand.c (expand_clobber): New function.
19977 (expand_gimple_stmt_1): Use it.
19978 * tree-vect-stmts.c (vect_clobber_variable): New function,
19980 (vectorizable_simd_clone_call): ...here.
19981 (vectorizable_store): Emit a clobber either side of an
19982 IFN_STORE_LANES sequence.
19983 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
19985 2018-05-09 Tom de Vries <tom@codesourcery.com>
19988 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
19989 (define_insn "trap_if_false"): Add exit after trap.
19991 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
19993 PR rtl-optimization/85638
19994 * bb-reorder.c: Include common/common-target.h.
19995 (create_forwarder_block): New function extracted from...
19996 (fix_up_crossing_landing_pad): ...here. Rename into...
19997 (dw2_fix_up_crossing_landing_pad): ...this.
19998 (sjlj_fix_up_crossing_landing_pad): New function.
19999 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
20000 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
20001 from both partitions and exit the loop after one iteration.
20003 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20006 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20008 (Basic PowerPC Built-in Functions): The new name of the
20009 subsection previously known as "PowerPC Built-in Functions".
20010 (Basic PowerPC Built-in Functions Available on all Configurations):
20012 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
20014 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
20016 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
20018 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
20021 2018-05-08 Jim Wilson <jimw@sifive.com>
20023 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
20024 (LD_EMUL_SUFFIX): New.
20025 (LINK_SPEC): Use it.
20027 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20029 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20031 (Basic PowerPC Built-in Functions): The new name of the
20032 subsection previously known as "PowerPC Built-in Functions".
20033 (Basic PowerPC Built-in Functions Available on all Configurations):
20035 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
20037 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
20039 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
20041 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
20044 2018-05-08 Jakub Jelinek <jakub@redhat.com>
20047 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
20048 after cmpelim optimization.
20050 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
20052 * config.gcc: Support "goldmont".
20053 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
20054 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20055 PROCESSOR_GOLDMONT.
20056 * config/i386/i386.c (m_GOLDMONT): Define.
20057 (processor_target_table): Add "goldmont".
20058 (PTA_GOLDMONT): Define.
20059 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
20060 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
20061 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
20062 (fold_builtin_cpu): Add "goldmont".
20063 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
20064 (ix86_option_override_internal): Add "goldmont".
20065 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
20066 (processor_type): Add PROCESSOR_GOLDMONT.
20067 * config/i386/i386.md: Add CPU "glm".
20068 * config/i386/glm.md: New file.
20069 * config/i386/x86-tune.def: Add m_GOLDMONT.
20070 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
20072 2018-05-08 Jakub Jelinek <jakub@redhat.com>
20075 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
20077 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
20078 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
20079 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
20082 * config/i386/i386.c (ix86_fold_builtin): Handle
20083 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
20086 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
20087 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
20089 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
20092 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
20093 (check_arch): Likewise.
20094 (check_fpu): Return the result rather than printing it.
20095 (end arch): Fix operator precedence.
20096 (end cpu): Likewise.
20097 (END): Print the result from check_fpu.
20099 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
20100 Alan Hayward <alan.hayward@arm.com>
20101 David Sherwood <david.sherwood@arm.com>
20103 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
20104 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
20105 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
20106 (*fcmuo<mode>_and): New patterns.
20108 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
20110 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
20111 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
20112 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
20113 (cmp_op, sve_imm_con): New code attributes.
20114 (SVE_COND_INT_CMP, imm_con): Delete.
20115 (cmp_op): Remove above unspecs from int attribute.
20116 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
20118 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
20119 comparison-specific unspecs.
20120 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
20121 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
20122 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
20123 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
20124 (*vec_fcm<cmp_op><mode>): Rename to...
20125 (*fcm<cmp_op><mode>): ...this and adjust likewise.
20126 (*vec_fcmuo<mode>): Rename to...
20127 (*fcmuo<mode>): ...this and adjust likewise.
20128 (*pred_fcm<cmp_op><mode>): New pattern.
20129 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
20130 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
20132 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
20134 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
20135 (aarch64_emit_sve_predicated_cond): New function.
20136 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
20137 (aarch64_emit_unspec_cond_or): Replace with...
20138 (aarch64_emit_sve_or_conds): ...this new function. Use
20139 aarch64_emit_sve_ptrue_op for the individual comparisons and
20140 aarch64_emit_binop to OR them together.
20141 (aarch64_emit_inverted_unspec_cond): Replace with...
20142 (aarch64_emit_sve_inverted_cond): ...this new function. Use
20143 aarch64_emit_sve_ptrue_op for the comparison and
20144 aarch64_emit_unop to invert the result.
20145 (aarch64_expand_sve_vec_cmp_float): Update after the above
20146 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
20148 2018-05-07 Nathan Sidwell <nathan@acm.org>
20150 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
20151 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
20152 (Backwards Compatibility): Likewise.
20154 2018-05-07 Luis Machado <luis.machado@linaro.org>
20158 2018-05-07 Luis Machado <luis.machado@linaro.org>
20160 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20161 <prefetch_dynamic_strides>: New const bool field.
20162 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20163 prefetch_dynamic_strides.
20164 (exynosm1_prefetch_tune): Likewise.
20165 (thunderxt88_prefetch_tune): Likewise.
20166 (thunderx_prefetch_tune): Likewise.
20167 (thunderx2t99_prefetch_tune): Likewise.
20168 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
20170 (aarch64_override_options_internal): Update to set
20171 PARAM_PREFETCH_DYNAMIC_STRIDES.
20172 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20173 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20174 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20175 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20176 prefetch-dynamic-strides setting.
20178 2018-05-07 Luis Machado <luis.machado@linaro.org>
20180 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20181 <minimum_stride>: New const int field.
20182 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20183 minimum_stride field.
20184 (exynosm1_prefetch_tune): Likewise.
20185 (thunderxt88_prefetch_tune): Likewise.
20186 (thunderx_prefetch_tune): Likewise.
20187 (thunderx2t99_prefetch_tune): Likewise.
20188 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
20189 (aarch64_override_options_internal): Update to set
20190 PARAM_PREFETCH_MINIMUM_STRIDE.
20191 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20192 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20193 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20194 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20195 stride is constant and is below the minimum stride threshold.
20197 2018-05-07 Luis Machado <luis.machado@linaro.org>
20199 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
20202 2018-05-07 Luis Machado <luis.machado@linaro.org>
20204 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20205 <prefetch_dynamic_strides>: New const bool field.
20206 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20207 prefetch_dynamic_strides.
20208 (exynosm1_prefetch_tune): Likewise.
20209 (thunderxt88_prefetch_tune): Likewise.
20210 (thunderx_prefetch_tune): Likewise.
20211 (thunderx2t99_prefetch_tune): Likewise.
20212 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
20214 (aarch64_override_options_internal): Update to set
20215 PARAM_PREFETCH_DYNAMIC_STRIDES.
20216 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20217 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20218 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20219 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20220 prefetch-dynamic-strides setting.
20222 2018-05-07 Luis Machado <luis.machado@linaro.org>
20224 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20225 <minimum_stride>: New const int field.
20226 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20227 minimum_stride field.
20228 (exynosm1_prefetch_tune): Likewise.
20229 (thunderxt88_prefetch_tune): Likewise.
20230 (thunderx_prefetch_tune): Likewise.
20231 (thunderx2t99_prefetch_tune): Likewise.
20232 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
20233 (aarch64_override_options_internal): Update to set
20234 PARAM_PREFETCH_MINIMUM_STRIDE.
20235 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20236 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20237 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20238 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20239 stride is constant and is below the minimum stride threshold.
20241 2018-05-06 Jakub Jelinek <jakub@redhat.com>
20244 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
20245 the type is addressable. Don't force op into register if it has
20248 2018-05-05 Roland McGrath <mcgrathr@google.com>
20251 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
20252 any section for which we don't know a specific type it should have,
20253 regardless of name. Previously this was done only for the exact
20254 names ".init_array", ".fini_array", and ".preinit_array".
20255 (default_elf_asm_named_section): Add comment about
20256 relationship with default_section_type_flags and SECTION_NOTYPE.
20257 (get_section): Don't consider it a type conflict if one side has
20258 SECTION_NOTYPE and the other doesn't, as long as neither has the
20259 SECTION_BSS et al used in the default_section_type_flags logic.
20261 2018-05-05 Tom de Vries <tom@codesourcery.com>
20264 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
20265 (workaround_barsyncs): New function.
20266 (nvptx_reorg): Use workaround_barsyncs.
20267 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
20268 (define_expand "nvptx_membar_cta"): New define_expand.
20269 (define_insn "*nvptx_membar_cta"): New insn.
20271 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20273 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
20274 To improve optimization opportunities.
20275 * builtin-types.def: The new needed builtin types for the above.
20277 2018-05-04 Richard Biener <rguenther@suse.de>
20279 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
20280 * gimple-ssa-store-merging.c
20281 (imm_store_chain_info::output_merged_store): Remove redundant create,
20282 release split_store vector contents on failure.
20283 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
20284 scalar stmt vector on cache hit.
20286 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
20288 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20290 * config.gcc (powerpc-xilinx-eabi*): Remove.
20291 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
20293 (fusion_addis_mem_combo_load): Ditto.
20294 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
20296 (rs6000_cpu_cpp_builtins): Ditto.
20297 * config/rs6000/rs6000-linux.c
20298 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
20299 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
20300 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
20302 (rs6000_setup_reg_addr_masks): Ditto.
20303 (rs6000_init_hard_regno_mode_ok): Ditto.
20304 (rs6000_option_override_internal): Ditto.
20305 (legitimate_lo_sum_address_p): Ditto.
20306 (rs6000_legitimize_address): Ditto.
20307 (rs6000_legitimize_reload_address): Ditto.
20308 (rs6000_legitimate_address_p): Ditto.
20309 (abi_v4_pass_in_fpr): Ditto.
20310 (setup_incoming_varargs): Ditto.
20311 (rs6000_gimplify_va_arg): Ditto.
20312 (rs6000_split_multireg_move): Ditto.
20313 (rs6000_savres_strategy): Ditto.
20314 (rs6000_emit_prologue_components): Ditto.
20315 (rs6000_emit_epilogue_components): Ditto.
20316 (rs6000_emit_prologue): Ditto.
20317 (rs6000_emit_epilogue): Ditto.
20318 (rs6000_elf_file_end): Ditto.
20319 (rs6000_function_value): Ditto.
20320 (rs6000_libcall_value): Ditto.
20321 * config/rs6000/rs6000.h: Ditto.
20322 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
20323 (TARGET_MINMAX): ... this. New.
20324 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
20325 * config/rs6000/rs6000.md: Remove Xilinx FP support.
20326 (*movsi_internal1_single): Delete.
20327 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
20328 mfpu=, mxilinx-fpu): Delete.
20329 * config/rs6000/singlefp.h: Delete.
20330 * config/rs6000/sysv4.h: Remove Xilinx FP support.
20331 * config/rs6000/t-rs6000: Ditto.
20332 * config/rs6000/t-xilinx: Delete.
20333 * config/rs6000/titan.md: Adjust for fp_type removal.
20334 * config/rs6000/vsx.md: Remove Xilinx FP support.
20335 (VStype_simple): Delete.
20336 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
20337 * config/rs6000/xfpu.h: Delete.
20338 * config/rs6000/xfpu.md: Delete.
20339 * config/rs6000/xilinx.h: Delete.
20340 * config/rs6000/xilinx.opt: Delete.
20341 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
20342 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
20344 2018-05-04 Tom de Vries <tom@codesourcery.com>
20347 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
20350 2018-05-04 Richard Biener <rguenther@suse.de>
20352 PR middle-end/85627
20353 * tree-complex.c (update_complex_assignment): We are always in SSA form.
20354 (expand_complex_div_wide): Likewise.
20355 (expand_complex_operations_1): Likewise.
20356 (expand_complex_libcall): Preserve EH info of the original stmt.
20357 (tree_lower_complex): Handle removed blocks.
20358 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
20359 on complex multiplication and division libcall builtins.
20361 2018-05-04 Richard Biener <rguenther@suse.de>
20363 PR middle-end/85574
20364 * fold-const.c (negate_expr_p): Restrict negation of operand
20365 zero of a division to when we know that can happen without
20367 (fold_negate_expr_1): Likewise.
20369 2018-05-04 Jakub Jelinek <jakub@redhat.com>
20372 * real.h (real_nextafter): Declare.
20373 * real.c (real_nextafter): New function.
20374 * fold-const-call.c (fold_const_nextafter): New function.
20375 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
20376 CASE_CFN_NEXTTOWARD.
20377 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
20378 even when arg1_mode is different from arg0_mode.
20380 2018-05-03 Nathan Sidwell <nathan@acm.org>
20382 * doc/extend.texi (Deprecated Features): Remove
20383 -ffriend-injection.
20384 (Backwards Compatibility): Likewise.
20385 * doc/invoke.texi (C++ Language Options): Likewise.
20386 (C++ Dialect Options): Likewise.
20388 2018-05-03 Jakub Jelinek <jakub@redhat.com>
20391 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
20392 _mm512_mask_mullox_epi64): New intrinsics.
20394 2018-05-03 Tom de Vries <tom@codesourcery.com>
20397 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20398 dump files): Add offload-tree.
20400 2018-05-03 Richard Biener <rguenther@suse.de>
20402 PR tree-optimization/85615
20403 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
20404 to loops not nested in BBs loop father to avoid creating multi-entry
20407 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20409 PR tree-optimization/70291
20410 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
20411 arguments. Change return type to tree. Emit libcall as a new
20412 statement rather than replacing existing one when inplace_p is true.
20413 (expand_complex_multiplication_components): New function.
20414 (expand_complex_multiplication): Expand floating-point complex
20415 multiplication using the above.
20416 (expand_complex_division): Rename inner_type parameter to type.
20417 Update expand_complex_libcall call-site.
20418 (expand_complex_operations_1): Update expand_complex_multiplication
20419 and expand_complex_division call-sites.
20421 2018-05-02 Jakub Jelinek <jakub@redhat.com>
20424 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
20425 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
20426 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
20427 the highest significant bit of the shift count mask is clear. In
20428 check whether and[sq]i3 is needed verify that all significant bits
20429 of the shift count other than the highest are set.
20431 2018-05-02 Tom de Vries <tom@codesourcery.com>
20434 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
20435 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
20436 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
20437 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
20438 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
20439 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
20440 __builtin_goacc_parlevel_size.
20442 2018-05-02 Richard Biener <rguenther@suse.de>
20444 PR tree-optimization/85597
20445 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
20446 do not use split vect_get_vec_defs call but call vect_get_slp_defs
20449 2018-05-02 Tom de Vries <tom@codesourcery.com>
20452 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20453 dump files): Add ltrans-tree.
20455 2018-05-02 Tom de Vries <tom@codesourcery.com>
20458 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20459 dump files): Add wpa-ipa.
20461 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
20463 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
20464 powerpc*-*-linux*paired* target.
20465 * config/rs6000/750cl.h: Delete.
20466 * config/rs6000/paired.h: Delete.
20467 * config/rs6000/paired.md: Delete.
20468 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
20470 * config/rs6000/rs6000-builtin.def: Remove paired float support.
20471 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20472 comment. Remove paired float support.
20473 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
20474 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
20476 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
20477 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
20479 * config/rs6000/rs6000.c: Remove paired float support.
20480 (paired_expand_vector_init, paired_expand_vector_move,
20481 paired_emit_vector_compare, paired_emit_vector_cond_expr,
20482 (paired_expand_lv_builtin, paired_expand_stv_builtin,
20483 paired_expand_builtin, paired_expand_predicate_builtin,
20484 paired_init_builtins): Delete.
20485 * config/rs6000/rs6000.h: Remove paired float support.
20486 * config/rs6000/rs6000.md: Remove paired float support.
20487 (move_from_CR_ov_bit): Delete.
20488 * config/rs6000/rs6000.opt (mpaired): Delete.
20489 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
20490 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
20492 2018-05-02 Richard Biener <rguenther@suse.de>
20494 PR middle-end/85567
20495 * gimplify.c (gimplify_save_expr): When in SSA form allow
20496 SAVE_EXPRs to compute to SSA vars.
20498 2018-05-02 Jakub Jelinek <jakub@redhat.com>
20501 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
20502 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
20503 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
20504 clobber operands[2], instead use a new pseudo. Formatting fixes.
20506 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
20508 PR tree-optimization/85586
20509 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
20510 exit early for statements in the same group if the accesses are
20513 2018-05-02 Tom de Vries <tom@codesourcery.com>
20516 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
20519 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
20521 PR tree-optimization/85143
20522 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
20524 2018-05-01 Tom de Vries <tom@codesourcery.com>
20527 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
20528 not found" error message.
20530 2018-05-01 Tom de Vries <tom@codesourcery.com>
20533 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
20534 * vec.c (test_ordered_remove_if): New function.
20535 (vec_c_tests): Call test_ordered_remove_if.
20536 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
20537 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
20538 * tree-vect-patterns.c (vect_pattern_recog_1): Use
20539 VEC_ORDERED_REMOVE_IF.
20541 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20543 PR tree-optimization/82665
20544 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
20545 pointer subtraction where arguments come from a memchr call.
20547 2018-05-01 Jakub Jelinek <jakub@redhat.com>
20549 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
20550 --push-state --as-needed and --pop-state instead of --as-needed and
20551 --no-as-needed if ld supports it.
20552 * configure: Regenerated.
20555 * doc/install.texi2html: Replace _002d with - and _002a with * in
20556 generated html files using sed.
20558 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20561 * gcc-rich-location.c (blank_line_before_p): New function.
20562 (use_new_line): New function.
20563 (gcc_rich_location::add_fixit_insert_formatted): New function.
20564 * gcc-rich-location.h
20565 (gcc_rich_location::add_fixit_insert_formatted): New function.
20567 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20569 * selftest.c (assert_streq): Rename "expected" and "actual" to
20570 "val1" and "val2". Extend NULL-handling to cover both inputs
20571 symmetrically, while still requiring both to be non-NULL for a pass.
20572 * selftest.h (assert_streq): Rename "expected" and "actual" to
20574 (ASSERT_EQ): Likewise.
20575 (ASSERT_EQ_AT): Likewise.
20576 (ASSERT_KNOWN_EQ): Likewise.
20577 (ASSERT_KNOWN_EQ_AT): Likewise.
20578 (ASSERT_NE): Likewise.
20579 (ASSERT_MAYBE_NE): Likewise.
20580 (ASSERT_MAYBE_NE_AT): Likewise.
20581 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
20582 the assertion to pass.
20583 (ASSERT_STREQ_AT): Likewise.
20585 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
20587 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
20588 interaction with -pie.
20590 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20592 * selftest.h: Fix alphabetization of per-source-file selftest
20595 2018-04-30 Jason Merrill <jason@redhat.com>
20597 PR c++/61982 - dead stores to destroyed objects.
20598 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
20601 2018-04-30 Jason Merrill <jason@redhat.com>
20603 * tree.c (build_clobber): New.
20604 * tree.h: Declare it.
20605 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
20607 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20609 * diagnostic-show-locus.c (layout::layout): Update for
20610 location_get_source_line returning a char_span.
20611 (struct char_span): Move to input.h.
20612 (struct correction): Update for fields in char_span becoming
20614 (struct source_line): Update for location_get_source_line
20615 returning a char_span.
20616 (layout::print_line): Likewise.
20617 * edit-context.c (edited_file::print_content): Likewise.
20618 (edited_file::print_diff_hunk): Likewise.
20619 (edited_file::print_run_of_changed_lines): Likewise.
20620 (edited_file::get_num_lines): Likewise.
20621 (edited_line::edited_line): Likewise.
20622 * final.c (asm_show_source): Likewise.
20623 * input.c (location_get_source_line): Convert return type
20624 from const char * to char_span, losing the final "line_len"
20626 (dump_location_info): Update for the above.
20627 (get_substring_ranges_for_loc): Likewise. Use a char_span
20628 when handling the literal within the line.
20629 (test_reading_source_line): Update for location_get_source_line
20630 returning a char_span.
20631 * input.h (class char_span): Move here from
20632 diagnostic-show-locus.c, converting from a struct to a class.
20633 Make data members private.
20634 (char_span::operator bool): New.
20635 (char_span::length): New.
20636 (char_span::get_buffer): New.
20637 (char_span::operator[]): New.
20638 (char_span::subspan): Make const.
20639 (char_span::xstrdup): New.
20640 (location_get_source_line): Convert return type from const char *
20641 to char_span, losing the final "line_size" param.
20643 2018-04-30 Jan Hubicka <jh@suse.cz>
20645 * lto-wrapper.c (ltrans_priorities): New static var.
20646 (cmp_priority): New.
20647 (run_gcc): Read priorities and if doing parallel build order
20648 the Makefile by them.
20650 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20652 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
20654 2018-04-30 Richard Biener <rguenther@suse.de>
20656 * tree-cfg.c (verify_address): Remove base argument, add
20657 flag whether to check TREE_ADDRESSABLE and do that.
20658 (verify_expr): Remove.
20659 (verify_types_in_gimple_reference): Add pieces from verify_expr.
20660 (verify_gimple_assign_single): Likewise.
20661 (verify_gimple_switch): Likewise.
20662 (verify_expr_location_1): Dereference tp once. Add (disabled)
20663 piece from verify_expr.
20664 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
20666 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20668 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
20670 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20672 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
20673 (small_data_pattern): Likewise.
20674 (arc_rewrite_small_data): Likewise.
20675 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
20676 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
20677 (get_symbol_alignment): New function.
20678 (legitimate_small_data_address_p): Likewise.
20679 (legitimate_scaled_address): Update, call
20680 legitimate_small_data_address_p.
20681 (output_sdata): New static variable.
20682 (arc_print_operand): Update how we handle small data operands.
20683 (arc_print_operand_address): Likewise.
20684 (arc_legitimate_address_p): Update, use
20685 legitimate_small_data_address_p.
20686 (arc_rewrite_small_data_p): Remove.
20687 (arc_rewrite_small_data_1): Likewise.
20688 (arc_rewrite_small_data): Likewise.
20689 (small_data_pattern): Likewise.
20690 (compact_sda_memory_operand): Update to use
20691 legitimate_small_data_address_p and get_symbol_alignment.
20692 (prepare_move_operands): Don't rewite sdata pattern.
20693 (prepare_extend_operands): Remove.
20694 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
20696 (zero_extendqisi2): Likewise.
20697 (zero_extendhisi2): Likewise.
20698 (extendqihi2): Likewise.
20699 (extendqisi2): Likewise.
20700 (extendhisi2): Likewise.
20701 (addsi3): Likewise.
20702 (subsi3): Likewise.
20703 (andsi3): Likewise.
20704 * config/arc/constraints.md (Usd): Change it to memory constraint.
20706 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20708 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
20709 as source of std instructions.
20710 * config/arc/arc.md (movsi_insn): Update pattern predicate to
20711 allow 6-bit constants as source for store instructions.
20712 (movdi_insn): Update instruction pattern to allow 6-bit constants
20713 as source for store instructions.
20715 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
20717 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
20719 2018-04-30 Nathan Sidwell <nathan@acm.org>
20720 Sandra Loosemore <sandra@codesourcery.com>
20722 * dumpfile.c (dump_open): Allow '-' for stdout.
20723 * doc/invoke.texi (Developer Options): Document dump filename
20724 determination early. Document stdin/stdout selection.
20726 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
20728 Microblaze Target: PIC data text relative
20730 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
20731 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
20733 * config/microblaze/microblaze.h (microblaze_constant_address_p):
20734 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
20735 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
20736 New addressing mode for data-text relative position indepenedent code.
20737 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
20738 'ADDRESS_SYMBOLIC_TXT_REL'.
20739 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
20740 (microblaze_legitimate_pic_operand): Exclude function calls from
20741 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
20742 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
20744 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
20745 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
20746 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
20747 for 'address + offset'.
20748 (microblaze_expand_prologue): Add new function prologue call for
20750 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
20751 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
20752 table in case of TARGET_PIC_DATA_TEXT_REL.
20753 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
20754 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
20755 Add new macros 'UNSPEC_TEXT',
20756 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
20757 + exclude function calls from 'UNSPEC_PLT' in case of data text
20759 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
20760 new target hook for generating address diff vector tables in case of
20762 * doc/tm.texi : Regenerate.
20763 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
20764 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
20765 of addr diff vector generation.
20766 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
20767 target hook definition.
20768 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
20769 Add default function for generate_pic_addr_diff_vec -> flag_pic.
20770 * doc/invoke.texi (Add new pic option): Add new microblaze pic
20771 option for data text relative.
20773 2018-04-30 Richard Biener <rguenther@suse.de>
20775 * tree-chrec.h (evolution_function_is_constant_p): Remove
20777 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
20779 2018-04-30 Richard Biener <rguenther@suse.de>
20782 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
20784 2018-04-30 Richard Biener <rguenther@suse.de>
20786 PR tree-optimization/28364
20787 PR tree-optimization/85275
20788 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
20789 copying first exit test.
20791 2018-04-28 Mark Wielaard <mark@klomp.org>
20793 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
20794 dwarf_version >= 5.
20795 (dwarf_AT): Handle DW_AT_addr_base.
20796 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
20798 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
20801 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
20802 (*ashl<dwi>3_doubleword_mask_1): Ditto.
20803 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
20804 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
20806 2018-04-28 Richard Biener <rguenther@suse.de>
20808 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
20809 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
20810 to reflect use. Only add interesting stmts.
20812 2018-04-27 Martin Jambor <mjambor@suse.cz>
20815 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
20816 the jump function allows for passing through aggregate values.
20818 2018-04-27 David Malcolm <dmalcolm@redhat.com>
20820 * input.h (in_system_header_at): Convert from macro to inline
20822 (from_macro_expansion_at): Likewise.
20823 (from_macro_definition_at): Likewise.
20825 2018-04-27 Jeff Law <law@redhat.com>
20827 * config.gcc: Mark tile* targets as deprecated/obsolete.
20829 2018-04-27 Richard Biener <rguenther@suse.de>
20831 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
20834 2018-04-27 Richard Biener <rguenther@suse.de>
20836 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
20838 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
20840 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
20841 with Yd constraint. Set "preferred_for_speed" attribute from
20842 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
20843 with Yd constraint.
20844 (*movdi_internal): Ditto.
20845 (movti_interunit splitters): Remove
20846 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
20847 (movdi_interunit splitters): Ditto.
20848 * config/i386/constraints.md (Ye): Remove.
20849 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
20851 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20854 * config/aarch64/constraints.md (Usg): Limit to 31.
20855 (Usj): Limit to 63.
20857 2018-04-27 Jakub Jelinek <jakub@redhat.com>
20859 PR tree-optimization/85529
20860 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
20861 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
20862 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
20863 zero extension or masking of the MSB bit.
20864 (optimize_range_tests): Add FIRST_BB argument, pass it through
20865 to optimize_range_tests_var_bound.
20866 (maybe_optimize_range_tests, reassociate_bb): Adjust
20867 optimize_range_tests callers.
20869 2018-04-26 Richard Biener <rguenther@suse.de>
20870 Jakub Jelinek <jakub@redhat.com>
20872 * cgraph.h (symbol_table): Just declare debug method here.
20873 * symtab.c (symbol_table::debug): Define.
20875 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
20877 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
20879 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
20881 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
20882 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
20883 (*movdi_internal): Substitute Yi and Yj constraint with x
20884 and Ym and Yn constraint with y constraint. Update "isa"
20885 attribute and set "preferred_for_speed" attribute from
20886 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
20887 (*movsi_internal): Ditto.
20888 (*movdf_internal): Ditto.
20889 (*movsf_internal): Ditto.
20890 (*zero_extendsidi2): Ditto.
20891 * config/i386/sse.md (vec_set<mode>_0): Ditto.
20892 (sse2_loadld): Ditto.
20893 (*vec_extract<ssevecmodelower>_0): Ditto.
20894 (*vec_extractv4si_0_zext_sse4): Ditto.
20895 (vec_concatv2di): Ditto.
20896 (*vec_dup<mode>): Ditto.
20897 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
20898 * config/i386/constraints.md (Yi): Remove.
20903 2018-04-26 Nathan Sidwell <nathan@acm.org>
20905 * dumpfile.c (dump_open): New.
20906 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
20907 (dump_finish): Detect stdio/stderr by value not name.
20909 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
20911 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
20913 2018-04-26 Tom de Vries <tom@codesourcery.com>
20916 * config/nvptx/nvptx.c (verify_neutering_jumps)
20917 (verify_neutering_labels): New function
20918 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
20920 2018-04-26 Tom de Vries <tom@codesourcery.com>
20923 * config/nvptx/nvptx.c (needs_neutering_p): New function.
20924 (nvptx_single): Use needs_neutering_p to skip over insns that do not
20927 2018-04-26 Richard Biener <rguenther@suse.de>
20928 Tom de Vries <tom@codesourcery.com>
20931 * lto-streamer-out.c (output_function): Fixup loops if required to match
20932 discovery done in the reader.
20934 2018-04-26 Richard Biener <rguenther@suse.de>
20936 PR tree-optimization/85116
20937 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
20938 have a loop exit from the single latch predecessor. Remove
20939 case of header with just condition.
20940 (ch_base::copy_headers): Exclude infinite loops from any
20942 (pass_ch::execute): Record exits.
20944 2018-04-26 Richard Biener <rguenther@suse.de>
20946 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
20947 prologue cost vector and pass it to vect_get_load_cost.
20948 (vect_get_peeling_costs_all_drs): Likewise.
20949 (vect_peeling_hash_get_lowest_cost): Likewise.
20950 (vect_enhance_data_refs_alignment): Likewise.
20952 2018-04-26 Richard Biener <rguenther@suse.de>
20954 PR middle-end/85450
20955 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
20956 checking of integer<->pointer conversions.
20957 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
20958 sign-/zero-extending pointer types.
20959 (expand_omp_for_static_chunk): Likewise.
20961 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
20962 Jean Lee <xiaoyur347@gmail.com>
20964 * config/mips/mips.c (mips_asan_shadow_offset): New function.
20965 (TARGET_ASAN_SHADOW_OFFSET): Define.
20966 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
20967 true for -fsanitize=address.
20969 2018-04-25 Mark Wielaard <mark@klomp.org>
20971 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
20974 2018-04-25 Jakub Jelinek <jakub@redhat.com>
20976 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
20977 than "alu", remove explicit "memory" and "imm_disp" attributes.
20978 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
20980 PR middle-end/85414
20981 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
20982 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
20983 gen_lowpart_no_emit.
20985 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
20988 * config/i386/i386.c (ix86_expand_builtin): Change memory
20989 operand to XI, extend p0 to Pmode.
20990 * config/i386/i386.md: Change unspec volatile and operand
20991 1 mode to XI, change operand 0 mode to P.
20993 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20995 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
20996 GET_MODE_MASK before any checking.
20997 (nds32_can_use_bset_p): Likewise.
20998 (nds32_can_use_btgl_p): Likewise.
21000 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21002 * config/nds32/nds32-doubleword.md: New define_split pattern for
21003 illegal register number.
21005 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21007 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
21009 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21011 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
21013 2018-04-25 Richard Biener <rguenther@suse.de>
21015 * lto-streamer.h (LTO_major_version): Bump to 8.
21017 2018-04-25 Jakub Jelinek <jakub@redhat.com>
21019 * BASE-VER: Set to 9.0.0.
21021 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
21023 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
21024 in __abskf2 and __powikf2.
21026 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21029 * config/aarch64/constraints.md (Usg, Usj): New constraints.
21030 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
21031 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
21032 Use the above on operand 2. Reindent.
21033 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
21035 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
21038 * common/config/i386/i386-common.c (ix86_handle_option): Don't
21040 * config/i386/i386.opt (mcet): Removed.
21041 * doc/install.texi: Remove -mcet documentation.
21042 * doc/invoke.texi: Likewise.
21044 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
21047 * doc/install.texi: Remove -mcet from bootstrap-cet.
21049 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21052 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
21053 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
21057 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
21058 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
21059 containing a CONST_VECTOR.
21061 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
21063 * doc/install.texi: Update newlib dependency for nvptx.
21065 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21068 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
21069 instead of INTVAL when shifting x left.
21071 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
21073 PR tree-optimization/85478
21074 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
21075 vect_grouped_store_supported for single element vectors.
21077 2018-04-24 Richard Biener <rguenther@suse.de>
21080 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
21081 load cost increase to the case of non-constant step.
21083 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21086 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
21087 destination if any_malformed_asm.
21089 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
21091 PR middle-end/85496
21092 * expr.c (store_field): In the bitfield case, if the value comes from
21093 a function call and is returned in registers by means of a PARALLEL,
21094 do not change the mode of the temporary unless BLKmode and VOIDmode.
21096 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
21098 PR rtl-optimization/85423
21099 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
21100 dependencies to debug insns when the previous insn is non-debug.
21102 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
21104 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
21105 enums into a single definition.
21106 (fls): Fix predicates and printing.
21109 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
21111 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
21112 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
21113 and short u6 immediate.
21114 (check_if_valid_sleep_operand): Remove.
21115 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
21117 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21119 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
21120 flag_always_save_lp condition.
21121 * config/nds32/nds32.opt (malways-save-lp): New option.
21123 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21125 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
21126 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
21127 * config/nds32/nds32.h
21128 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
21129 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
21131 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21133 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
21134 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
21136 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21137 Chung-Ju Wu <jasonwucj@gmail.com>
21139 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
21141 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
21142 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
21144 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21146 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
21148 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21150 * config/nds32/nds32-protos.h (nds32_data_alignment,
21151 nds32_local_alignment): Declare.
21152 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
21153 nds32_local_alignment): New functions.
21154 (TARGET_CONSTANT_ALIGNMENT): Define.
21155 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
21157 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21159 * config/nds32/nds32.c
21160 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
21161 (TARGET_MODES_TIEABLE_P): Likewise.
21163 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21165 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
21166 level Ofast and Og.
21168 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
21169 Chung-Ju Wu <jasonwucj@gmail.com>
21171 * config/nds32/constants.md (unspec_volatile_element): Add enum values
21172 for unaligned access.
21173 * config/nds32/nds32-intrinsic.c: Implementation of expanding
21175 * config/nds32/nds32-intrinsic.md: Likewise.
21176 * config/nds32/nds32_intrinsic.h: Likewise.
21177 * config/nds32/nds32.h (nds32_builtins): Likewise.
21178 * config/nds32/nds32.opt (munaligned-access): New option.
21179 * config/nds32/nds32.c (nds32_asm_file_start): Display
21180 flag_unaligned_access status.
21182 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
21184 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
21185 -mno-relax is present.
21186 * config/riscv/linux.h (LINK_SPEC): Ditto.
21188 2018-04-20 Martin Sebor <msebor@redhat.com>
21191 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
21193 (gimple_fold_builtin_stxcpy_chk): Same.
21194 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
21196 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
21199 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
21200 __powikf2 when long double is IEEE 128-bit.
21202 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
21204 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
21205 step to make sure stack always aligned.
21207 2018-04-20 Carl Love <cel@us.ibm.com>
21210 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
21211 size check for arg0.
21213 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
21214 Tom de Vries <tom@codesourcery.com>
21217 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
21218 Emit insns for calls too.
21219 (nvptx_find_par): Always look for worker-level predecessor insn.
21220 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
21222 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
21223 (nvptx_process_pars): Propagate frames for calls.
21225 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
21228 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
21230 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
21231 (ix86_handle_option): Don't handle OPT_mibt.
21232 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
21234 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
21236 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
21237 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
21238 (ix86_target_macros): Define __CET__ with flag_cf_protection
21239 for -fcf-protection.
21240 * config/i386/i386.c (isa2_opts): Remove -mibt.
21241 * config/i386/i386.h (TARGET_IBT): Removed.
21242 (TARGET_IBT_P): Likewise.
21243 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
21244 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
21245 * config/i386/i386.opt (mcet): Update help message.
21246 (mshstk): Likewise.
21248 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
21249 -mcet as an alias for -mshstk.
21251 2018-04-20 Richard Biener <rguenther@suse.de>
21253 PR middle-end/85475
21254 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
21255 complexity by forcing a single use of the multiply operand.
21257 2018-04-20 Martin Jambor <mjambor@suse.cz>
21260 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
21261 recursion dependency to only apply to non-clones.
21263 2018-04-20 Martin Jambor <mjambor@suse.cz>
21266 * ipa-cp.c (create_specialized_node): Check that clones of
21267 self-recursive edges exist during IPA-CP.
21269 2018-04-19 Toon Moene <toon@moene.org>
21271 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
21274 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21276 PR tree-optimization/85467
21277 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
21278 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
21279 VECTOR_CST element to type.
21281 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21284 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
21285 * config/i386/i386.md (builtin_setjmp_setup): Removed.
21286 (builtin_longjmp): Likewise.
21287 (save_stack_nonlocal): New pattern.
21288 (restore_stack_nonlocal): Likewise.
21290 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21293 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
21294 Replace ASM_OUTPUT_LABEL with fprintf.
21296 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21299 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
21300 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
21301 * config/i386/i386-c.c (ix86_target_macros_internal): Also
21302 define __IBT__ and __SHSTK__ for -fcf-protection.
21303 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
21305 (ix86_trampoline_init): Likewise.
21306 (x86_output_mi_thunk): Likewise.
21307 (ix86_notrack_prefixed_insn_p): Likewise.
21308 (ix86_option_override_internal): Don't disallow -fcf-protection.
21309 * config/i386/i386.md (rdssp<mode>): Also enable for
21311 (incssp<mode>): Likewise.
21312 (nop_endbr): Likewise.
21313 * config/i386/i386.opt (mcet): Change help message to built-in
21316 (mshstk): Likewise.
21317 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
21318 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
21319 enable CET built-in functions.
21321 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
21323 * common/config/i386/i386-common.c
21324 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
21325 OPTION_MASK_ISA_MOVDIRI_UNSET,
21326 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
21327 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
21328 * config.gcc (movdirintrin.h): New header.
21329 * config/i386/cpuid.h (bit_MOVDIRI,
21330 bit_MOVDIR64B): New bits.
21331 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
21333 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
21334 (VOID, PVOID, PCVOID)): New function types.
21335 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
21336 __builtin_ia32_directstoreu_u64,
21337 __builtin_ia32_movdir64b): New builtins.
21338 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
21339 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
21341 (ix86_valid_target_attribute_inner_p): Ditto.
21342 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
21343 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
21344 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
21345 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
21346 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
21347 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
21348 (movdiri<mode>, movdir64b_<mode>): New.
21349 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
21350 * config/i386/immintrin.h: Include movdirintrin.h.
21351 * config/i386/movdirintrin.h: New file.
21352 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
21354 2018-04-19 Richard Biener <rguenther@suse.de>
21356 PR middle-end/85455
21357 * cfg.c (clear_bb_flags): When loop state says we have
21358 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
21360 2018-04-19 Richard Biener <rguenther@suse.de>
21362 PR tree-optimization/84737
21363 * tree-vect-data-refs.c (vect_copy_ref_info): New function
21364 copying restrict info.
21365 (vect_setup_realignment): Use it.
21366 * tree-vectorizer.h (vect_copy_ref_info): Declare.
21367 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
21368 the first DR to all generated stores.
21369 (vectorizable_load): Likewise for loads.
21371 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21373 PR tree-optimization/85446
21374 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
21375 the integral and pointer types to have the same precision.
21377 * doc/install.texi: Document --disable-cet being the default and
21380 2018-04-18 Martin Liska <mliska@suse.cz>
21382 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
21385 2018-04-18 Martin Liska <mliska@suse.cz>
21388 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21391 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21392 arguments if they are comparable.
21394 2018-04-18 Martin Liska <mliska@suse.cz>
21397 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21400 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21403 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
21406 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
21407 ENDBR after calling __morestack.
21409 2018-04-18 David Malcolm <dmalcolm@redhat.com>
21412 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
21413 by using gcc_base_ver to generate a gcc_driver_version, and use
21414 it when generating GCC_DRIVER_NAME.
21415 * configure: Regenerate.
21417 2018-04-18 Jakub Jelinek <jakub@redhat.com>
21420 * config.gcc: Obsolete powerpc*-*-*spe*.
21422 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21425 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
21426 (stabstr_D): Change type of unum from unsigned int to
21427 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
21430 2018-04-17 Jim Wilson <jimw@sifive.com>
21433 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
21434 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
21435 Set arg_pointer_offset after using pretend_args_size.
21437 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21439 PR rtl-optimization/85431
21440 * dse.c (record_store): Ignore zero width stores.
21443 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
21444 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
21445 __builtin_stack_restore rather than after it.
21446 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
21447 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
21448 argument instead of virtual_dynamic_stack_rtx.
21450 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
21452 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
21454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21455 Add note to error message to explain internal mapping of overloaded
21456 built-in function name to non-overloaded built-in function name.
21457 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
21460 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21463 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
21464 where the inputs overlap with the output.
21466 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21468 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
21469 (=v, v) alternative and explicit "memory" attribute.
21470 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
21471 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21473 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
21474 "sselog1" type instead of "sselog".
21475 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
21476 "sselog". Remove explicit "memory" attribute.
21477 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
21478 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21480 (vec_extract_hi_v32hi): Merge all alternatives into one, use
21481 "sselog1" type instead of "sselog". Remove explicit "memory"
21483 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
21484 use "sselog1" type instead of "sselog". Remove explicit "memory"
21486 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
21487 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21489 (vec_extract_hi_v64qi): Merge all alternatives into one, use
21490 "sselog1" type instead of "sselog". Remove explicit "memory"
21492 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
21493 use "sselog1" type instead of "sselog". Remove explicit "memory"
21497 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
21499 PR middle-end/85414
21500 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
21503 2018-04-17 Martin Jambor <mjambor@suse.cz>
21506 * ipa-cp.c (create_specialized_node): Call
21507 expand_all_artificial_thunks if necessary.
21509 2018-04-17 Martin Liska <mliska@suse.cz>
21512 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
21513 in message, remote space in between '_G' and '('.
21515 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21518 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
21519 avx512f_vmcmp<mode>3<round_saeonly_name>,
21520 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
21521 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
21522 avx512f_rndscale<mode><round_saeonly_name>,
21523 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
21524 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
21525 Use %<iptr>2 instead of %2 for -masm=intel.
21526 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
21527 avx512f_vcvttss2usi<round_saeonly_name>,
21528 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
21530 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
21531 avx512f_vcvttsd2usi<round_saeonly_name>,
21532 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
21533 Use %q1 instead of %1 for -masm=intel.
21534 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
21535 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
21536 of %3 for -masm=intel.
21537 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
21539 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
21541 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
21543 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
21544 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
21546 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
21548 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
21549 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
21550 %g1 and one with %0 and %1.
21551 (avx512er_vmrcp28<mode><round_saeonly_name>,
21552 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
21553 %1 for -masm=intel.
21554 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
21555 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
21556 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
21557 of %0 and %{%4%} for -masm=intel.
21558 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
21559 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
21560 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
21561 order of %0 and %{%5%}%{z%} for -masm=intel.
21563 2018-04-17 Jan Hubicka <jh@suse.cz>
21566 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
21568 2018-04-17 Martin Liska <mliska@suse.cz>
21571 * multiple_target.c (create_dispatcher_calls): Set apostrophes
21572 for target_clone error message. Make default implementation
21573 clone to be a local declaration.
21574 (separate_attrs): Add new argument and check for an empty
21576 (expand_target_clones): Handle it.
21577 (ipa_target_clone): Make redirection just for target_clones
21580 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
21581 Tom de Vries <tom@codesourcery.com>
21583 PR middle-end/84955
21584 * omp-expand.c (expand_oacc_for): Add dummy false branch for
21585 tiled basic blocks without omp continue statements.
21587 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
21590 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
21591 vec_extract expression as having side effects to make sure it gets
21594 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
21597 * config/i386/i386.c (get_builtin_code_for_version): Check
21600 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
21603 * config.gcc: Support "skylake".
21604 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21606 * config/i386/i386.c (m_SKYLAKE): Define.
21607 (processor_target_table): Add "skylake".
21608 (ix86_option_override_internal): Add "skylake".
21609 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
21610 PROCESSOR_CANNONLAKE.
21611 (get_builtin_code_for_version): Fix priority for
21612 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
21613 PROCESSOR_SKYLAKE-AVX512.
21614 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
21615 (processor_type): Add PROCESSOR_SKYLAKE.
21617 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
21618 Jason Merrill <jason@redhat.com>
21621 * convert.c (convert_to_integer_1): Use direct recursion for
21622 enumeral types and types with a precision less than the number
21623 of bits in their mode.
21625 2018-04-16 Julia Koval <julia.koval@intel.com>
21628 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
21629 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
21631 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
21634 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
21635 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
21636 and -mno-direct-move.
21638 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
21641 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
21642 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
21643 Ensure negative shifts result in {0}.
21645 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
21647 PR rtl-optimization/79916
21648 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
21649 regs (if any) to define how to gnerate SD moves when LRA is in
21652 2018-04-13 Jakub Jelinek <jakub@redhat.com>
21654 PR rtl-optimization/85393
21655 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
21656 * except.c (expand_dw2_landing_pad_for_region): Make static.
21657 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
21658 a label and unconditional jump to old_bb, rather than
21659 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
21662 PR rtl-optimization/85376
21663 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
21664 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
21665 instead of a specific value.
21667 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
21668 Bin Cheng <bin.cheng@arm.com>
21670 PR tree-optimization/82965
21671 PR tree-optimization/83991
21672 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
21673 by_profile_only parameter.
21674 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
21675 information if the loop was predicted to iterate too many times.
21676 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
21678 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
21681 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
21684 2018-04-13 Martin Liska <mliska@suse.cz>
21685 Jakub Jelinek <jakub@redhat.com>
21687 PR middle-end/81657
21688 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
21689 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
21690 * builtins.c (expand_builtin_memory_copy_args): Use
21691 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
21692 handle dest_addr == pc_rtx.
21694 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21697 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
21698 asked to not generate direct moves.
21699 (fix_trunc<mode>si2_stfiwx): Similar.
21700 (fix_trunc<mode>si2_internal): Similar.
21702 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21705 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
21706 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
21707 lookup if dest in some wider mode is known to be const0_rtx and
21708 if so, record permanent equivalence for it to be ZERO_EXTEND of
21709 the narrower mode destination.
21711 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
21713 * lto-streamer-out.c (output_function): Revert 259346.
21714 * omp-expand.c (expand_oacc_for): Likewise.
21716 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
21718 PR rtl-optimization/85354
21719 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
21720 * sel-sched.c (sel_global_init): ... here.
21722 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
21725 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
21726 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
21727 mode for PE-COFF targets.
21728 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
21729 (i386_pe_asm_lto_end): Likewise.
21730 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
21731 (TARGET_ASM_LTO_END): Likewise.
21732 * config/i386/winnt.c (saved_debug_info_level): New static variable.
21733 (i386_pe_asm_lto_start): New function.
21734 (i386_pe_asm_lto_end): Likewise.
21736 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
21737 Richard Biener <rguenther@suse.de>
21739 PR middle-end/84955
21740 * lto-streamer-out.c (output_function): Fix CFG loop state before
21742 * omp-expand.c (expand_oacc_for): Handle calls to internal
21743 functions like regular functions.
21745 2018-04-12 Richard Biener <rguenther@suse.de>
21748 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
21749 for the early LTO debug to properly generate references to it
21750 during DIE emission. Do not re-use that for the skeleton for
21752 (dwarf2out_early_finish): Likewise.
21754 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21757 * config/i386/sse.md
21758 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
21759 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
21760 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
21761 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
21762 and output is a reg, avoid creating invalid lowpart subreg, but
21763 instead split into a 512-bit move. Don't split if not AVX512VL,
21764 input is xmm16+ reg and output is a mem.
21765 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
21766 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
21767 xmm16+ reg and output is a mem.
21769 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21771 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
21772 also for flag_dwarf2_cfi_asm.
21774 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21776 PR rtl-optimization/85342
21777 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
21778 a bool scalar var inside of the loop instead. Don't try to update
21779 recog_data.operand after failed apply_change_group.
21781 2018-04-12 Tom de Vries <tom@codesourcery.com>
21784 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
21785 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
21786 array with flexible array member as array without given dimension.
21787 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
21788 argument for undefined param to true.
21790 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
21793 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
21794 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
21795 from PowerPC section.
21796 * config/rs6000/sysv4.opt (mcall-): Improve help text.
21797 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
21798 help text that is too long.
21799 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
21800 help text that is too long.
21801 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
21802 help text that is too long.
21804 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
21806 * config/alpha/alpha.md (stack_probe_internal): Rename
21807 from "probe_stack". Update all callers.
21809 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21811 PR rtl-optimization/84566
21812 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
21813 sched_macro_fuse_insns.
21815 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21818 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
21819 (compute_block_dependences): ... from here.
21821 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21823 PR tree-optimization/85331
21824 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
21825 from int to HOST_WIDE_INT.
21827 2018-04-11 Martin Jambor <mjambor@suse.cz>
21830 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
21831 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
21832 not the same as the source val.
21833 (cgraph_edge_brings_value_p): New parameter.
21834 (gather_edges_for_value): Pass destination value to
21835 cgraph_edge_brings_value_p.
21836 (perhaps_add_new_callers): Likewise.
21837 (get_info_about_necessary_edges): Likewise and exclude values brought
21838 only by self-recursive edges.
21839 (create_specialized_node): Redirect only clones of self-calling edges.
21840 (+self_recursive_pass_through_p): New function.
21841 (find_more_scalar_values_for_callers_subset): Use it.
21842 (find_aggregate_values_for_callers_subset): Likewise.
21843 (known_aggs_to_agg_replacement_list): Removed.
21844 (decide_whether_version_node): Re-calculate known constants for all
21845 remaining context clones.
21847 2018-04-11 Richard Biener <rguenther@suse.de>
21850 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
21851 from early DWARF output.
21852 (dwarf2out_early_finish): Output line info unconditionally into
21853 early DWARF and add reference to it.
21855 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21858 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
21859 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
21860 other than V2DFmode using iptr mode attribute.
21861 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
21863 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21865 PR rtl-optimization/84659
21866 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
21868 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21871 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
21873 (output_loc_list): Pass address of a dummy size variable even in the
21874 locview handling loop.
21875 (index_location_lists): Add comment on why skip_loc_list_entry can't
21878 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
21881 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
21884 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
21887 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
21888 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
21889 and -mstring-compare-inline-limit.
21891 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21894 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
21895 for stack clash protection in a register whenever we need it to be in
21898 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21900 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
21901 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
21903 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21906 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
21908 (mlong-double-): Ditto.
21909 * config/rs6000/sysv4.opt (msdata=): Ditto.
21910 (mtls-size=): Ditto.
21912 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21914 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
21915 erroneous entries for
21916 "vector int vec_ldl (int, long int *)", and
21917 "vector unsigned int vec_ldl (int, unsigned long int *)".
21918 Add comments and entries for
21919 "vector bool char vec_ldl (int, bool char *)",
21920 "vector bool short vec_ldl (int, bool short *)",
21921 "vector bool int vec_ldl (int, bool int *)",
21922 "vector bool long long vec_ldl (int, bool long long *)",
21923 "vector pixel vec_ldl (int, pixel *)",
21924 "vector long long vec_ldl (int, long long *)",
21925 "vector unsigned long long vec_ldl (int, unsigned long long *)".
21926 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
21927 type tree bool_long_long_type_node and correct definition of
21928 bool_V2DI_type_node to make reference to this new type tree.
21929 (rs6000_mangle_type): Replace erroneous reference to
21930 bool_long_type_node with bool_long_long_type_node.
21931 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
21932 comments to emphasize sign distinctions for char and int types and
21933 replace RS6000_BTI_bool_long constant with
21934 RS6000_BTI_bool_long_long constant. Also add comment to restrict
21935 use of RS6000_BTI_pixel.
21936 (bool_long_type_node): Remove this macro definition.
21937 (bool_long_long_type_node): New macro definition
21939 2018-04-10 Jakub Jelinek <jakub@redhat.com>
21941 PR rtl-optimization/85300
21942 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
21943 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
21944 simplify_unary_operation fails.
21946 2018-04-10 Martin Liska <mliska@suse.cz>
21948 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
21949 cgraph_edge and ipa_ref.
21951 2018-04-10 Jakub Jelinek <jakub@redhat.com>
21955 * config/i386/sse.md
21956 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
21957 computation of the VEC_MERGE selector from mask.
21958 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
21959 Fix decoding of the VEC_MERGE selector into mask.
21961 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
21963 PR tree-optimization/85286
21964 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
21966 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
21968 * final.c (final_1): Set insn_last_address as well as
21969 insn_current_address.
21971 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21974 * explow.c (emit_stack_probe): Call validize_mem on memory location
21975 before passing it to gen_probe_stack. Create address operand and
21976 legitimize it for the probe_stack_address case.
21978 2018-04-09 Jan Hubicka <jh@suse.cz>
21981 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
21982 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
21983 * tree.c (free_lang_data_in_type): Fix handling of binfos;
21985 (free_lang_data): Rebuild type inheritance graph.
21987 2018-04-09 Martin Sebor <msebor@redhat.com>
21989 * invoke.texi (-finline-small-functions): Mention other optimization
21991 (-findirect-inlining, -fpartial-inlining): Same.
21992 (-finline-functions-called-once): Same.
21993 (-freorder-blocks-and-partition): Same.
21995 2018-04-09 Jan Hubicka <jh@suse.cz>
21998 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
21999 jumps; choose last target that matches the criteria (i.e.
22000 no partition changes for non-crossing jumps).
22001 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
22002 support for redirecting crossing jumps to non-crossing.
22004 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
22006 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
22007 also for naked functions.
22009 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
22011 * config/arc/arc.md (add_shift): New pattern.
22012 (add_shift2): Likewise.
22013 (sub_shift): Likewise.
22014 (sub_shift_cmp0_noout): Likewise.
22015 (compare_si_ashiftsi): Likewise.
22016 (xbfu_cmp0_noout): New combine pattern.
22017 (xbfu_cmp0"): Likewise.
22018 (movsi_set_cc_insn): Place the predicable variant first.
22019 (commutative_binary_cmp0_noout): Remove clobber.
22020 (commutative_binary_cmp0): New pattern.
22021 (noncommutative_binary_cmp0): Likewise.
22022 (noncommutative_binary_cmp0_noout): Likewise.
22023 (noncommutative_binary_comparison_result_used): Removed.
22024 (rsub_cmp0): New pattern.
22025 (rsub_cmp0_noout): Likewise.
22026 (extzvsi): Changed, keep only meaningful variants.
22027 (SQH, SEZ): New iterators.
22028 (SQH_postfix): New mode attribute.
22029 (SEZ_prefix): New code attribute.
22030 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
22031 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
22032 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
22033 of numerical value.
22034 (noncommutative_operator): Check the availability of barrel
22037 2018-04-09 Richard Biener <rguenther@suse.de>
22039 PR tree-optimization/85284
22040 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
22041 Only use the niter constraining form of simple_iv when the exit
22042 is always executed.
22044 2018-04-09 Tom de Vries <tom@codesourcery.com>
22047 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22048 (define_expand "*memory_barrier"): New define_expand.
22049 (define_insn "memory_barrier"): New insn.
22051 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22053 PR rtl-optimization/80463
22054 PR rtl-optimization/83972
22055 PR rtl-optimization/83480
22057 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
22058 correct producer for the insn.
22059 (tidy_control_flow): Fixup seqnos in case of debug insns.
22061 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22063 PR rtl-optimization/83913
22065 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
22066 different sched-times when merging exprs.
22068 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22070 PR rtl-optimization/83962
22072 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
22073 tidy_fallthru_edge and tidy_control_flow.
22075 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22077 PR rtl-optimization/83530
22079 * sel-sched.c (force_next_insn): New global variable.
22080 (remove_insn_for_debug): When force_next_insn is true, also leave only
22081 next insn in the ready list.
22082 (sel_sched_region): When the region wasn't scheduled, make another pass
22083 over it with force_next_insn set to 1.
22085 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
22087 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
22089 * config/nds32/constants.md (unspec_volatile_element): Add enum values
22090 for interrupt control.
22091 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
22092 functions for interrupt control.
22093 * config/nds32/nds32-intrinsic.md: Likewise.
22094 * config/nds32/nds32_intrinsic.h: Likewise.
22095 * config/nds32/nds32.h (nds32_builtins): Likewise.
22097 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
22099 * config/nds32/nds32.c (nds32_init_machine_status,
22100 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
22101 strict_aligned_p field.
22102 (nds32_expand_to_rtl_hook): New function.
22103 (TARGET_EXPAND_TO_RTL_HOOK): Define.
22104 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
22106 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22107 Chung-Ju Wu <jasonwucj@gmail.com>
22109 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
22110 * config/nds32/nds32-n7.md: New file.
22111 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
22112 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
22114 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
22115 * config/nds32/nds32.md (pipeline_model): Add n7.
22116 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
22117 * config/nds32/pipelines.md: Include n7 settings.
22119 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22120 Chung-Ju Wu <jasonwucj@gmail.com>
22122 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
22123 * config/nds32/nds32-e8.md: New file.
22124 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
22125 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
22127 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
22128 * config/nds32/nds32.md (pipeline_model): Add e8.
22129 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
22130 * config/nds32/pipelines.md: Include e8 settings.
22132 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22133 Chung-Ju Wu <jasonwucj@gmail.com>
22135 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
22136 * config/nds32/nds32-n8.md: New file.
22137 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
22138 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
22140 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
22141 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
22142 * config/nds32/nds32.md (pipeline_model): Add n8.
22143 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
22144 * config/nds32/pipelines.md: Include n8 settings.
22146 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22147 Chung-Ju Wu <jasonwucj@gmail.com>
22149 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
22150 * config/nds32/nds32-n9-2r1w.md: New file.
22151 * config/nds32/nds32-n9-3r2w.md: New file.
22152 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
22153 nds32_register_ports): New or modify for cpu n9.
22154 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
22156 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
22157 * config/nds32/nds32-utils.c: New file.
22158 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
22159 TARGET_MUL_SLOW): Define.
22160 * config/nds32/nds32.md (pipeline_model): New attribute.
22161 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
22162 New options that support cpu n9.
22163 * config/nds32/pipelines.md: Include n9 settings.
22164 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
22166 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
22168 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
22169 information if necessary.
22170 (output_cond_branch_compare_zero): Likewise.
22171 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
22172 (nds32_target_alignment): Refine for alignment.
22173 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
22174 (FUNCTION_BOUNDARY): Modify.
22175 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
22177 * config/nds32/nds32.opt (malways-align, malign-functions): New.
22179 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
22181 * config/nds32/constants.md (unspec_volatile_element): Add values for
22182 TLB operation and data prefetch.
22183 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
22184 functions for TLB operation and data prefetch.
22185 * config/nds32/nds32-intrinsic.md: Likewise.
22186 * config/nds32/nds32_intrinsic.h: Likewise.
22187 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
22188 (nds32_print_operand): Likewise.
22189 * config/nds32/nds32.h (nds32_builtins): Likewise.
22191 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22192 Andrew Pinski <pinsika@gcc.gnu.org>
22194 PR middle-end/82976
22195 * match.pd: Use constant_boolean_node of correct type instead of
22196 boolean_true_node or boolean_false_node for simplifying
22197 pointer comparisons to zero.
22199 2018-04-07 Jakub Jelinek <jakub@redhat.com>
22201 PR tree-optimization/80021
22202 * tree.c (verify_type_variant): Make error call in verify_variant_match
22203 translatable and remove final full stop.
22205 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22207 * config/nds32/constants.md (unspec_volatile_element): Add
22208 UNSPEC_VOLATILE_EH_RETURN.
22209 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
22210 nds32_output_stack_pop): Support dwarf exception handling process.
22211 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
22212 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
22213 exception handling process.
22214 (nds32_compute_stack_frame): Likewise.
22215 (nds32_return_addr_rtx): Likewise.
22216 (nds32_initial_elimination_offset): Likewise.
22217 (nds32_expand_prologue): Likewise.
22218 (nds32_expand_epilogue): Likewise.
22219 (nds32_dynamic_chain_address): New function.
22220 * config/nds32/nds32.h (machine_function): Add fields for dwarf
22221 exception handling.
22222 (DYNAMIC_CHAIN_ADDRESS): Define.
22223 (EH_RETURN_DATA_REGNO): Define.
22224 (EH_RETURN_STACKADJ_RTX): Define.
22225 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
22226 patterns for dwarf exception handling.
22228 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22230 * config/nds32/nds32.h: Clean up obsolete macros.
22232 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22234 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22235 Add enum values for particular instructions.
22236 * config/nds32/nds32-intrinsic.c: Implementation of expanding
22237 particular intrinsic functions.
22238 * config/nds32/nds32-intrinsic.md: Likewise.
22239 * config/nds32/nds32_intrinsic.h: Likewise.
22240 * config/nds32/nds32.h (nds32_builtins): Likewise.
22241 * config/nds32/nds32.md (type): Add pbsad and pbsada.
22242 (btst, ave): New patterns for particular instructions.
22244 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22246 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22247 Add enum values for atomic load/store and memory sync.
22248 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
22250 * config/nds32/nds32-intrinsic.md: Likewise.
22251 * config/nds32/nds32_intrinsic.h: Likewise.
22252 * config/nds32/nds32.h (nds32_builtins): Likewise.
22254 2018-04-07 Jakub Jelinek <jakub@redhat.com>
22256 PR tree-optimization/85257
22257 * fold-const.c (native_encode_vector): If not all elts could fit
22258 and off is -1, return 0 rather than offset.
22259 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
22260 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
22261 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
22262 adjust buffer in native_interpret_expr call.
22264 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22266 * config/nds32/constants.md (unspec_volatile_element): Add cache
22267 control enum values.
22268 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
22269 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
22270 * config/nds32/nds32.c (nds32_cctl_names): New.
22271 (nds32_print_operand): Handle cache control register names.
22272 * config/nds32/nds32.h (nds32_builtins): New enum values.
22273 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
22275 * config/nds32/nds32.md (type): Add mmu.
22276 * config/nds32/pipelines.md (simple_insn): Add mmu.
22278 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22280 * config/nds32/nds32.md (type): Remove call.
22281 * config/nds32/pipelines.md (simple_insn): Likewise.
22283 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22285 * config/nds32/constants.md (unspec_volatile_element): Add
22286 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
22287 UNSPEC_VOLATILE_FMFCFG.
22288 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
22289 description for fmfcfg and fmfcsr.
22290 (bdesc_1arg): Add fmtcsr.
22291 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
22292 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
22293 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
22294 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
22295 unspec_fmfcfg): New patterns.
22296 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
22297 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
22298 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
22299 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
22300 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
22301 __nds32__fmfcfg): Define.
22303 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22305 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
22306 intrinsic register names.
22307 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
22308 intrinsic register enum values and macros.
22310 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22312 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
22313 for load/store addressing form.
22314 (nds32_print_operand_address): Likewise.
22316 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
22319 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
22320 based on LABEL_REF. Remove useless assertion.
22321 (pic_address_needs_scratch): Fix formatting.
22322 (sparc_legitimize_pic_address): Minor tweaks.
22323 (sparc_delegitimize_address): Adjust assertion accordingly.
22324 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
22325 into symbolic_operand.
22326 (movsi_high_pic_label_ref): Likewise.
22327 (movsi_lo_sum_pic_label_ref): Likewise.
22328 (movdi_pic_label_ref): Likewise.
22329 (movdi_high_pic_label_ref): Likewise.
22330 (movdi_lo_sum_pic_label_ref): Likewise.
22332 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
22334 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
22335 custom LIB_SPEC setup.
22337 2018-04-06 Ruslan Bukin <br@bsdpad.com>
22338 Kito Cheng <kito.cheng@gmail.com>
22340 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
22341 * config/riscv/freebsd.h: New.
22343 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22345 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
22346 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
22349 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22350 Kito Cheng <kito.cheng@gmail.com>
22352 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
22353 nds32_output_call, nds32_symbol_binds_local_p): New functions.
22354 * config/nds32/nds32-protos.h (nds32_output_call,
22355 nds32_output_return): Declare.
22356 * config/nds32/nds32.md: Refine all the call and return patterns.
22358 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22361 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
22362 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
22364 PR rtl-optimization/84872
22365 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
22366 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
22367 EDGE_CROSSING edge.
22369 2018-04-06 Tamar Christina <tamar.christina@arm.com>
22371 * expr.c (copy_blkmode_to_reg): Revert 254862.
22372 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
22374 2018-04-06 Richard Biener <rguenther@suse.de>
22376 PR middle-end/85244
22377 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
22378 after seeing a component reference with an adjacent field. Treat
22379 refs to arrays at struct end of external decls similar to
22380 refs to unconstrained commons.
22382 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22385 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
22386 look through SAVE_EXPRs with non-side-effects argument. Adjust
22388 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
22391 2018-04-06 Richard Biener <rguenther@suse.de>
22393 PR middle-end/85180
22394 * alias.c (find_base_term): New wrapper around find_base_term
22395 unwinding CSELIB_VAL_PTR changes.
22396 (find_base_term): Do not restore CSELIB_VAL_PTR during the
22399 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22401 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
22403 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
22404 constant definitions.
22405 ("nop"): lr 0,0 -> nopr r0
22406 ("nop_lr0", "nop_lr1"): New insn definitions.
22408 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22410 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
22411 NDS32_V3PUSH_AVAILABLE_P macro.
22413 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
22414 Chung-Ju Wu <jasonwucj@gmail.com>
22416 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
22417 (nds32*-*-*): Add float and fpu_config into supported_defaults.
22418 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
22419 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
22420 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
22421 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
22422 * config/nds32/constraints.md: New constraints and checking for hard
22423 float configuration.
22424 * config/nds32/iterators.md: New mode iterator and attribute for hard
22425 float configuration.
22426 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
22428 * config/nds32/nds32-fpu.md: New file.
22429 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
22430 deal with hard float code generation.
22431 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
22433 (abi_type, float_reg_number): New enum type.
22434 * config/nds32/nds32-predicates.c: New predicates for hard float.
22435 * config/nds32/nds32-protos.h: Declare functions for hard float.
22436 * config/nds32/nds32.c: Implementation for hard float configuration.
22437 * config/nds32/nds32.h: Definitions for hard float configuration.
22438 * config/nds32/nds32.md: Include hard float machine description and
22439 modify patterns for hard float configuration.
22440 * config/nds32/nds32.opt: New options for hard float configuration.
22441 * config/nds32/predicates.md: New predicates for hard float
22444 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22446 * common/config/nds32/nds32-common.c
22447 (nds32_option_optimization_table): Enable -mreleax-hint by default.
22449 2018-04-05 Jakub Jelinek <jakub@redhat.com>
22451 PR middle-end/85195
22452 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
22453 CONSTRUCTOR_ELT (ctor, ...)->value.
22455 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
22458 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
22460 2018-04-05 Tom de Vries <tom@codesourcery.com>
22463 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
22466 2018-04-05 Shiva Chen <shiva0217@gmail.com>
22467 Kito Cheng <kito.cheng@gmail.com>
22469 * config/nds32/constraints.md (U33): Fine-tune checking condition.
22470 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
22471 * config/nds32/nds32.h (nds32_16bit_address_type): Add
22472 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
22474 2018-04-05 Shiva Chen <shiva0217@gmail.com>
22475 Kito Cheng <kito.cheng@gmail.com>
22477 * config/nds32/constraints.md (Ufe): New memory constraint.
22478 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
22479 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
22480 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
22482 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
22483 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
22485 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22487 * config/nds32/nds32.md: Use optimize_size in the condition for
22488 alu-shift instructions.
22490 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22492 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
22494 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22496 * config/nds32/nds32.md (negsi2): Refine pattern.
22498 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
22499 Chung-Ju Wu <jasonwucj@gmail.com>
22501 * config/nds32/iterators.md (shift_rotate): New code iterator.
22502 (shift): New code attribute.
22503 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
22504 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
22505 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
22506 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
22507 bit-wise operations.
22508 (andsi3, *andsi3): Ditto.
22509 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
22510 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
22511 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
22512 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
22513 nds32_ior_operand, nds32_xor_operand): New predicates.
22515 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22517 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
22518 (addsi3, subsi3): ... this.
22520 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22522 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
22524 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22526 * config/nds32/nds32.md: Adjust indention.
22528 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
22530 * config/nds32/nds32.md (feature): New attribute.
22532 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22534 * config/nds32/nds32.md (subtype): New attribute.
22536 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22539 * config/arm/arm-builtins.c (arm_expand_builtin): Change
22540 expansion to perform a bitwise AND of the argument followed by a
22541 boolean negation of the result.
22543 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
22545 PR rtl-optimization/84878
22546 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
22547 the basic block. Assert the use reference is not artificial and that
22548 it has an associated insn.
22550 2018-04-04 Michael Matz <matz@suse.de>
22552 * builtins.c (compute_objsize): Pass correct operand
22553 to array_at_struct_end_p.
22555 2018-04-04 Richard Biener <rguenther@suse.de>
22558 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
22559 from contexts for DINFO_LEVEL_TERSE and below.
22561 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22563 * config/nds32/nds32-doubleword.md (move_<mode>): Require
22564 resiter_operand condition.
22565 * config/nds32/nds32.md (*move<mode>): Ditto.
22567 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22568 Monk Chiang <sh.chiang04@gmail.com>
22570 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
22572 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22574 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
22576 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22577 Kito Cheng <kito.cheng@gmail.com>
22579 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
22580 nds32_cond_code_str, output_cond_branch,
22581 output_cond_branch_compare_zero, nds32_expand_cbranch,
22582 nds32_expand_cstore, nds32_expand_movcc,
22583 nds32_output_cbranchsi4_equality_zero,
22584 nds32_output_cbranchsi4_equality_reg,
22585 nds32_output_cbranchsi4_equality_reg_or_const_int,
22586 nds32_output_cbranchsi4_greater_less_zero: New functions.
22587 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
22588 nds32_expand_cstore, nds32_expand_movcc,
22589 nds32_output_cbranchsi4_equality_zero,
22590 nds32_output_cbranchsi4_equality_reg,
22591 nds32_output_cbranchsi4_equality_reg_or_const_int,
22592 nds32_output_cbranchsi4_greater_less_zero): Declare.
22593 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
22594 nds32_rimm11s_operand): New predicates.
22595 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
22596 * config/nds32/nds32.md: Rewrite all the branch and conditional move
22599 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22601 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
22602 * config/nds32/nds32.md: Ditto.
22603 * config/nds32/pipelines.md: Ditto.
22605 2018-04-04 Richard Biener <rguenther@suse.de>
22607 PR tree-optimization/85168
22608 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
22609 propagating abnormals.
22611 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22613 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
22615 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22616 Kito Cheng <kito.cheng@gmail.com>
22618 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
22619 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
22620 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
22621 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
22622 * config/nds32/nds32.md (sibcall_internal): New.
22623 (sibcall_register): Remove.
22624 (sibcall_immediate): Remove.
22625 (sibcall_value_internal): New.
22626 (sibcall_value_register): Remove.
22627 (sibcall_value_immediate): Remove.
22628 * config/nds32/predicates.md (nds32_general_register_operand): New.
22629 (nds32_call_address_operand): New.
22631 2018-04-03 Jakub Jelinek <jakub@redhat.com>
22633 PR rtl-optimization/85167
22634 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
22635 bb_defs if *split_p, instead preinitialize it to NULL.
22637 PR tree-optimization/85156
22638 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
22639 evaluating the argument multiple times.
22641 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
22643 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
22645 (_mm_cvtpd_ps): Likewise.
22646 (_mm_cvttpd_epi32): Likewise.
22647 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
22648 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
22649 vector, pixel, and bool following altivec.h include.
22651 2018-04-03 Martin Sebor <msebor@redhat.com>
22653 * doc/extend.texi (Common Function Attributes): Clarify.
22654 (const attribute): Likewise.
22655 (pure attribute): Likewise.
22657 2018-04-03 Jakub Jelinek <jakub@redhat.com>
22660 * config/i386/i386.c (ix86_expand_vector_set): Use
22661 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
22663 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
22665 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
22666 instructions when changing rounding bits to preserve precision bits
22667 in the x87 control word.
22669 2018-04-03 Martin Liska <mliska@suse.cz>
22671 PR tree-optimization/82491
22672 * rtl.h (strip_offset_and_add): Replace += suboffset with
22673 poly_uint64 () + suboffset.
22675 2018-03-29 Martin Liska <mliska@suse.cz>
22676 Martin Jambor <mjambor@suse.cz>
22679 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
22680 param_type is not an integral or pointer type.
22682 2018-04-03 Richard Biener <rguenther@suse.de>
22684 * sese.h (recompute_all_dominators): Remove.
22686 2018-04-02 Martin Sebor <msebor@redhat.com>
22688 * doc/invoke.texi (-Wrestrict): Fix typos.
22690 2018-04-02 Jim Wilson <jimw@sifive.com>
22692 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
22693 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
22694 (<optab>di3, <optab>si3_extend): Likewise.
22695 (<optab>si3_mask, <optab>si3_mask_1): New.
22696 (<optab>di3_mask, <optab>di3_mask_1): New.
22697 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
22698 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
22699 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
22701 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
22703 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
22706 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
22708 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
22709 (nds32_canonicalize_comparison): New function.
22711 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22712 Kito Cheng <kito.cheng@gmail.com>
22713 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22715 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
22716 * config/nds32/constants.md (unspec_volatile_element): Add
22717 UNSPEC_VOLATILE_RELAX_GROUP.
22718 * config/nds32/nds32-relax-opt.c: New file.
22719 * config/nds32/nds32-predicates.c
22720 (nds32_symbol_load_store_p): New function.
22721 * config/nds32/nds32-protos.h
22722 (nds32_symbol_load_store_p): Declare function.
22723 (make_pass_nds32_relax_opt): Declare new rtl pass function.
22724 * config/nds32/nds32.c
22725 (nds32_register_pass): New function to register pass.
22726 (nds32_register_passes): New function to register passes.
22727 * config/nds32/nds32.md (relax_group): New pattern.
22728 * config/nds32/nds32.opt (mrelax-hint): New option.
22729 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
22731 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
22733 * config/nds32/t-nds32: Modify files dependency.
22735 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22737 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
22738 (PROFILE_HOOK): Define its implementation.
22740 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22742 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
22743 type and 32-bit size.
22745 2018-04-01 Jakub Jelinek <jakub@redhat.com>
22747 PR middle-end/85090
22748 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
22749 (V_128_256): New mode iterator.
22750 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
22751 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
22752 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
22754 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
22755 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
22757 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22760 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
22761 NaN inputs correctly.
22763 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
22766 * config/rs6000/vsx.md (??r): New mode attribute.
22767 (*vsx_mov<mode>_64bit): Use it.
22768 (*vsx_mov<mode>_32bit): Likewise.
22770 2018-03-30 Martin Sebor <msebor@redhat.com>
22772 PR tree-optimization/84818
22773 * builtins.c (check_access): Use warning_n.
22775 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22778 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
22780 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
22783 2018-03-30 Julia Koval <julia.koval@intel.com>
22786 * x86-tune.def (movx, partial_reg_dependency): Enable for
22789 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
22791 PR inline-asm/84985
22792 * lra-constraints.c (process_alt_operands): Move setting
22793 this_alternative_matches below.
22795 2018-03-29 Martin Liska <mliska@suse.cz>
22798 * doc/invoke.texi: Document how LTO works with debug info.
22799 Describe auto-load support of binutils. Mention 'x86-64'
22800 as valid option value of -march option.
22802 2018-03-29 Jakub Jelinek <jakub@redhat.com>
22804 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
22807 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
22808 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
22809 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
22812 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
22815 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
22816 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
22817 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
22818 for RS6000_BTM_POWERPC64.
22819 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
22820 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
22821 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
22824 (DIVDEU): Likewise.
22826 2018-03-28 Carl Love <cel@us.ibm.com>
22829 2017-09-27 Carl Love <cel@us.ibm.com>
22831 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
22832 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
22833 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
22836 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22838 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
22839 instead of __vector bool.
22840 (_mm_max_pu8): Likewise.
22841 (_mm_min_pi16): Likewise.
22843 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
22846 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
22847 (DIVWEUO): Likewise.
22848 (DIVDEO): Likewise.
22849 (DIVDEUO): Likewise.
22850 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
22851 DIVWEUO and DIVDEUO.
22852 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
22853 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
22854 (div_extend): Likewise.
22855 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
22857 (__builtin_divweuo): Likewise.
22858 (__builtin_divdeo): Likewise.
22859 (__builtin_divdeuo): Likewise.
22861 2018-03-28 Jakub Jelinek <jakub@redhat.com>
22864 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
22865 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
22867 PR tree-optimization/82004
22868 * gimple-match-head.c (optimize_pow_to_exp): New function.
22869 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
22870 Don't fold to exp if optimize_pow_to_exp is false.
22872 2018-03-28 Martin Liska <mliska@suse.cz>
22875 * calls.c (initialize_argument_information): Fix trailing space.
22876 * common.opt: Fix typo and provide better explanation for
22877 -fsanitize-coverage option.
22878 * config/i386/i386.opt: Fix typo.
22880 2018-03-28 Jakub Jelinek <jakub@redhat.com>
22881 Martin Liska <mliska@suse.cz>
22884 * gimplify.c (asan_poison_variable): Don't do the check for
22885 gimplify_omp_ctxp here.
22886 (gimplify_decl_expr): Do it here.
22887 (gimplify_target_expr): Likewise.
22889 2018-03-28 Martin Liska <mliska@suse.cz>
22892 * config/i386/i386.c (ix86_function_arg_advance): Do not call
22893 chkp_type_bounds_count if MPX is not enabled.
22895 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
22897 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
22899 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22902 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
22903 function to create the function decl for complex long double
22904 multiply and divide for -mabi=ieeelongdouble.
22905 (init_float128_ieee): Call it.
22907 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
22910 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
22911 -fcf-protection=branch -mibt.
22912 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
22914 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22917 * config/arm/arm.c (arm_valid_symbolic_address): Handle
22918 arm_word_relocations.
22920 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
22923 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
22924 extern array declarations.
22926 2018-03-27 Richard Biener <rguenther@suse.de>
22928 PR middle-end/84067
22929 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
22930 explicit single_use checks.
22932 2018-03-27 Richard Biener <rguenther@suse.de>
22934 PR tree-optimization/85082
22935 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
22938 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22940 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
22941 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
22942 Turn on fasynchronous-unwind-tables and funwind-tables.
22944 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
22947 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
22948 (*bmi_blsr_<mode>_ccz): Ditto.
22950 2018-03-26 Tom de Vries <tom@codesourcery.com>
22952 PR tree-optimization/85063
22953 * omp-general.c (offloading_function_p): New function. Factor out
22955 * omp-offload.c (pass_omp_target_link::gate): ... here.
22956 * omp-general.h (offloading_function_p): Declare.
22957 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
22958 with attribute omp declare target for offloading functions.
22960 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
22962 PR tree-optimization/84005
22963 * tree-data-ref.h (get_base_for_alignment): Declare.
22964 * tree-data-ref.c (get_base_for_alignment_1): New function.
22965 (get_base_for_alignment): Likewise.
22966 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22967 get_base_for_alignment to find a suitable base object, instead
22968 of always using drb->base_address.
22970 2018-03-23 Jakub Jelinek <jakub@redhat.com>
22972 PR inline-asm/85022
22973 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
22974 known size by default.
22976 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
22978 PR inline-asm/85030
22979 * lra-constraints.c (process_alt_operands): Don't match BLKmode
22980 and non BLKmode operands.
22982 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22985 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
22986 Clean up attributes.
22988 2018-03-23 Richard Biener <rguenther@suse.de>
22991 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
22992 we are going to emit early debug for LTO.
22994 2018-03-23 Jakub Jelinek <jakub@redhat.com>
22996 PR inline-asm/85034
22997 * function.c (match_asm_constraints_1): Don't optimize if input
22998 doesn't satisfy general_operand predicate for output's mode.
23000 PR inline-asm/85022
23001 * alias.c (write_dependence_p): Don't require for x_canonicalized
23002 non-VOIDmode if x has VOIDmode.
23005 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
23006 just don't try to optimize it rather than assert it never happens.
23008 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
23010 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
23011 macro expansions for definition of ST_INTERNAL_<mode> and
23012 LD_INTERNAL_<mode> builtins.
23013 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
23015 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
23017 (altivec_expand_st_builtin): Likewise.
23018 (altivec_expand_builtin): Remove calls to deleted functions.
23019 (rs6000_address_for_altivec): Delete this function.
23020 * config/rs6000/vector.md: Remove expands for
23021 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
23023 2018-03-22 Sudakshina Das <sudi.das@arm.com>
23026 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
23027 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
23028 re-computing once computed.
23029 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
23030 (arm_init_machine_status): Initialize
23031 machine->static_chain_stack_bytes.
23033 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
23036 * doc/extend.texi: Add four new prototypes for vec_ld.
23037 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
23038 definitions for more logical presentation.
23039 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
23040 entries for V1TI variants of __builtin_altivec_ld builtin.
23041 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
23042 handling of V1TI variant of LVX icode pattern.
23043 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
23044 (rs6000_gimple_fold_builtin): Likewise.
23045 (altivec_init_builtins): Add code to define
23046 __builtin_altivec_lvx_v1ti function.
23048 2018-03-22 Jakub Jelinek <jakub@redhat.com>
23050 PR inline-asm/84941
23051 * function.c (match_asm_constraints_1): Don't do the optimization
23052 if input isn't a REG, SUBREG, MEM or constant.
23054 2018-03-22 Tom de Vries <tom@codesourcery.com>
23056 PR tree-optimization/84956
23057 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
23058 bb_has_abnormal_pred.
23060 2018-03-22 Jakub Jelinek <jakub@redhat.com>
23063 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
23064 DECL_INITIAL (decl) to decl at the end.
23065 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
23066 adjust the comment.
23068 2018-03-21 Joseph Myers <joseph@codesourcery.com>
23070 * doc/extend.texi (__builtin_tgmath): Document when complex
23071 integer types are treated as _Complex _Float64.
23073 2018-03-21 Tom de Vries <tom@codesourcery.com>
23075 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
23077 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23079 PR tree-optimization/84960
23080 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
23081 if it is ENTRY block, move them into single succ of ENTRY in that case.
23083 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
23085 PR tree-optimization/84811
23086 * poly-int.h (poly_span_traits): Remove the T3 parameter and
23087 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
23088 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
23089 (known_subrange_p): Update accordingly. Cast each value involved
23090 in the size comparison, rather than casting the result of the
23093 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23095 PR tree-optimization/84982
23096 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
23097 by flipping the least significant bit rather than all bits from
23098 bitpos to bitpos + bitsize - 1.
23100 2018-03-21 Nathan Sidwell <nathan@acm.org>
23102 * doc/extend.texi (Deprecated Features): Remove mention of
23103 long-deleted deprecations.
23105 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23108 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
23109 * configure: Regenerate.
23111 2018-03-21 Tom de Vries <tom@codesourcery.com>
23113 PR tree-optimization/83126
23114 * tree-parloops.c (num_phis): New function.
23115 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
23117 2018-03-21 Nathan Sidwell <nathan@acm.org>
23119 * doc/extend.texi (Deprecated Features): Update deprecated flags,
23120 mention anon-struct/union members and trailing attributes.
23122 2018-03-21 Bin Cheng <bin.cheng@arm.com>
23124 PR tree-optimization/84969
23125 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
23126 builtin memset partitions if they set different rhs values.
23128 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23130 PR rtl-optimization/84989
23131 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
23132 VEC_DUPLICATE with scalar result mode.
23134 2018-03-21 Martin Liska <mliska@suse.cz>
23137 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
23138 not intended return statement.
23140 2018-03-21 Martin Liska <mliska@suse.cz>
23143 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
23144 (chkp_find_bound_slots_1): Limit number of iterations.
23146 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
23149 * Minor grammar fixes for x86 options.
23151 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23154 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
23155 holding REG_CFA_RESTORE notes, instead turn them into a USE.
23157 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
23160 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
23161 (altivec_lvx_<mode>_1op): Likewise.
23162 (altivec_stvx_<mode>_2op): Likewise.
23163 (altivec_stvx_<mode>_1op): Likewise.
23164 (altivec_lvx_<VM2:mode>): New define_expand.
23165 (altivec_stvx_<VM2:mode>): Likewise.
23166 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
23167 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
23168 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
23169 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
23170 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
23171 (rs6000_gen_lvx): Likewise.
23172 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
23173 (altivec_expand_stv_builtin): Likewise.
23174 (altivec_expand_builtin): Likewise.
23175 * config/rs6000/vector.md: Likewise.
23177 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23180 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
23183 2018-03-20 Richard Biener <rguenther@suse.de>
23186 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
23187 sign-conversions as zero, fall back to standard scalar_stmt
23190 2018-03-20 Martin Liska <mliska@suse.cz>
23193 * predict.c (rebuild_frequencies): Handle case when we have
23194 PROFILE_ABSENT, but flag_guess_branch_prob is false.
23196 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23199 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
23200 flag_section_anchors.
23201 * varasm.c (use_blocks_for_decl_p): Remove hack for
23202 dw2_force_const_mem.
23205 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
23207 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
23208 be created, use lowpart_subreg of operands[0] rather than operands[0]
23210 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
23211 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
23212 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
23213 and n constraint instead of aarch64_shift_imm_di and Usd.
23214 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
23215 (*aarch64_<optab>_reg_minus<mode>3): ... this.
23217 2018-03-20 Sudakshina Das <sudi.das@arm.com>
23220 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
23221 to favor GPR over NEON registers.
23222 (<shift>di3_neon): Likewise.
23224 2018-03-20 Tom de Vries <tom@codesourcery.com>
23227 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
23228 (nvptx_process_pars): Emit bar.sync asap and alap.
23230 2018-03-20 Tom de Vries <tom@codesourcery.com>
23233 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
23234 seen_label if seen_label is already set.
23236 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23239 * config/i386/i386.c (fold_builtin_cpu): For features above 31
23240 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
23241 Use 1U instead of 1. Formatting fixes.
23244 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
23245 instead of TREE_TYPE (s1) for the return value.
23247 2018-03-19 Jakub Jelinek <jakub@redhat.com>
23249 PR tree-optimization/84946
23250 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
23251 bitsize + bitsize in poly_uint64 rather than poly_int64.
23254 * dwarf2asm.c: Include fold-const.c.
23255 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
23256 of decl rather than decl itself.
23258 PR rtl-optimization/84643
23259 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
23261 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
23264 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
23265 calling assemble_variable.
23267 2018-03-19 Sudakshina Das <sudi.das@arm.com>
23270 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
23271 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
23273 2018-03-19 Jim Wilson <jimw@sifive.com>
23276 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
23277 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
23278 (riscv_first_stack_step): Likewise.
23279 (riscv_option_override): Use STACK_BOUNDARY instead of
23280 MIN_STACK_BOUNDARY.
23281 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
23282 MIN_STACK_BOUNDARY.
23283 (BIGGEST_ALIGNMENT): Set to 128.
23284 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
23285 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
23288 2018-03-19 Richard Biener <rguenther@suse.de>
23290 PR tree-optimization/84933
23291 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
23292 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
23294 2018-03-19 Richard Biener <rguenther@suse.de>
23296 PR tree-optimization/84859
23297 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
23298 (cond_if_else_store_replacement): Perform sinking operation on
23299 single-store BBs regardless of MAX_STORES_TO_SINK setting.
23300 Generalize what a BB with a single eligible store is.
23302 2018-03-19 Richard Biener <rguenther@suse.de>
23304 PR tree-optimization/84929
23305 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
23306 chrec_is_positive against non-chrec arg.
23308 2018-03-19 Tamar Christina <tamar.christina@arm.com>
23311 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
23313 2018-03-18 Martin Liska <mliska@suse.cz>
23315 PR rtl-optimization/84635
23316 * regrename.c (build_def_use): Use matches_mode only when
23319 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
23321 PR tree-optimization/84913
23322 * tree-vect-loop.c (vectorizable_reduction): Don't try to
23323 vectorize chains of COND_EXPRs.
23325 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23327 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
23329 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23331 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
23333 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23335 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
23337 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23338 Kito Cheng <kito.cheng@gmail.com>
23340 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
23341 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
23342 (nds32_adjust_reg_alloc_order): New function.
23343 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
23345 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
23347 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
23348 nds32_print_operand, nds32_print_operand_address): Use
23349 HOST_WIDE_INT_PRINT_DEC instead.
23351 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23353 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
23355 2018-03-17 Jakub Jelinek <jakub@redhat.com>
23358 * config/i386/i386.c (initial_ix86_tune_features,
23359 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
23360 unsigned long long.
23361 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
23362 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
23363 rather than 1u << ix86_tune. Formatting fix.
23364 (ix86_option_override_internal): Change ix86_arch_mask from
23365 unsigned int to unsigned HOST_WIDE_INT, initialize to
23366 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
23367 (ix86_function_specific_restore): Likewise.
23369 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23372 * postreload.c (reload_combine_recognize_pattern): Perform
23373 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
23374 truncate_int_for_mode the result for the destination's mode.
23377 * hsa-gen.c (mem_type_for_type): Fix comment typo.
23378 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
23380 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23383 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
23386 * lra-assigns.c (lra_split_hard_reg_for): Don't use
23387 regno_allocno_class_array and sorted_pseudos.
23388 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
23389 insns where regno is used.
23391 2018-03-16 Martin Liska <mliska@suse.cz>
23394 * multiple_target.c (create_dispatcher_calls): Redirect
23395 reference in the symbol table.
23397 2018-03-16 Martin Liska <mliska@suse.cz>
23400 * multiple_target.c (create_dispatcher_calls): Redirect also
23403 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23407 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
23408 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
23409 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
23411 2018-03-16 Julia Koval <julia.koval@intel.com>
23413 * doc/invoke.texi (Skylake Server): Add CLWB.
23414 Cannonlake): Remove CLWB.
23416 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23418 PR tree-optimization/84841
23419 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
23421 (FLOAT_ONE_CONST_TYPE): Define.
23422 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
23423 (sort_by_operand_rank): Put entries with higher constant_type last
23424 rather than first to match comments.
23426 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
23428 * config/nios2/nios2.md (movsi_internal): Fix thinko in
23431 2018-03-15 Jakub Jelinek <jakub@redhat.com>
23434 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
23435 check and use address of target always.
23437 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
23440 * config/i386/i386.c (indirect_thunk_needed): Update comments.
23441 (indirect_thunk_bnd_needed): Likewise.
23442 (indirect_thunks_used): Likewise.
23443 (indirect_thunks_bnd_used): Likewise.
23444 (indirect_return_needed): New.
23445 (indirect_return_bnd_needed): Likewise.
23446 (output_indirect_thunk_function): Add a bool argument for
23448 (output_indirect_thunk_function): Don't generate alias for
23449 function return thunk.
23450 (ix86_code_end): Call output_indirect_thunk_function to generate
23451 function return thunks.
23452 (ix86_output_function_return): Set indirect_return_bnd_needed
23453 and indirect_return_needed instead of indirect_thunk_bnd_needed
23454 and indirect_thunk_needed.
23456 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
23458 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
23459 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
23460 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
23462 2018-03-15 David Malcolm <dmalcolm@redhat.com>
23463 Paul Hua <paul.hua.gm@gmail.com>
23466 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
23468 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
23470 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
23471 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
23472 resp. SFmode cases.
23474 2018-03-15 Tamar Christina <tamar.christina@arm.com>
23477 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
23478 instead of GET_MODE_SIZE when comparing Units.
23480 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
23483 * varasm.c (hash_section): Return an unchangeble hash value
23484 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
23485 Return !aarch64_can_use_per_function_literal_pools_p ().
23487 2018-03-15 Jakub Jelinek <jakub@redhat.com>
23490 * optabs.c (emit_conditional_move): Pass address of cmode's copy
23491 rather than address of cmode as last argument to prepare_cmp_insn.
23493 2018-03-15 Julia Koval <julia.koval@intel.com>
23495 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
23496 F_AVX512VNNI, F_AVX512BITALG): New.
23498 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
23501 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
23502 insn for floating-point loads and stores.
23504 2018-03-14 Carl Love <cel@us.ibm.com>
23506 * config/rs6000/rs6000-c.c: Add macro definitions for
23507 ALTIVEC_BUILTIN_VEC_PERMXOR.
23508 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
23509 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
23510 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
23511 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
23513 * config/doc/extend.texi: Add prototypes for vec_permxor.
23515 2018-03-14 David Malcolm <dmalcolm@redhat.com>
23518 * diagnostic-show-locus.c (class layout_point): Convert m_line
23519 from int to linenum_type.
23520 (line_span::comparator): Use linenum "compare" function when
23521 comparing line numbers.
23522 (test_line_span): New function.
23523 (layout_range::contains_point): Convert param "row" from int to
23525 (layout_range::intersects_line_p): Likewise.
23526 (layout::will_show_line_p): Likewise.
23527 (layout::print_source_line): Likewise.
23528 (layout::should_print_annotation_line_p): Likewise.
23529 (layout::print_annotation_line): Likewise.
23530 (layout::print_leading_fixits): Likewise.
23531 (layout::annotation_line_showed_range_p): Likewise.
23532 (struct line_corrections): Likewise for field m_row.
23533 (line_corrections::line_corrections): Likewise for param "row".
23534 (layout::print_trailing_fixits): Likewise.
23535 (layout::get_state_at_point): Likewise.
23536 (layout::get_x_bound_for_row): Likewise.
23537 (layout::print_line): Likewise.
23538 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
23539 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
23540 * input.c (selftest::test_linenum_comparisons): New function.
23541 (selftest::input_c_tests): Call it.
23542 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
23543 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
23544 * selftest.h (ASSERT_GT): New macro.
23545 (ASSERT_GT_AT): New macro.
23546 (ASSERT_LT): New macro.
23547 (ASSERT_LT_AT): New macro.
23549 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
23551 PR rtl-optimization/84780
23552 * combine.c (distribute_links): Don't make a link based on pc_rtx.
23554 2018-03-14 Martin Liska <mliska@suse.cz>
23556 * tree.c (record_node_allocation_statistics): Use
23557 get_stats_node_kind.
23558 (get_stats_node_kind): New function extracted from
23559 record_node_allocation_statistics.
23560 (free_node): Use get_stats_node_kind.
23562 2018-03-14 Richard Biener <rguenther@suse.de>
23564 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
23565 that the value-set of ANTIC_IN doesn't grow.
23568 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
23570 (BB_VISITED_WITH_VISITED_SUCCS): New define.
23571 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
23573 2018-03-14 Julia Koval <julia.koval@intel.com>
23575 * config.gcc (icelake-client, icelake-server): New.
23577 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
23578 (initial_ix86_arch_features): Ditto.
23579 (PTA_SKYLAKE): Add SGX.
23580 (PTA_ICELAKE): Remove.
23581 (PTA_ICELAKE_CLIENT): New.
23582 (PTA_ICELAKE_SERVER): New.
23583 (ix86_option_override_internal): Split up icelake on icelake client and
23585 (get_builtin_code_for_version): Ditto.
23586 (fold_builtin_cpu): Ditto.
23587 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
23588 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
23589 * config/i386/i386.h (processor_type): Ditto.
23590 * doc/invoke.texi: Ditto.
23592 2018-03-14 Jakub Jelinek <jakub@redhat.com>
23595 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
23596 INTEGER_CST offset, add it together with bitpos / 8 and
23597 sign extend based on POINTER_SIZE.
23601 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23604 * config/i386/constraints.md (Yc): New register constraint.
23605 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23606 Use Yc constraint for alternative 2 of operand 0. Remove
23607 preferred_for_speed attribute.
23609 2018-03-14 Richard Biener <rguenther@suse.de>
23611 PR tree-optimization/84830
23612 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
23613 with the old one to avoid oscillations.
23615 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
23618 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
23620 (assign_by_spills): Return a flag of reload assignment failure.
23621 Do not process the reload assignment failures. Do not spill other
23622 reload pseudos if they has the same reg class. Update n if
23624 (lra_assign): Add a return arg. Set up from the result of
23625 assign_by_spills call.
23626 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23627 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23628 usage_insns if it is not NULL.
23629 (spill_hard_reg_in_range): New function.
23630 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23631 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23632 function prototypes.
23633 (lra_assign): Change prototype.
23634 * lra.c (lra): Add code to deal with fails by splitting hard reg
23637 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
23639 * config/riscv/riscv.opt (mrelax): New option.
23640 * config/riscv/riscv.c (riscv_file_start): Emit ".option
23641 "norelax" when riscv_mrelax is disabled.
23642 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
23644 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23647 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
23648 reassociation for int modes.
23650 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23652 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
23653 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
23655 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
23656 * config/aarch64/aarch64-sve.md
23657 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
23658 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
23659 (*extend<mode><Vwide>2): Rename to...
23660 (aarch64_sve_extend<mode><Vwide>2): ...this.
23661 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
23662 renaming the old pattern to...
23663 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
23665 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
23666 define_expand, renaming the old pattern to...
23667 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
23668 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
23669 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
23670 account when deciding which SVE instruction the optab should use.
23671 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
23673 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23675 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
23676 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
23677 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
23678 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
23679 (tlsdesc_small_<mode>): Turn a define_expand and use
23680 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
23681 (tlsdesc_small_advsimd_<mode>): ...this.
23682 (tlsdesc_small_sve_<mode>): New pattern.
23684 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23686 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
23687 (UNSPEC_UMUL_HIGHPART): New constants.
23688 (MUL_HIGHPART): New int iteraor.
23689 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
23690 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
23692 (*<su>mul<mode>3_highpart): New define_insn.
23694 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23697 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23700 2018-03-13 Martin Liska <mliska@suse.cz>
23703 * (sem_item_optimizer::sem_item_optimizer): Initialize new
23705 (sem_item_optimizer::~sem_item_optimizer): Release it.
23706 (sem_item_optimizer::merge_classes): Register variable aliases.
23707 (sem_item_optimizer::fixup_pt_set): New function.
23708 (sem_item_optimizer::fixup_points_to_sets): Likewise.
23709 * ipa-icf.h: Declare new variables and functions.
23711 2018-03-13 Jakub Jelinek <jakub@redhat.com>
23713 PR middle-end/84834
23714 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
23715 integer_pow2p@2 and test integer_pow2p in condition.
23716 (A < 0 ? C : 0): Similarly for @1.
23718 PR middle-end/84831
23719 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
23720 characters starting at p contain '\0' character, don't look beyond
23724 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
23725 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
23728 * reg-stack.c (change_stack): Change update_end var from int to
23729 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
23730 also call set_block_for_insn on the newly added insns and rescan.
23733 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
23734 on the last operand.
23737 * tree.c (stabilize_reference_1): Return save_expr (e) for
23738 STATEMENT_LIST even if it doesn't have side-effects.
23740 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
23742 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
23744 2018-03-12 Renlin Li <renlin.li@arm.com>
23746 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
23747 aarch64_output_scalar_simd_mov_immediate.
23749 2018-03-12 Martin Sebor <msebor@redhat.com>
23751 PR tree-optimization/83456
23752 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
23753 for perfectly overlapping calls to memcpy.
23754 (gimple_fold_builtin_memory_chk): Same.
23755 (gimple_fold_builtin_strcpy): Handle no-warning.
23756 (gimple_fold_builtin_stxcpy_chk): Same.
23757 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
23759 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
23761 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
23762 parameter. Use it for SFmode.
23763 (rs6000_function_arg_advance_1): Adjust.
23764 (rs6000_function_arg): Adjust.
23765 (rs6000_gimplify_va_arg): Pass false for that new parameter.
23767 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
23769 PR rtl-optimization/84169
23770 PR rtl-optimization/84780
23771 * combine.c (can_combine_p): Check for a 2-insn combination whether
23772 the destination register is used between the two insns, too.
23774 2018-03-12 Richard Biener <rguenther@suse.de>
23776 PR tree-optimization/84803
23777 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
23778 for refs DR analysis didn't process.
23780 2018-03-12 Richard Biener <rguenther@suse.de>
23782 PR tree-optimization/84777
23783 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
23784 force-vectorize loops ignore whether we are optimizing for size.
23786 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
23788 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
23789 (TARGET_MD_ASM_ADJUST): Define.
23791 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
23792 Kito Cheng <kito.cheng@gmail.com>
23793 Chung-Ju Wu <jasonwucj@gmail.com>
23795 * config/nds32/nds32.c (nds32_compute_stack_frame,
23796 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
23797 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
23798 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
23799 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
23800 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
23801 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
23802 * config/nds32/nds32.md (prologue, epilogue): Use macro
23803 NDS32_V3PUSH_AVAILABLE_P to do checking.
23805 2018-03-11 Jakub Jelinek <jakub@redhat.com>
23808 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
23809 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
23810 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
23811 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
23812 addition of most attributes on !orig_type_die or the attribute not
23813 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
23815 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23816 Chung-Ju Wu <jasonwucj@gmail.com>
23818 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
23819 __NDS32_VH__ macro.
23820 * config/nds32/nds32.opt (mvh): New option.
23822 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23823 Chung-Ju Wu <jasonwucj@gmail.com>
23825 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
23827 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
23828 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
23831 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23832 Chung-Ju Wu <jasonwucj@gmail.com>
23834 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
23836 * config/nds32/nds32-multiple.md (strlensi): New pattern.
23837 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
23839 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
23840 Kito Cheng <kito.cheng@gmail.com>
23841 Chung-Ju Wu <jasonwucj@gmail.com>
23843 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
23844 UNSPEC_FFMISM and UNSPEC_FLMISM.
23845 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
23846 for ffb, ffmism and flmism.
23847 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
23848 (unspec_ffmism): Ditto.
23849 (unspec_flmism): Ditto.
23850 (nds32_expand_builtin_impl): Check if string extension is available.
23851 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
23852 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
23854 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
23857 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
23860 * lra-assigns.c (assign_by_spills): Return a flag of reload
23861 assignment failure. Do not process the reload assignment
23862 failures. Do not spill other reload pseudos if they has the same
23864 (lra_assign): Add a return arg. Set up from the result of
23865 assign_by_spills call.
23866 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23867 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23868 usage_insns if it is not NULL.
23869 (spill_hard_reg_in_range): New function.
23870 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23871 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23872 function prototypes.
23873 (lra_assign): Change prototype.
23874 * lra.c (lra): Add code to deal with fails by splitting hard reg
23877 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
23880 * config/i386/i386.opt: Replace Enforcment with Enforcement.
23882 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
23885 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
23886 (dw_val_node): Add val_symbolic_view.
23887 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
23888 (symview_upper_bound): New.
23889 (new_line_info_table): Initialize symviews_since_reset.
23890 (dwarf2out_source_line): Count symviews_since_reset and set
23891 symview_upper_bound.
23892 (dw_val_equal_p): Handle symview.
23893 (add_AT_symview): New.
23894 (print_dw_val): Handle symview.
23895 (attr_checksum, attr_checksum_ordered): Likewise.
23896 (same_dw_val_p, size_of_die): Likewise.
23897 (value_format, output_die): Likewise.
23898 (add_high_low_attributes): Use add_AT_symview for entry_view.
23899 (dwarf2out_finish): Reset symview_upper_bound, clear
23902 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
23905 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
23906 Add strict argument and use it.
23907 (rs6000_split_multireg_move): Update for new strict argument.
23908 (mem_operand_gpr): Disallow all non-offsettable addresses.
23909 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
23911 2018-03-09 Jakub Jelinek <jakub@redhat.com>
23914 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
23915 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
23916 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
23919 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
23920 decl, use remap_type if we want to use the type.
23922 2018-03-09 Martin Sebor <msebor@redhat.com>
23924 PR tree-optimization/84526
23925 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23927 (builtin_access::generic_overlap): Be prepared to handle non-array
23930 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
23932 PR rtl-optimization/84682
23933 * lra-constraints.c (process_address_1): Check is_address flag
23934 for address constraints.
23935 (process_alt_operands): Likewise.
23936 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
23937 preprocess_constraints.
23938 * recog.h (preprocess_constraints): Add oploc parameter.
23940 * recog.c (preprocess_constraints): Test address_operand for
23941 CT_ADDRESS constraints.
23943 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
23946 * lra-assigns.c (assign_by_spills): Return a flag of reload
23947 assignment failure. Do not process the reload assignment
23948 failures. Do not spill other reload pseudos if they has the same
23950 (lra_assign): Add a return arg. Set up from the result of
23951 assign_by_spills call.
23952 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23953 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23954 usage_insns if it is not NULL.
23955 (spill_hard_reg_in_range): New function.
23956 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23957 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23958 function prototypes.
23959 (lra_assign): Change prototype.
23960 * lra.c (lra): Add code to deal with fails by splitting hard reg
23963 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23966 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
23967 Accept complain bool parameter. Only emit errors if it is true.
23968 (arm_parse_cpu_option_name): Likewise.
23969 (arm_target_thumb_only): Adjust callers of the above.
23970 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
23971 prototype to take a default true bool parameter.
23972 (arm_parse_arch_option_name): Likewise.
23974 2018-03-09 David Malcolm <dmalcolm@redhat.com>
23975 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
23979 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
23980 * configure: Regenerate.
23981 * configure.ac ("linker --version-script option"): New.
23982 ("linker soname option"): New.
23984 2018-03-09 Richard Biener <rguenther@suse.de>
23986 PR tree-optimization/84775
23987 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
23988 immediate uses of predicate stmts and mark them modified.
23991 PR tree-optimization/84178
23992 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
23994 (version_loop_for_if_conversion): Delay update_ssa call.
23995 (tree_if_conversion): Delay update_ssa until after predicate
23998 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
24001 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
24002 when the function accesses prior frames.
24004 2018-03-08 Jakub Jelinek <jakub@redhat.com>
24007 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
24008 gen_llsym, otherwise call maybe_gen_llsym.
24010 PR inline-asm/84742
24011 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
24012 has ',' character inside of it.
24014 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24017 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
24018 as clobbering CC_REGNUM.
24020 2018-03-08 Richard Biener <rguenther@suse.de>
24022 PR middle-end/84552
24023 * tree-scalar-evolution.c: Include tree-into-ssa.h.
24024 (follow_copies_to_constant): Do not follow SSA names registered
24027 2018-03-08 Richard Biener <rguenther@suse.de>
24029 PR tree-optimization/84178
24030 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
24032 (version_loop_for_if_conversion): Delay update_ssa call.
24033 (tree_if_conversion): Delay update_ssa until after predicate
24036 2018-03-08 David Malcolm <dmalcolm@redhat.com>
24038 PR tree-optimization/84178
24039 * tree-if-conv.c (release_bb_predicate): Remove the
24040 the assertion that the stmts have NULL use_ops.
24041 Discard the statements, asserting that they haven't
24042 yet been added to a BB.
24044 2018-03-08 Richard Biener <rguenther@suse.de>
24046 PR tree-optimization/84746
24047 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
24048 (phi_translate): Pass in destination ANTIC_OUT set.
24049 (phi_translate_1): Likewise. For a simplified result lookup
24050 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
24051 (phi_translate_set): Adjust.
24052 (do_pre_regular_insertion): Likewise.
24053 (do_pre_partial_partial_insertion): Likewise.
24055 2018-03-08 Martin Liska <mliska@suse.cz>
24057 PR gcov-profile/84735
24058 * doc/gcov.texi: Document usage of profile files.
24059 * gcov-io.h: Document changes in the format.
24061 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
24065 * dwarf2out.c (struct dw_line_info_table): Update comments for
24067 (FORCE_RESET_NEXT_VIEW): New.
24068 (FORCE_RESETTING_VIEW_P): New.
24069 (RESETTING_VIEW_P): Check for -1 too.
24070 (ZERO_VIEW_P): Likewise.
24071 (new_line_info_table): Force-reset next view.
24072 (dwarf2out_begin_function): Likewise.
24073 (dwarf2out_source_line): Simplify zero_view_p initialization.
24074 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
24075 view directly. Omit view when omitting .loc at line 0.
24077 2018-03-08 Jakub Jelinek <jakub@redhat.com>
24079 PR tree-optimization/84740
24080 * tree-switch-conversion.c (process_switch): Call build_constructors
24081 only if info.phi_count is non-zero.
24083 PR tree-optimization/84739
24084 * tree-tailcall.c (find_tail_calls): Check call arguments against
24085 DECL_ARGUMENTS (current_function_decl) rather than
24086 DECL_ARGUMENTS (func) when checking for tail recursion.
24088 2018-03-07 Jakub Jelinek <jakub@redhat.com>
24090 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
24091 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
24092 Volker Reichelt's entry and add entries for people that perform
24093 GCC fuzzy testing and report numerous bugs.
24095 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24098 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
24099 readonly data in sdata, if that is disabled.
24100 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
24101 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
24102 -mreadonly-in-sdata option.
24104 2018-03-07 Martin Sebor <msebor@redhat.com>
24106 PR tree-optimization/84468
24107 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
24108 basic block when looking for nul assignment.
24110 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
24113 * except.h (output_function_exception_table): Adjust prototype.
24114 * except.c (output_function_exception_table): Remove FNNAME parameter
24115 and add SECTION parameter. Ouput one part of the table at a time.
24116 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
24117 the first part of the exception table and emit unwind directives.
24118 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
24119 (i386_pe_seh_cold_init): Likewise.
24120 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
24121 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
24122 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
24123 (ix86_output_call_insn): Emit a nop in one more case for SEH.
24124 * config/i386/winnt.c: Include except.h.
24125 (struct seh_frame_state): Add reg_offset, after_prologue and
24126 in_cold_section fields.
24127 (i386_pe_seh_end_prologue): Set seh->after_prologue.
24128 (i386_pe_seh_cold_init): New function.
24129 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
24130 to seh->in_cold_section.
24131 (seh_emit_push): Record the offset of the push.
24132 (seh_emit_save): Record the offet of the save.
24133 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
24134 Test seh->after_prologue to disregard the epilogue.
24135 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
24136 (i386_pe_end_cold_function): New function.
24138 2018-03-07 Jakub Jelinek <jakub@redhat.com>
24141 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
24142 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
24145 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
24147 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
24148 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
24150 PR middle-end/84723
24151 * multiple_target.c: Include tree-inline.h and intl.h.
24152 (expand_target_clones): Diagnose and fail if node->definition and
24153 !tree_versionable_function_p (node->decl).
24155 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
24157 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
24159 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
24160 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
24161 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
24163 2018-03-06 Jakub Jelinek <jakub@redhat.com>
24166 * combine.c (try_combine): Use reg_or_subregno instead of handling
24167 just paradoxical SUBREGs and REGs.
24169 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
24171 * config/arc/arc.c (arc_finalize_pic): Remove function.
24172 (arc_must_save_register): We use single base PIC register, remove
24173 checks to save/restore the PIC register.
24174 (arc_expand_prologue): Likewise.
24175 * config/arc/arc-protos.h (arc_set_default_type_attributes):
24177 (arc_verify_short): Likewise.
24178 (arc_attr_type): Likewise.
24179 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
24180 (walk_stores): Likewise.
24181 (arc_address_cost): Make it static.
24182 (arc_verify_short): Likewise.
24183 (branch_dest): Likewise.
24184 (arc_attr_type): Likewise.
24185 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
24186 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
24187 (arc_final_prescan_insn): Remove inserting the nops due to
24188 hardware hazards. It is done in reorg step.
24189 (insn_length_variant_t): Remove.
24190 (insn_length_parameters_t): Likewise.
24191 (arc_insn_length_parameters): Likewise.
24192 (arc_get_insn_variants): Likewise.
24193 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
24195 2018-03-06 Jakub Jelinek <jakub@redhat.com>
24197 PR inline-asm/84683
24198 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
24201 PR tree-optimization/84687
24202 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
24204 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
24206 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24208 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
24209 Rename to ppc_speculation_barrier.
24210 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
24211 __builtin_ppc_speculation_barrier.
24213 2018-03-05 Jakub Jelinek <jakub@redhat.com>
24216 * combine.c (combine_simplify_rtx): Don't try to simplify if
24217 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
24220 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
24222 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
24223 to 32 bytes when compiling for POWER9.
24225 2018-03-05 Jakub Jelinek <jakub@redhat.com>
24228 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
24229 regparm >= 3 with no arg reg available also for calls with
24230 flag_force_indirect_call. Pass decl to ix86_function_regparm.
24233 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
24235 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
24237 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
24240 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
24242 2018-03-05 Richard Biener <rguenther@suse.de>
24244 PR tree-optimization/84486
24245 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
24246 When inserting a __builtin_assume_aligned call set the LHS
24247 SSA name alignment info accordingly.
24249 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
24251 PR tree-optimization/84114
24252 * config/aarch64/aarch64.c (aarch64_reassociation_width)
24253 Avoid reassociation of FLOAT_MODE addition.
24255 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
24257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
24258 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
24259 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
24260 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
24261 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
24262 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
24263 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
24265 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
24266 __builtin_ia32_wbinvd): New builtins.
24267 (SPECIAL_ARGS2): New.
24268 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
24269 (SPECIAL_ARGS2): New.
24270 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
24271 (ix86_valid_target_attribute_inner_p): Ditto.
24272 (ix86_init_mmx_sse_builtins): Add special_args2.
24273 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
24274 TARGET_WBNOINVD_P): New.
24275 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
24276 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
24277 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
24278 * config/i386/immintrin.h (_wbinvd): New intrinsic.
24279 * config/i386/pconfigintrin.h: New file.
24280 * config/i386/wbnoinvdintrin.h: Ditto.
24281 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
24283 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
24285 2018-03-05 Richard Biener <rguenther@suse.de>
24287 PR tree-optimization/84670
24288 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
24290 (BB_VISITED_WITH_VISITED_SUCCS): New define.
24291 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
24292 (compute_antic_aux): Only assert the number of values in ANTIC_IN
24293 doesn't grow if all successors (recursively) were visited at least
24296 2018-03-05 Richard Biener <rguenther@suse.de>
24298 PR tree-optimization/84650
24299 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
24300 if executed in the loop pipeline.
24302 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
24304 * doc/configfiles.texi (Configuration Files): Move info about
24305 conditionalizing $target-protos.h to...
24306 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
24307 differs from $target-protos.h.
24309 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
24310 Chung-Ju Wu <jasonwucj@gmail.com>
24312 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
24313 * config/nds32/nds32-multiple.md (setmemsi): Define.
24314 * config/nds32/nds32-memory-manipulation.c
24315 (nds32_gen_dup_4_byte_to_word_value): New.
24316 (emit_setmem_word_loop): New.
24317 (emit_setmem_byte_loop): New.
24318 (nds32_expand_setmem_loop): New.
24319 (nds32_expand_setmem_loop_v3m): New.
24320 (nds32_expand_setmem_unroll): New.
24321 (nds32_expand_setmem): New.
24323 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24324 Chung-Ju Wu <jasonwucj@gmail.com>
24326 * config/nds32/nds32-memory-manipulation.c
24327 (nds32_emit_load_store): New.
24328 (nds32_emit_post_inc_load_store): New.
24329 (nds32_emit_mem_move): New.
24330 (nds32_emit_mem_move_block): New.
24331 (nds32_expand_movmemsi_loop_unknown_size): New.
24332 (nds32_expand_movmemsi_loop_known_size): New.
24333 (nds32_expand_movmemsi_loop): New.
24334 (nds32_expand_movmemsi_unroll): New.
24335 (nds32_expand_movmemqi): Rename ...
24336 (nds32_expand_movmemsi): ... to this.
24337 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
24338 (movmemsi): ... to this.
24339 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
24340 (nds32_expand_movmemsi): ... to this.
24342 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24343 Monk Chiang <sh.chiang04@gmail.com>
24344 Chung-Ju Wu <jasonwucj@gmail.com>
24346 * config/nds32/nds32-protos.h
24347 (nds32_expand_load_multiple): New arguments.
24348 (nds32_expand_store_multiple): Ditto.
24349 (nds32_valid_multiple_load_store): Rename ...
24350 (nds32_valid_multiple_load_store_p): ... to this.
24351 * config/nds32/nds32-memory-manipulation.c
24352 (nds32_expand_load_multiple): Refine implementation.
24353 (nds32_expand_store_multiple): Ditto.
24354 * config/nds32/nds32-multiple.md
24355 (load_multiple): Update nds32_expand_load_multiple interface.
24356 (store_multiple): Update nds32_expand_store_multiple interface.
24357 * config/nds32/nds32-predicates.c
24358 (nds32_valid_multiple_load_store): Rename ...
24359 (nds32_valid_multiple_load_store_p): ... to this and refine
24361 * config/nds32/predicates.md
24362 (nds32_load_multiple_and_update_address_operation): New predicate.
24363 (nds32_store_multiple_and_update_address_operation): New predicate.
24365 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24366 Chung-Ju Wu <jasonwucj@gmail.com>
24368 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
24369 (combo): New attribute.
24370 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
24372 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
24374 * config/nds32/nds32.opt: Change -mcmodel= default value.
24376 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
24377 Monk Chiang <sh.chiang04@gmail.com>
24378 Chung-Ju Wu <jasonwucj@gmail.com>
24380 * config/nds32/constants.md (unspec_element): New enum.
24381 * config/nds32/constraints.md (Umw): New constraint.
24382 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
24383 * config/nds32/nds32-intrinsic.md: Likewise.
24384 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
24385 (nds32_valid_smw_lwm_base_p): New.
24386 (nds32_output_smw_single_word): New.
24387 (nds32_output_lmw_single_word): New.
24388 (nds32_expand_unaligned_load): New.
24389 (nds32_expand_unaligned_store): New.
24390 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
24391 (nds32_output_smw_single_word): Declare.
24392 (nds32_output_lmw_single_word): Declare.
24393 (nds32_expand_unaligned_load): Declare.
24394 (nds32_expand_unaligned_store): Declare.
24395 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
24396 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
24397 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
24398 NDS32_BUILTIN_UASTORE_DW.
24399 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
24402 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
24403 Kito Cheng <kito.cheng@gmail.com>
24404 Chung-Ju Wu <jasonwucj@gmail.com>
24406 * config/nds32/nds32-intrinsic.c
24407 (nds32_expand_builtin_null_ftype_reg): Delete.
24408 (nds32_expand_builtin_reg_ftype_imm): Ditto.
24409 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
24410 (nds32_read_argument): New.
24411 (nds32_legitimize_target): Ditto.
24412 (nds32_legitimize_argument): Ditto.
24413 (nds32_check_constant_argument): Ditto.
24414 (nds32_expand_unop_builtin): Ditto.
24415 (nds32_expand_unopimm_builtin): Ditto.
24416 (nds32_expand_binop_builtin): Ditto.
24417 (nds32_builtin_decl_impl): Ditto.
24418 (builtin_description): Ditto.
24419 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
24420 (nds32_init_builtins_impl): Ditto.
24421 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
24422 (nds32_builtin_decl): New.
24423 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
24424 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
24426 2018-03-02 Jeff Law <law@redhat.com>
24428 * reorg.c (stop_search_p): Handle DEBUG_INSN.
24429 (redundant_insn, fill_simple_delay_slots): Likewise.
24430 (fill_slots_from_thread): Likewise.
24431 * resource.c (mark_referenced_resources): Likewise.
24432 (mark_set_resources, find_dead_or_set_registers): Likewise.
24434 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24436 * substring-locations.h (format_warning_va): Formatting fix for
24437 ATTRIBUTE_GCC_DIAG.
24438 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
24440 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
24441 * substring-locations.c: Include intl.h.
24442 (format_warning_va): Turned into small wrapper around
24443 format_warning_n_va, renamed to ...
24444 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
24445 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
24447 (format_warning_at_substring_n): New function.
24448 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
24449 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
24450 format_warning_at_substring with just a shorter name instead of
24451 const function pointer.
24452 (fmtwarn_n): New function.
24453 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
24454 appropriate, get rid of all the fmtstr temporaries, move conditionals
24455 with G_() wrapped string literals directly into fmtwarn arguments,
24456 cast dir.len to (int), formatting fixes.
24458 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
24460 * doc/invoke.texi: Remove "Cilk Plus" references.
24462 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24463 Richard Biener <rguenther@suse.de>
24466 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
24467 for error or warning attributes if CALL_FROM_THUNK_P is set.
24470 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24473 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
24474 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
24477 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
24478 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
24480 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
24481 instead of -1U in last predictors element's probability member.
24483 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
24486 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
24487 arguments if they are comparable.
24489 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
24491 PR tree-optimization/84634
24492 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
24493 masks and masked_loop_p with a single loop_masks, making sure it's
24494 null for bb vectorization.
24496 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
24498 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
24499 (vect_analyze_data_ref_access): Use loop->safe_len rather than
24500 loop->force_vectorize to check whether there is no alias.
24502 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24505 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
24507 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
24509 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
24510 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
24511 instead of a loop around prev_real_insn.
24512 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
24515 PR inline-asm/84625
24516 * config/i386/i386.c (ix86_print_operand): Use conditional
24517 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
24520 2018-03-02 Richard Biener <rguenther@suse.de>
24522 PR tree-optimization/84427
24523 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
24524 (bitmap_set_subtract_values): Rewrite to handle multiple
24527 (prune_clobbered_mems): Likewise.
24528 (phi_translate): Take edge instead of pred/phiblock.
24529 (phi_translate_1): Likewise.
24530 (phi_translate_set): Likewise. Insert all translated
24531 exprs for a value into the set, keeping possibly multiple
24532 expressions per value.
24533 (compute_antic_aux): Adjust for phi_translate changes.
24534 When intersecting union the expressions and prune those
24535 not in the final value set, keeping possibly multiple
24536 expressions per value. Do not use value-insertion
24537 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
24538 all expressions. Add verification that the value-sets
24539 only shrink during iteration.
24540 (compute_partial_antic_aux): Adjust for the phi_translate changes.
24541 (do_pre_regular_insertion): Likewise.
24542 (do_pre_partial_partial_insertion): Likewise.
24544 2018-03-02 Richard Biener <rguenther@suse.de>
24547 * config/darwin.c (saved_debug_info_level): New static global.
24548 (darwin_asm_lto_start): Disable debug info generation for LTO out.
24549 (darwin_asm_lto_end): Restore debug info generation settings.
24551 2018-03-01 Martin Liska <mliska@suse.cz>
24554 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
24555 volatile arguments.
24557 2018-03-01 Richard Biener <rguenther@suse.de>
24560 * dwarf2out.c (gen_variable_die): Properly handle late VLA
24561 type annotation with LTO when debug was disabled at compile-time.
24563 2018-03-01 Matthew Fortune <mfortune@gmail.com>
24565 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
24567 (mips_final_postscan_insn): Likewise.
24569 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
24571 PR rtl-optimization/84528
24572 * alias.c (init_alias_target): Add commentary.
24573 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
24574 a unique base value if the frame pointer is not eliminated
24575 to the stack pointer.
24577 2018-03-01 Tom de Vries <tom@codesourcery.com>
24579 PR rtl-optimization/83327
24580 * lra-int.h (hard_regs_spilled_into): Declare.
24581 * lra.c (hard_regs_spilled_into): Define.
24582 (init_reg_info): Init hard_regs_spilled_into.
24583 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
24584 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
24585 (process_bb_lives): Handle hard_regs_spilled_into.
24586 (lra_create_live_ranges_1): Before doing liveness propagation, clear
24587 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
24589 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
24591 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
24592 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
24593 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
24594 * config/rs6000/aix72.h: New file.
24596 2018-02-28 Jakub Jelinek <jakub@redhat.com>
24598 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
24599 instead of warning_at with conditional singular and plural messages
24603 * stor-layout.c (update_alignment_for_field): For
24604 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
24605 && !DECL_PACKED (field), do the alignment update, just use
24606 only desired_align instead of MAX (type_align, desired_align)
24608 (place_field): Don't do known_align < desired_align handling
24609 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
24610 is non-NULL, instead do it after rli->prev_field handling and
24611 only if not within a bitfield word. For DECL_PACKED (field)
24612 use type_align of BITS_PER_UNIT.
24614 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
24616 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
24617 superfluous parentheses and trailing spaces.
24619 2018-02-28 Richard Biener <rguenther@suse.de>
24621 PR tree-optimization/84584
24622 * graphite-scop-detection.c (scop_detection::add_scop): Discard
24623 SCoPs with fake exit edge.
24625 2018-02-28 Martin Liska <mliska@suse.cz>
24628 * timevar.c (timer::print): Fix format to properly print 100%
24631 2018-02-28 Richard Biener <rguenther@suse.de>
24633 PR middle-end/84607
24634 * genmatch.c (capture_info::walk_match): Do not mark
24635 captured expressions without operands as expr_p given
24636 they act more like predicates and should be subject to
24637 "lost tail" side-effect preserving.
24639 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
24641 PR rtl-optimization/81611
24642 * auto-inc-dec.c (attempt_change): Move dead note from
24643 mem_insn if it's the next use of regno
24644 (find_address): Take address use of reg holding
24645 non-incremented value. Add parm to limit search to the named
24647 (merge_in_block): Attempt to use a mem insn that is the next
24648 use of the original regno.
24650 2018-02-27 Martin Sebor <msebor@redhat.com>
24653 * doc/invoke.texi (-Wmissing-attributes): New option.
24654 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
24656 2018-02-27 Martin Sebor <msebor@redhat.com>
24658 PR translation/84207
24659 * diagnostic-core.h (warning_n, error_n, inform_n): Change
24660 n argument to unsigned HOST_WIDE_INT.
24661 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
24662 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
24663 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
24664 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
24666 2018-02-27 Richard Biener <rguenther@suse.de>
24668 PR tree-optimization/84512
24669 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
24670 Do not use the estimate returned from record_stmt_cost for
24671 the scalar iteration cost but sum properly using add_stmt_cost.
24673 2018-02-27 Richard Biener <rguenther@suse.de>
24675 PR tree-optimization/84466
24676 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
24677 Adjust last change to less strictly validate use operands.
24679 2018-02-27 Martin Liska <mliska@suse.cz>
24681 PR gcov-profile/84548
24682 * gcov.c (process_file): Allow partial overlap and consider it
24683 also as group functions.
24684 (output_lines): Properly calculate range of lines for a group.
24686 2018-02-27 Martin Liska <mliska@suse.cz>
24688 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
24689 'ggc' suffixes. Change first column width.
24690 (timer::print): Fix formatting of the column.
24692 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
24694 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
24695 preserve inline entry blocks for the sake of debug inline
24696 entry point markers alone.
24697 (remove_unused_locals): Suggest in comments a better place to
24698 force the preservation of inline entry blocks that are
24699 otherwise unused, but do not preserve them.
24701 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24703 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
24705 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24708 * config/i386/constraints.md (Bs): Replace
24709 ix86_indirect_branch_register with
24710 TARGET_INDIRECT_BRANCH_REGISTER.
24712 * config/i386/i386.md (indirect_jump): Likewise.
24713 (tablejump): Likewise.
24714 (*sibcall_memory): Likewise.
24715 (*sibcall_value_memory): Likewise.
24716 Peepholes of indirect call and jump via memory: Likewise.
24717 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
24718 (*sibcall_value_GOT_32): Likewise.
24719 * config/i386/predicates.md (indirect_branch_operand): Likewise.
24720 (GOT_memory_operand): Likewise.
24721 (call_insn_operand): Likewise.
24722 (sibcall_insn_operand): Likewise.
24723 (GOT32_symbol_operand): Likewise.
24724 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
24726 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
24728 PR rtl-optimization/83496
24729 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
24730 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
24731 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
24732 redundant insn, if any.
24733 (relax_delay_slots): Likewise.
24734 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
24736 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
24738 PR tree-optimization/83965
24739 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
24740 that grouped statements are part of a reduction chain. Return
24741 true if the statement is not marked as a reduction itself but
24742 is part of a group.
24743 (vect_recog_dot_prod_pattern): Don't check whether the statement
24744 is part of a group here.
24745 (vect_recog_sad_pattern): Likewise.
24746 (vect_recog_widen_sum_pattern): Likewise.
24748 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
24751 * final.c (rest_of_clean_state): Also look for calls inside sequences.
24753 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24756 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
24758 (ix86_output_indirect_function_return): New prototype.
24759 (ix86_split_simple_return_pop_internal): Likewise.
24760 * config/i386/i386.c (indirect_return_via_cx): New.
24761 (indirect_return_via_cx_bnd): Likewise.
24762 (indirect_thunk_name): Handle return va CX_REG.
24763 (output_indirect_thunk_function): Create alias for
24764 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
24765 (ix86_output_indirect_jmp): Remove the bool argument.
24766 (ix86_output_indirect_function_return): New function.
24767 (ix86_split_simple_return_pop_internal): Likewise.
24768 * config/i386/i386.md (*indirect_jump): Don't pass false
24769 to ix86_output_indirect_jmp.
24770 (*tablejump_1): Likewise.
24771 (simple_return_pop_internal): Change it to define_insn_and_split.
24772 Call ix86_split_simple_return_pop_internal to split it for
24773 -mfunction-return=.
24774 (simple_return_indirect_internal): Call
24775 ix86_output_indirect_function_return instead of
24776 ix86_output_indirect_jmp.
24778 2018-02-26 Jakub Jelinek <jakub@redhat.com>
24781 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
24782 memset and value initialization afterwards.
24784 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
24786 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
24788 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24791 * common/config/aarch64/aarch64-common.c
24792 (aarch_option_optimization_table[]): Switch
24793 off fomit-frame-pointer
24795 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
24796 Chung-Ju Wu <jasonwucj@gmail.com>
24798 * config/nds32/nds32-multiple.md (load_multiple): Disallow
24800 (store_multiple): Ditto.
24802 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
24804 * config.gcc: Add --with-cpu support for nds32 target.
24805 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
24806 * config/nds32/nds32.opt: Add -mcpu= option.
24808 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
24810 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
24811 isel=yes): Warn for these deprecated options.
24813 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
24815 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
24816 ISA_2_5_MASKS_EMBEDDED.
24818 2018-02-23 Jakub Jelinek <jakub@redhat.com>
24820 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
24821 p->max as pointers rather than using iterative_hash_expr.
24823 2018-02-23 Carl Love <cel@us.ibm.com>
24825 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
24826 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
24828 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
24829 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
24830 P8V_BUILTIN_VEC_VUNSIGNED2.
24832 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
24835 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
24836 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
24838 * lra-constraints.c (curr_insn_transform): Set up
24839 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
24841 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
24842 * lra-eliminations.c (spill_pseudos): Ditto.
24843 (process_insn_for_elimination): Ditto.
24844 * lra-lives.c (reg_early_clobber_p): Use the new macros.
24845 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
24846 LRA_NON_CLOBBERED_ALT.
24848 2018-02-22 Martin Sebor <msebor@redhat.com>
24850 PR tree-optimization/84480
24851 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
24852 to maybe_diag_stxncpy_trunc. Call it.
24853 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
24854 from gimple_fold_builtin_strcpy. Print inlining stack.
24855 (handle_builtin_stxncpy): Print inlining stack.
24856 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
24858 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
24861 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
24862 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
24863 and -fcheck-pointer-bounds are used together.
24864 (indirect_thunk_prefix): New enum.
24865 (indirect_thunk_need_prefix): New function.
24866 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
24867 "_nt" instead of "_bnd" for NOTRACK prefix.
24868 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
24869 (output_indirect_thunk_function): Likewise.
24871 (ix86_code_end): Update output_indirect_thunk_function calls.
24872 (ix86_output_indirect_branch_via_reg): Replace
24873 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
24874 (ix86_output_indirect_branch_via_push): Likewise.
24875 (ix86_output_function_return): Likewise.
24876 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
24877 incompatible with -fcf-protection=branch and
24878 -fcheck-pointer-bounds.
24880 2018-02-22 Steve Ellcey <sellcey@cavium.com>
24883 * config/aarch64/aarch64.c (aarch64_print_address_internal):
24884 Change gcc_assert call to output_operand_lossage.
24886 2018-02-22 Steve Ellcey <sellcey@cavium.com>
24888 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
24890 2018-02-22 DJ Delorie <dj@redhat.com>
24891 Sebastian Perta <sebastian.perta@renesas.com>
24892 Oleg Endo <olegendo@gcc.gnu.org>
24894 * config/rx/rx.c (rx_rtx_costs): New function.
24895 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
24897 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24899 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
24901 2018-02-22 Martin Liska <mliska@suse.cz>
24904 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
24905 Add "native" as a possible value.
24907 2018-02-22 Martin Liska <mliska@suse.cz>
24910 * config/i386/i386.c (ix86_option_override_internal):
24911 Add "native" as a possible value for -march and -mtune.
24913 2018-02-22 Jakub Jelinek <jakub@redhat.com>
24916 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
24917 to all type variants.
24919 PR tree-optimization/84503
24920 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
24921 width as info->bitpos + info->bitsize - start.
24922 (merged_store_group::merge_overlapping): Simplify width computation.
24923 (check_no_overlap): New function.
24924 (imm_store_chain_info::try_coalesce_bswap): Compute expected
24925 start + width and last_order of the group, fail if check_no_overlap
24927 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
24928 to group if check_no_overlap fails.
24930 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
24932 * config/rs6000/altivec.md: Delete contraint arguments to
24933 define_expand, define_split, and define_peephole2, and in
24934 define_insn_and_split if always unused.
24935 * config/rs6000/darwin.md: Ditto.
24936 * config/rs6000/dfp.md: Ditto.
24937 * config/rs6000/rs6000.md: Ditto.
24938 * config/rs6000/sync.md: Ditto.
24939 * config/rs6000/vector.md: Ditto.
24940 * config/rs6000/vsx.md: Ditto.
24942 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
24944 * config/rs6000/altivec.md: Write output control strings as braced
24945 blocks instead of double-quoted strings.
24946 * config/rs6000/darwin.md: Ditto.
24947 * config/rs6000/rs6000.md: Ditto.
24948 * config/rs6000/vector.md: Ditto.
24949 * config/rs6000/vsx.md: Ditto.
24951 2018-02-21 Jason Merrill <jason@redhat.com>
24953 PR c++/84314 - ICE with templates and fastcall attribute.
24954 * attribs.c (build_type_attribute_qual_variant): Remove assert.
24956 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
24958 * ipa-cp.c (determine_versionability): Fix comment typos.
24960 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
24963 * ipa-cp.c (determine_versionability): Do not version functions caling
24966 2018-02-21 Martin Liska <mliska@suse.cz>
24969 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
24970 Add "native" as a possible value.
24971 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
24972 the macro when native cpu detection is available.
24974 2018-02-21 Martin Liska <mliska@suse.cz>
24977 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
24978 Add "native" as a possible value.
24979 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
24980 when native cpu detection is available.
24982 2018-02-21 Jakub Jelinek <jakub@redhat.com>
24983 Martin Sebor <msebor@redhat.com>
24985 PR tree-optimization/84478
24986 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
24988 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
24989 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
24990 support which is conservatively correct, for 2 only stay conservative
24991 for maxlen. Formatting and comment capitalization fixes. Add STRICT
24992 argument to the 2 argument get_range_strlen, adjust 6 arg
24993 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
24995 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
24996 (gimple_fold_builtin_strlen): Pass true as last argument to
24999 2018-02-20 Martin Sebor <msebor@redhat.com>
25001 PR middle-end/84095
25002 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
25003 (builtin_memref::set_base_and_offset): Same. Handle inner references.
25004 (builtin_memref::builtin_memref): Factor out parts into
25005 set_base_and_offset and call it.
25007 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
25009 PR middle-end/84406
25010 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
25011 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
25012 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
25013 search at the associated MODE_INT.
25015 2018-02-20 Jeff Law <law@redhat.com>
25017 PR middle-end/82123
25018 PR tree-optimization/81592
25019 PR middle-end/79257
25020 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
25021 for range data rather than using global data.
25022 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
25023 range data rather than using global data.
25024 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
25025 pass it to children as needed.
25026 (struct directive::fmtresult): Similarly.
25027 (struct directive::set_width): Similarly.
25028 (struct directive::set_precision): Similarly.
25029 (format_integer, format_directive, parse_directive): Similarly.
25030 (format_none): Accept unnamed vr_values parameter.
25031 (format_percent, format_floating, format_character): Similarly.
25032 (format_string, format_plain): Similarly.
25033 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
25034 the EVRP range analyzer for range data rather than using global data.
25035 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
25036 gimple-ssa-evrp-analyze.h
25037 (class sprintf_dom_walker): Add after_dom_children member function.
25038 Add evrp_range_analyzer member.
25039 (sprintf_dom_walker::before_dom_children): Call into the EVRP
25040 range analyzer as needed.
25041 (sprintf_dom_walker::after_dom_children): New member function.
25042 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
25044 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
25045 (evrp_range_analyzer::pop_to_marker): Likewise.
25047 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
25049 PR tree-optimization/84419
25050 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
25051 with the required type if its current type is compatible but
25054 2018-02-20 Jakub Jelinek <jakub@redhat.com>
25056 PR middle-end/82004
25057 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
25058 after vectorization.
25060 2018-02-20 Martin Liska <mliska@suse.cz>
25063 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
25064 possible values if we don't have a hint.
25066 2018-02-20 Martin Liska <mliska@suse.cz>
25070 * final.c (shorten_branches): Build align_tab array with one
25072 * opts.c (finish_options): Add alignment option limit check.
25073 (MAX_CODE_ALIGN): Likewise.
25074 (MAX_CODE_ALIGN_VALUE): Likewise.
25075 * doc/invoke.texi: Document maximum allowed option value for
25076 all -falign-* options.
25078 2018-02-19 Jakub Jelinek <jakub@redhat.com>
25081 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
25082 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
25083 * var-tracking.c (emit_note_insn_var_location): Remove all references
25084 to NOTE_INSN_CALL_ARG_LOCATION.
25085 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
25086 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
25087 Use copy_rtx_if_shared.
25088 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
25089 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
25090 (dwarf2out_var_location): Remove handling of
25091 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
25093 * final.c (final_scan_insn): Remove all references to
25094 NOTE_INSN_CALL_ARG_LOCATION.
25095 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
25096 before dumping final insns.
25097 * except.c (emit_note_eh_region_end): Remove all references to
25098 NOTE_INSN_CALL_ARG_LOCATION.
25099 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
25100 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
25101 * config/arc/arc.c (hwloop_optimize): Likewise.
25102 * config/arm/arm.c (create_fix_barrier): Likewise.
25103 * config/s390/s390.c (s390_chunkify_start): Likewise.
25104 * config/sh/sh.c (find_barrier): Likewise.
25105 * config/i386/i386.c (rest_of_insert_endbranch,
25106 ix86_seh_fixup_eh_fallthru): Likewise.
25107 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
25108 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
25109 * config/frv/frv.c (frv_function_prologue): Likewise.
25110 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
25112 (note_outside_basic_block_p): Remove all references to
25113 NOTE_INSN_CALL_ARG_LOCATION.
25114 * gengtype.c (adjust_field_rtx_def): Likewise.
25115 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
25117 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
25118 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
25121 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
25124 PR tree-optimization/84452
25125 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
25126 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
25129 2018-02-19 Martin Liska <mliska@suse.cz>
25132 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
25134 2018-02-19 Martin Liska <mliska@suse.cz>
25135 Richard Sandiford <richard.sandiford@linaro.org>
25137 PR tree-optimization/82491
25138 * gimple-fold.c (get_base_constructor): Make earlier bail out
25141 2018-02-19 Carl Love <cel@us.ibm.com>
25143 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
25144 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
25146 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
25147 P8V_BUILTIN_VEC_NEG.
25149 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
25151 * config/rl78/rl78.md (movdf): New define expand.
25153 2018-02-19 Martin Liska <mliska@suse.cz>
25156 * doc/invoke.texi: Fix typo.
25157 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
25159 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
25161 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
25162 handle rs6000_single_float and rs6000_double_float specially for
25165 2018-02-16 Jeff Law <law@redhat.com>
25167 * config/rx/rx.c (add_pop_cfi_notes): New function.;
25168 (pop_regs): Use it.
25170 2018-02-16 Jakub Jelinek <jakub@redhat.com>
25173 * ipa-inline.c (inline_small_functions): Fix a typo.
25175 2018-02-16 Nathan Sidwell <nathan@acm.org>
25177 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
25179 2018-02-16 Carl Love <cel@us.ibm.com>
25181 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
25182 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
25183 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
25184 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
25185 expansion to P8V_BUILTIN_VEC_FLOAT2.
25187 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
25189 PR rtl-optimization/70023
25190 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
25191 src_regno into account.
25193 2018-02-16 Carl Love <cel@us.ibm.com>
25195 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
25196 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
25197 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
25198 * config/rs6000/rs6000.c: Remove case statements for
25199 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
25200 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
25201 and P9V_BUILTIN_VEC_VINSERT4B.
25202 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
25203 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
25204 * config/rs6000/vsx.md:
25205 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
25208 2018-02-16 Carl Love <cel@us.ibm.com>
25210 * config/rs6000/altivec.h: Add builtin names vec_extract4b
25212 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
25214 * config/rs6000/rs6000-c.c: Add the definitions for
25215 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
25216 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
25217 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
25218 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
25219 definition for insert4b and define insn *insert3b_internal.
25220 * doc/extend.texi: Add documentation for vec_extract4b.
25222 2018-02-16 Nathan Sidwell <nathan@acm.org>
25224 * doc/extend.texi (Backwards Compatibility): Mention friend
25225 injection. Note for-scope is deprecated.
25226 * doc/invoke.texi (-ffriend-injection): Deprecate.
25228 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
25230 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
25231 that moved to I2, also allow destinations that are a paradoxical
25232 subreg (instead of a normal reg).
25234 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
25237 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
25240 2018-02-16 Richard Biener <rguenther@suse.de>
25242 PR tree-optimization/84037
25243 PR tree-optimization/84016
25245 * config/i386/i386.c (ix86_builtin_vectorization_cost):
25246 Adjust vec_construct for the fact we need additional higher latency
25247 128bit inserts for AVX256 and AVX512 vector builds.
25248 (ix86_add_stmt_cost): Scale vector construction cost for
25251 2018-02-16 Richard Biener <rguenther@suse.de>
25253 PR tree-optimization/84417
25254 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
25255 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
25256 (non_rewritable_lvalue_p): Likewise, use poly-ints.
25258 2018-02-16 Martin Liska <mliska@suse.cz>
25261 * internal-fn.def (ASAN_CHECK): Set proper flags.
25262 (ASAN_MARK): Likewise.
25264 2018-02-16 Julia Koval <julia.koval@intel.com>
25266 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
25267 from PTA_CANNONLAKE.
25269 2018-02-16 Jakub Jelinek <jakub@redhat.com>
25272 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
25273 Use ++iter rather than iter++ for std::list iterators.
25274 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
25275 defer deleting them until all nodes in the forest are processed. Do
25276 free even leaf nodes. Change to_process into auto_vec.
25279 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
25280 * vec.h (vec_default_construct): Use memset instead of placement new
25281 if BROKEN_VALUE_INITIALIZATION is defined.
25282 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
25283 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
25286 PR rtl-optimization/83723
25287 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
25288 * lra.c (lra_substitute_pseudo): Likewise. If true, use
25289 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
25291 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
25293 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
25295 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
25297 PR rtl-optimization/81443
25298 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
25299 from inner REGs to paradoxical SUBREGs.
25301 2018-02-16 Richard Biener <rguenther@suse.de>
25303 PR tree-optimization/84399
25304 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
25305 For operands we can analyze at their definition make sure we can
25306 analyze them at each use as well.
25308 2018-02-16 Richard Biener <rguenther@suse.de>
25310 PR tree-optimization/84190
25311 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
25312 volatile accesses if the decl isn't volatile.
25314 2018-02-15 Jason Merrill <jason@redhat.com>
25316 PR c++/84314 - ICE with templates and fastcall attribute.
25317 * attribs.c (build_type_attribute_qual_variant): Don't clobber
25318 TYPE_CANONICAL on an existing type.
25320 2018-02-15 Jakub Jelinek <jakub@redhat.com>
25322 PR tree-optimization/84383
25323 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
25324 dstoff nor call operand_equal_p if dstbase is NULL.
25326 PR tree-optimization/84334
25327 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
25328 also a CONSTANT_CLASS_P, punt.
25330 2018-02-14 Jim Wilson <jimw@sifive.com>
25332 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
25333 first SMALL_OPERAND check. New local min_second_step. Move assert
25334 to where locals are set. Add TARGET_RVC support.
25335 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
25337 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
25339 * doc/invoke.texi: Correct -Wformat-overflow code sample.
25341 2018-02-14 Martin Sebor <msebor@redhat.com>
25343 PR tree-optimization/83698
25344 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
25345 arrays constrain the offset range to their bounds.
25346 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
25347 (builtin_access::overlap): Avoid setting the size of overlap if it's
25349 (maybe_diag_overlap): Also consider arrays when deciding what values
25350 of offsets to include in diagnostics.
25352 2018-02-14 Martin Sebor <msebor@redhat.com>
25355 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
25356 that correspond to the kind of a declaration.
25358 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
25361 * config/pa/pa.md: Load address of PIC label using the linkage table
25362 if the label is nonlocal.
25364 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
25366 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
25367 warning message if user requests -maltivec=be.
25368 * doc/invoke.texi: Document deprecation of -maltivec=be.
25370 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
25373 * config/rs6000/rs6000-c.c: Update definitions for
25374 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
25375 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
25377 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25380 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
25381 add _get_ssp intrinsics. Remove argument from
25382 __builtin_ia32_rdssp[d|q].
25383 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
25384 * config/i386/i386-builtin.def: Remove argument from
25385 __builtin_ia32_rdssp[d|q].
25386 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
25387 ix86_expand_special_args_builtin for _rdssp[d|q].
25388 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
25389 Clear register before usage.
25390 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
25391 Add documentation for new _get_ssp and _inc_ssp intrinsics.
25393 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
25395 PR tree-optimization/84357
25396 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
25397 operand 1 of an ARRAY_REF too.
25399 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
25402 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
25403 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
25405 (set_of_reg): New struct.
25406 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
25407 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
25408 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
25410 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
25411 Split into bitclr, bitset, bitinvert patterns if appropriate.
25412 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
25413 use rx_fuse_in_memory_bitop.
25414 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
25415 to named insn, correct maximum insn length.
25417 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
25420 * machmode.def: Define a complex mode for PARTIAL_INT.
25421 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
25423 * doc/rtl.texi: Document CSPImode.
25424 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
25426 (msp430_hard_regno_nregs_with_padding): Likewise.
25428 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
25431 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
25433 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25435 PR rtl-optimization/84169
25436 * combine.c (try_combine): New variable split_i2i3. Set it to true if
25437 we generated a parallel as new i3 and we split that to new i2 and i3
25438 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
25439 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
25440 those to i2, not i1. Partially rewrite this scan code.
25442 2018-02-13 Jakub Jelinek <jakub@redhat.com>
25445 * stor-layout.c (place_field): For variable length fields, adjust
25446 offset_align afterwards not just based on the field's alignment,
25447 but also on the size.
25449 PR middle-end/84309
25450 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
25451 of exps and logs in the use_exp2 case.
25453 2018-02-13 Jeff Law <law@redhat.com>
25455 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
25456 entry for "vector".
25458 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
25461 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
25465 * common.opt (gas-loc-support, gas-locview-support): New.
25466 (ginline-points, ginternal-reset-location-views): New.
25467 * doc/invoke.texi: Document them. Use @itemx where intended.
25468 (gvariable-location-views): Adjust.
25469 * target.def (reset_location_view): New.
25470 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
25471 (TARGET_RESET_LOCATION_VIEW): New.
25472 * doc/tm.texi: Rebuilt.
25473 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
25474 (dwarf2out_default_as_locview_support): New.
25475 (output_asm_line_debug_info): Use option variables.
25476 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
25477 (output_loc_list): Likewise.
25478 (add_high_low_attributes): Check option variables.
25479 Don't output entry view attribute in strict mode.
25480 (gen_inlined_subroutine_die): Check option variables.
25481 (dwarf2out_inline_entry): Likewise.
25482 (init_sections_and_labels): Likewise.
25483 (dwarf2out_early_finish): Likewise.
25484 (maybe_reset_location_view): New, from...
25485 (dwarf2out_var_location): ... here. Call it.
25486 * debug.h (dwarf2out_default_as_loc_support): Declare.
25487 (dwarf2out_default_as_locview_support): Declare.
25488 * hooks.c (hook_int_rtx_insn_0): New.
25489 * hooks.h (hook_int_rtx_insn_0): Declare.
25490 * toplev.c (process_options): Take -gas-loc-support and
25491 -gas-locview-support from dwarf2out. Enable
25492 -gvariable-location-views by default only with locview
25493 assembler support. Enable -ginternal-reset-location-views by
25494 default only if the target defines the corresponding hook.
25495 Enable -ginline-points by default if location views are
25496 enabled; force it disabled if statement frontiers are
25498 * tree-inline.c (expand_call_inline): Check option variables.
25499 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
25501 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
25503 PR tree-optimization/84321
25504 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
25505 handling. Also check whether the anti-range contains any values
25506 that satisfy the mask; switch to a VR_RANGE if not.
25508 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
25511 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
25513 2018-02-13 Martin Jambor <mjambor@suse.cz>
25516 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
25517 of call statements, also set location of a load to a temporary.
25519 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
25521 * config/rl78/rl78.c (add_vector_labels): New function.
25522 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
25523 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
25524 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
25525 which checks that no arguments are passed.
25526 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
25527 * doc/extend.texi: Documentation for the new attribute.
25529 2018-02-13 Andreas Schwab <schwab@suse.de>
25531 * config/riscv/linux.h (CPP_SPEC): Define.
25533 2018-02-13 Jakub Jelinek <jakub@redhat.com>
25536 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
25537 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
25538 OPTION_MASK_ISA_AES as first argument to def_builtin_const
25539 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
25540 instead of OPTION_MASK_ISA_PCLMUL as first argument to
25541 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
25542 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
25543 temporarily for AES and PCLMUL builtins.
25545 PR tree-optimization/84339
25546 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
25547 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
25550 PR middle-end/84309
25551 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
25552 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
25553 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
25555 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
25557 * omp-simd-clone.h: New file.
25558 * omp-simd-clone.c: Include omp-simd-clone.h.
25559 (expand_simd_clones): No longer static.
25560 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
25561 cgraph.h and omp-simd-clone.h.
25562 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
25563 (vect_recog_widen_shift_pattern): Formatting fix.
25564 (vect_pattern_recog_1): Don't check optab for calls.
25567 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
25568 operands[2] into a REG before using gen_lowpart on it.
25570 2018-02-12 Jeff Law <law@redhat.com>
25573 * config/sh/sh.c (find_barrier): Consider a sibling call
25576 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
25577 successfully back substituting a reg.
25579 2018-02-12 Richard Biener <rguenther@suse.de>
25581 PR tree-optimization/84037
25582 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
25583 parameter, move visited init to caller.
25584 (vect_slp_analyze_operations): Separate cost from validity
25585 check, initialize visited once for all instances.
25586 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
25588 * tree-vect-stmts.c (vect_model_simple_cost): Make early
25590 (vect_model_promotion_demotion_cost): Likewise.
25591 (vectorizable_bswap): Guard cost modeling with !slp_node
25592 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
25594 (vectorizable_call): Likewise.
25595 (vectorizable_conversion): Likewise.
25596 (vectorizable_assignment): Likewise.
25597 (vectorizable_shift): Likewise.
25598 (vectorizable_operation): Likewise.
25599 (vectorizable_store): Likewise.
25600 (vectorizable_load): Likewise.
25601 (vectorizable_condition): Likewise.
25602 (vectorizable_comparison): Likewise.
25604 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
25607 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
25608 (ASAN_MARK): Fix fnspec to account for return value, change pointer
25609 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
25611 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
25613 PR middle-end/83665
25614 * params.def (inline-min-speedup): Increase from 8 to 15.
25615 (max-inline-insns-auto): Decrease from 40 to 30.
25616 * ipa-split.c (consider_split): Add some buffer for function to
25617 be considered inlining candidate.
25618 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
25621 2018-02-12 Richard Biener <rguenther@suse.de>
25623 PR tree-optimization/84037
25624 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
25625 matched stmts if we cannot swap the non-matched ones.
25627 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
25629 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
25630 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
25631 _mm_maskz_scalef_round_ss): New intrinsics.
25632 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
25633 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
25634 __builtin_ia32_scalefss_round): Remove.
25635 (__builtin_ia32_scalefsd_mask_round,
25636 __builtin_ia32_scalefss_mask_round): New intrinsics.
25637 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
25638 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
25639 ((match_operand:VF_128 2 "<round_nimm_predicate>"
25640 "<round_constraint>")): Changed to ...
25641 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
25642 "<round_scalar_constraint>")): ... this.
25643 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
25644 %0, %1, %2<round_op3>}"): Changed to ...
25645 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
25646 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
25647 %2<round_scalar_mask_op3>}"): ... this.
25648 * config/i386/subst.md (round_scalar_nimm_predicate): New.
25650 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
25652 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
25653 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
25654 (_mm_maskz_sqrt_round_ss): New intrinsics.
25655 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
25656 (__builtin_ia32_sqrtsd_mask_round)
25657 (__builtin_ia32_sqrtss_mask_round): New builtins.
25658 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
25659 (__builtin_ia32_sqrtss_round): Remove.
25660 (__builtin_ia32_sqrtsd_mask_round)
25661 (__builtin_ia32_sqrtss_mask_round): New builtins.
25662 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
25663 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
25664 ((match_operand:VF_128 1 "vector_operand"
25665 "xBm,<round_constraint>")): Changed to ...
25666 ((match_operand:VF_128 1 "vector_operand"
25667 "xBm,<round_scalar_constraint>")): ... this.
25668 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
25669 %0, %2, %<iptr>1<round_op3>}): Changed to ...
25670 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
25671 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
25672 %<iptr>1<round_scalar_mask_op3>}): ... this.
25673 ((set_attr "prefix" "<round_prefix>")): Changed to ...
25674 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
25676 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
25679 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
25680 Cast vec_cmpeq result to correct type.
25681 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
25682 Cast vec_cmpgt result to correct type.
25684 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
25686 * final.c (final_scan_insn_1): Renamed from...
25687 (final_scan_insn): ... this. New wrapper, to recover
25688 seen from the outermost call in recursive ones.
25689 * config/sparc/sparc.c (output_return): Drop seen from call.
25690 (output_sibcall): Likewise.
25691 * config/visium/visium.c (output_branch): Likewise.
25693 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
25695 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
25698 2018-02-10 Alan Modra <amodra@gmail.com>
25701 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
25702 Specify LR as an input.
25704 2018-02-10 Jakub Jelinek <jakub@redhat.com>
25707 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
25708 remove_member_access_dummy_vars): New functions.
25709 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
25710 lower_omp_1, execute_lower_omp): Use them.
25712 PR rtl-optimization/84308
25713 * shrink-wrap.c (spread_components): Release todo vector.
25715 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
25717 PR rtl-optimization/57193
25718 * ira-color.c (struct allocno_color_data): Add member
25719 conflict_allocno_hard_prefs.
25720 (update_conflict_allocno_hard_prefs): New.
25721 (bucket_allocno_compare_func): Add a preference based on
25722 conflict_allocno_hard_prefs.
25723 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
25724 (color_allocnos): Remove a dead code. Initiate
25725 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
25727 2018-02-09 Jakub Jelinek <jakub@redhat.com>
25730 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
25731 constraint from =wa to wa. Avoid a subreg on the output operand,
25732 instead use a pseudo and subreg it in a move.
25733 (p9_xxbrd_<mode>): Changed to ...
25734 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
25735 (p9_xxbrd_v2df): New expander.
25736 (p9_xxbrw_<mode>): Changed to ...
25737 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
25738 (p9_xxbrw_v4sf): New expander.
25740 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
25742 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
25744 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
25747 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
25748 multiply in 32-bit mode.
25749 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
25750 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
25753 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
25755 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
25756 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
25757 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
25758 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
25760 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
25763 * dwarf2out.c (is_trivial_indirect_ref): New function.
25764 (dwarf2out_late_global_decl): Do not generate a location
25765 attribute for variables that have a non-trivial DECL_VALUE_EXPR
25766 and that are not defined in the current unit.
25768 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
25770 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
25771 instead of a libcall for UNORDERED.
25773 2018-02-09 Tamar Christina <tamar.christina@arm.com>
25776 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
25777 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
25779 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25782 * config/s390/s390.c (s390_set_current_function): Invoke
25783 s390_indirect_branch_settings also if fndecl didn't change.
25785 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
25787 * config/rs6000/rs6000.md (blockage): Set length to zero.
25789 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
25791 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
25793 2018-02-09 Jakub Jelinek <jakub@redhat.com>
25796 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
25797 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
25801 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
25802 PARALLEL incoming that failed vt_get_decl_and_offset check.
25804 PR middle-end/84237
25805 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
25806 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
25808 (get_variable_section): For decls in named .bss* sections pass true as
25809 second argument to bss_initializer_p.
25811 2018-02-09 Marek Polacek <polacek@redhat.com>
25812 Jakub Jelinek <jakub@redhat.com>
25815 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
25816 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
25817 Sync some changes from cxx_fold_indirect_ref.
25819 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
25821 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
25823 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
25824 (BLOCK_INLINE_ENTRY_LABEL): New.
25825 (dwarf2out_var_location): Disregard inline entry markers.
25826 (inline_entry_data): New struct.
25827 (inline_entry_data_hasher): New hashtable type.
25828 (inline_entry_data_hasher::hash): New.
25829 (inline_entry_data_hasher::equal): New.
25830 (inline_entry_data_table): New variable.
25831 (add_high_low_attributes): Add DW_AT_entry_pc and
25832 DW_AT_GNU_entry_view attributes if a pending entry is found
25833 in inline_entry_data_table. Add old entry_pc attribute only
25834 if debug nonbinding markers are disabled.
25835 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
25836 markers are enabled.
25837 (block_within_block_p, dwarf2out_inline_entry): New.
25838 (dwarf2out_finish): Check that no entries remained in
25839 inline_entry_data_table.
25840 * final.c (reemit_insn_block_notes): Handle inline entry notes.
25841 (final_scan_insn, notice_source_line): Likewise.
25842 (rest_of_clean_state): Skip inline entry markers.
25843 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
25845 * gimple.c (gimple_build_debug_inline_entry): New.
25846 * gimple.h (enum gimple_debug_subcode): Add
25847 GIMPLE_DEBUG_INLINE_ENTRY.
25848 (gimple_build_debug_inline_entry): Declare.
25849 (gimple_debug_inline_entry_p): New.
25850 (gimple_debug_nonbind_marker_p): Adjust.
25851 * insn-notes.def (INLINE_ENTRY): New.
25852 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
25853 inline entry marker notes.
25854 (print_insn): Likewise.
25855 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
25856 (INSN_DEBUG_MARKER_KIND): Likewise.
25857 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
25858 * tree-inline.c (expand_call_inline): Build and insert
25859 debug_inline_entry stmt.
25860 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
25861 inline entry blocks early, if nonbind markers are enabled.
25862 (dump_scope_block): Dump fragment info.
25863 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
25864 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
25865 (gimple_build_debug_inline_entry): New.
25866 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
25867 Enable/disable inline entry points too.
25868 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
25869 (DEBUG_INSN): Describe inline entry markers.
25871 * common.opt (gvariable-location-views): New.
25872 (gvariable-location-views=incompat5): New.
25873 * config.in: Rebuilt.
25874 * configure: Rebuilt.
25875 * configure.ac: Test assembler for view support.
25876 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
25877 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
25878 * dwarf2out.c (var_loc_view): New typedef.
25879 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
25880 (dwarf2out_locviews_in_attribute): New.
25881 (dwarf2out_locviews_in_loclist): New.
25882 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
25883 (enum dw_line_info_opcode): Add LI_adv_address.
25884 (struct dw_line_info_table): Add view.
25885 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
25886 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
25887 (zero_view_p): New variable.
25888 (ZERO_VIEW_P): New macro.
25889 (output_asm_line_debug_info): New.
25890 (struct var_loc_node): Add view.
25891 (add_AT_view_list, AT_loc_list): New.
25892 (add_var_loc_to_decl): Add view param. Test it against last.
25893 (new_loc_list): Add view params. Record them.
25894 (AT_loc_list_ptr): Handle loc and view lists.
25895 (view_list_to_loc_list_val_node): New.
25896 (print_dw_val): Handle dw_val_class_view_list.
25897 (size_of_die): Likewise.
25898 (value_format): Likewise.
25899 (loc_list_has_views): New.
25900 (gen_llsym): Set vl_symbol too.
25901 (maybe_gen_llsym, skip_loc_list_entry): New.
25902 (dwarf2out_maybe_output_loclist_view_pair): New.
25903 (output_loc_list): Output view list or entries too.
25904 (output_view_list_offset): New.
25905 (output_die): Handle dw_val_class_view_list.
25906 (output_dwarf_version): New.
25907 (output_compilation_unit_header): Use it.
25908 (output_skeleton_debug_sections): Likewise.
25909 (output_rnglists, output_line_info): Likewise.
25910 (output_pubnames, output_aranges): Update version comments.
25911 (output_one_line_info_table): Output view numbers in asm comments.
25912 (dw_loc_list): Determine current endview, pass it to new_loc_list.
25913 Call maybe_gen_llsym.
25914 (loc_list_from_tree_1): Adjust.
25915 (add_AT_location_description): Create view list attribute if
25916 needed, check it's absent otherwise.
25917 (convert_cfa_to_fb_loc_list): Adjust.
25918 (maybe_emit_file): Call output_asm_line_debug_info for test.
25919 (dwarf2out_var_location): Reset views as needed. Precompute
25920 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
25921 attribute. Set view.
25922 (new_line_info_table): Reset next view.
25923 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
25924 (dwarf2out_source_line): Likewise. Output view resets and labels to
25925 the assembler, or select appropriate line info opcodes.
25926 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
25927 (optimize_string_length): Catch it. Adjust.
25928 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
25929 dw_val_class_view_list, and remove it if no longer needed.
25930 (hash_loc_list): Hash view numbers.
25931 (loc_list_hasher::equal): Compare them.
25932 (optimize_location_lists): Check whether a view list symbol is
25933 needed, and whether the locview attribute is present, and
25934 whether they match. Remove the locview attribute if no longer
25936 (index_location_lists): Call skip_loc_list_entry for test.
25937 (dwarf2out_finish): Call output_asm_line_debug_info for test.
25938 Use output_dwarf_version.
25939 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
25940 (struct dw_val_node): Add val_view_list.
25941 * final.c (SEEN_NEXT_VIEW): New.
25942 (set_next_view_needed): New.
25943 (clear_next_view_needed): New.
25944 (maybe_output_next_view): New.
25945 (final_start_function): Rename to...
25946 (final_start_function_1): ... this. Take pointer to FIRST,
25947 add SEEN parameter. Emit param bindings in the initial view.
25948 (final_start_function): Reintroduce SEEN-less interface.
25949 (final): Rename to...
25950 (final_1): ... this. Take SEEN parameter. Output final pending
25951 next view at the end.
25952 (final): Reintroduce seen-less interface.
25953 (final_scan_insn): Output pending next view before switching
25954 sections or ending a block. Mark the next view as needed when
25955 outputting variable locations. Notify debug backend of section
25956 changes, and of location view changes.
25957 (rest_of_handle_final): Adjust.
25958 * toplev.c (process_options): Autodetect value for debug variable
25959 location views option. Warn on incompat5 without -gdwarf-5.
25960 * doc/invoke.texi (gvariable-location-views): New.
25961 (gvariable-location-views=incompat5): New.
25962 (gno-variable-location-views): New.
25964 2018-02-08 David Malcolm <dmalcolm@redhat.com>
25966 PR tree-optimization/84136
25967 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
25968 that the result of find_edge is non-NULL.
25970 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
25973 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
25974 storing integer register in SImode. Fix cost of 256 and 512
25975 byte aligned SSE register store.
25977 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
25979 * config/i386/i386.c (ix86_multiplication_cost): Fix
25980 multiplication cost for TARGET_AVX512DQ.
25982 2018-02-08 Marek Polacek <polacek@redhat.com>
25984 PR tree-optimization/84238
25985 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
25988 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25990 PR tree-optimization/84265
25991 * tree-vect-stmts.c (vectorizable_store): Don't treat
25992 VMAT_CONTIGUOUS accesses as grouped.
25993 (vectorizable_load): Likewise.
25995 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25997 PR tree-optimization/81635
25998 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
25999 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
26000 (test_round_for_mask): New functions.
26001 (wide_int_cc_tests): Call test_round_for_mask.
26002 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
26003 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
26004 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
26005 range returned by get_range_info.
26007 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
26010 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
26011 * symtab.c: Include builtins.h
26012 (symtab_node::output_to_lto_symbol_table_p): Move here
26013 from lto-streamer-out.c:output_symbol_p.
26014 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
26015 (output_symbol_p): Move all logic to symtab.c
26016 (produce_symtab): Update.
26018 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26020 * config/s390/s390-opts.h (enum indirect_branch): Define.
26021 * config/s390/s390-protos.h (s390_return_addr_from_memory)
26022 (s390_indirect_branch_via_thunk)
26023 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
26024 (enum s390_indirect_branch_type): Define.
26025 * config/s390/s390.c (struct s390_frame_layout, struct
26026 machine_function): Remove.
26027 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
26028 (indirect_branch_table_label_no, indirect_branch_table_name):
26030 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
26031 (enum s390_indirect_branch_option): Define.
26032 (s390_return_addr_from_memory): New function.
26033 (s390_handle_string_attribute): New function.
26034 (s390_attribute_table): Add new attribute handler.
26035 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
26036 (s390_indirect_branch_via_thunk): New function.
26037 (s390_indirect_branch_via_inline_thunk): New function.
26038 (s390_function_ok_for_sibcall): When jumping via thunk disallow
26039 sibling call optimization for non z10 compiles.
26040 (s390_emit_call): Force indirect branch target to be a single
26041 register. Add r1 clobber for non-z10 compiles.
26042 (s390_emit_epilogue): Emit return jump via return_use expander.
26043 (s390_reorg): Handle JUMP_INSNs as execute targets.
26044 (s390_option_override_internal): Perform validity checks for the
26045 new command line options.
26046 (s390_indirect_branch_attrvalue): New function.
26047 (s390_indirect_branch_settings): New function.
26048 (s390_set_current_function): Invoke s390_indirect_branch_settings.
26049 (s390_output_indirect_thunk_function): New function.
26050 (s390_code_end): Implement target hook.
26051 (s390_case_values_threshold): Implement target hook.
26052 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
26054 * config/s390/s390.h (struct s390_frame_layout)
26055 (struct machine_function): Move here from s390.c.
26056 (TARGET_INDIRECT_BRANCH_NOBP_RET)
26057 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
26058 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
26059 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
26060 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
26061 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
26062 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
26063 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
26064 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
26065 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
26066 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
26067 (mnemonic attribute): Add values which aren't recognized
26069 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
26070 pattern for branch conversion. Fix mnemonic attribute.
26071 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
26072 indirect branch via thunk if requested.
26073 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
26074 ("*indirect_jump"): Disable for branch conversion using out of
26076 ("indirect_jump_via_thunk<mode>_z10")
26077 ("indirect_jump_via_thunk<mode>")
26078 ("indirect_jump_via_inlinethunk<mode>_z10")
26079 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
26080 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
26081 ("casesi_jump_via_inlinethunk<mode>_z10")
26082 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
26083 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
26084 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
26085 ("*indirect2_jump"): Disable for branch conversion.
26086 ("casesi_jump"): Turn into expander and expand patterns for branch
26088 ("return_use"): New expander.
26089 ("*return"): Emit return via thunk and rename it to ...
26090 ("*return<mode>"): ... this one.
26091 * config/s390/s390.opt: Add new options and and enum for the
26094 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26096 * lra-constraints.c (match_reload): Unconditionally use
26097 gen_lowpart_SUBREG, rather than selecting between that
26098 and equivalent gen_rtx_SUBREG code.
26100 2018-02-08 Richard Biener <rguenther@suse.de>
26102 PR tree-optimization/84233
26103 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
26104 changed flag instead of boguously re-using phi_inserted.
26106 2018-02-08 Martin Jambor <mjambor@suse.cz>
26108 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
26109 static local variables.
26111 2018-02-08 Richard Biener <rguenther@suse.de>
26113 PR tree-optimization/84278
26114 * tree-vect-stmts.c (vectorizable_store): When looking for
26115 smaller vector types to perform grouped strided loads/stores
26116 make sure the mode is supported by the target.
26117 (vectorizable_load): Likewise.
26119 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
26121 * config/aarch64/aarch64.c (aarch64_components_for_bb):
26122 Increase LDP/STP opportunities by adding adjacent callee-saves.
26124 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
26126 PR rtl-optimization/84068
26127 PR rtl-optimization/83459
26128 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
26130 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
26132 PR tree-optimization/84224
26133 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
26134 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
26135 non-zero arguments.
26137 2018-02-07 Iain Sandoe <iain@codesourcery.com>
26140 * config/rs6000/altivec.md (*restore_world): Remove LR use.
26141 * config/rs6000/predicates.md (restore_world_operation): Adjust op
26142 count, remove one USE.
26144 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
26146 * doc/install.texi (Configuration): Document the
26147 --with-long-double-format={ibm,ieee} PowerPC configuration
26151 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
26152 Convert from define_expand to be define_insn_and_split. Rework
26153 float/double/_Float128 conversions to QI/HI/SImode to work with
26154 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
26155 conversions to QI/HImode types did a store and then a load to
26156 truncate the value. For conversions to VSX registers, don't split
26157 the insn, instead emit the code directly. Use the code iterator
26158 any_fix to combine signed and unsigned conversions.
26159 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
26160 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
26161 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
26162 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
26163 (fix_<mode>di2_hw): Likewise.
26164 (fixuns_<mode>di2_hw): Likewise.
26165 (fix_<mode>si2_hw): Likewise.
26166 (fixuns_<mode>si2_hw): Likewise.
26167 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
26168 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
26169 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
26170 fix<uns>_trunc<SFDF:mode>si2_p8.
26171 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
26172 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
26173 (fix<uns>_<mode>_mem): Likewise.
26174 (fctiw<u>z_<mode>_mem): Likewise.
26175 (fix<uns>_<mode>_mem): Likewise.
26176 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
26177 the register allocator from doing a direct move to the GPRs to do
26178 a store, and instead use the ISA 3.0 store byte/half-word from
26179 vector register instruction. For IEEE 128-bit floating point,
26180 also optimize stores of 32-bit ints.
26181 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
26183 2018-02-07 Alan Hayward <alan.hayward@arm.com>
26185 * genextract.c (push_pathstr_operand): New function to support
26187 (walk_rtx): Call push_pathstr_operand.
26188 (print_path): Support [a-zA-Z].
26190 2018-02-07 Richard Biener <rguenther@suse.de>
26192 PR tree-optimization/84037
26193 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
26194 (cse_and_gimplify_to_preheader): Declare.
26195 (vect_get_place_in_interleaving_chain): Likewise.
26196 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26198 (_loop_vec_info::~_loop_vec_info): Delete it.
26199 (cse_and_gimplify_to_preheader): New function.
26200 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
26201 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
26202 (vectorizable_load): Likewise. For grouped stores always base
26203 the IV on the first element.
26204 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
26205 condition before gimplifying.
26207 2018-02-07 Jakub Jelinek <jakub@redhat.com>
26209 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
26210 *DIV_EXPR and *MOD_EXPR.
26212 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
26215 * config/i386/i386.c (ix86_option_override_internal): Mask out
26216 the CF_SET bit when checking -fcf-protection.
26218 2018-02-07 Tom de Vries <tom@codesourcery.com>
26221 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
26224 2018-02-07 Richard Biener <rguenther@suse.de>
26226 PR tree-optimization/84204
26227 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
26230 PR tree-optimization/84205
26231 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
26232 special-case isl_ast_op_zdiv_r.
26234 PR tree-optimization/84223
26235 * graphite-scop-detection.c (gather_bbs::before_dom_children):
26236 Only add conditions from within the region.
26237 (gather_bbs::after_dom_children): Adjust.
26239 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
26242 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
26243 * config/avr/avr.md: Only post-reload split REG-REG moves if
26244 either register is GENERAL_REG_P.
26246 2018-02-07 Jakub Jelinek <jakub@redhat.com>
26248 PR tree-optimization/84235
26249 * tree-ssa-scopedtables.c
26250 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
26251 if the subtraction is performed in floating point type where NaNs are
26252 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
26253 build 1. Formatting fix.
26255 2018-02-06 Jakub Jelinek <jakub@redhat.com>
26258 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
26259 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
26260 and skip it regardless of bb boundaries. Use CALL_P macro,
26261 don't test INSN_P (insn) together with CALL_P or JUMP_P check
26262 unnecessarily, formatting fix.
26264 2018-02-06 Michael Collison <michael.collison@arm.com>
26266 * config/arm/thumb2.md:
26267 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
26268 (*thumb_mov_notscc): Ditto.
26270 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26273 * config/rs6000/rs6000.md (su code attribute): Use "u" for
26274 unsigned_fix, not "s".
26276 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26278 * configure.ac (gcc_fn_eh_frame_ro): New function.
26279 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
26280 correct .eh_frame permissions.
26281 * configure: Regenerate.
26283 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
26285 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
26286 irrelevant options.
26288 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26290 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26291 Display warning message for -mno-speculate-indirect-jumps.
26293 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
26295 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
26297 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
26299 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
26301 PR tree-optimization/84225
26302 * tree-eh.c (find_trapping_overflow): Only call
26303 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
26305 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26308 * config/i386/i386.c: Reimplement the check of possible options
26309 -mibt/-mshstk conbination. Change error messages.
26310 * doc/invoke.texi: Fix a typo: remove extra '='.
26312 2018-02-06 Marek Polacek <polacek@redhat.com>
26314 PR tree-optimization/84228
26315 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
26317 2018-02-06 Tamar Christina <tamar.christina@arm.com>
26320 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
26321 emitted arch directives.
26322 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
26323 __ARM_FEATURE_COPROC before changing architectures.
26325 2018-02-06 Richard Biener <rguenther@suse.de>
26327 * config/i386/i386.c (print_reg): Fix typo.
26328 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
26330 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
26332 * configure: Regenerate.
26334 2018-02-05 Martin Sebor <msebor@redhat.com>
26336 PR tree-optimization/83369
26337 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
26340 2018-02-05 Martin Liska <mliska@suse.cz>
26342 * doc/invoke.texi: Cherry-pick upstream r323995.
26344 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
26346 * ira.c (ira_init_register_move_cost): Adjust comment.
26348 2018-02-05 Martin Liska <mliska@suse.cz>
26350 PR gcov-profile/84137
26351 * doc/gcov.texi: Fix typo in documentation.
26353 2018-02-05 Martin Liska <mliska@suse.cz>
26355 PR gcov-profile/83879
26356 * doc/gcov.texi: Document necessity of --dynamic-list-data when
26357 using dlopen functionality.
26359 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
26361 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
26362 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
26363 _mm_maskz_range_ss, _mm_mask_range_round_ss,
26364 _mm_maskz_range_round_ss): New intrinsics.
26365 (__builtin_ia32_rangesd128_round)
26366 (__builtin_ia32_rangess128_round): Remove.
26367 (__builtin_ia32_rangesd128_mask_round,
26368 __builtin_ia32_rangess128_mask_round): New builtins.
26369 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
26370 __builtin_ia32_rangess128_round): Remove.
26371 (__builtin_ia32_rangesd128_mask_round,
26372 __builtin_ia32_rangess128_mask_round): New builtins.
26373 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
26374 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
26375 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
26376 "<round_saeonly_constraint>")): Changed to ...
26377 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
26378 "<round_saeonly_scalar_constraint>")): ... this.
26379 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
26380 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
26381 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
26382 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
26383 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
26385 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
26387 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
26389 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
26390 Remove all values except native, 8540 and 8548.
26392 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
26394 * config/i386/i386.c (ix86_output_function_return): Pass
26395 INVALID_REGNUM, instead of -1, as invalid register number to
26396 indirect_thunk_name and output_indirect_thunk.
26398 2018-02-02 Julia Koval <julia.koval@intel.com>
26400 * config.gcc: Add -march=icelake.
26401 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
26402 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
26403 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
26404 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
26405 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
26406 (processor_target_table): Add icelake.
26407 (ix86_option_override_internal): Handle new PTAs.
26408 (get_builtin_code_for_version): Handle icelake.
26409 (M_INTEL_COREI7_ICELAKE): New.
26410 (fold_builtin_cpu): Handle icelake.
26411 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
26412 * doc/invoke.texi: Add -march=icelake.
26414 2018-02-02 Julia Koval <julia.koval@intel.com>
26416 * config/i386/i386.c (ix86_option_override_internal): Change flags type
26417 to wide_int_bitmask.
26418 * wide-int-bitmask.h: New.
26420 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26423 * config/i386/i386.md: Replace Pmode with word_mode in
26424 builtin_setjmp_setup and builtin_longjmp to support x32.
26426 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
26430 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
26432 (rs6000_supported_cpu_names): New static variable.
26433 (linux_cpu_translation_table): Likewise.
26434 (elf_platform) <cpu>: Define new static variable and use it.
26435 Translate kernel AT_PLATFORM name to canonical name if needed.
26436 Error if platform name is unknown.
26438 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
26441 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
26443 2018-02-01 Jeff Law <law@redhat.com>
26446 * config/i386/i386.c (release_scratch_register_on_entry): Add new
26447 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
26448 the scratch if RELEASE_VIA_POP is false.
26449 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
26450 If we have to save a temporary register, decrement SIZE appropriately.
26451 Pass new arguments to release_scratch_register_on_entry.
26452 (ix86_adjust_stack_and_probe): Likewise.
26453 (ix86_emit_probe_stack_range): Pass new arguments to
26454 release_scratch_register_on_entry.
26456 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
26458 PR rtl-optimization/84157
26459 * combine.c (change_zero_ext): Use REG_P predicate in
26460 front of HARD_REGISTER_P predicate.
26462 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
26464 * config/avr/avr.c (avr_option_override): Move disabling of
26465 -fdelete-null-pointer-checks to...
26466 * common/config/avr/avr-common.c (avr_option_optimization_table):
26469 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26471 PR tree-optimization/81635
26472 * tree-data-ref.c (split_constant_offset_1): For types that
26473 wrap on overflow, try to use range info to prove that wrapping
26476 2018-02-01 Renlin Li <renlin.li@arm.com>
26479 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
26480 TAILCALL_ADDR_REGS.
26481 (aarch64_register_move_cost): Likewise.
26482 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
26483 TAILCALL_ADDR_REGS.
26484 (REG_CLASS_NAMES): Likewise.
26485 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
26486 TAILCALL_ADDR_REGS. Remove IP registers.
26487 * config/aarch64/aarch64.md (Ucs): Update register constraint.
26489 2018-02-01 Richard Biener <rguenther@suse.de>
26491 * domwalk.h (dom_walker::dom_walker): Add additional constructor
26492 for specifying RPO order and allow NULL for that.
26493 * domwalk.c (dom_walker::dom_walker): Likewise.
26494 (dom_walker::walk): Handle NULL RPO order.
26495 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
26497 (rewrite_update_dom_walker): Likewise.
26498 (mark_def_dom_walker): Likewise.
26500 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26502 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
26503 (aarch64_maybe_expand_sve_subreg_move): Declare.
26504 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
26505 * config/aarch64/predicates.md (aarch64_any_register_operand): New
26507 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
26508 that are semantically a reverse operation.
26509 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
26510 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
26511 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
26513 (aarch64_can_change_mode_class): For big-endian, forbid changes
26514 between two SVE modes if they have different element sizes.
26516 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26518 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
26519 the TImode handling for big-endian targets.
26521 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26523 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
26524 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
26526 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
26527 Remove BSWAP handing for big-endian targets and use the form of
26528 LD1RQ appropariate for the mode.
26530 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26532 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
26533 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
26534 duplicated element.
26536 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26539 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
26540 check for operands that need to go through aarch64_sve_reload_be.
26542 2018-02-01 Jakub Jelinek <jakub@redhat.com>
26544 PR tree-optimization/81661
26545 PR tree-optimization/84117
26546 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
26547 * tree-eh.c: Include gimplify.h.
26548 (find_trapping_overflow, replace_trapping_overflow,
26549 rewrite_to_non_trapping_overflow): New functions.
26550 * tree-vect-loop.c: Include tree-eh.h.
26551 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
26552 * tree-data-ref.c: Include tree-eh.h.
26553 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
26555 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
26557 PR rtl-optimization/84123
26558 * combine.c (change_zero_ext): Check if hard register satisfies
26559 can_change_dest_mode before calling gen_lowpart_SUBREG.
26561 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
26564 * ira.c (ira_init_register_move_cost): Remove assert.
26566 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26568 PR rtl-optimization/84071
26569 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
26570 * doc/tm.texi: Regenerate.
26572 2018-01-31 Richard Biener <rguenther@suse.de>
26574 PR tree-optimization/84132
26575 * tree-data-ref.c (analyze_miv_subscript): Properly
26576 check whether evolution_function_is_affine_multivariate_p
26577 before calling gcd_of_steps_may_divide_p.
26579 2018-01-31 Julia Koval <julia.koval@intel.com>
26582 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26583 * config/i386/i386.md (rdpid_rex64) New.
26584 (rdpid): Make 32bit only.
26586 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
26589 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
26590 an IDENTIFIER_NODE for FUNCTION_TYPE's.
26592 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26595 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26597 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26599 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26601 PR rtl-optimization/84071
26602 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
26603 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
26605 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
26607 * config/arc/arc.c (arc_handle_aux_attribute): New function.
26608 (arc_attribute_table): Add 'aux' attribute.
26609 (arc_in_small_data_p): Consider aux like variables.
26610 (arc_is_aux_reg_p): New function.
26611 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
26612 (arc_get_aux_arg): New function.
26613 (prepare_move_operands): Handle aux-register access.
26614 (arc_handle_aux_attribute): New function.
26615 * doc/extend.texi (ARC Variable attributes): Add subsection.
26617 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
26619 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
26620 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
26621 (arc_attribute_table): Add 'uncached' attribute.
26622 (arc_print_operand): Print '.di' flag for uncached memory
26624 (arc_in_small_data_p): Do not consider for small data the uncached
26626 (arc_is_uncached_mem_p): New function.
26627 * config/arc/predicates.md (compact_store_memory_operand): Check
26628 for uncached memory accesses.
26629 (nonvol_nonimm_operand): Likewise.
26630 * doc/extend.texi (ARC Type Attribute): New subsection.
26632 2018-01-31 Jakub Jelinek <jakub@redhat.com>
26635 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
26636 falign-loops=): Add Optimization flag.
26638 2018-01-30 Jeff Law <law@redhat.com>
26641 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
26642 INT_REGISTERS_SAVED. Check it prior to calling
26643 get_scratch_register_on_entry.
26644 (ix86_adjust_stack_and_probe): Similarly.
26645 (ix86_emit_probe_stack_range): Similarly.
26646 (ix86_expand_prologue): Corresponding changes.
26648 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26651 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
26652 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
26654 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
26657 * lra-constraints.c (curr_insn_transform): Process AND in the
26660 2018-01-30 Jakub Jelinek <jakub@redhat.com>
26662 PR rtl-optimization/83986
26663 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
26664 dependence against last_pending_memory_flush in addition to
26665 pending_jump_insns.
26667 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
26669 PR tree-optimization/81611
26670 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
26673 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26676 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
26679 2018-01-30 Richard Biener <rguenther@suse.de>
26680 Jakub Jelinek <jakub@redhat.com>
26682 PR tree-optimization/84111
26683 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
26684 inner loops added during recursion, as they don't have up-to-date
26687 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
26690 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
26691 (can_inline_edge_by_limits_p): ... here.
26692 (can_early_inline_edge_p, check_callers,
26693 update_caller_keys, update_callee_keys, recursive_inlining,
26694 add_new_edges_to_heap, speculation_useful_p,
26695 inline_small_functions,
26696 inline_small_functions, flatten_function,
26697 inline_to_all_callers_1): Update.
26699 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
26701 * profile-count.c (profile_count::combine_with_ipa_count): Handle
26704 2018-01-30 Richard Biener <rguenther@suse.de>
26706 PR tree-optimization/83008
26707 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
26708 invariant and constant vector uses in stmts when they need
26709 more than one stmt.
26711 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26714 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
26715 * configure: Regenerate.
26717 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26719 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
26721 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
26722 Use gen_rtx_REG rather than gen_lowpart.
26724 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26726 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
26727 rather than 0 when creating partial subregs.
26729 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26731 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
26734 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
26737 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
26738 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
26739 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
26740 flags. This restores the settings used before the 2017-07-24.
26741 Turning off pre increment/decrement/modify allows IVOPTS to
26742 optimize DF/SF loops where the index is an int.
26744 2018-01-29 Richard Biener <rguenther@suse.de>
26745 Kelvin Nilsen <kelvin@gcc.gnu.org>
26748 * tree-vect-stmts.c (vectorizable_call): Don't call
26749 targetm.vectorize_builtin_md_vectorized_function if callee is
26752 2018-01-22 Carl Love <cel@us.ibm.com>
26754 * doc/extend.tex: Fix typo in second arg in
26755 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
26757 2018-01-29 Richard Biener <rguenther@suse.de>
26759 PR tree-optimization/84086
26760 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
26761 (flush_ssaname_freelist): When SSA names were released reset
26762 the SCEV hash table.
26764 2018-01-29 Richard Biener <rguenther@suse.de>
26766 PR tree-optimization/84057
26767 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
26768 removed paths when removing edges.
26770 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
26772 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
26773 -mfunction-return=@var{choice}.
26775 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
26777 PR diagnostic/84034
26778 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
26779 Handle CR like TAB.
26780 (layout::print_source_line): Likewise.
26781 (test_get_line_width_without_trailing_whitespace): Add test cases.
26783 2018-01-27 Jakub Jelinek <jakub@redhat.com>
26785 PR middle-end/84040
26786 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
26789 2018-01-26 Jim Wilson <jimw@sifive.com>
26791 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
26793 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
26796 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26798 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
26799 and CMP + SUB-immediate -> SUBS.
26801 2018-01-26 Martin Sebor <msebor@redhat.com>
26803 PR tree-optimization/83896
26804 * tree-ssa-strlen.c (get_string_len): Rename...
26805 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
26806 Avoid assuming length is constant.
26807 (handle_char_store): Use HOST_WIDE_INT for string length.
26809 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
26812 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
26813 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
26815 2018-01-26 Richard Biener <rguenther@suse.de>
26817 PR rtl-optimization/84003
26818 * dse.c (record_store): Only record redundant stores when
26819 the earlier store aliases at least all accesses the later one does.
26821 2018-01-26 Jakub Jelinek <jakub@redhat.com>
26823 PR rtl-optimization/83985
26824 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
26825 REG_CFA_RESTORE insns.
26826 (delete_unmarked_insns): Don't ignore separate shrink wrapping
26827 REG_CFA_RESTORE insns here.
26830 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
26831 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
26833 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26835 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
26836 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
26837 (arc_init): Likewise.
26838 (arc_override_options): Likewise.
26839 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
26841 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
26843 * config/arc/arc.h (TARGET_DBNZ): Define.
26844 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
26845 properly set the tune attribute.
26846 (dbnz): Use TARGET_DBNZ guard.
26847 * config/arc/arc.opt (mtune): Add core3 option.
26849 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26851 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
26852 recognize new pic like addresses.
26853 (arc_delegitimize_address): Clean up.
26855 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26857 * config/arc/arc-arches.def: Option mrf16 valid for all
26859 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
26860 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
26861 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
26862 * config/arc/arc-tables.opt: Regenerate.
26863 * config/arc/arc.c (arc_conditional_register_usage): Handle
26864 reduced register file case.
26865 (arc_file_start): Set must have build attributes.
26866 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
26867 mrf16 option value.
26868 * config/arc/arc.opt (mrf16): Add new option.
26869 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
26870 * config/arc/genmultilib.awk: Handle new mrf16 option.
26871 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
26872 * config/arc/t-multilib: Regenerate.
26873 * doc/invoke.texi (ARC Options): Document mrf16 option.
26875 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26877 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
26878 * config/arc/arc.c (arc_handle_secure_attribute): New function.
26879 (arc_attribute_table): Add 'secure_call' attribute.
26880 (arc_print_operand): Print secure call operand.
26881 (arc_function_ok_for_sibcall): Don't optimize tail calls when
26883 (arc_is_secure_call_p): New function. * config/arc/arc.md
26884 (call_i): Add support for sjli instruction.
26885 (call_value_i): Likewise.
26886 * config/arc/constraints.md (Csc): New constraint.
26888 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26889 John Eric Martin <John.Martin@emmicro-us.com>
26891 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
26892 * config/arc/arc.c (_arc_jli_section): New struct.
26893 (arc_jli_section): New type.
26894 (rc_jli_sections): New static variable.
26895 (arc_handle_jli_attribute): New function.
26896 (arc_attribute_table): Add jli_always and jli_fixed attribute.
26897 (arc_file_end): New function.
26898 (TARGET_ASM_FILE_END): Define.
26899 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
26900 (arc_add_jli_section): New function.
26901 (jli_call_scan): Likewise.
26902 (arc_reorg): Call jli_call_scan.
26903 (arc_output_addsi): Remove 'S' from printing asm operand.
26904 (arc_is_jli_call_p): New function.
26905 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
26907 (movhi_insn): Likewise.
26908 (movsi_insn): Likewise.
26909 (movsi_set_cc_insn): Likewise.
26910 (loadqi_update): Likewise.
26911 (load_zeroextendqisi_update): Likewise.
26912 (load_signextendqisi_update): Likewise.
26913 (loadhi_update): Likewise.
26914 (load_zeroextendhisi_update): Likewise.
26915 (load_signextendhisi_update): Likewise.
26916 (loadsi_update): Likewise.
26917 (loadsf_update): Likewise.
26918 (movsicc_insn): Likewise.
26919 (bset_insn): Likewise.
26920 (bxor_insn): Likewise.
26921 (bclr_insn): Likewise.
26922 (bmsk_insn): Likewise.
26923 (bicsi3_insn): Likewise.
26924 (cmpsi_cc_c_insn): Likewise.
26925 (movsi_ne): Likewise.
26926 (movsi_cond_exec): Likewise.
26927 (clrsbsi2): Likewise.
26928 (norm_f): Likewise.
26938 (call_i): Remove 'S' asm letter, add jli instruction.
26939 (call_value_i): Likewise.
26940 * config/arc/arc.op (mjli-always): New option.
26941 * config/arc/constraints.md (Cji): New constraint.
26942 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
26944 (subsf3_fpx): Likewise.
26945 (mulsf3_fpx): Likewise.
26946 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
26948 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
26949 function attrbutes.
26950 * doc/invoke.texi (ARC): Document mjli-always option.
26952 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
26954 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
26955 avoid addition with 0 and use incw and decw where possible.
26957 2018-01-26 Richard Biener <rguenther@suse.de>
26959 PR tree-optimization/81082
26960 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
26961 association if it requires casting to unsigned.
26962 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
26963 from fold_plusminus_mult_expr to catch important cases late when
26964 range info is available.
26966 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26968 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
26969 * configure.ac (hidden_linkonce): New test.
26970 * configure: Regenerate.
26971 * config.in: Regenerate.
26973 2018-01-26 Julia Koval <julia.koval@intel.com>
26975 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
26976 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
26977 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
26978 _mm_mask_bitshuffle_epi64_mask): Fix type.
26979 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
26980 USI_FTYPE_V4DI_V4DI_USI): Remove.
26981 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
26982 __builtin_ia32_vpshufbitqmb256_mask,
26983 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
26984 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
26985 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
26987 2018-01-26 Alan Modra <amodra@gmail.com>
26990 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
26991 UNSPEC_VBPERMQ. Sort other unspecs.
26993 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
26995 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
26997 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
26999 PR middle-end/83055
27000 * predict.c (drop_profile): Do not push/pop cfun; update also
27002 (handle_missing_profiles): Fix logic looking for zero profiles.
27004 2018-01-25 Jakub Jelinek <jakub@redhat.com>
27006 PR middle-end/83977
27007 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
27008 on functions with #pragma omp declare simd or functions with simd
27010 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
27011 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
27012 Remove trailing \n from warning_at calls.
27014 2018-01-25 Tom de Vries <tom@codesourcery.com>
27017 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
27018 for neutered workers.
27020 2018-01-24 Joseph Myers <joseph@codesourcery.com>
27023 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
27024 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
27026 2018-01-24 Jeff Law <law@redhat.com>
27029 * i386.c (get_probe_interval): Move to earlier point.
27030 (ix86_compute_frame_layout): If -fstack-clash-protection and
27031 the frame is larger than the probe interval, then use pushes
27032 to save registers rather than reg->mem moves.
27033 (ix86_expand_prologue): Remove conditional for int_registers_saved
27036 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
27039 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
27040 min/max for never referenced object.
27042 2018-01-24 Jakub Jelinek <jakub@redhat.com>
27044 PR middle-end/83977
27045 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
27047 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
27048 attributes from DECL_ATTRIBUTES (decl) without affecting
27049 DECL_ATTRIBUTES (current_function_decl).
27050 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
27051 functions with non-NULL DECL_ABSTRACT_ORIGIN.
27053 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
27055 PR tree-optimization/83979
27056 * fold-const.c (fold_comparison): Use constant_boolean_node
27057 instead of boolean_{true,false}_node.
27059 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
27061 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
27064 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27066 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27067 Simplify the clause that sets the length attribute.
27068 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27069 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
27070 clause that sets the length attribute.
27071 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27073 2018-01-24 Tom de Vries <tom@codesourcery.com>
27076 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
27077 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
27078 Add strict parameter.
27079 (prevent_branch_around_nothing): Insert dummy insn between branch to
27080 label and label with no ptx insn inbetween.
27081 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
27083 2018-01-24 Tom de Vries <tom@codesourcery.com>
27086 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
27087 for neutered threads in warp.
27088 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
27090 2018-01-24 Richard Biener <rguenther@suse.de>
27092 PR tree-optimization/83176
27093 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
27096 2018-01-24 Richard Biener <rguenther@suse.de>
27098 PR tree-optimization/82819
27099 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
27100 code generating pluses that are no-ops in the target precision.
27102 2018-01-24 Richard Biener <rguenther@suse.de>
27104 PR middle-end/84000
27105 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
27107 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27109 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
27110 to merge probabilities.
27111 * predict.c (probably_never_executed): Also mark as cold functions
27112 with global 0 profile and guessed local profile.
27113 * profile-count.c (profile_probability::combine_with_count): New
27115 * profile-count.h (profile_probability::operator*,
27116 profile_probability::operator*=, profile_probability::operator/,
27117 profile_probability::operator/=): Reduce precision to adjusted
27118 and set value to guessed on contradictory divisions.
27119 (profile_probability::combine_with_freq): Remove.
27120 (profile_probability::combine_wiht_count): Declare.
27121 (profile_count::force_nonzero):: Set to adjusted.
27122 (profile_count::probability_in):: Set quality to adjusted.
27123 * tree-ssa-tail-merge.c (replace_block_by): Use
27124 combine_with_count.
27126 2018-01-23 Andrew Waterman <andrew@sifive.com>
27127 Jim Wilson <jimw@sifive.com>
27129 * config/riscv/riscv.c (riscv_stack_boundary): New.
27130 (riscv_option_override): Set riscv_stack_boundary. Handle
27131 riscv_preferred_stack_boundary_arg.
27132 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
27133 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
27134 (STACK_BOUNDARY): Set to riscv_stack_boundary.
27135 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
27136 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
27137 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
27139 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
27142 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
27143 of struct ix86_frame.
27144 (ix86_expand_epilogue): Likewise. Add a local variable for
27145 the reg_save_offset field in struct ix86_frame.
27147 2018-01-23 Bin Cheng <bin.cheng@arm.com>
27149 PR tree-optimization/82604
27150 * tree-loop-distribution.c (enum partition_kind): New enum item
27151 PKIND_PARTIAL_MEMSET.
27152 (partition_builtin_p): Support above new enum item.
27153 (generate_code_for_partition): Ditto.
27154 (compute_access_range): Differentiate cases that equality can be
27155 proven at all loops, the innermost loops or no loops.
27156 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
27157 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
27158 (finalize_partitions, distribute_loop): Don't fuse partition of
27159 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
27160 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
27161 parloop is enabled.
27163 2018-01-23 Martin Liska <mliska@suse.cz>
27165 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
27166 order to ignore the predictor.
27167 (PRED_POLYMORPHIC_CALL): Likewise.
27168 (PRED_RECURSIVE_CALL): Likewise.
27170 2018-01-23 Martin Liska <mliska@suse.cz>
27172 * tree-profile.c (tree_profiling): Print function header to
27173 aware reader which function we are working on.
27174 * value-prof.c (gimple_find_values_to_profile): Do not print
27175 not interesting value histograms.
27177 2018-01-23 Martin Liska <mliska@suse.cz>
27179 * profile-count.h (enum profile_quality): Add
27180 profile_uninitialized as the first value. Do not number values
27181 as they are zero based.
27182 (profile_count::verify): Update sanity check.
27183 (profile_probability::verify): Likewise.
27185 2018-01-23 Nathan Sidwell <nathan@acm.org>
27187 * doc/invoke.texi (ffor-scope): Deprecate.
27189 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27191 PR tree-optimization/83510
27192 * domwalk.c (set_all_edges_as_executable): New function.
27193 (dom_walker::dom_walker): Convert bool param
27194 "skip_unreachable_blocks" to enum reachability. Move setup of
27195 edge flags to set_all_edges_as_executable and only do it when
27196 reachability is REACHABLE_BLOCKS.
27197 * domwalk.h (enum dom_walker::reachability): New enum.
27198 (dom_walker::dom_walker): Convert bool param
27199 "skip_unreachable_blocks" to enum reachability.
27200 (set_all_edges_as_executable): New decl.
27201 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
27202 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
27204 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
27205 but converting true to REACHABLE_BLOCKS.
27206 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
27208 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
27209 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
27210 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
27212 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
27213 if check_all_array_refs will be called.
27215 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27217 * tree.c (selftest::test_location_wrappers): Add more test
27220 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27222 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
27223 (selftest::test_bit_in_range): Likewise.
27225 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
27228 * doc/sourcebuild.texi (vect_float): Say that the selector
27229 only describes the situation when -funsafe-math-optimizations is on.
27230 (vect_float_strict): Document.
27232 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
27234 PR tree-optimization/83965
27235 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
27236 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
27237 instead of checking only for a reduction.
27238 (vect_recog_widen_sum_pattern): Likewise.
27240 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27242 * predict.c (probably_never_executed): Only use precise profile info.
27243 (compute_function_frequency): Skip after inlining hack since we now
27244 have quality checking.
27246 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27248 * profile-count.h (profile_probability::very_unlikely,
27249 profile_probability::unlikely, profile_probability::even): Set
27250 precision to guessed.
27252 2018-01-23 Richard Biener <rguenther@suse.de>
27254 PR tree-optimization/83963
27255 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27256 Properly terminate dominator walk when crossing the exit edge not
27257 when visiting its source block.
27259 2018-01-23 Jakub Jelinek <jakub@redhat.com>
27262 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
27263 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
27265 2018-01-22 Jakub Jelinek <jakub@redhat.com>
27267 PR tree-optimization/83957
27268 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
27269 semicolon after for body surrounded by braces.
27271 PR tree-optimization/83081
27272 * profile-count.h (profile_probability::split): New method.
27273 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
27274 Use profile_probability::split.
27275 (do_compare_rtx_and_jump): Fix adjustment of probabilities
27276 when splitting a single conditional jump into 2.
27278 2018-01-22 David Malcolm <dmalcolm@redhat.com>
27280 PR tree-optimization/69452
27281 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
27284 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27286 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
27287 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
27288 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
27290 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27292 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
27294 * config/rl78/rl78.md (movdi): New define_expand.
27295 * config/rl78/rl78.c (rl78_split_movdi): New function.
27297 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27300 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
27302 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
27303 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
27304 128-bit to produce an UNSPEC move to get the double word with the
27305 signbit and then a shift directly to do signbit.
27306 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
27307 implementation with a new version that just does either a direct
27308 move or a regular move. Move memory interface to separate insns.
27309 Move insns so they are next to the expander.
27310 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
27311 with signbit move. Split big and little endian case.
27312 (signbit<mode>2_dm_mem_le): Likewise.
27313 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
27314 (signbit<mode>2_dm2): Likewise.
27316 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27318 * config/rl78/rl78.md (anddi3): New define_expand.
27320 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27322 * config/rl78/rl78.md (umindi3): New define_expand.
27324 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27326 * config/rl78/rl78.md (smindi3): New define_expand.
27328 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27330 * config/rl78/rl78.md (smaxdi3): New define_expand.
27332 2018-01-22 Carl Love <cel@us.ibm.com>
27334 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
27335 LVX_V1TI): Add macro expansion.
27336 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
27337 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
27338 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
27339 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
27340 Change check to determine if the instruction is a byte reversing
27341 entry. Fix typo in comment.
27342 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
27343 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
27344 Add def_builtin calls for new builtins.
27345 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
27346 Add define_insn expansion.
27348 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27350 * config/rl78/rl78.md (umaxdi3): New define_expand.
27352 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27354 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
27355 for non-QImode registers.
27357 2018-01-22 Richard Biener <rguenther@suse.de>
27359 PR tree-optimization/83963
27360 * graphite-scop-detection.c (scop_detection::get_sese): Delay
27361 including the loop exit block.
27362 (scop_detection::merge_sese): Likewise.
27363 (scop_detection::add_scop): Do it here instead.
27365 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27367 * doc/sourcebuild.texi (arm_softfloat): Document.
27369 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
27372 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
27373 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
27374 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
27376 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27377 David Edelsohn <dje.gcc@gmail.com>
27380 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27381 Change "crset eq" to "crset 2".
27382 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27383 (*call_indirect_aix<mode>_nospec): Likewise.
27384 (*call_value_indirect_aix<mode>_nospec): Likewise.
27385 (*call_indirect_elfv2<mode>_nospec): Likewise.
27386 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
27387 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
27388 change assembly output from . to $.
27389 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27390 (indirect_jump<mode>_nospec): Change assembly output from . to $.
27391 (*tablejump<mode>_internal1_nospec): Likewise.
27393 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
27396 * config/sh/sh_optimize_sett_clrt.cc:
27397 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
27399 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
27401 PR tree-optimization/83940
27402 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
27403 offset_dt to vect_constant_def rather than vect_unknown_def_type.
27404 (vect_check_load_store_mask): Add a mask_dt_out parameter and
27405 use it to pass back the definition type.
27406 (vect_check_store_rhs): Likewise rhs_dt_out.
27407 (vect_build_gather_load_calls): Add a mask_dt argument and use
27408 it instead of a call to vect_is_simple_use.
27409 (vectorizable_store): Update calls to vect_check_load_store_mask
27410 and vect_check_store_rhs. Use the dt returned by the latter instead
27411 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
27412 instead of calls to vect_is_simple_use. Pass the scalar rather
27413 than the vector operand to vect_is_simple_use when handling
27414 second and subsequent copies of an rhs value.
27415 (vectorizable_load): Update calls to vect_check_load_store_mask
27416 and vect_build_gather_load_calls. Use the cached mask_dt and
27417 gs_info.offset_dt instead of calls to vect_is_simple_use.
27419 2018-01-20 Jakub Jelinek <jakub@redhat.com>
27421 PR middle-end/83945
27422 * tree-emutls.c: Include gimplify.h.
27423 (lower_emutls_2): New function.
27424 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
27425 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
27426 it before further processing.
27429 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
27430 UINTVAL (trueop1) instead of INTVAL (op1).
27432 2018-01-19 Jakub Jelinek <jakub@redhat.com>
27436 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
27437 INCOMING_FRAME_SP_OFFSET if not defined.
27438 (scan_trace): Add ENTRY argument. If true and
27439 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
27440 emit a note to adjust the CFA offset.
27441 (create_cfi_notes): Adjust scan_trace callers.
27442 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
27443 INCOMING_FRAME_SP_OFFSET in the CIE.
27444 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
27445 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
27447 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
27448 * doc/tm.texi: Regenerated.
27450 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27452 PR rtl-optimization/83147
27453 * lra-constraints.c (remove_inheritance_pseudos): Use
27454 lra_substitute_pseudo_within_insn.
27456 2018-01-19 Tom de Vries <tom@codesourcery.com>
27457 Cesar Philippidis <cesar@codesourcery.com>
27460 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
27462 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
27465 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
27466 spaces for function labels.
27468 2018-01-19 Martin Liska <mliska@suse.cz>
27470 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
27471 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
27472 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
27473 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
27474 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
27475 (PRED_CONST_RETURN): Change from 69 to 65.
27476 (PRED_NULL_RETURN): Change from 91 to 71.
27477 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
27478 (PRED_LOOP_GUARD): Change from 66 to 73.
27480 2018-01-19 Martin Liska <mliska@suse.cz>
27482 * predict.c (predict_insn_def): Add new assert.
27483 (struct branch_predictor): Change type to signed integer.
27484 (test_prediction_value_range): Amend test to cover
27485 PROB_UNINITIALIZED.
27486 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
27487 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
27488 (PRED_LOOP_ITERATIONS_MAX): Likewise.
27489 (PRED_LOOP_IV_COMPARE): Likewise.
27490 * predict.h (PROB_UNINITIALIZED): Define new constant.
27492 2018-01-19 Martin Liska <mliska@suse.cz>
27494 * predict.c (dump_prediction): Add new format for
27495 analyze_brprob.py script which is enabled with -details
27497 * profile-count.h (precise_p): New function.
27499 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
27501 PR tree-optimization/83922
27502 * tree-vect-loop.c (vect_verify_full_masking): Return false if
27503 there are no statements that need masking.
27504 (vect_active_double_reduction_p): New function.
27505 (vect_analyze_loop_operations): Use it when handling phis that
27506 are not in the loop header.
27508 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
27510 PR tree-optimization/83914
27511 * tree-vect-loop.c (vectorizable_induction): Don't convert
27512 init_expr or apply the peeling adjustment for inductions
27513 that are nested within the vectorized loop.
27515 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27517 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
27520 2018-01-18 Jakub Jelinek <jakub@redhat.com>
27524 * function.h (gimplify_parameters): Add gimple_seq * argument.
27525 * function.c: Include gimple.h and options.h.
27526 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
27527 for the added local temporaries if needed.
27528 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
27529 if there are any parameter cleanups, wrap whole body into a
27530 try/finally with the cleanups.
27532 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27535 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27536 Use GET_MODE_CLASS for scalar floating point.
27538 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
27542 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
27543 Fix call of call_cgraph_insertion_hooks.
27545 2018-01-18 Martin Sebor <msebor@redhat.com>
27547 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
27549 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
27552 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
27555 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
27558 * common.opt: (-ffile-prefix-map): New option.
27559 * opts.c (common_handle_option): Defer it.
27560 * opts-global.c (handle_common_deferred_options): Handle it.
27561 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
27562 * file-prefix-map.h: New file.
27563 (remap_debug_filename, add_debug_prefix_map): ...here.
27564 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
27565 * final.c (debug_prefix_map, add_debug_prefix_map
27566 remap_debug_filename): Move to...
27567 * file-prefix-map.c: New file.
27568 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
27569 generalize, get rid of alloca(), use strrchr() instead of strchr().
27570 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
27571 Implement in terms of add_prefix_map().
27572 (remap_macro_filename, remap_debug_filename): Implement in term of
27574 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
27575 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
27576 * dbxout.c: Include file-prefix-map.h.
27577 * varasm.c: Likewise.
27578 * vmsdbgout.c: Likewise.
27579 * xcoffout.c: Likewise.
27580 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
27581 * doc/cppopts.texi (-fmacro-prefix-map): Document.
27582 * doc/invoke.texi (-ffile-prefix-map): Document.
27583 (-fdebug-prefix-map): Update description.
27585 2018-01-18 Martin Liska <mliska@suse.cz>
27587 * config/i386/i386.c (indirect_thunk_name): Document that also
27589 (output_indirect_thunk): Document why both instructions
27590 (pause and lfence) are generated.
27592 2018-01-18 Richard Biener <rguenther@suse.de>
27594 PR tree-optimization/83887
27595 * graphite-scop-detection.c
27596 (scop_detection::get_nearest_dom_with_single_entry): Remove.
27597 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
27598 (scop_detection::merge_sese): Re-implement with a flood-fill
27599 algorithm that properly finds a SESE region if it exists.
27601 2018-01-18 Jakub Jelinek <jakub@redhat.com>
27604 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
27605 pointer_diff optimizations use view_convert instead of convert.
27607 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27609 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27610 Generate different code for -mno-speculate-indirect-jumps.
27611 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27612 (*call_indirect_aix<mode>): Disable for
27613 -mno-speculate-indirect-jumps.
27614 (*call_indirect_aix<mode>_nospec): New define_insn.
27615 (*call_value_indirect_aix<mode>): Disable for
27616 -mno-speculate-indirect-jumps.
27617 (*call_value_indirect_aix<mode>_nospec): New define_insn.
27618 (*sibcall_nonlocal_sysv<mode>): Generate different code for
27619 -mno-speculate-indirect-jumps.
27620 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27622 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
27624 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
27625 long double type, set the flags for noting the default long double
27626 type, even if we don't pass or return a long double type.
27628 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
27631 * ipa-inline.c (flatten_function): Do not overwrite final inlining
27634 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
27636 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
27637 support for merge[hl].
27638 (fold_mergehl_helper): New helper function.
27639 (tree-vector-builder.h): New #include for tree_vector_builder usage.
27640 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
27641 (altivec_vmrglw_direct): Add xxmrglw insn.
27643 2018-01-17 Andrew Waterman <andrew@sifive.com>
27645 * config/riscv/riscv.c (riscv_conditional_register_usage): If
27646 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
27648 2018-01-17 David Malcolm <dmalcolm@redhat.com>
27651 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
27652 call the lto_location_cache before reading the
27653 DECL_SOURCE_LOCATION of the types.
27655 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27656 Richard Sandiford <richard.sandiford@linaro.org>
27658 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
27659 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
27660 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
27661 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
27662 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
27664 * config/aarch64/constraints.md (aarch64_movti_operand):
27666 * config/aarch64/predicates.md (Uti): Add new constraint.
27668 2018-01-17 Carl Love <cel@us.ibm.com>
27670 * config/rs6000/vsx.md (define_expand xl_len_r,
27671 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
27672 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
27674 (define_expand, define_insn): Move the shift left from the
27675 define_insn to the define_expand for lxvl and stxvl instructions.
27676 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
27677 and XL_LEN_R definitions to PURE.
27679 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
27681 * config/i386/i386.c (indirect_thunk_name): Declare regno
27682 as unsigned int. Compare regno with INVALID_REGNUM.
27683 (output_indirect_thunk): Ditto.
27684 (output_indirect_thunk_function): Ditto.
27685 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
27686 in the call to output_indirect_thunk_function.
27688 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
27690 PR middle-end/83884
27691 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
27692 rather than the size of inner_type to determine the stack slot size
27693 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
27695 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
27698 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
27701 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27703 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
27704 endian Linux systems to optionally enable multilibs for selecting
27705 the long double type if the user configured an explicit type.
27706 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
27707 have no long double multilibs if not defined.
27708 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
27709 warn if the user used -mabi={ieee,ibm}longdouble and we built
27710 multilibs for long double.
27711 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
27712 appropriate multilib option.
27713 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
27715 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
27716 for building long double multilibs.
27717 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
27719 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
27721 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
27724 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
27726 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
27729 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
27732 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
27735 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
27737 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
27738 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
27740 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
27742 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
27743 different rtl trees depending on TARGET_64BIT.
27744 (rs6000_gen_lvx): Likewise.
27746 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
27748 * config/visium/visium.md (nop): Tweak comment.
27749 (hazard_nop): Likewise.
27751 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27753 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
27754 -mspeculate-indirect-jumps.
27755 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
27756 for -mno-speculate-indirect-jumps.
27757 (*call_indirect_elfv2<mode>_nospec): New define_insn.
27758 (*call_value_indirect_elfv2<mode>): Disable for
27759 -mno-speculate-indirect-jumps.
27760 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
27761 (indirect_jump): Emit different RTL for
27762 -mno-speculate-indirect-jumps.
27763 (*indirect_jump<mode>): Disable for
27764 -mno-speculate-indirect-jumps.
27765 (*indirect_jump<mode>_nospec): New define_insn.
27766 (tablejump): Emit different RTL for
27767 -mno-speculate-indirect-jumps.
27768 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
27769 (tablejumpsi_nospec): New define_expand.
27770 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
27771 (tablejumpdi_nospec): New define_expand.
27772 (*tablejump<mode>_internal1): Disable for
27773 -mno-speculate-indirect-jumps.
27774 (*tablejump<mode>_internal1_nospec): New define_insn.
27775 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
27778 2018-01-16 Artyom Skrobov tyomitch@gmail.com
27780 * caller-save.c (insert_save): Drop unnecessary parameter. All
27783 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27784 Richard Biener <rguenth@suse.de>
27787 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
27788 return early, inline manually is_gimple_sizepos. Make sure if we
27789 call gimplify_expr we don't end up with a gimple constant.
27790 * tree.c (variably_modified_type_p): Don't return true for
27791 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
27792 * gimplify.h (is_gimple_sizepos): Remove.
27794 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27796 PR tree-optimization/83857
27797 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
27798 vectorizable_live_operation for pure SLP statements.
27799 (vectorizable_live_operation): Handle PHIs.
27801 2018-01-16 Richard Biener <rguenther@suse.de>
27803 PR tree-optimization/83867
27804 * tree-vect-stmts.c (vect_transform_stmt): Precompute
27805 nested_in_vect_loop_p since the scalar stmt may get invalidated.
27807 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27810 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
27811 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
27812 If off is not INTEGER_CST, issue a may not be aligned warning
27813 rather than isn't aligned. Use isn%'t rather than isn't.
27814 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
27816 <case MULT_EXPR>: Improve the case when bottom and one of the
27817 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
27818 operand, in that case check if the other operand is multiple of
27819 bottom divided by the INTEGER_CST operand.
27821 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27824 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
27825 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
27826 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
27827 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
27828 * config/pa/pa.c (pa_function_arg_advance): Likewise.
27829 (pa_function_arg, pa_arg_partial_bytes): Likewise.
27830 (pa_function_arg_size): New function.
27832 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27834 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
27835 in a separate statement.
27837 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27839 PR tree-optimization/83847
27840 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
27841 group gathers and scatters.
27843 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27845 PR rtl-optimization/86620
27846 * params.def (max-sched-ready-insns): Bump minimum value to 1.
27848 PR rtl-optimization/83213
27849 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
27850 to last if both are JUMP_INSNs.
27852 PR tree-optimization/83843
27853 * gimple-ssa-store-merging.c
27854 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
27855 store_immediate_info for bswap/nop orig_stores.
27857 2018-01-15 Andrew Waterman <andrew@sifive.com>
27859 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
27861 <UDIV>: Increase cost if !TARGET_DIV.
27863 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27865 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
27866 (define_attr "cr_logical_3op"): New.
27867 (cceq_ior_compare): Adjust.
27868 (cceq_ior_compare_complement): Adjust.
27869 (*cceq_rev_compare): Adjust.
27870 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
27871 (is_cracked_insn): Adjust.
27872 (insn_must_be_first_in_group): Adjust.
27873 * config/rs6000/40x.md: Adjust.
27874 * config/rs6000/440.md: Adjust.
27875 * config/rs6000/476.md: Adjust.
27876 * config/rs6000/601.md: Adjust.
27877 * config/rs6000/603.md: Adjust.
27878 * config/rs6000/6xx.md: Adjust.
27879 * config/rs6000/7450.md: Adjust.
27880 * config/rs6000/7xx.md: Adjust.
27881 * config/rs6000/8540.md: Adjust.
27882 * config/rs6000/cell.md: Adjust.
27883 * config/rs6000/e300c2c3.md: Adjust.
27884 * config/rs6000/e500mc.md: Adjust.
27885 * config/rs6000/e500mc64.md: Adjust.
27886 * config/rs6000/e5500.md: Adjust.
27887 * config/rs6000/e6500.md: Adjust.
27888 * config/rs6000/mpc.md: Adjust.
27889 * config/rs6000/power4.md: Adjust.
27890 * config/rs6000/power5.md: Adjust.
27891 * config/rs6000/power6.md: Adjust.
27892 * config/rs6000/power7.md: Adjust.
27893 * config/rs6000/power8.md: Adjust.
27894 * config/rs6000/power9.md: Adjust.
27895 * config/rs6000/rs64.md: Adjust.
27896 * config/rs6000/titan.md: Adjust.
27898 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27900 * config/i386/predicates.md (indirect_branch_operand): Rewrite
27901 ix86_indirect_branch_register logic.
27903 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27905 * config/i386/constraints.md (Bs): Update
27906 ix86_indirect_branch_register check. Don't check
27907 ix86_indirect_branch_register with GOT_memory_operand.
27909 * config/i386/predicates.md (GOT_memory_operand): Don't check
27910 ix86_indirect_branch_register here.
27911 (GOT32_symbol_operand): Likewise.
27913 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27915 * config/i386/predicates.md (constant_call_address_operand):
27916 Rewrite ix86_indirect_branch_register logic.
27917 (sibcall_insn_operand): Likewise.
27919 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27921 * config/i386/constraints.md (Bs): Replace
27922 ix86_indirect_branch_thunk_register with
27923 ix86_indirect_branch_register.
27925 * config/i386/i386.md (indirect_jump): Likewise.
27926 (tablejump): Likewise.
27927 (*sibcall_memory): Likewise.
27928 (*sibcall_value_memory): Likewise.
27929 Peepholes of indirect call and jump via memory: Likewise.
27930 * config/i386/i386.opt: Likewise.
27931 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27932 (GOT_memory_operand): Likewise.
27933 (call_insn_operand): Likewise.
27934 (sibcall_insn_operand): Likewise.
27935 (GOT32_symbol_operand): Likewise.
27937 2018-01-15 Jakub Jelinek <jakub@redhat.com>
27939 PR middle-end/83837
27940 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
27941 type rather than type addr's type points to.
27942 (expand_omp_atomic_mutex): Likewise.
27943 (expand_omp_atomic): Likewise.
27945 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27948 * config/i386/i386.c (output_indirect_thunk_function): Use
27949 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
27950 for __x86_return_thunk.
27952 2018-01-15 Richard Biener <rguenther@suse.de>
27954 PR middle-end/83850
27955 * expmed.c (extract_bit_field_1): Fix typo.
27957 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27960 * config/arm/iterators.md (VF): New mode iterator.
27961 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
27962 Remove integer-related logic from pattern.
27963 (neon_vabd<mode>_3): Likewise.
27965 2018-01-15 Jakub Jelinek <jakub@redhat.com>
27967 PR middle-end/82694
27968 * common.opt (fstrict-overflow): No longer an alias.
27969 (fwrapv-pointer): New option.
27970 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
27971 also for pointer types based on flag_wrapv_pointer.
27972 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
27973 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
27974 opts->x_flag_wrapv got set.
27975 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
27976 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
27977 POINTER_TYPE_OVERFLOW_UNDEFINED.
27978 * match.pd: Likewise in address comparison pattern.
27979 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
27981 2018-01-15 Richard Biener <rguenther@suse.de>
27984 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
27985 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
27986 Reset type names to their identifier if their TYPE_DECL doesn't
27987 have linkage (and thus is used for ODR and devirt).
27988 (save_debug_info_for_decl): Remove.
27989 (save_debug_info_for_type): Likewise.
27990 (add_tree_to_fld_list): Adjust.
27991 * tree-pretty-print.c (dump_generic_node): Make dumping of
27992 type names more robust.
27994 2018-01-15 Richard Biener <rguenther@suse.de>
27996 * BASE-VER: Bump to 8.0.1.
27998 2018-01-14 Martin Sebor <msebor@redhat.com>
28001 * builtins.c (check_access): Avoid warning when the no-warning bit
28004 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
28006 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
28007 * ira-color (allocno_hard_regs_compare): Likewise.
28009 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
28012 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
28013 Use .pushsection/.popsection.
28015 2018-01-14 Martin Sebor <msebor@redhat.com>
28018 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
28020 2018-01-14 Jakub Jelinek <jakub@redhat.com>
28022 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
28023 entry from extra_headers.
28024 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
28025 extra_headers, make the list bitwise identical to the i?86-*-* one.
28027 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28029 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
28030 -mcmodel=large with -mindirect-branch=thunk,
28031 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
28032 -mfunction-return=thunk-extern.
28033 * doc/invoke.texi: Document -mcmodel=large is incompatible with
28034 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
28035 -mfunction-return=thunk and -mfunction-return=thunk-extern.
28037 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28039 * config/i386/i386.c (print_reg): Print the name of the full
28040 integer register without '%'.
28041 (ix86_print_operand): Handle 'V'.
28042 * doc/extend.texi: Document 'V' modifier.
28044 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28046 * config/i386/constraints.md (Bs): Disallow memory operand for
28047 -mindirect-branch-register.
28049 * config/i386/predicates.md (indirect_branch_operand): Likewise.
28050 (GOT_memory_operand): Likewise.
28051 (call_insn_operand): Likewise.
28052 (sibcall_insn_operand): Likewise.
28053 (GOT32_symbol_operand): Likewise.
28054 * config/i386/i386.md (indirect_jump): Call convert_memory_address
28055 for -mindirect-branch-register.
28056 (tablejump): Likewise.
28057 (*sibcall_memory): Likewise.
28058 (*sibcall_value_memory): Likewise.
28059 Disallow peepholes of indirect call and jump via memory for
28060 -mindirect-branch-register.
28061 (*call_pop): Replace m with Bw.
28062 (*call_value_pop): Likewise.
28063 (*sibcall_pop_memory): Replace m with Bs.
28064 * config/i386/i386.opt (mindirect-branch-register): New option.
28065 * doc/invoke.texi: Document -mindirect-branch-register option.
28067 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28069 * config/i386/i386-protos.h (ix86_output_function_return): New.
28070 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
28071 set function_return_type.
28072 (indirect_thunk_name): Add ret_p to indicate thunk for function
28074 (output_indirect_thunk_function): Pass false to
28075 indirect_thunk_name.
28076 (ix86_output_indirect_branch_via_reg): Likewise.
28077 (ix86_output_indirect_branch_via_push): Likewise.
28078 (output_indirect_thunk_function): Create alias for function
28079 return thunk if regno < 0.
28080 (ix86_output_function_return): New function.
28081 (ix86_handle_fndecl_attribute): Handle function_return.
28082 (ix86_attribute_table): Add function_return.
28083 * config/i386/i386.h (machine_function): Add
28084 function_return_type.
28085 * config/i386/i386.md (simple_return_internal): Use
28086 ix86_output_function_return.
28087 (simple_return_internal_long): Likewise.
28088 * config/i386/i386.opt (mfunction-return=): New option.
28089 (indirect_branch): Mention -mfunction-return=.
28090 * doc/extend.texi: Document function_return function attribute.
28091 * doc/invoke.texi: Document -mfunction-return= option.
28093 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28095 * config/i386/i386-opts.h (indirect_branch): New.
28096 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
28097 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
28098 with local indirect jump when converting indirect call and jump.
28099 (ix86_set_indirect_branch_type): New.
28100 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
28101 (indirectlabelno): New.
28102 (indirect_thunk_needed): Likewise.
28103 (indirect_thunk_bnd_needed): Likewise.
28104 (indirect_thunks_used): Likewise.
28105 (indirect_thunks_bnd_used): Likewise.
28106 (INDIRECT_LABEL): Likewise.
28107 (indirect_thunk_name): Likewise.
28108 (output_indirect_thunk): Likewise.
28109 (output_indirect_thunk_function): Likewise.
28110 (ix86_output_indirect_branch_via_reg): Likewise.
28111 (ix86_output_indirect_branch_via_push): Likewise.
28112 (ix86_output_indirect_branch): Likewise.
28113 (ix86_output_indirect_jmp): Likewise.
28114 (ix86_code_end): Call output_indirect_thunk_function if needed.
28115 (ix86_output_call_insn): Call ix86_output_indirect_branch if
28117 (ix86_handle_fndecl_attribute): Handle indirect_branch.
28118 (ix86_attribute_table): Add indirect_branch.
28119 * config/i386/i386.h (machine_function): Add indirect_branch_type
28120 and has_local_indirect_jump.
28121 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
28123 (tablejump): Likewise.
28124 (*indirect_jump): Use ix86_output_indirect_jmp.
28125 (*tablejump_1): Likewise.
28126 (simple_return_indirect_internal): Likewise.
28127 * config/i386/i386.opt (mindirect-branch=): New option.
28128 (indirect_branch): New.
28131 (thunk-inline): Likewise.
28132 (thunk-extern): Likewise.
28133 * doc/extend.texi: Document indirect_branch function attribute.
28134 * doc/invoke.texi: Document -mindirect-branch= option.
28136 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
28139 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
28141 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
28143 * ipa-inline.c (want_inline_small_function_p): Return false if
28144 inlining has already failed with CIF_FINAL_ERROR.
28145 (update_caller_keys): Call want_inline_small_function_p before
28147 (update_callee_keys): Likewise.
28149 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28151 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
28153 (rs6000_quadword_masked_address_p): Likewise.
28154 (quad_aligned_load_p): Likewise.
28155 (quad_aligned_store_p): Likewise.
28156 (const_load_sequence_p): Add comment to describe the outer-most loop.
28157 (mimic_memory_attributes_and_flags): New function.
28158 (rs6000_gen_stvx): Likewise.
28159 (replace_swapped_aligned_store): Likewise.
28160 (rs6000_gen_lvx): Likewise.
28161 (replace_swapped_aligned_load): Likewise.
28162 (replace_swapped_load_constant): Capitalize argument name in
28163 comment describing this function.
28164 (rs6000_analyze_swaps): Add a third pass to search for vector loads
28165 and stores that access quad-word aligned addresses and replace
28166 with stvx or lvx instructions when appropriate.
28167 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
28168 New function prototype.
28169 (rs6000_quadword_masked_address_p): Likewise.
28170 (rs6000_gen_lvx): Likewise.
28171 (rs6000_gen_stvx): Likewise.
28172 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
28173 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
28174 when memory address is aligned.
28175 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
28176 this split to select lvx instruction when memory address is aligned.
28177 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
28178 instruction when memory address is aligned.
28179 (*vsx_le_perm_load_v16qi): Likewise.
28180 (four unnamed splitters): Modify to select the stvx instruction
28181 when memory is aligned.
28183 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
28185 * predict.c (determine_unlikely_bbs): Handle correctly BBs
28186 which appears in the queue multiple times.
28188 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28189 Alan Hayward <alan.hayward@arm.com>
28190 David Sherwood <david.sherwood@arm.com>
28192 * tree-vectorizer.h (vec_lower_bound): New structure.
28193 (_loop_vec_info): Add check_nonzero and lower_bounds.
28194 (LOOP_VINFO_CHECK_NONZERO): New macro.
28195 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
28196 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
28197 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
28198 fields. Make seg_len the distance travelled, not including the
28200 (dr_direction_indicator): Declare.
28201 (dr_zero_step_indicator): Likewise.
28202 (dr_known_forward_stride_p): Likewise.
28203 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
28205 (runtime_alias_check_p): Allow runtime alias checks with
28207 (operator ==): Compare access_size and align.
28208 (prune_runtime_alias_test_list): Rework for new distinction between
28209 the access_size and seg_len.
28210 (create_intersect_range_checks_index): Likewise. Cope with polynomial
28212 (get_segment_min_max): New function.
28213 (create_intersect_range_checks): Use it.
28214 (dr_step_indicator): New function.
28215 (dr_direction_indicator): Likewise.
28216 (dr_zero_step_indicator): Likewise.
28217 (dr_known_forward_stride_p): Likewise.
28218 * tree-loop-distribution.c (data_ref_segment_size): Return
28219 DR_STEP * (niters - 1).
28220 (compute_alias_check_pairs): Update call to the dr_with_seg_len
28222 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
28223 (vect_preserves_scalar_order_p): New function, split out from...
28224 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
28225 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
28226 (vect_vfa_access_size): New function.
28227 (vect_vfa_align): Likewise.
28228 (vect_compile_time_alias): Take access_size_a and access_b arguments.
28229 (dump_lower_bound): New function.
28230 (vect_check_lower_bound): Likewise.
28231 (vect_small_gap_p): Likewise.
28232 (vectorizable_with_step_bound_p): Likewise.
28233 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
28234 depencies if the vectorization factor is 1. Convert the checks
28235 for nonzero steps into checks on the bounds of DR_STEP. Try using
28236 a bunds check for variable steps if the minimum required step is
28237 relatively small. Update calls to the dr_with_seg_len
28238 constructor and to vect_compile_time_alias.
28239 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
28241 (vect_loop_versioning): Call it.
28242 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
28244 (vect_estimate_min_profitable_iters): Account for any bounds checks.
28246 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28247 Alan Hayward <alan.hayward@arm.com>
28248 David Sherwood <david.sherwood@arm.com>
28250 * doc/sourcebuild.texi (vect_scatter_store): Document.
28251 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
28253 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
28255 * genopinit.c (main): Add supports_vec_scatter_store and
28256 supports_vec_scatter_store_cached to target_optabs.
28257 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
28258 IFN_MASK_SCATTER_STORE.
28259 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
28261 * internal-fn.h (internal_store_fn_p): Declare.
28262 (internal_fn_stored_value_index): Likewise.
28263 * internal-fn.c (scatter_store_direct): New macro.
28264 (expand_scatter_store_optab_fn): New function.
28265 (direct_scatter_store_optab_supported_p): New macro.
28266 (internal_store_fn_p): New function.
28267 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
28268 IFN_MASK_SCATTER_STORE.
28269 (internal_fn_mask_index): Likewise.
28270 (internal_fn_stored_value_index): New function.
28271 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
28272 for scatter stores.
28273 * optabs-query.h (supports_vec_scatter_store_p): Declare.
28274 * optabs-query.c (supports_vec_scatter_store_p): New function.
28275 * tree-vectorizer.h (vect_get_store_rhs): Declare.
28276 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
28277 true for scatter stores.
28278 (vect_gather_scatter_fn_p): Handle scatter stores too.
28279 (vect_check_gather_scatter): Consider using scatter stores if
28280 supports_vec_scatter_store_p.
28281 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
28282 scatter stores too.
28283 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28284 internal_fn_stored_value_index.
28285 (check_load_store_masking): Handle scatter stores too.
28286 (vect_get_store_rhs): Make public.
28287 (vectorizable_call): Use internal_store_fn_p.
28288 (vectorizable_store): Handle scatter store internal functions.
28289 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
28290 when deciding whether the end of the group has been reached.
28291 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
28292 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
28293 (mask_scatter_store<mode>): New insns.
28295 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28296 Alan Hayward <alan.hayward@arm.com>
28297 David Sherwood <david.sherwood@arm.com>
28299 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
28300 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
28301 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
28303 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
28304 Use vect_truncate_gather_scatter_offset if we can't treat the
28305 operation as a normal gather load or scatter store.
28306 (get_group_load_store_type): Take the gather_scatter_info
28307 as argument. Try using a gather load or scatter store for
28308 single-element groups.
28309 (get_load_store_type): Update calls to get_group_load_store_type
28310 and vect_use_strided_gather_scatters_p.
28312 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28313 Alan Hayward <alan.hayward@arm.com>
28314 David Sherwood <david.sherwood@arm.com>
28316 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
28317 optional tree argument.
28318 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
28320 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
28321 but continue to use the current value as a fallback.
28322 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
28323 to compare the updates.
28324 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
28325 (get_load_store_type): Use it when handling a strided access.
28326 (vect_get_strided_load_store_ops): New function.
28327 (vect_get_data_ptr_increment): Likewise.
28328 (vectorizable_load): Handle strided gather loads. Always pass
28329 a step to vect_create_data_ref_ptr and bump_vector_ptr.
28331 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28332 Alan Hayward <alan.hayward@arm.com>
28333 David Sherwood <david.sherwood@arm.com>
28335 * doc/md.texi (gather_load@var{m}): Document.
28336 (mask_gather_load@var{m}): Likewise.
28337 * genopinit.c (main): Add supports_vec_gather_load and
28338 supports_vec_gather_load_cached to target_optabs.
28339 * optabs-tree.c (init_tree_optimization_optabs): Use
28340 ggc_cleared_alloc to allocate target_optabs.
28341 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
28342 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
28344 * internal-fn.h (internal_load_fn_p): Declare.
28345 (internal_gather_scatter_fn_p): Likewise.
28346 (internal_fn_mask_index): Likewise.
28347 (internal_gather_scatter_fn_supported_p): Likewise.
28348 * internal-fn.c (gather_load_direct): New macro.
28349 (expand_gather_load_optab_fn): New function.
28350 (direct_gather_load_optab_supported_p): New macro.
28351 (direct_internal_fn_optab): New function.
28352 (internal_load_fn_p): Likewise.
28353 (internal_gather_scatter_fn_p): Likewise.
28354 (internal_fn_mask_index): Likewise.
28355 (internal_gather_scatter_fn_supported_p): Likewise.
28356 * optabs-query.c (supports_at_least_one_mode_p): New function.
28357 (supports_vec_gather_load_p): Likewise.
28358 * optabs-query.h (supports_vec_gather_load_p): Declare.
28359 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
28360 and memory_type field.
28361 (NUM_PATTERNS): Bump to 15.
28362 * tree-vect-data-refs.c: Include internal-fn.h.
28363 (vect_gather_scatter_fn_p): New function.
28364 (vect_describe_gather_scatter_call): Likewise.
28365 (vect_check_gather_scatter): Try using internal functions for
28366 gather loads. Recognize existing calls to a gather load function.
28367 (vect_analyze_data_refs): Consider using gather loads if
28368 supports_vec_gather_load_p.
28369 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
28370 (vect_get_gather_scatter_offset_type): Likewise.
28371 (vect_convert_mask_for_vectype): Likewise.
28372 (vect_add_conversion_to_patterm): Likewise.
28373 (vect_try_gather_scatter_pattern): Likewise.
28374 (vect_recog_gather_scatter_pattern): New pattern recognizer.
28375 (vect_vect_recog_func_ptrs): Add it.
28376 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28377 internal_fn_mask_index and internal_gather_scatter_fn_p.
28378 (check_load_store_masking): Take the gather_scatter_info as an
28379 argument and handle gather loads.
28380 (vect_get_gather_scatter_ops): New function.
28381 (vectorizable_call): Check internal_load_fn_p.
28382 (vectorizable_load): Likewise. Handle gather load internal
28384 (vectorizable_store): Update call to check_load_store_masking.
28385 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
28386 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
28387 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
28388 (aarch64_gather_scale_operand_d): New predicates.
28389 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
28390 (mask_gather_load<mode>): New insns.
28392 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28393 Alan Hayward <alan.hayward@arm.com>
28394 David Sherwood <david.sherwood@arm.com>
28396 * optabs.def (fold_left_plus_optab): New optab.
28397 * doc/md.texi (fold_left_plus_@var{m}): Document.
28398 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
28399 * internal-fn.c (fold_left_direct): Define.
28400 (expand_fold_left_optab_fn): Likewise.
28401 (direct_fold_left_optab_supported_p): Likewise.
28402 * fold-const-call.c (fold_const_fold_left): New function.
28403 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
28404 * tree-parloops.c (valid_reduction_p): New function.
28405 (gather_scalar_reductions): Use it.
28406 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
28407 (vect_finish_replace_stmt): Declare.
28408 * tree-vect-loop.c (fold_left_reduction_fn): New function.
28409 (needs_fold_left_reduction_p): New function, split out from...
28410 (vect_is_simple_reduction): ...here. Accept reductions that
28411 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
28412 (vect_force_simple_reduction): Also store the reduction type in
28413 the assignment's STMT_VINFO_REDUC_TYPE.
28414 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
28415 (merge_with_identity): New function.
28416 (vect_expand_fold_left): Likewise.
28417 (vectorize_fold_left_reduction): Likewise.
28418 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
28419 scalar phi in place for it. Check for target support and reject
28420 cases that would reassociate the operation. Defer the transform
28421 phase to vectorize_fold_left_reduction.
28422 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
28423 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
28424 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
28426 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28428 * tree-if-conv.c (predicate_mem_writes): Remove redundant
28429 call to ifc_temp_var.
28431 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28432 Alan Hayward <alan.hayward@arm.com>
28433 David Sherwood <david.sherwood@arm.com>
28435 * target.def (legitimize_address_displacement): Take the original
28436 offset as a poly_int.
28437 * targhooks.h (default_legitimize_address_displacement): Update
28439 * targhooks.c (default_legitimize_address_displacement): Likewise.
28440 * doc/tm.texi: Regenerate.
28441 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
28442 as an argument, moving assert of ad->disp == ad->disp_term to...
28443 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
28444 Try calling targetm.legitimize_address_displacement before expanding
28445 the address rather than afterwards, and adjust for the new interface.
28446 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
28447 Match the new hook interface. Handle SVE addresses.
28448 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
28449 new hook interface.
28451 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28453 * Makefile.in (OBJS): Add early-remat.o.
28454 * target.def (select_early_remat_modes): New hook.
28455 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
28456 * doc/tm.texi: Regenerate.
28457 * targhooks.h (default_select_early_remat_modes): Declare.
28458 * targhooks.c (default_select_early_remat_modes): New function.
28459 * timevar.def (TV_EARLY_REMAT): New timevar.
28460 * passes.def (pass_early_remat): New pass.
28461 * tree-pass.h (make_pass_early_remat): Declare.
28462 * early-remat.c: New file.
28463 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
28465 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
28467 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28468 Alan Hayward <alan.hayward@arm.com>
28469 David Sherwood <david.sherwood@arm.com>
28471 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
28472 vfm1 with a bound_epilog parameter.
28473 (vect_do_peeling): Update calls accordingly, and move the prologue
28474 call earlier in the function. Treat the base bound_epilog as 0 for
28475 fully-masked loops and retain vf - 1 for other loops. Add 1 to
28476 this base when peeling for gaps.
28477 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
28478 with fully-masked loops.
28479 (vect_estimate_min_profitable_iters): Handle the single peeled
28480 iteration in that case.
28482 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28483 Alan Hayward <alan.hayward@arm.com>
28484 David Sherwood <david.sherwood@arm.com>
28486 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
28487 single-element interleaving even if the size is not a power of 2.
28488 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
28489 accesses for single-element interleaving if the group size is
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 * doc/md.texi (fold_extract_last_@var{m}): Document.
28497 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
28498 * optabs.def (fold_extract_last_optab): New optab.
28499 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
28500 * internal-fn.c (fold_extract_direct): New macro.
28501 (expand_fold_extract_optab_fn): Likewise.
28502 (direct_fold_extract_optab_supported_p): Likewise.
28503 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
28504 * tree-vect-loop.c (vect_model_reduction_cost): Handle
28505 EXTRACT_LAST_REDUCTION.
28506 (get_initial_def_for_reduction): Do not create an initial vector
28507 for EXTRACT_LAST_REDUCTION reductions.
28508 (vectorizable_reduction): Leave the scalar phi in place for
28509 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
28510 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
28511 epilogue code for EXTRACT_LAST_REDUCTION and defer the
28512 transform phase to vectorizable_condition.
28513 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
28515 (vect_finish_stmt_generation): ...here.
28516 (vect_finish_replace_stmt): New function.
28517 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
28518 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
28520 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
28522 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28523 Alan Hayward <alan.hayward@arm.com>
28524 David Sherwood <david.sherwood@arm.com>
28526 * doc/md.texi (extract_last_@var{m}): Document.
28527 * optabs.def (extract_last_optab): New optab.
28528 * internal-fn.def (EXTRACT_LAST): New internal function.
28529 * internal-fn.c (cond_unary_direct): New macro.
28530 (expand_cond_unary_optab_fn): Likewise.
28531 (direct_cond_unary_optab_supported_p): Likewise.
28532 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
28533 loops using EXTRACT_LAST.
28534 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
28535 (extract_last_<mode>): ...this optab.
28536 (vec_extract<mode><Vel>): Update accordingly.
28538 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28539 Alan Hayward <alan.hayward@arm.com>
28540 David Sherwood <david.sherwood@arm.com>
28542 * target.def (empty_mask_is_expensive): New hook.
28543 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
28544 * doc/tm.texi: Regenerate.
28545 * targhooks.h (default_empty_mask_is_expensive): Declare.
28546 * targhooks.c (default_empty_mask_is_expensive): New function.
28547 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
28548 if the target says that empty masks are expensive.
28549 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
28551 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
28553 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28554 Alan Hayward <alan.hayward@arm.com>
28555 David Sherwood <david.sherwood@arm.com>
28557 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
28558 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
28559 (vect_use_loop_mask_for_alignment_p): New function.
28560 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
28561 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
28562 niters_skip argument. Make sure that the first niters_skip elements
28563 of the first iteration are inactive.
28564 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
28565 Update call to vect_set_loop_masks_directly.
28566 (get_misalign_in_elems): New function, split out from...
28567 (vect_gen_prolog_loop_niters): ...here.
28568 (vect_update_init_of_dr): Take a code argument that specifies whether
28569 the adjustment should be added or subtracted.
28570 (vect_update_init_of_drs): Likewise.
28571 (vect_prepare_for_masked_peels): New function.
28572 (vect_do_peeling): Skip prologue peeling if we're using a mask
28573 instead. Update call to vect_update_inits_of_drs.
28574 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28576 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
28577 alignment. Do not include the number of peeled iterations in
28578 the minimum threshold in that case.
28579 (vectorizable_induction): Adjust the start value down by
28580 LOOP_VINFO_MASK_SKIP_NITERS iterations.
28581 (vect_transform_loop): Call vect_prepare_for_masked_peels.
28582 Take the number of skipped iterations into account when calculating
28584 * tree-vect-stmts.c (vect_gen_while_not): New function.
28586 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28587 Alan Hayward <alan.hayward@arm.com>
28588 David Sherwood <david.sherwood@arm.com>
28590 * doc/sourcebuild.texi (vect_fully_masked): Document.
28591 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
28592 default value to 0.
28593 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
28595 (vect_analyze_loop_2): ...here. Don't check the vectorization
28596 factor against the number of loop iterations if the loop is
28599 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28600 Alan Hayward <alan.hayward@arm.com>
28601 David Sherwood <david.sherwood@arm.com>
28603 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
28604 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
28605 (dump_groups): Update accordingly.
28606 (iv_use::mem_type): New member variable.
28607 (address_p): New function.
28608 (record_use): Add a mem_type argument and initialize the new
28610 (record_group_use): Add a mem_type argument. Use address_p.
28611 Remove obsolete null checks of base_object. Update call to record_use.
28612 (find_interesting_uses_op): Update call to record_group_use.
28613 (find_interesting_uses_cond): Likewise.
28614 (find_interesting_uses_address): Likewise.
28615 (get_mem_type_for_internal_fn): New function.
28616 (find_address_like_use): Likewise.
28617 (find_interesting_uses_stmt): Try find_address_like_use before
28618 calling find_interesting_uses_op.
28619 (addr_offset_valid_p): Use the iv mem_type field as the type
28620 of the addressed memory.
28621 (add_autoinc_candidates): Likewise.
28622 (get_address_cost): Likewise.
28623 (split_small_address_groups_p): Use address_p.
28624 (split_address_groups): Likewise.
28625 (add_iv_candidate_for_use): Likewise.
28626 (autoinc_possible_for_pair): Likewise.
28627 (rewrite_groups): Likewise.
28628 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
28629 (determine_group_iv_cost): Update after split of USE_ADDRESS.
28630 (get_alias_ptr_type_for_ptr_address): New function.
28631 (rewrite_use_address): Rewrite address uses in calls that were
28632 identified by find_address_like_use.
28634 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28635 Alan Hayward <alan.hayward@arm.com>
28636 David Sherwood <david.sherwood@arm.com>
28638 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
28640 * gimple-expr.h (is_gimple_addressable: Likewise.
28641 * gimple-expr.c (is_gimple_address): Likewise.
28642 * internal-fn.c (expand_call_mem_ref): New function.
28643 (expand_mask_load_optab_fn): Use it.
28644 (expand_mask_store_optab_fn): Likewise.
28646 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28647 Alan Hayward <alan.hayward@arm.com>
28648 David Sherwood <david.sherwood@arm.com>
28650 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
28651 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
28652 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
28653 (cond_umax@var{mode}): Document.
28654 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
28655 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
28656 (cond_umin_optab, cond_umax_optab): New optabs.
28657 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
28658 (COND_IOR, COND_XOR): New internal functions.
28659 * internal-fn.h (get_conditional_internal_fn): Declare.
28660 * internal-fn.c (cond_binary_direct): New macro.
28661 (expand_cond_binary_optab_fn): Likewise.
28662 (direct_cond_binary_optab_supported_p): Likewise.
28663 (get_conditional_internal_fn): New function.
28664 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
28665 Cope with reduction statements that are vectorized as calls rather
28667 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
28668 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
28669 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
28670 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
28671 (UNSPEC_COND_EOR): New unspecs.
28672 (optab): Add mappings for them.
28673 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
28674 (sve_int_op, sve_fp_op): New int attributes.
28676 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28677 Alan Hayward <alan.hayward@arm.com>
28678 David Sherwood <david.sherwood@arm.com>
28680 * optabs.def (while_ult_optab): New optab.
28681 * doc/md.texi (while_ult@var{m}@var{n}): Document.
28682 * internal-fn.def (WHILE_ULT): New internal function.
28683 * internal-fn.h (direct_internal_fn_supported_p): New override
28684 that takes two types as argument.
28685 * internal-fn.c (while_direct): New macro.
28686 (expand_while_optab_fn): New function.
28687 (convert_optab_supported_p): Likewise.
28688 (direct_while_optab_supported_p): New macro.
28689 * wide-int.h (wi::udiv_ceil): New function.
28690 * tree-vectorizer.h (rgroup_masks): New structure.
28691 (vec_loop_masks): New typedef.
28692 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
28693 and fully_masked_p.
28694 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
28695 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
28696 (vect_max_vf): New function.
28697 (slpeel_make_loop_iterate_ntimes): Delete.
28698 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
28699 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
28700 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
28701 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
28702 internal-fn.h, stor-layout.h and optabs-query.h.
28703 (vect_set_loop_mask): New function.
28704 (add_preheader_seq): Likewise.
28705 (add_header_seq): Likewise.
28706 (interleave_supported_p): Likewise.
28707 (vect_maybe_permute_loop_masks): Likewise.
28708 (vect_set_loop_masks_directly): Likewise.
28709 (vect_set_loop_condition_masked): Likewise.
28710 (vect_set_loop_condition_unmasked): New function, split out from
28711 slpeel_make_loop_iterate_ntimes.
28712 (slpeel_make_loop_iterate_ntimes): Rename to..
28713 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
28714 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
28715 (vect_do_peeling): Update call accordingly.
28716 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
28718 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28719 mask_compare_type, can_fully_mask_p and fully_masked_p.
28720 (release_vec_loop_masks): New function.
28721 (_loop_vec_info): Use it to free the loop masks.
28722 (can_produce_all_loop_masks_p): New function.
28723 (vect_get_max_nscalars_per_iter): Likewise.
28724 (vect_verify_full_masking): Likewise.
28725 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
28726 retries, and free the mask rgroups before retrying. Check loop-wide
28727 reasons for disallowing fully-masked loops. Make the final decision
28728 about whether use a fully-masked loop or not.
28729 (vect_estimate_min_profitable_iters): Do not assume that peeling
28730 for the number of iterations will be needed for fully-masked loops.
28731 (vectorizable_reduction): Disable fully-masked loops.
28732 (vectorizable_live_operation): Likewise.
28733 (vect_halve_mask_nunits): New function.
28734 (vect_double_mask_nunits): Likewise.
28735 (vect_record_loop_mask): Likewise.
28736 (vect_get_loop_mask): Likewise.
28737 (vect_transform_loop): Handle the case in which the final loop
28738 iteration might handle a partial vector. Call vect_set_loop_condition
28739 instead of slpeel_make_loop_iterate_ntimes.
28740 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
28741 (check_load_store_masking): New function.
28742 (prepare_load_store_mask): Likewise.
28743 (vectorizable_store): Handle fully-masked loops.
28744 (vectorizable_load): Likewise.
28745 (supportable_widening_operation): Use vect_halve_mask_nunits for
28747 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
28748 (vect_gen_while): New function.
28749 * config/aarch64/aarch64.md (umax<mode>3): New expander.
28750 (aarch64_uqdec<mode>): New insn.
28752 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28753 Alan Hayward <alan.hayward@arm.com>
28754 David Sherwood <david.sherwood@arm.com>
28756 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
28757 (reduc_xor_scal_optab): New optabs.
28758 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
28759 (reduc_xor_scal_@var{m}): Document.
28760 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
28761 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
28762 internal functions.
28763 * fold-const-call.c (fold_const_call): Handle them.
28764 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
28765 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
28766 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
28767 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
28768 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
28769 (UNSPEC_XORV): New unspecs.
28770 (optab): Add entries for them.
28771 (BITWISEV): New int iterator.
28772 (bit_reduc_op): New int attributes.
28774 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28775 Alan Hayward <alan.hayward@arm.com>
28776 David Sherwood <david.sherwood@arm.com>
28778 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
28779 * internal-fn.def (VEC_SHL_INSERT): New internal function.
28780 * optabs.def (vec_shl_insert_optab): New optab.
28781 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
28782 (duplicate_and_interleave): Likewise.
28783 * tree-vect-loop.c: Include internal-fn.h.
28784 (neutral_op_for_slp_reduction): New function, split out from
28785 get_initial_defs_for_reduction.
28786 (get_initial_def_for_reduction): Handle option 2 for variable-length
28787 vectors by loading the neutral value into a vector and then shifting
28788 the initial value into element 0.
28789 (get_initial_defs_for_reduction): Replace the code argument with
28790 the neutral value calculated by neutral_op_for_slp_reduction.
28791 Use gimple_build_vector for constant-length vectors.
28792 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
28793 but the first group_size elements have a neutral value.
28794 Use duplicate_and_interleave otherwise.
28795 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
28796 Update call to get_initial_defs_for_reduction. Handle SLP
28797 reductions for variable-length vectors by creating one vector
28798 result for each scalar result, with the elements associated
28799 with other scalar results stubbed out with the neutral value.
28800 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
28801 Require IFN_VEC_SHL_INSERT for double reductions on
28802 variable-length vectors, or SLP reductions that have
28803 a neutral value. Require can_duplicate_and_interleave_p
28804 support for variable-length unchained SLP reductions if there
28805 is no neutral value, such as for MIN/MAX reductions. Also require
28806 the number of vector elements to be a multiple of the number of
28807 SLP statements when doing variable-length unchained SLP reductions.
28808 Update call to vect_create_epilog_for_reduction.
28809 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
28810 and remove initial values.
28811 (duplicate_and_interleave): Make public.
28812 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
28813 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
28815 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28816 Alan Hayward <alan.hayward@arm.com>
28817 David Sherwood <david.sherwood@arm.com>
28819 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
28820 (can_duplicate_and_interleave_p): New function.
28821 (vect_get_and_check_slp_defs): Take the vector of statements
28822 rather than just the current one. Remove excess parentheses.
28823 Restriction rejectinon of vect_constant_def and vect_external_def
28824 for variable-length vectors to boolean types, or types for which
28825 can_duplicate_and_interleave_p is false.
28826 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
28827 (duplicate_and_interleave): New function.
28828 (vect_get_constant_vectors): Use gimple_build_vector for
28829 constant-length vectors and suitable variable-length constant
28830 vectors. Use duplicate_and_interleave for other variable-length
28831 vectors. Don't defer the update when inserting new statements.
28833 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28834 Alan Hayward <alan.hayward@arm.com>
28835 David Sherwood <david.sherwood@arm.com>
28837 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
28838 min_profitable_iters doesn't go negative.
28840 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28841 Alan Hayward <alan.hayward@arm.com>
28842 David Sherwood <david.sherwood@arm.com>
28844 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
28845 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
28846 * optabs.def (vec_mask_load_lanes_optab): New optab.
28847 (vec_mask_store_lanes_optab): Likewise.
28848 * internal-fn.def (MASK_LOAD_LANES): New internal function.
28849 (MASK_STORE_LANES): Likewise.
28850 * internal-fn.c (mask_load_lanes_direct): New macro.
28851 (mask_store_lanes_direct): Likewise.
28852 (expand_mask_load_optab_fn): Handle masked operations.
28853 (expand_mask_load_lanes_optab_fn): New macro.
28854 (expand_mask_store_optab_fn): Handle masked operations.
28855 (expand_mask_store_lanes_optab_fn): New macro.
28856 (direct_mask_load_lanes_optab_supported_p): Likewise.
28857 (direct_mask_store_lanes_optab_supported_p): Likewise.
28858 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
28860 (vect_load_lanes_supported): Likewise.
28861 * tree-vect-data-refs.c (strip_conversion): New function.
28862 (can_group_stmts_p): Likewise.
28863 (vect_analyze_data_ref_accesses): Use it instead of checking
28864 for a pair of assignments.
28865 (vect_store_lanes_supported): Take a masked_p parameter.
28866 (vect_load_lanes_supported): Likewise.
28867 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
28868 vect_store_lanes_supported and vect_load_lanes_supported.
28869 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
28870 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
28871 parameter. Don't allow gaps for masked accesses.
28872 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
28873 and vect_load_lanes_supported.
28874 (get_load_store_type): Take a masked_p parameter and update
28875 call to get_group_load_store_type.
28876 (vectorizable_store): Update call to get_load_store_type.
28877 Handle IFN_MASK_STORE_LANES.
28878 (vectorizable_load): Update call to get_load_store_type.
28879 Handle IFN_MASK_LOAD_LANES.
28881 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28882 Alan Hayward <alan.hayward@arm.com>
28883 David Sherwood <david.sherwood@arm.com>
28885 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
28887 * config/aarch64/aarch64-protos.h
28888 (aarch64_sve_struct_memory_operand_p): Declare.
28889 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
28890 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
28891 (VPRED, vpred): Handle SVE structure modes.
28892 * config/aarch64/constraints.md (Utx): New constraint.
28893 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
28894 (aarch64_sve_struct_nonimmediate_operand): New predicates.
28895 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
28896 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
28897 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
28898 structure modes. Split into pieces after RA.
28899 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
28900 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
28902 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
28903 SVE structure modes.
28904 (aarch64_classify_address): Likewise.
28905 (sizetochar): Move earlier in file.
28906 (aarch64_print_operand): Handle SVE register lists.
28907 (aarch64_array_mode): New function.
28908 (aarch64_sve_struct_memory_operand_p): Likewise.
28909 (TARGET_ARRAY_MODE): Redefine.
28911 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28912 Alan Hayward <alan.hayward@arm.com>
28913 David Sherwood <david.sherwood@arm.com>
28915 * target.def (array_mode): New target hook.
28916 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
28917 * doc/tm.texi: Regenerate.
28918 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
28919 * hooks.c (hook_optmode_mode_uhwi_none): New function.
28920 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
28921 targetm.array_mode.
28922 * stor-layout.c (mode_for_array): Likewise. Support polynomial
28925 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28926 Alan Hayward <alan.hayward@arm.com>
28927 David Sherwood <david.sherwood@arm.com>
28929 * fold-const.c (fold_binary_loc): Check the argument types
28930 rather than the result type when testing for a vector operation.
28932 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28934 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
28935 * doc/tm.texi: Regenerate.
28937 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28938 Alan Hayward <alan.hayward@arm.com>
28939 David Sherwood <david.sherwood@arm.com>
28941 * doc/invoke.texi (-msve-vector-bits=): Document new option.
28942 (sve): Document new AArch64 extension.
28943 * doc/md.texi (w): Extend the description of the AArch64
28944 constraint to include SVE vectors.
28945 (Upl, Upa): Document new AArch64 predicate constraints.
28946 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
28948 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
28949 (msve-vector-bits=): New option.
28950 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
28951 SVE when these are disabled.
28952 (sve): New extension.
28953 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
28954 modes. Adjust their number of units based on aarch64_sve_vg.
28955 (MAX_BITSIZE_MODE_ANY_MODE): Define.
28956 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
28957 aarch64_addr_query_type.
28958 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
28959 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
28960 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
28961 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
28962 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
28963 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
28964 (aarch64_simd_imm_zero_p): Delete.
28965 (aarch64_check_zero_based_sve_index_immediate): Declare.
28966 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
28967 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
28968 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
28969 (aarch64_sve_float_mul_immediate_p): Likewise.
28970 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
28971 rather than an rtx.
28972 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
28973 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
28974 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
28975 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
28976 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
28977 (aarch64_regmode_natural_size): Likewise.
28978 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
28979 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
28981 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
28982 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
28983 for VG and the SVE predicate registers.
28984 (V_ALIASES): Add a "z"-prefixed alias.
28985 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
28986 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
28987 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
28988 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
28989 (REG_CLASS_NAMES): Add entries for them.
28990 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
28991 and the predicate registers.
28992 (aarch64_sve_vg): Declare.
28993 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
28994 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
28995 (REGMODE_NATURAL_SIZE): Define.
28996 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
28998 * config/aarch64/aarch64.c: Include cfgrtl.h.
28999 (simd_immediate_info): Add a constructor for series vectors,
29000 and an associated step field.
29001 (aarch64_sve_vg): New variable.
29002 (aarch64_dbx_register_number): Handle VG and the predicate registers.
29003 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
29004 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
29005 (VEC_ANY_DATA, VEC_STRUCT): New constants.
29006 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
29007 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
29008 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
29009 (aarch64_get_mask_mode): New functions.
29010 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
29011 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
29012 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
29013 predicate modes and predicate registers. Explicitly restrict
29014 GPRs to modes of 16 bytes or smaller. Only allow FP registers
29015 to store a vector mode if it is recognized by
29016 aarch64_classify_vector_mode.
29017 (aarch64_regmode_natural_size): New function.
29018 (aarch64_hard_regno_caller_save_mode): Return the original mode
29020 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
29021 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
29022 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
29023 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
29025 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
29026 does not overlap dest if the function is frame-related. Handle
29028 (aarch64_split_add_offset): New function.
29029 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
29030 them aarch64_add_offset.
29031 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
29032 and update call to aarch64_sub_sp.
29033 (aarch64_add_cfa_expression): New function.
29034 (aarch64_expand_prologue): Pass extra temporary registers to the
29035 functions above. Handle the case in which we need to emit new
29036 DW_CFA_expressions for registers that were originally saved
29037 relative to the stack pointer, but now have to be expressed
29038 relative to the frame pointer.
29039 (aarch64_output_mi_thunk): Pass extra temporary registers to the
29041 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
29042 IP0 and IP1 values for SVE frames.
29043 (aarch64_expand_vec_series): New function.
29044 (aarch64_expand_sve_widened_duplicate): Likewise.
29045 (aarch64_expand_sve_const_vector): Likewise.
29046 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
29047 Handle SVE constants. Use emit_move_insn to move a force_const_mem
29048 into the register, rather than emitting a SET directly.
29049 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
29050 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
29051 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
29052 (offset_9bit_signed_scaled_p): New functions.
29053 (aarch64_replicate_bitmask_imm): New function.
29054 (aarch64_bitmask_imm): Use it.
29055 (aarch64_cannot_force_const_mem): Reject expressions involving
29056 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
29057 (aarch64_classify_index): Handle SVE indices, by requiring
29058 a plain register index with a scale that matches the element size.
29059 (aarch64_classify_address): Handle SVE addresses. Assert that
29060 the mode of the address is VOIDmode or an integer mode.
29061 Update call to aarch64_classify_symbol.
29062 (aarch64_classify_symbolic_expression): Update call to
29063 aarch64_classify_symbol.
29064 (aarch64_const_vec_all_in_range_p): New function.
29065 (aarch64_print_vector_float_operand): Likewise.
29066 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
29067 "vN" for FP registers with SVE modes. Handle (const ...) vectors
29068 and the FP immediates 1.0 and 0.5.
29069 (aarch64_print_address_internal): Handle SVE addresses.
29070 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
29071 (aarch64_regno_regclass): Handle predicate registers.
29072 (aarch64_secondary_reload): Handle big-endian reloads of SVE
29074 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
29075 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
29076 (aarch64_convert_sve_vector_bits): New function.
29077 (aarch64_override_options): Use it to handle -msve-vector-bits=.
29078 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
29079 rather than an rtx.
29080 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
29081 Handle SVE vector and predicate modes. Accept VL-based constants
29082 that need only one temporary register, and VL offsets that require
29083 no temporary registers.
29084 (aarch64_conditional_register_usage): Mark the predicate registers
29085 as fixed if SVE isn't available.
29086 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
29087 Return true for SVE vector and predicate modes.
29088 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
29089 rather than an unsigned int. Handle SVE modes.
29090 (aarch64_preferred_simd_mode): Update call accordingly. Handle
29092 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
29094 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
29095 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
29096 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
29097 (aarch64_sve_float_mul_immediate_p): New functions.
29098 (aarch64_sve_valid_immediate): New function.
29099 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
29100 Explicitly reject structure modes. Check for INDEX constants.
29101 Handle PTRUE and PFALSE constants.
29102 (aarch64_check_zero_based_sve_index_immediate): New function.
29103 (aarch64_simd_imm_zero_p): Delete.
29104 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
29105 vector modes. Accept constants in the range of CNT[BHWD].
29106 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
29107 ask for an Advanced SIMD mode.
29108 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
29109 (aarch64_simd_vector_alignment): Handle SVE predicates.
29110 (aarch64_vectorize_preferred_vector_alignment): New function.
29111 (aarch64_simd_vector_alignment_reachable): Use it instead of
29113 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
29114 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
29116 (MAX_VECT_LEN): Delete.
29117 (expand_vec_perm_d): Add a vec_flags field.
29118 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
29119 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
29120 (aarch64_evpc_ext): Don't apply a big-endian lane correction
29122 (aarch64_evpc_rev): Rename to...
29123 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
29124 (aarch64_evpc_rev_global): New function.
29125 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
29126 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
29128 (aarch64_evpc_sve_tbl): New function.
29129 (aarch64_expand_vec_perm_const_1): Update after rename of
29130 aarch64_evpc_rev. Handle SVE permutes too, trying
29131 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
29132 than aarch64_evpc_tbl.
29133 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
29134 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
29135 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
29136 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
29137 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
29138 (aarch64_expand_sve_vcond): New functions.
29139 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
29140 of aarch64_vector_mode_p.
29141 (aarch64_dwarf_poly_indeterminate_value): New function.
29142 (aarch64_compute_pressure_classes): Likewise.
29143 (aarch64_can_change_mode_class): Likewise.
29144 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
29145 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
29146 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
29147 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
29148 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
29149 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
29150 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
29151 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
29153 (Dn, Dl, Dr): Accept const as well as const_vector.
29154 (Dz): Likewise. Compare against CONST0_RTX.
29155 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
29156 of "vector" where appropriate.
29157 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
29158 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
29159 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
29160 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
29161 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
29162 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
29163 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
29164 (v_int_equiv): Extend to SVE modes.
29165 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
29167 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
29168 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
29169 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
29170 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
29171 (SVE_COND_FP_CMP): New int iterators.
29172 (perm_hilo): Handle the new unpack unspecs.
29173 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
29175 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
29176 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
29177 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
29178 (aarch64_equality_operator, aarch64_constant_vector_operand)
29179 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
29180 (aarch64_sve_nonimmediate_operand): Likewise.
29181 (aarch64_sve_general_operand): Likewise.
29182 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
29183 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
29184 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
29185 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
29186 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
29187 (aarch64_sve_float_arith_immediate): Likewise.
29188 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
29189 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
29190 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
29191 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
29192 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
29193 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
29194 (aarch64_sve_float_arith_operand): Likewise.
29195 (aarch64_sve_float_arith_with_sub_operand): Likewise.
29196 (aarch64_sve_float_mul_operand): Likewise.
29197 (aarch64_sve_vec_perm_operand): Likewise.
29198 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
29199 (aarch64_mov_operand): Accept const_poly_int and const_vector.
29200 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
29201 as well as const_vector.
29202 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
29203 in file. Use CONST0_RTX and CONSTM1_RTX.
29204 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
29205 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
29206 Use aarch64_simd_imm_zero.
29207 * config/aarch64/aarch64-sve.md: New file.
29208 * config/aarch64/aarch64.md: Include it.
29209 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
29210 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
29211 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
29212 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
29213 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
29214 (sve): New attribute.
29215 (enabled): Disable instructions with the sve attribute unless
29217 (movqi, movhi): Pass CONST_POLY_INT operaneds through
29218 aarch64_expand_mov_immediate.
29219 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
29220 CNT[BHSD] immediates.
29221 (movti): Split CONST_POLY_INT moves into two halves.
29222 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
29223 Split additions that need a temporary here if the destination
29224 is the stack pointer.
29225 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
29226 (*add<mode>3_poly_1): New instruction.
29227 (set_clobber_cc): New expander.
29229 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29231 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
29232 parameter and use it instead of GET_MODE_SIZE (innermode). Use
29233 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
29234 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
29235 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
29236 Change innermode from fixed_mode_size to machine_mode.
29237 (simplify_subreg): Update call accordingly. Handle a constant-sized
29238 subreg of a variable-length CONST_VECTOR.
29240 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29241 Alan Hayward <alan.hayward@arm.com>
29242 David Sherwood <david.sherwood@arm.com>
29244 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
29245 (add_offset_to_base): New function, split out from...
29246 (create_mem_ref): ...here. When handling a scale other than 1,
29247 check first whether the address is valid without the offset.
29248 Add it into the base if so, leaving the index and scale as-is.
29250 2018-01-12 Jakub Jelinek <jakub@redhat.com>
29253 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
29254 fold_for_warn before checking if arg2 is INTEGER_CST.
29256 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
29258 * config/rs6000/predicates.md (load_multiple_operation): Delete.
29259 (store_multiple_operation): Delete.
29260 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
29261 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
29262 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
29263 guarded by TARGET_STRING.
29264 (rs6000_output_load_multiple): Delete.
29265 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29266 OPTION_MASK_STRING / TARGET_STRING handling.
29267 (print_operand) <'N', 'O'>: Add comment that these are unused now.
29268 (const rs6000_opt_masks) <"string">: Change mask to 0.
29269 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
29270 (MASK_STRING): Delete.
29271 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
29273 (load_multiple): Delete.
29280 (store_multiple): Delete.
29287 (movmemsi_8reg): Delete.
29288 (corresponding unnamed define_insn): Delete.
29289 (movmemsi_6reg): Delete.
29290 (corresponding unnamed define_insn): Delete.
29291 (movmemsi_4reg): Delete.
29292 (corresponding unnamed define_insn): Delete.
29293 (movmemsi_2reg): Delete.
29294 (corresponding unnamed define_insn): Delete.
29295 (movmemsi_1reg): Delete.
29296 (corresponding unnamed define_insn): Delete.
29297 * config/rs6000/rs6000.opt (mno-string): New.
29298 (mstring): Replace by deprecation warning stub.
29299 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
29301 2018-01-12 Jakub Jelinek <jakub@redhat.com>
29303 * regrename.c (regrename_do_replace): If replacing the same
29304 reg multiple times, try to reuse last created gen_raw_REG.
29307 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
29308 main to workaround a bug in GDB.
29310 2018-01-12 Tom de Vries <tom@codesourcery.com>
29313 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
29315 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
29317 PR rtl-optimization/80481
29318 * ira-color.c (get_cap_member): New function.
29319 (allocnos_conflict_by_live_ranges_p): Use it.
29320 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
29321 (setup_slot_coalesced_allocno_live_ranges): Ditto.
29323 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
29326 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
29327 (*saddl_se_1): Ditto.
29328 (*ssubsi_1): Ditto.
29329 (*ssubl_se_1): Ditto.
29331 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29333 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
29334 rather than wi::to_widest for DR_INITs.
29335 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
29336 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
29337 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
29339 (vect_analyze_group_access_1): Note that here.
29341 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29343 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
29344 polynomial type sizes.
29346 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29348 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
29349 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
29350 (gimple_add_tmp_var): Likewise.
29352 2018-01-12 Martin Liska <mliska@suse.cz>
29354 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
29355 (gimple_alloc_sizes): Likewise.
29356 (dump_gimple_statistics): Use PRIu64 in printf format.
29357 * gimple.h: Change uint64_t to int.
29359 2018-01-12 Martin Liska <mliska@suse.cz>
29361 * tree-core.h: Use uint64_t instead of int.
29362 * tree.c (tree_node_counts): Likewise.
29363 (tree_node_sizes): Likewise.
29364 (dump_tree_statistics): Use PRIu64 in printf format.
29366 2018-01-12 Martin Liska <mliska@suse.cz>
29368 * Makefile.in: As qsort_chk is implemented in vec.c, add
29369 vec.o to linkage of gencfn-macros.
29370 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
29371 passing the info to record_node_allocation_statistics.
29372 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
29374 * ggc-common.c (struct ggc_usage): Add operator== and use
29375 it in operator< and compare function.
29376 * mem-stats.h (struct mem_usage): Likewise.
29377 * vec.c (struct vec_usage): Remove operator< and compare
29378 function. Can be simply inherited.
29380 2018-01-12 Martin Jambor <mjambor@suse.cz>
29383 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
29384 * tree-ssa-math-opts.c: Include domwalk.h.
29385 (convert_mult_to_fma_1): New function.
29386 (fma_transformation_info): New type.
29387 (fma_deferring_state): Likewise.
29388 (cancel_fma_deferring): New function.
29389 (result_of_phi): Likewise.
29390 (last_fma_candidate_feeds_initial_phi): Likewise.
29391 (convert_mult_to_fma): Added deferring logic, split actual
29392 transformation to convert_mult_to_fma_1.
29393 (math_opts_dom_walker): New type.
29394 (math_opts_dom_walker::after_dom_children): New method, body moved
29395 here from pass_optimize_widening_mul::execute, added deferring logic
29397 (pass_optimize_widening_mul::execute): Moved most of code to
29398 math_opts_dom_walker::after_dom_children.
29399 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
29400 * config/i386/i386.c (ix86_option_override_internal): Added
29401 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
29403 2018-01-12 Richard Biener <rguenther@suse.de>
29406 * dwarf2out.c (gen_variable_die): Do not reset old_die for
29407 inline instance vars.
29409 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
29412 * config/rx/rx.c (rx_is_restricted_memory_address):
29413 Handle SUBREG case.
29415 2018-01-12 Richard Biener <rguenther@suse.de>
29417 PR tree-optimization/80846
29418 * target.def (split_reduction): New target hook.
29419 * targhooks.c (default_split_reduction): New function.
29420 * targhooks.h (default_split_reduction): Declare.
29421 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
29422 target requests first reduce vectors by combining low and high
29424 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
29425 (get_vectype_for_scalar_type_and_size): Export.
29426 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
29427 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
29428 * doc/tm.texi: Regenerate.
29429 * config/i386/i386.c (ix86_split_reduction): Implement
29430 TARGET_VECTORIZE_SPLIT_REDUCTION.
29432 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29435 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
29436 in PIC mode except for TARGET_VXWORKS_RTP.
29437 * config/sparc/sparc.c: Include cfgrtl.h.
29438 (TARGET_INIT_PIC_REG): Define.
29439 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
29440 (sparc_pic_register_p): New predicate.
29441 (sparc_legitimate_address_p): Use it.
29442 (sparc_legitimize_pic_address): Likewise.
29443 (sparc_delegitimize_address): Likewise.
29444 (sparc_mode_dependent_address_p): Likewise.
29445 (gen_load_pcrel_sym): Remove 4th parameter.
29446 (load_got_register): Adjust call to above. Remove obsolete stuff.
29447 (sparc_expand_prologue): Do not call load_got_register here.
29448 (sparc_flat_expand_prologue): Likewise.
29449 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
29450 (sparc_use_pseudo_pic_reg): New function.
29451 (sparc_init_pic_reg): Likewise.
29452 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
29453 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
29455 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
29457 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29458 Add item for branch_cost.
29460 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29462 PR rtl-optimization/83565
29463 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
29464 not extend the result to a larger mode for rotate operations.
29465 (num_sign_bit_copies1): Likewise.
29467 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29470 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
29472 Use values-Xc.o for -pedantic.
29473 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
29475 2018-01-12 Martin Liska <mliska@suse.cz>
29478 * ipa-devirt.c (final_warning_record::grow_type_warnings):
29480 (possible_polymorphic_call_targets): Use it.
29481 (ipa_devirt): Likewise.
29483 2018-01-12 Martin Liska <mliska@suse.cz>
29485 * profile-count.h (enum profile_quality): Use 0 as invalid
29486 enum value of profile_quality.
29488 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
29490 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
29491 -mext-string options.
29493 2018-01-12 Richard Biener <rguenther@suse.de>
29495 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
29496 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
29497 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
29499 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
29501 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29503 * configure.ac (--with-long-double-format): Add support for the
29504 configuration option to change the default long double format on
29506 * config.gcc (powerpc*-linux*-*): Likewise.
29507 * configure: Regenerate.
29508 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
29509 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
29510 used without modification.
29512 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29514 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
29515 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
29516 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
29517 MISC_BUILTIN_SPEC_BARRIER.
29518 (rs6000_init_builtins): Likewise.
29519 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
29521 (speculation_barrier): New define_insn.
29522 * doc/extend.texi: Document __builtin_speculation_barrier.
29524 2018-01-11 Jakub Jelinek <jakub@redhat.com>
29527 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
29528 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
29529 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
29531 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
29532 integral modes instead of "ss" and "sd".
29533 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
29534 vectors with 32-bit and 64-bit elements.
29535 (vecdupssescalarmodesuffix): New mode attribute.
29536 (vec_dup<mode>): Use it.
29538 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
29541 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
29542 frame if argument is passed on stack.
29544 2018-01-11 Jakub Jelinek <jakub@redhat.com>
29547 * ree.c (combine_reaching_defs): Optimize also
29548 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
29549 reg2=any_extend(exp); reg1=reg2;, formatting fix.
29551 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
29553 PR middle-end/83189
29554 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
29556 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
29558 PR middle-end/83718
29559 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
29560 after they are computed.
29562 2018-01-11 Bin Cheng <bin.cheng@arm.com>
29564 PR tree-optimization/83695
29565 * gimple-loop-linterchange.cc
29566 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
29567 reset cached scev information after interchange.
29568 (pass_linterchange::execute): Remove call to scev_reset_htab.
29570 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29572 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
29573 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
29574 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
29575 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
29576 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
29577 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
29578 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
29579 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
29580 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
29581 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
29582 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
29583 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
29584 (V_lane_reg): Likewise.
29585 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
29587 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
29588 (vfmal_lane_low<mode>_intrinsic,
29589 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
29590 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
29591 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
29592 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
29593 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
29594 vfmsl_lane_high<mode>_intrinsic): New define_insns.
29596 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29598 * config/arm/arm-cpus.in (fp16fml): New feature.
29599 (ALL_SIMD): Add fp16fml.
29600 (armv8.2-a): Add fp16fml as an option.
29601 (armv8.3-a): Likewise.
29602 (armv8.4-a): Add fp16fml as part of fp16.
29603 * config/arm/arm.h (TARGET_FP16FML): Define.
29604 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
29606 * config/arm/arm-modes.def (V2HF): Define.
29607 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
29608 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
29609 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
29610 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
29611 vfmsl_low, vfmsl_high): New set of builtins.
29612 * config/arm/iterators.md (PLUSMINUS): New code iterator.
29613 (vfml_op): New code attribute.
29614 (VFMLHALVES): New int iterator.
29615 (VFML, VFMLSEL): New mode attributes.
29616 (V_reg): Define mapping for V2HF.
29617 (V_hi, V_lo): New mode attributes.
29618 (VF_constraint): Likewise.
29619 (vfml_half, vfml_half_selector): New int attributes.
29620 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
29622 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
29623 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
29625 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
29626 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
29627 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
29628 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
29630 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
29631 Document new effective target and option set.
29633 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29635 * config/arm/arm-cpus.in (armv8_4): New feature.
29636 (ARMv8_4a): New fgroup.
29637 (armv8.4-a): New arch.
29638 * config/arm/arm-tables.opt: Regenerate.
29639 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
29640 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
29641 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
29642 Add matching rules for -march=armv8.4-a and extensions.
29643 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
29645 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
29648 * config/rx/rx.md (BW): New mode attribute.
29649 (sync_lock_test_and_setsi): Add mode suffix to insn output.
29651 2018-01-11 Richard Biener <rguenther@suse.de>
29653 PR tree-optimization/83435
29654 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
29655 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
29656 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
29658 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29659 Alan Hayward <alan.hayward@arm.com>
29660 David Sherwood <david.sherwood@arm.com>
29662 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
29664 (aarch64_classify_address): Initialize it. Track polynomial offsets.
29665 (aarch64_print_address_internal): Use it to check for a zero offset.
29667 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29668 Alan Hayward <alan.hayward@arm.com>
29669 David Sherwood <david.sherwood@arm.com>
29671 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
29672 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
29673 Return a poly_int64 rather than a HOST_WIDE_INT.
29674 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
29675 rather than a HOST_WIDE_INT.
29676 * config/aarch64/aarch64.h (aarch64_frame): Protect with
29677 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
29678 hard_fp_offset, frame_size, initial_adjust, callee_offset and
29679 final_offset from HOST_WIDE_INT to poly_int64.
29680 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
29681 to_constant when getting the number of units in an Advanced SIMD
29683 (aarch64_builtin_vectorized_function): Check for a constant number
29685 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
29687 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
29688 attribute instead of GET_MODE_NUNITS.
29689 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
29690 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
29691 GET_MODE_SIZE for fixed-size registers.
29692 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
29693 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
29694 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
29695 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
29696 (aarch64_print_operand, aarch64_print_address_internal)
29697 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
29698 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
29699 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
29700 Handle polynomial GET_MODE_SIZE.
29701 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
29702 wider than SImode without modification.
29703 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
29704 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
29705 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
29706 passing and returning SVE modes.
29707 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
29708 rather than GEN_INT.
29709 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
29710 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
29711 (aarch64_allocate_and_probe_stack_space): Likewise.
29712 (aarch64_layout_frame): Cope with polynomial offsets.
29713 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
29714 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
29715 polynomial offsets.
29716 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
29717 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
29718 poly_int64 rather than a HOST_WIDE_INT.
29719 (aarch64_get_separate_components, aarch64_process_components)
29720 (aarch64_expand_prologue, aarch64_expand_epilogue)
29721 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
29722 (aarch64_anchor_offset): New function, split out from...
29723 (aarch64_legitimize_address): ...here.
29724 (aarch64_builtin_vectorization_cost): Handle polynomial
29725 TYPE_VECTOR_SUBPARTS.
29726 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
29728 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
29729 number of elements from the PARALLEL rather than the mode.
29730 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
29731 rather than GET_MODE_BITSIZE.
29732 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
29733 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
29734 (aarch64_expand_vec_perm_const_1): Handle polynomial
29735 d->perm.length () and d->perm elements.
29736 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
29737 Apply to_constant to d->perm elements.
29738 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
29739 polynomial CONST_VECTOR_NUNITS.
29740 (aarch64_move_pointer): Take amount as a poly_int64 rather
29742 (aarch64_progress_pointer): Avoid temporary variable.
29743 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
29744 the mode attribute instead of GET_MODE.
29746 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29747 Alan Hayward <alan.hayward@arm.com>
29748 David Sherwood <david.sherwood@arm.com>
29750 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
29751 x exists before using it.
29752 (aarch64_add_constant_internal): Rename to...
29753 (aarch64_add_offset_1): ...this. Replace regnum with separate
29754 src and dest rtxes. Handle the case in which they're different,
29755 including when the offset is zero. Replace scratchreg with an rtx.
29756 Use 2 additions if there is no spare register into which we can
29757 move a 16-bit constant.
29758 (aarch64_add_constant): Delete.
29759 (aarch64_add_offset): Replace reg with separate src and dest
29760 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
29761 Use aarch64_add_offset_1.
29762 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
29763 an rtx rather than an int. Take the delta as a poly_int64
29764 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
29765 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
29766 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
29767 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
29768 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
29769 and aarch64_add_sp.
29770 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
29771 aarch64_add_constant.
29773 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29775 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
29776 Use scalar_float_mode.
29778 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29780 * config/aarch64/aarch64-simd.md
29781 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
29782 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
29783 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
29784 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
29785 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
29786 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
29787 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
29788 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
29789 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
29790 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
29792 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29795 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
29796 targ_options->x_arm_arch_string is non NULL.
29798 2018-01-11 Tamar Christina <tamar.christina@arm.com>
29800 * config/aarch64/aarch64.h
29801 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
29803 2018-01-11 Sudakshina Das <sudi.das@arm.com>
29806 * expmed.c (emit_store_flag_force): Swap if const op0
29807 and change VOIDmode to mode of op0.
29809 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29811 PR rtl-optimization/83761
29812 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
29813 than bytes to mode_for_size.
29815 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29817 PR middle-end/83189
29818 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
29819 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
29822 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29824 PR middle-end/83575
29825 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
29826 when in layout mode.
29827 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
29828 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
29831 2018-01-10 Michael Collison <michael.collison@arm.com>
29833 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
29834 * config/aarch64/aarch64-option-extension.def: Add
29835 AARCH64_OPT_EXTENSION of 'fp16fml'.
29836 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29837 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
29838 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
29839 * config/aarch64/constraints.md (Ui7): New constraint.
29840 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
29841 (VFMLA_SEL_W): Ditto.
29844 (VFMLA16_LOW): New int iterator.
29845 (VFMLA16_HIGH): Ditto.
29846 (UNSPEC_FMLAL): New unspec.
29847 (UNSPEC_FMLSL): Ditto.
29848 (UNSPEC_FMLAL2): Ditto.
29849 (UNSPEC_FMLSL2): Ditto.
29850 (f16mac): New code attribute.
29851 * config/aarch64/aarch64-simd-builtins.def
29852 (aarch64_fmlal_lowv2sf): Ditto.
29853 (aarch64_fmlsl_lowv2sf): Ditto.
29854 (aarch64_fmlalq_lowv4sf): Ditto.
29855 (aarch64_fmlslq_lowv4sf): Ditto.
29856 (aarch64_fmlal_highv2sf): Ditto.
29857 (aarch64_fmlsl_highv2sf): Ditto.
29858 (aarch64_fmlalq_highv4sf): Ditto.
29859 (aarch64_fmlslq_highv4sf): Ditto.
29860 (aarch64_fmlal_lane_lowv2sf): Ditto.
29861 (aarch64_fmlsl_lane_lowv2sf): Ditto.
29862 (aarch64_fmlal_laneq_lowv2sf): Ditto.
29863 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
29864 (aarch64_fmlalq_lane_lowv4sf): Ditto.
29865 (aarch64_fmlsl_lane_lowv4sf): Ditto.
29866 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
29867 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
29868 (aarch64_fmlal_lane_highv2sf): Ditto.
29869 (aarch64_fmlsl_lane_highv2sf): Ditto.
29870 (aarch64_fmlal_laneq_highv2sf): Ditto.
29871 (aarch64_fmlsl_laneq_highv2sf): Ditto.
29872 (aarch64_fmlalq_lane_highv4sf): Ditto.
29873 (aarch64_fmlsl_lane_highv4sf): Ditto.
29874 (aarch64_fmlalq_laneq_highv4sf): Ditto.
29875 (aarch64_fmlsl_laneq_highv4sf): Ditto.
29876 * config/aarch64/aarch64-simd.md:
29877 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
29878 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
29879 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
29880 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
29881 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
29882 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
29883 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
29884 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
29885 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
29886 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
29887 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
29888 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
29889 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
29890 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
29891 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
29892 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
29893 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
29894 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
29895 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
29896 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
29897 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
29898 (vfmlsl_low_u32): Ditto.
29899 (vfmlalq_low_u32): Ditto.
29900 (vfmlslq_low_u32): Ditto.
29901 (vfmlal_high_u32): Ditto.
29902 (vfmlsl_high_u32): Ditto.
29903 (vfmlalq_high_u32): Ditto.
29904 (vfmlslq_high_u32): Ditto.
29905 (vfmlal_lane_low_u32): Ditto.
29906 (vfmlsl_lane_low_u32): Ditto.
29907 (vfmlal_laneq_low_u32): Ditto.
29908 (vfmlsl_laneq_low_u32): Ditto.
29909 (vfmlalq_lane_low_u32): Ditto.
29910 (vfmlslq_lane_low_u32): Ditto.
29911 (vfmlalq_laneq_low_u32): Ditto.
29912 (vfmlslq_laneq_low_u32): Ditto.
29913 (vfmlal_lane_high_u32): Ditto.
29914 (vfmlsl_lane_high_u32): Ditto.
29915 (vfmlal_laneq_high_u32): Ditto.
29916 (vfmlsl_laneq_high_u32): Ditto.
29917 (vfmlalq_lane_high_u32): Ditto.
29918 (vfmlslq_lane_high_u32): Ditto.
29919 (vfmlalq_laneq_high_u32): Ditto.
29920 (vfmlslq_laneq_high_u32): Ditto.
29921 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
29922 (AARCH64_FL_FOR_ARCH8_4): New.
29923 (AARCH64_ISA_F16FML): New ISA flag.
29924 (TARGET_F16FML): New feature flag for fp16fml.
29925 (doc/invoke.texi): Document new fp16fml option.
29927 2018-01-10 Michael Collison <michael.collison@arm.com>
29929 * config/aarch64/aarch64-builtins.c:
29930 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
29931 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29932 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
29933 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
29934 (AARCH64_ISA_SHA3): New ISA flag.
29935 (TARGET_SHA3): New feature flag for sha3.
29936 * config/aarch64/iterators.md (sha512_op): New int attribute.
29937 (CRYPTO_SHA512): New int iterator.
29938 (UNSPEC_SHA512H): New unspec.
29939 (UNSPEC_SHA512H2): Ditto.
29940 (UNSPEC_SHA512SU0): Ditto.
29941 (UNSPEC_SHA512SU1): Ditto.
29942 * config/aarch64/aarch64-simd-builtins.def
29943 (aarch64_crypto_sha512hqv2di): New builtin.
29944 (aarch64_crypto_sha512h2qv2di): Ditto.
29945 (aarch64_crypto_sha512su0qv2di): Ditto.
29946 (aarch64_crypto_sha512su1qv2di): Ditto.
29947 (aarch64_eor3qv8hi): Ditto.
29948 (aarch64_rax1qv2di): Ditto.
29949 (aarch64_xarqv2di): Ditto.
29950 (aarch64_bcaxqv8hi): Ditto.
29951 * config/aarch64/aarch64-simd.md:
29952 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
29953 (aarch64_crypto_sha512su0qv2di): Ditto.
29954 (aarch64_crypto_sha512su1qv2di): Ditto.
29955 (aarch64_eor3qv8hi): Ditto.
29956 (aarch64_rax1qv2di): Ditto.
29957 (aarch64_xarqv2di): Ditto.
29958 (aarch64_bcaxqv8hi): Ditto.
29959 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
29960 (vsha512h2q_u64): Ditto.
29961 (vsha512su0q_u64): Ditto.
29962 (vsha512su1q_u64): Ditto.
29963 (veor3q_u16): Ditto.
29964 (vrax1q_u64): Ditto.
29965 (vxarq_u64): Ditto.
29966 (vbcaxq_u16): Ditto.
29967 * config/arm/types.md (crypto_sha512): New type attribute.
29968 (crypto_sha3): Ditto.
29969 (doc/invoke.texi): Document new sha3 option.
29971 2018-01-10 Michael Collison <michael.collison@arm.com>
29973 * config/aarch64/aarch64-builtins.c:
29974 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
29975 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29976 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
29977 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
29978 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
29979 (AARCH64_ISA_SM4): New ISA flag.
29980 (TARGET_SM4): New feature flag for sm4.
29981 * config/aarch64/aarch64-simd-builtins.def
29982 (aarch64_sm3ss1qv4si): Ditto.
29983 (aarch64_sm3tt1aq4si): Ditto.
29984 (aarch64_sm3tt1bq4si): Ditto.
29985 (aarch64_sm3tt2aq4si): Ditto.
29986 (aarch64_sm3tt2bq4si): Ditto.
29987 (aarch64_sm3partw1qv4si): Ditto.
29988 (aarch64_sm3partw2qv4si): Ditto.
29989 (aarch64_sm4eqv4si): Ditto.
29990 (aarch64_sm4ekeyqv4si): Ditto.
29991 * config/aarch64/aarch64-simd.md:
29992 (aarch64_sm3ss1qv4si): Ditto.
29993 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
29994 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
29995 (aarch64_sm4eqv4si): Ditto.
29996 (aarch64_sm4ekeyqv4si): Ditto.
29997 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
29998 (sm3part_op): Ditto.
29999 (CRYPTO_SM3TT): Ditto.
30000 (CRYPTO_SM3PART): Ditto.
30001 (UNSPEC_SM3SS1): New unspec.
30002 (UNSPEC_SM3TT1A): Ditto.
30003 (UNSPEC_SM3TT1B): Ditto.
30004 (UNSPEC_SM3TT2A): Ditto.
30005 (UNSPEC_SM3TT2B): Ditto.
30006 (UNSPEC_SM3PARTW1): Ditto.
30007 (UNSPEC_SM3PARTW2): Ditto.
30008 (UNSPEC_SM4E): Ditto.
30009 (UNSPEC_SM4EKEY): Ditto.
30010 * config/aarch64/constraints.md (Ui2): New constraint.
30011 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
30012 * config/arm/types.md (crypto_sm3): New type attribute.
30013 (crypto_sm4): Ditto.
30014 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
30015 (vsm3tt1aq_u32): Ditto.
30016 (vsm3tt1bq_u32): Ditto.
30017 (vsm3tt2aq_u32): Ditto.
30018 (vsm3tt2bq_u32): Ditto.
30019 (vsm3partw1q_u32): Ditto.
30020 (vsm3partw2q_u32): Ditto.
30021 (vsm4eq_u32): Ditto.
30022 (vsm4ekeyq_u32): Ditto.
30023 (doc/invoke.texi): Document new sm4 option.
30025 2018-01-10 Michael Collison <michael.collison@arm.com>
30027 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
30028 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
30029 (AARCH64_FL_FOR_ARCH8_4): New.
30030 (AARCH64_FL_V8_4): New flag.
30031 (doc/invoke.texi): Document new armv8.4-a option.
30033 2018-01-10 Michael Collison <michael.collison@arm.com>
30035 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30036 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
30037 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
30038 * config/aarch64/aarch64-option-extension.def: Add
30039 AARCH64_OPT_EXTENSION of 'sha2'.
30040 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
30041 (crypto): Disable sha2 and aes if crypto disabled.
30042 (crypto): Enable aes and sha2 if enabled.
30043 (simd): Disable sha2 and aes if simd disabled.
30044 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
30046 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
30047 (TARGET_SHA2): New feature flag for sha2.
30048 (TARGET_AES): New feature flag for aes.
30049 * config/aarch64/aarch64-simd.md:
30050 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
30051 conditional on TARGET_AES.
30052 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
30053 (aarch64_crypto_sha1hsi): Make pattern conditional
30055 (aarch64_crypto_sha1hv4si): Ditto.
30056 (aarch64_be_crypto_sha1hv4si): Ditto.
30057 (aarch64_crypto_sha1su1v4si): Ditto.
30058 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
30059 (aarch64_crypto_sha1su0v4si): Ditto.
30060 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
30061 (aarch64_crypto_sha256su0v4si): Ditto.
30062 (aarch64_crypto_sha256su1v4si): Ditto.
30063 (doc/invoke.texi): Document new aes and sha2 options.
30065 2018-01-10 Martin Sebor <msebor@redhat.com>
30067 PR tree-optimization/83781
30068 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
30071 2018-01-11 Martin Sebor <msebor@gmail.com>
30072 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30074 PR tree-optimization/83501
30075 PR tree-optimization/81703
30077 * tree-ssa-strlen.c (get_string_cst): Rename...
30078 (get_string_len): ...to this. Handle global constants.
30079 (handle_char_store): Adjust.
30081 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
30082 Jim Wilson <jimw@sifive.com>
30084 * config/riscv/riscv-protos.h (riscv_output_return): New.
30085 * config/riscv/riscv.c (struct machine_function): New naked_p field.
30086 (riscv_attribute_table, riscv_output_return),
30087 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
30088 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
30089 (riscv_compute_frame_info): Only compute frame->mask if not a naked
30091 (riscv_expand_prologue): Add early return for naked function.
30092 (riscv_expand_epilogue): Likewise.
30093 (riscv_function_ok_for_sibcall): Return false for naked function.
30094 (riscv_set_current_function): New.
30095 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
30096 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
30097 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
30098 * doc/extend.texi (RISC-V Function Attributes): New.
30100 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
30102 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
30103 check for 128-bit long double before checking TCmode.
30104 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
30105 128-bit long doubles before checking TFmode or TCmode.
30106 (FLOAT128_IBM_P): Likewise.
30108 2018-01-10 Martin Sebor <msebor@redhat.com>
30110 PR tree-optimization/83671
30111 * builtins.c (c_strlen): Unconditionally return zero for the empty
30113 Use -Warray-bounds for warnings.
30114 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
30115 for non-constant array indices with COMPONENT_REF, arrays of
30116 arrays, and pointers to arrays.
30117 (gimple_fold_builtin_strlen): Determine and set length range for
30118 non-constant character arrays.
30120 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
30122 PR middle-end/81897
30123 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
30126 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
30128 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
30130 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
30133 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
30134 VECTOR_MEM_ALTIVEC_OR_VSX_P.
30135 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
30136 indexed_or_indirect_operand predicate.
30137 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
30138 (*vsx_le_perm_load_v8hi): Likewise.
30139 (*vsx_le_perm_load_v16qi): Likewise.
30140 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
30141 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
30142 (*vsx_le_perm_store_v8hi): Likewise.
30143 (*vsx_le_perm_store_v16qi): Likewise.
30144 (eight unnamed splitters): Likewise.
30146 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
30148 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
30149 * config/rs6000/emmintrin.h: Likewise.
30150 * config/rs6000/mmintrin.h: Likewise.
30151 * config/rs6000/xmmintrin.h: Likewise.
30153 2018-01-10 David Malcolm <dmalcolm@redhat.com>
30156 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
30158 * tree.c (tree_nop_conversion): Return true for location wrapper
30160 (maybe_wrap_with_location): New function.
30161 (selftest::check_strip_nops): New function.
30162 (selftest::test_location_wrappers): New function.
30163 (selftest::tree_c_tests): Call it.
30164 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
30165 (maybe_wrap_with_location): New decl.
30166 (EXPR_LOCATION_WRAPPER_P): New macro.
30167 (location_wrapper_p): New inline function.
30168 (tree_strip_any_location_wrapper): New inline function.
30170 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
30173 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
30174 stack_realign_offset for the largest alignment of stack slot
30176 (ix86_find_max_used_stack_alignment): New function.
30177 (ix86_finalize_stack_frame_flags): Use it. Set
30178 max_used_stack_alignment if we don't realign stack.
30179 * config/i386/i386.h (machine_function): Add
30180 max_used_stack_alignment.
30182 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
30184 * config/arm/arm.opt (-mbranch-cost): New option.
30185 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
30188 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
30191 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
30192 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
30194 2018-01-10 Richard Biener <rguenther@suse.de>
30197 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
30198 early out so it also covers the case where we have a non-NULL
30201 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
30203 PR tree-optimization/83753
30204 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
30205 for non-strided grouped accesses if the number of elements is 1.
30207 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30210 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
30211 * i386.h (TARGET_USE_GATHER): Define.
30212 * x86-tune.def (X86_TUNE_USE_GATHER): New.
30214 2018-01-10 Martin Liska <mliska@suse.cz>
30217 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
30218 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
30220 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
30221 CLEANUP_NO_PARTITIONING is not set.
30223 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
30225 * doc/rtl.texi: Remove documentation of (const ...) wrappers
30226 for vectors, as a partial revert of r254296.
30227 * rtl.h (const_vec_p): Delete.
30228 (const_vec_duplicate_p): Don't test for vector CONSTs.
30229 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
30230 * expmed.c (make_tree): Likewise.
30233 * common.md (E, F): Use CONSTANT_P instead of checking for
30235 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
30236 checking for CONST_VECTOR.
30238 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30240 PR middle-end/83575
30241 * predict.c (force_edge_cold): Handle in more sane way edges
30242 with no prediction.
30244 2018-01-09 Carl Love <cel@us.ibm.com>
30246 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
30248 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
30249 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
30250 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
30251 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
30252 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
30253 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
30254 * config/rs6000/rs6000-protos.h: Add extern defition for
30255 rs6000_generate_float2_double_code.
30256 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
30258 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
30259 (float2_v2df): Add define_expand.
30261 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
30264 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
30265 op_mode in the force_to_mode call.
30267 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30269 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
30270 instead of checking each element individually.
30271 (aarch64_evpc_uzp): Likewise.
30272 (aarch64_evpc_zip): Likewise.
30273 (aarch64_evpc_ext): Likewise.
30274 (aarch64_evpc_rev): Likewise.
30275 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
30276 instead of checking each element individually. Return true without
30278 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
30279 whether all selected elements come from the same input, instead of
30280 checking each element individually. Remove calls to gen_rtx_REG,
30281 start_sequence and end_sequence and instead assert that no rtl is
30284 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30286 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
30287 order of HIGH and CONST checks.
30289 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30291 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
30292 if the destination isn't an SSA_NAME.
30294 2018-01-09 Richard Biener <rguenther@suse.de>
30296 PR tree-optimization/83668
30297 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
30299 (canonicalize_loop_form): ... here, renamed from ...
30300 (canonicalize_loop_closed_ssa_form): ... this and amended to
30301 swap successor edges for loop exit blocks to make us use
30302 the RPO order we need for initial schedule generation.
30304 2018-01-09 Joseph Myers <joseph@codesourcery.com>
30306 PR tree-optimization/64811
30307 * match.pd: When optimizing comparisons with Inf, avoid
30308 introducing or losing exceptions from comparisons with NaN.
30310 2018-01-09 Martin Liska <mliska@suse.cz>
30313 * asan.c (shadow_mem_size): Add gcc_assert.
30315 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
30317 Don't save registers in main().
30320 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
30321 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
30322 * config/avr/avr.c (avr_set_current_function): Don't error if
30323 naked, OS_task or OS_main are specified at the same time.
30324 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
30326 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
30328 * common/config/avr/avr-common.c (avr_option_optimization_table):
30329 Switch on -mmain-is-OS_task for optimizing compilations.
30331 2018-01-09 Richard Biener <rguenther@suse.de>
30333 PR tree-optimization/83572
30334 * graphite.c: Include cfganal.h.
30335 (graphite_transform_loops): Connect infinite loops to exit
30336 and remove fake edges at the end.
30338 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30340 * ipa-inline.c (edge_badness): Revert accidental checkin.
30342 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30345 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
30346 symbols; not inline clones.
30348 2018-01-09 Jakub Jelinek <jakub@redhat.com>
30351 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
30352 hard registers. Formatting fixes.
30354 PR preprocessor/83722
30355 * gcc.c (try_generate_repro): Pass
30356 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
30357 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
30360 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
30361 Kito Cheng <kito.cheng@gmail.com>
30363 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
30364 (riscv_leaf_function_p): Delete.
30365 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
30367 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30369 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
30371 (do_ifelse): New function.
30372 (do_isel): New function.
30373 (do_sub3): New function.
30374 (do_add3): New function.
30375 (do_load_mask_compare): New function.
30376 (do_overlap_load_compare): New function.
30377 (expand_compare_loop): New function.
30378 (expand_block_compare): Call expand_compare_loop() when appropriate.
30379 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
30380 option description.
30381 (-mblock-compare-inline-loop-limit): New option.
30383 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30386 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
30387 Reverse order of second and third operands in first alternative.
30388 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
30389 of first and second elements in UNSPEC_VPERMR vector.
30390 (altivec_expand_vec_perm_le): Likewise.
30392 2018-01-08 Jeff Law <law@redhat.com>
30394 PR rtl-optimizatin/81308
30395 * tree-switch-conversion.c (cfg_altered): New file scoped static.
30396 (process_switch): If group_case_labels makes a change, then set
30398 (pass_convert_switch::execute): If a switch is converted, then
30399 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
30401 PR rtl-optimization/81308
30402 * recog.c (split_all_insns): Conditionally cleanup the CFG after
30405 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
30407 PR target/83663 - Revert r255946
30408 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
30409 generation for cases where splatting a value is not useful.
30410 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
30411 across a vec_duplicate and a paradoxical subreg forming a vector
30412 mode to a vec_concat.
30414 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30416 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
30417 -march=armv8.3-a variants.
30418 * config/arm/t-multilib: Likewise.
30419 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
30421 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30423 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
30425 (cceq_ior_compare_complement): Give it a name so I can use it, and
30426 change boolean_or_operator predicate to boolean_operator so it can
30427 be used to generate a crand.
30428 (eqne): New code iterator.
30429 (bd/bd_neg): New code_attrs.
30430 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
30431 a single define_insn.
30432 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
30433 decrement (bdnzt/bdnzf/bdzt/bdzf).
30434 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
30435 with the new names of the branch decrement patterns, and added the
30436 names of the branch decrement conditional patterns.
30438 2018-01-08 Richard Biener <rguenther@suse.de>
30440 PR tree-optimization/83563
30441 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
30444 2018-01-08 Richard Biener <rguenther@suse.de>
30446 PR middle-end/83713
30447 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
30449 2018-01-08 Richard Biener <rguenther@suse.de>
30451 PR tree-optimization/83685
30452 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
30453 references to abnormals.
30455 2018-01-08 Richard Biener <rguenther@suse.de>
30458 * dwarf2out.c (output_indirect_strings): Handle empty
30459 skeleton_debug_str_hash.
30460 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
30462 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
30464 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
30465 (emit_store_direct): Likewise.
30466 (arc_trampoline_adjust_address): Likewise.
30467 (arc_asm_trampoline_template): New function.
30468 (arc_initialize_trampoline): Use asm_trampoline_template.
30469 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
30470 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
30471 * config/arc/arc.md (flush_icache): Delete pattern.
30473 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
30475 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
30476 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
30479 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
30482 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
30483 by not USED_FOR_TARGET.
30484 (make_pass_resolve_sw_modes): Likewise.
30486 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
30488 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
30491 2018-01-08 Richard Biener <rguenther@suse.de>
30493 PR middle-end/83580
30494 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
30496 2018-01-08 Richard Biener <rguenther@suse.de>
30498 PR middle-end/83517
30499 * match.pd ((t * 2) / 2) -> t): Add missing :c.
30501 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
30503 PR middle-end/81897
30504 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
30505 basic blocks with a small number of successors.
30506 (convert_control_dep_chain_into_preds): Improve handling of
30508 (dump_predicates): Split apart into...
30509 (dump_pred_chain): ...here...
30510 (dump_pred_info): ...and here.
30511 (can_one_predicate_be_invalidated_p): Add debugging printfs.
30512 (can_chain_union_be_invalidated_p): Improve check for invalidation
30514 (uninit_uses_cannot_happen): Avoid unnecessary if
30515 convert_control_dep_chain_into_preds yielded nothing.
30517 2018-01-06 Martin Sebor <msebor@redhat.com>
30519 PR tree-optimization/83640
30520 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
30521 subtracting negative offset from size.
30522 (builtin_access::overlap): Adjust offset bounds of the access to fall
30523 within the size of the object if possible.
30525 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
30527 PR rtl-optimization/83699
30528 * expmed.c (extract_bit_field_1): Restrict the vector usage of
30529 extract_bit_field_as_subreg to cases in which the extracted
30530 value is also a vector.
30532 * lra-constraints.c (process_alt_operands): Test for the equivalence
30533 substitutions when detecting a possible reload cycle.
30535 2018-01-06 Jakub Jelinek <jakub@redhat.com>
30538 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
30539 by default if flag_selective_schedling{,2}. Formatting fixes.
30541 PR rtl-optimization/83682
30542 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
30543 if it has non-VECTOR_MODE element mode.
30544 (vec_duplicate_p): Likewise.
30546 PR middle-end/83694
30547 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
30548 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
30550 2018-01-05 Jakub Jelinek <jakub@redhat.com>
30553 * config/i386/i386-builtin.def
30554 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
30555 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
30556 Require also OPTION_MASK_ISA_AVX512F in addition to
30557 OPTION_MASK_ISA_GFNI.
30558 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
30559 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
30560 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
30561 to OPTION_MASK_ISA_GFNI.
30562 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
30563 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
30564 OPTION_MASK_ISA_AVX512BW.
30565 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
30566 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
30567 addition to OPTION_MASK_ISA_GFNI.
30568 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
30569 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
30570 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
30571 to OPTION_MASK_ISA_GFNI.
30572 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
30573 a requirement for all ISAs rather than any of them with a few
30575 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
30577 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
30578 bitmasks to be enabled with 3 exceptions, instead of requiring any
30579 enabled ISA with lots of exceptions.
30580 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
30581 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
30582 Change avx512bw in isa attribute to avx512f.
30583 * config/i386/sgxintrin.h: Add license boilerplate.
30584 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
30585 to __AVX512F__ and __AVX512VL to __AVX512VL__.
30586 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
30587 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
30589 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
30590 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
30591 temporarily sse2 rather than sse if not enabled already.
30594 * config/i386/sse.md (VI248_VLBW): Rename to ...
30595 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
30596 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
30597 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
30598 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
30599 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
30600 mode iterator instead of VI248_VLBW.
30602 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
30604 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
30605 (record_modified): Skip clobbers; add debug output.
30606 (param_change_prob): Use sreal frequencies.
30608 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
30610 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
30611 punt for user-aligned variables.
30613 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
30615 * tree-chrec.c (chrec_contains_symbols): Return true for
30618 2018-01-05 Sudakshina Das <sudi.das@arm.com>
30621 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
30622 of (x|y) == x for BICS pattern.
30624 2018-01-05 Jakub Jelinek <jakub@redhat.com>
30626 PR tree-optimization/83605
30627 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
30628 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
30631 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
30633 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
30634 * config/epiphany/rtems.h: New file.
30636 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30637 Uros Bizjak <ubizjak@gmail.com>
30640 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
30641 QIreg_operand instead of register_operand predicate.
30642 * config/i386/i386.c (ix86_rop_should_change_byte_p,
30643 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
30644 comments instead of -fmitigate[-_]rop.
30646 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30649 * cgraphunit.c (symbol_table::compile): Switch to text_section
30650 before calling assembly_start debug hook.
30651 * run-rtl-passes.c (run_rtl_passes): Likewise.
30654 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30656 * tree-vrp.c (extract_range_from_binary_expr_1): Check
30657 range_int_cst_p rather than !symbolic_range_p before calling
30658 extract_range_from_multiplicative_op_1.
30660 2018-01-04 Jeff Law <law@redhat.com>
30662 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
30663 redundant test in assertion.
30665 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30667 * doc/rtl.texi: Document machine_mode wrapper classes.
30669 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30671 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
30672 using tree_to_uhwi.
30674 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30676 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
30677 the VEC_PERM_EXPR fold to fail.
30679 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30682 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
30683 to switched_sections.
30685 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30688 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
30689 test for d.testing.
30691 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
30694 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
30695 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
30697 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30700 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
30701 is BLKmode and bitpos not zero or mode change is needed.
30703 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30706 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
30709 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
30712 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
30713 instead of MULT rtx. Update all corresponding splitters.
30714 (*saddl_se): Ditto.
30715 (*ssub<modesuffix>): Ditto.
30716 (*ssubl_se): Ditto.
30717 (*cmp_sadd_di): Update split patterns.
30718 (*cmp_sadd_si): Ditto.
30719 (*cmp_sadd_sidi): Ditto.
30720 (*cmp_ssub_di): Ditto.
30721 (*cmp_ssub_si): Ditto.
30722 (*cmp_ssub_sidi): Ditto.
30723 * config/alpha/predicates.md (const23_operand): New predicate.
30724 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
30725 Look for ASHIFT, not MULT inner operand.
30726 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
30728 2018-01-04 Martin Liska <mliska@suse.cz>
30730 PR gcov-profile/83669
30731 * gcov.c (output_intermediate_file): Add version to intermediate
30733 * doc/gcov.texi: Document new field 'version' in intermediate
30734 file format. Fix location of '-k' option of gcov command.
30736 2018-01-04 Martin Liska <mliska@suse.cz>
30739 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
30741 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30743 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
30745 2018-01-03 Martin Sebor <msebor@redhat.com>
30747 PR tree-optimization/83655
30748 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
30749 checking calls with invalid arguments.
30751 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30753 * tree-vect-stmts.c (vect_get_store_rhs): New function.
30754 (vectorizable_mask_load_store): Delete.
30755 (vectorizable_call): Return false for masked loads and stores.
30756 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
30757 instead of gimple_assign_rhs1.
30758 (vectorizable_load): Handle IFN_MASK_LOAD.
30759 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
30761 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30763 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
30765 (vectorizable_mask_load_store): ...here.
30766 (vectorizable_load): ...and here.
30768 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30770 * tree-vect-stmts.c (vect_build_all_ones_mask)
30771 (vect_build_zero_merge_argument): New functions, split out from...
30772 (vectorizable_load): ...here.
30774 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30776 * tree-vect-stmts.c (vect_check_store_rhs): New function,
30778 (vectorizable_mask_load_store): ...here.
30779 (vectorizable_store): ...and here.
30781 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30783 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
30785 (vectorizable_mask_load_store): ...here.
30787 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30789 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
30790 (vect_model_store_cost): Take a vec_load_store_type instead of a
30792 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
30793 (vect_model_store_cost): Take a vec_load_store_type instead of a
30795 (vectorizable_mask_load_store): Update accordingly.
30796 (vectorizable_store): Likewise.
30797 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
30799 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30801 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
30802 IFN_MASK_LOAD calls here rather than...
30803 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
30805 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30806 Alan Hayward <alan.hayward@arm.com>
30807 David Sherwood <david.sherwood@arm.com>
30809 * expmed.c (extract_bit_field_1): For vector extracts,
30810 fall back to extract_bit_field_as_subreg if vec_extract
30813 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30814 Alan Hayward <alan.hayward@arm.com>
30815 David Sherwood <david.sherwood@arm.com>
30817 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
30818 they are variable or constant sized.
30819 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
30820 slots for constant-sized data.
30822 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30823 Alan Hayward <alan.hayward@arm.com>
30824 David Sherwood <david.sherwood@arm.com>
30826 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
30827 handling COND_EXPRs with boolean comparisons, try to find a better
30828 basis for the mask type than the boolean itself.
30830 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30832 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
30833 is calculated and how it can be overridden.
30834 * genmodes.c (max_bitsize_mode_any_mode): New variable.
30835 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
30837 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
30840 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30841 Alan Hayward <alan.hayward@arm.com>
30842 David Sherwood <david.sherwood@arm.com>
30844 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
30845 Remove the mode argument.
30846 (aarch64_simd_valid_immediate): Remove the mode and inverse
30848 * config/aarch64/iterators.md (bitsize): New iterator.
30849 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
30850 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
30851 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
30852 aarch64_simd_valid_immediate.
30853 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
30854 (aarch64_reg_or_bic_imm): Likewise.
30855 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
30856 with an insn_type enum and msl with a modifier_type enum.
30857 Replace element_width with a scalar_mode. Change the shift
30858 to unsigned int. Add constructors for scalar_float_mode and
30859 scalar_int_mode elements.
30860 (aarch64_vect_float_const_representable_p): Delete.
30861 (aarch64_can_const_movi_rtx_p)
30862 (aarch64_simd_scalar_immediate_valid_for_move)
30863 (aarch64_simd_make_constant): Update call to
30864 aarch64_simd_valid_immediate.
30865 (aarch64_advsimd_valid_immediate_hs): New function.
30866 (aarch64_advsimd_valid_immediate): Likewise.
30867 (aarch64_simd_valid_immediate): Remove mode and inverse
30868 arguments. Rewrite to use the above. Use const_vec_duplicate_p
30869 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
30870 and aarch64_float_const_representable_p on the result.
30871 (aarch64_output_simd_mov_immediate): Remove mode argument.
30872 Update call to aarch64_simd_valid_immediate and use of
30873 simd_immediate_info.
30874 (aarch64_output_scalar_simd_mov_immediate): Update call
30877 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30878 Alan Hayward <alan.hayward@arm.com>
30879 David Sherwood <david.sherwood@arm.com>
30881 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
30882 (mode_nunits): Likewise CONST_MODE_NUNITS.
30883 * machmode.def (ADJUST_NUNITS): Document.
30884 * genmodes.c (mode_data::need_nunits_adj): New field.
30885 (blank_mode): Update accordingly.
30886 (adj_nunits): New variable.
30887 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
30889 (emit_mode_size_inline): Set need_bytesize_adj for all modes
30890 listed in adj_nunits.
30891 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
30892 listed in adj_nunits. Don't emit case statements for such modes.
30893 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
30894 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
30895 nothing if adj_nunits is nonnull.
30896 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
30897 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
30898 (emit_mode_fbit): Update use of print_maybe_const_decl.
30899 (emit_move_size): Likewise. Treat the array as non-const
30901 (emit_mode_adjustments): Handle adj_nunits.
30903 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30905 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
30906 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
30907 (VECTOR_MODES): Use it.
30908 (make_vector_modes): Take the prefix as an argument.
30910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30911 Alan Hayward <alan.hayward@arm.com>
30912 David Sherwood <david.sherwood@arm.com>
30914 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
30915 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
30916 for MODE_VECTOR_BOOL.
30917 * machmode.def (VECTOR_BOOL_MODE): Document.
30918 * genmodes.c (VECTOR_BOOL_MODE): New macro.
30919 (make_vector_bool_mode): New function.
30920 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
30922 * lto-streamer-in.c (lto_input_mode_table): Likewise.
30923 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
30925 * stor-layout.c (int_mode_for_mode): Likewise.
30926 * tree.c (build_vector_type_for_mode): Likewise.
30927 * varasm.c (output_constant_pool_2): Likewise.
30928 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
30929 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
30930 for MODE_VECTOR_BOOL.
30931 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
30932 of mode class checks.
30933 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
30934 instead of a list of mode class checks.
30935 (expand_vector_scalar_condition): Likewise.
30936 (type_for_widest_vector_mode): Handle BImode as an inner mode.
30938 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30939 Alan Hayward <alan.hayward@arm.com>
30940 David Sherwood <david.sherwood@arm.com>
30942 * machmode.h (mode_size): Change from unsigned short to
30944 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
30945 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
30946 or if measurement_type is not polynomial.
30947 (fixed_size_mode::includes_p): Check for constant-sized modes.
30948 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
30949 return a poly_uint16 rather than an unsigned short.
30950 (emit_mode_size): Change the type of mode_size from unsigned short
30951 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
30952 (emit_mode_adjustments): Cope with polynomial vector sizes.
30953 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30955 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30957 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
30958 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
30959 * caller-save.c (setup_save_areas): Likewise.
30960 (replace_reg_with_saved_mem): Likewise.
30961 * calls.c (emit_library_call_value_1): Likewise.
30962 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
30963 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
30964 (gen_lowpart_for_combine): Likewise.
30965 * convert.c (convert_to_integer_1): Likewise.
30966 * cse.c (equiv_constant, cse_insn): Likewise.
30967 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
30968 (cselib_subst_to_values): Likewise.
30969 * dce.c (word_dce_process_block): Likewise.
30970 * df-problems.c (df_word_lr_mark_ref): Likewise.
30971 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
30972 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
30973 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
30974 (rtl_for_decl_location): Likewise.
30975 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
30976 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
30977 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
30978 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
30979 (expand_expr_real_1): Likewise.
30980 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
30981 (pad_below): Likewise.
30982 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
30983 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
30984 * ira.c (get_subreg_tracking_sizes): Likewise.
30985 * ira-build.c (ira_create_allocno_objects): Likewise.
30986 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
30987 (ira_sort_regnos_for_alter_reg): Likewise.
30988 * ira-costs.c (record_operand_costs): Likewise.
30989 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
30990 (resolve_simple_move): Likewise.
30991 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
30992 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
30993 (lra_constraints): Likewise.
30994 (CONST_POOL_OK_P): Reject variable-sized modes.
30995 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
30996 (add_pseudo_to_slot, lra_spill): Likewise.
30997 * omp-low.c (omp_clause_aligned_alignment): Likewise.
30998 * optabs-query.c (get_best_extraction_insn): Likewise.
30999 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
31000 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
31001 (expand_mult_highpart, valid_multiword_target_p): Likewise.
31002 * recog.c (offsettable_address_addr_space_p): Likewise.
31003 * regcprop.c (maybe_mode_change): Likewise.
31004 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
31005 * regrename.c (build_def_use): Likewise.
31006 * regstat.c (dump_reg_info): Likewise.
31007 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
31008 (find_reloads, find_reloads_subreg_address): Likewise.
31009 * reload1.c (eliminate_regs_1): Likewise.
31010 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
31011 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
31012 (simplify_binary_operation_1, simplify_subreg): Likewise.
31013 * targhooks.c (default_function_arg_padding): Likewise.
31014 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
31015 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
31016 (verify_gimple_assign_ternary): Likewise.
31017 * tree-inline.c (estimate_move_cost): Likewise.
31018 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31019 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
31020 (get_address_cost_ainc): Likewise.
31021 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31022 (vect_supportable_dr_alignment): Likewise.
31023 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31024 (vectorizable_reduction): Likewise.
31025 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
31026 (vectorizable_operation, vectorizable_load): Likewise.
31027 * tree.c (build_same_sized_truth_vector_type): Likewise.
31028 * valtrack.c (cleanup_auto_inc_dec): Likewise.
31029 * var-tracking.c (emit_note_insn_var_location): Likewise.
31030 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
31031 (ADDR_VEC_ALIGN): Likewise.
31033 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31034 Alan Hayward <alan.hayward@arm.com>
31035 David Sherwood <david.sherwood@arm.com>
31037 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
31039 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
31040 or if measurement_type is polynomial.
31041 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
31042 * combine.c (make_extraction): Likewise.
31043 * dse.c (find_shift_sequence): Likewise.
31044 * dwarf2out.c (mem_loc_descriptor): Likewise.
31045 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
31046 (extract_bit_field, extract_low_bits): Likewise.
31047 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
31048 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
31049 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
31050 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
31051 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
31052 * reload.c (find_reloads): Likewise.
31053 * reload1.c (alter_reg): Likewise.
31054 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
31055 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
31056 * tree-if-conv.c (predicate_mem_writes): Likewise.
31057 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
31058 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
31059 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
31060 * valtrack.c (dead_debug_insert_temp): Likewise.
31061 * varasm.c (mergeable_constant_section): Likewise.
31062 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
31064 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31065 Alan Hayward <alan.hayward@arm.com>
31066 David Sherwood <david.sherwood@arm.com>
31068 * expr.c (expand_assignment): Cope with polynomial mode sizes
31069 when assigning to a CONCAT.
31071 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31072 Alan Hayward <alan.hayward@arm.com>
31073 David Sherwood <david.sherwood@arm.com>
31075 * machmode.h (mode_precision): Change from unsigned short to
31077 (mode_to_precision): Return a poly_uint16 rather than an unsigned
31079 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
31080 or if measurement_type is not polynomial.
31081 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
31082 in which the mode is already known to be a scalar_int_mode.
31083 * genmodes.c (emit_mode_precision): Change the type of mode_precision
31084 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
31086 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31087 for GET_MODE_PRECISION.
31088 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31089 for GET_MODE_PRECISION.
31090 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
31092 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
31093 (expand_field_assignment, make_extraction): Likewise.
31094 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
31095 (get_last_value): Likewise.
31096 * convert.c (convert_to_integer_1): Likewise.
31097 * cse.c (cse_insn): Likewise.
31098 * expr.c (expand_expr_real_1): Likewise.
31099 * lra-constraints.c (simplify_operand_subreg): Likewise.
31100 * optabs-query.c (can_atomic_load_p): Likewise.
31101 * optabs.c (expand_atomic_load): Likewise.
31102 (expand_atomic_store): Likewise.
31103 * ree.c (combine_reaching_defs): Likewise.
31104 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
31105 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
31106 * tree.h (type_has_mode_precision_p): Likewise.
31107 * ubsan.c (instrument_si_overflow): Likewise.
31109 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31110 Alan Hayward <alan.hayward@arm.com>
31111 David Sherwood <david.sherwood@arm.com>
31113 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
31114 polynomial numbers of units.
31115 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
31116 (valid_vector_subparts_p): New function.
31117 (build_vector_type): Remove temporary shim and take the number
31118 of units as a poly_uint64 rather than an int.
31119 (build_opaque_vector_type): Take the number of units as a
31120 poly_uint64 rather than an int.
31121 * tree.c (build_vector_from_ctor): Handle polynomial
31122 TYPE_VECTOR_SUBPARTS.
31123 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
31124 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
31125 (build_vector_from_val): If the number of units is variable,
31126 use build_vec_duplicate_cst for constant operands and
31127 VEC_DUPLICATE_EXPR otherwise.
31128 (make_vector_type): Remove temporary is_constant ().
31129 (build_vector_type, build_opaque_vector_type): Take the number of
31130 units as a poly_uint64 rather than an int.
31131 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
31133 * cfgexpand.c (expand_debug_expr): Likewise.
31134 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
31135 (store_constructor, expand_expr_real_1): Likewise.
31136 (const_scalar_mask_from_tree): Likewise.
31137 * fold-const-call.c (fold_const_reduction): Likewise.
31138 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
31139 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
31140 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
31141 (fold_relational_const): Likewise.
31142 (native_interpret_vector): Likewise. Change the size from an
31143 int to an unsigned int.
31144 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
31145 TYPE_VECTOR_SUBPARTS.
31146 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
31147 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
31148 duplicating a non-constant operand into a variable-length vector.
31149 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
31150 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
31151 * ipa-icf.c (sem_variable::equals): Likewise.
31152 * match.pd: Likewise.
31153 * omp-simd-clone.c (simd_clone_subparts): Likewise.
31154 * print-tree.c (print_node): Likewise.
31155 * stor-layout.c (layout_type): Likewise.
31156 * targhooks.c (default_builtin_vectorization_cost): Likewise.
31157 * tree-cfg.c (verify_gimple_comparison): Likewise.
31158 (verify_gimple_assign_binary): Likewise.
31159 (verify_gimple_assign_ternary): Likewise.
31160 (verify_gimple_assign_single): Likewise.
31161 * tree-pretty-print.c (dump_generic_node): Likewise.
31162 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31163 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
31164 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
31165 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
31166 (vect_shift_permute_load_chain): Likewise.
31167 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
31168 (expand_vector_condition, optimize_vector_constructor): Likewise.
31169 (lower_vec_perm, get_compute_type): Likewise.
31170 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31171 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
31172 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
31173 (vect_recog_mask_conversion_pattern): Likewise.
31174 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
31175 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
31176 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31177 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
31178 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
31179 (vectorizable_shift, vectorizable_operation, vectorizable_store)
31180 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
31181 (supportable_widening_operation): Likewise.
31182 (supportable_narrowing_operation): Likewise.
31183 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
31185 * varasm.c (output_constant): Likewise.
31187 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31188 Alan Hayward <alan.hayward@arm.com>
31189 David Sherwood <david.sherwood@arm.com>
31191 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
31192 so that both the length == 3 and length != 3 cases set up their
31193 own permute vectors. Add comments explaining why we know the
31194 number of elements is constant.
31195 (vect_permute_load_chain): Likewise.
31197 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31198 Alan Hayward <alan.hayward@arm.com>
31199 David Sherwood <david.sherwood@arm.com>
31201 * machmode.h (mode_nunits): Change from unsigned char to
31203 (ONLY_FIXED_SIZE_MODES): New macro.
31204 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
31205 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
31206 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
31208 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
31209 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
31210 or if measurement_type is not polynomial.
31211 * genmodes.c (ZERO_COEFFS): New macro.
31212 (emit_mode_nunits_inline): Make mode_nunits_inline return a
31214 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
31215 Use ZERO_COEFFS when emitting initializers.
31216 * data-streamer.h (bp_pack_poly_value): New function.
31217 (bp_unpack_poly_value): Likewise.
31218 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31219 for GET_MODE_NUNITS.
31220 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31221 for GET_MODE_NUNITS.
31222 * tree.c (make_vector_type): Remove temporary shim and make
31223 the real function take the number of units as a poly_uint64
31224 rather than an int.
31225 (build_vector_type_for_mode): Handle polynomial nunits.
31226 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
31227 * emit-rtl.c (const_vec_series_p_1): Likewise.
31228 (gen_rtx_CONST_VECTOR): Likewise.
31229 * fold-const.c (test_vec_duplicate_folding): Likewise.
31230 * genrecog.c (validate_pattern): Likewise.
31231 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
31232 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
31233 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
31234 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
31235 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
31236 * rtlanal.c (subreg_get_info): Likewise.
31237 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31238 (vect_grouped_load_supported): Likewise.
31239 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
31240 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
31241 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
31242 (simplify_const_unary_operation, simplify_binary_operation_1)
31243 (simplify_const_binary_operation, simplify_ternary_operation)
31244 (test_vector_ops_duplicate, test_vector_ops): Likewise.
31245 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
31246 instead of CONST_VECTOR_NUNITS.
31247 * varasm.c (output_constant_pool_2): Likewise.
31248 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
31249 explicit-encoded elements in the XVEC for variable-length vectors.
31251 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31253 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
31255 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31256 Alan Hayward <alan.hayward@arm.com>
31257 David Sherwood <david.sherwood@arm.com>
31259 * coretypes.h (fixed_size_mode): Declare.
31260 (fixed_size_mode_pod): New typedef.
31261 * builtins.h (target_builtins::x_apply_args_mode)
31262 (target_builtins::x_apply_result_mode): Change type to
31263 fixed_size_mode_pod.
31264 * builtins.c (apply_args_size, apply_result_size, result_vector)
31265 (expand_builtin_apply_args_1, expand_builtin_apply)
31266 (expand_builtin_return): Update accordingly.
31268 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31270 * cse.c (hash_rtx_cb): Hash only the encoded elements.
31271 * cselib.c (cselib_hash_rtx): Likewise.
31272 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
31273 CONST_VECTOR encoding.
31275 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31276 Jeff Law <law@redhat.com>
31279 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
31280 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
31281 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
31282 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
31285 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
31286 explicitly probe *sp in a noreturn function if there were any callee
31287 register saves or frame pointer is needed.
31289 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31292 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
31293 BLKmode for ternary, binary or unary expressions.
31296 * var-tracking.c (delete_vta_debug_insn): New inline function.
31297 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
31298 insns from get_insns () to NULL instead of each bb separately.
31299 Use delete_vta_debug_insn. No longer static.
31300 (vt_debug_insns_local, variable_tracking_main_1): Adjust
31301 delete_vta_debug_insns callers.
31302 * rtl.h (delete_vta_debug_insns): Declare.
31303 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
31304 instead of variable_tracking_main.
31306 2018-01-03 Martin Sebor <msebor@redhat.com>
31308 PR tree-optimization/83603
31309 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
31310 arguments past the endof the argument list in functions declared
31311 without a prototype.
31312 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
31313 Avoid checking when arguments are null.
31315 2018-01-03 Martin Sebor <msebor@redhat.com>
31318 * doc/extend.texi (attribute const): Fix a typo.
31319 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
31320 issuing -Wsuggest-attribute for void functions.
31322 2018-01-03 Martin Sebor <msebor@redhat.com>
31324 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
31325 offset_int::from instead of wide_int::to_shwi.
31326 (maybe_diag_overlap): Remove assertion.
31327 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
31328 * gimple-ssa-sprintf.c (format_directive): Same.
31329 (parse_directive): Same.
31330 (sprintf_dom_walker::compute_format_length): Same.
31331 (try_substitute_return_value): Same.
31333 2018-01-03 Jeff Law <law@redhat.com>
31335 PR middle-end/83654
31336 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
31337 non-constant residual for zero at runtime and avoid probing in
31338 that case. Reorganize code for trailing problem to mirror handling
31341 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31343 PR tree-optimization/83501
31344 * tree-ssa-strlen.c (get_string_cst): New.
31345 (handle_char_store): Call get_string_cst.
31347 2018-01-03 Martin Liska <mliska@suse.cz>
31349 PR tree-optimization/83593
31350 * tree-ssa-strlen.c: Include tree-cfg.h.
31351 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
31352 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
31353 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
31355 (strlen_dom_walker::before_dom_children): Call
31356 gimple_purge_dead_eh_edges. Dump tranformation with details
31358 (strlen_dom_walker::before_dom_children): Update call by adding
31359 new argument cleanup_eh.
31360 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
31362 2018-01-03 Martin Liska <mliska@suse.cz>
31365 * cif-code.def (VARIADIC_THUNK): New enum value.
31366 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
31369 2018-01-03 Jan Beulich <jbeulich@suse.com>
31371 * sse.md (mov<mode>_internal): Tighten condition for when to use
31372 vmovdqu<ssescalarsize> for TI and OI modes.
31374 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31376 Update copyright years.
31378 2018-01-03 Martin Liska <mliska@suse.cz>
31381 * ipa-visibility.c (function_and_variable_visibility): Skip
31382 functions with noipa attribure.
31384 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31386 * gcc.c (process_command): Update copyright notice dates.
31387 * gcov-dump.c (print_version): Ditto.
31388 * gcov.c (print_version): Ditto.
31389 * gcov-tool.c (print_version): Ditto.
31390 * gengtype.c (create_file): Ditto.
31391 * doc/cpp.texi: Bump @copying's copyright year.
31392 * doc/cppinternals.texi: Ditto.
31393 * doc/gcc.texi: Ditto.
31394 * doc/gccint.texi: Ditto.
31395 * doc/gcov.texi: Ditto.
31396 * doc/install.texi: Ditto.
31397 * doc/invoke.texi: Ditto.
31399 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31401 * vector-builder.h (vector_builder::m_full_nelts): Change from
31402 unsigned int to poly_uint64.
31403 (vector_builder::full_nelts): Update prototype accordingly.
31404 (vector_builder::new_vector): Likewise.
31405 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
31406 (vector_builder::operator ==): Likewise.
31407 (vector_builder::finalize): Likewise.
31408 * int-vector-builder.h (int_vector_builder::int_vector_builder):
31409 Take the number of elements as a poly_uint64 rather than an
31411 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
31412 from unsigned int to poly_uint64.
31413 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
31414 (vec_perm_indices::new_vector): Likewise.
31415 (vec_perm_indices::length): Likewise.
31416 (vec_perm_indices::nelts_per_input): Likewise.
31417 (vec_perm_indices::input_nelts): Likewise.
31418 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
31419 number of elements per input as a poly_uint64 rather than an
31420 unsigned int. Use the original encoding for variable-length
31421 vectors, rather than clamping each individual element.
31422 For the second and subsequent elements in each pattern,
31423 clamp the step and base before clamping their sum.
31424 (vec_perm_indices::series_p): Handle polynomial element counts.
31425 (vec_perm_indices::all_in_range_p): Likewise.
31426 (vec_perm_indices_to_tree): Likewise.
31427 (vec_perm_indices_to_rtx): Likewise.
31428 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
31429 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
31430 (tree_vector_builder::new_binary_operation): Handle polynomial
31431 element counts. Return false if we need to know the number
31432 of elements at compile time.
31433 * fold-const.c (fold_vec_perm): Punt if the number of elements
31434 isn't known at compile time.
31436 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31438 * vec-perm-indices.h (vec_perm_builder): Change element type
31439 from HOST_WIDE_INT to poly_int64.
31440 (vec_perm_indices::element_type): Update accordingly.
31441 (vec_perm_indices::clamp): Handle polynomial element_types.
31442 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
31443 (vec_perm_indices::all_in_range_p): Likewise.
31444 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
31446 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
31447 polynomial vec_perm_indices element types.
31448 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
31449 * fold-const.c (fold_vec_perm): Likewise.
31450 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
31451 * tree-vect-generic.c (lower_vec_perm): Likewise.
31452 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
31453 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
31454 element type to HOST_WIDE_INT.
31456 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31457 Alan Hayward <alan.hayward@arm.com>
31458 David Sherwood <david.sherwood@arm.com>
31460 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
31461 rather than an int. Use plus_constant.
31462 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
31463 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
31465 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31466 Alan Hayward <alan.hayward@arm.com>
31467 David Sherwood <david.sherwood@arm.com>
31469 * calls.c (emit_call_1, expand_call): Change struct_value_size from
31470 a HOST_WIDE_INT to a poly_int64.
31472 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31473 Alan Hayward <alan.hayward@arm.com>
31474 David Sherwood <david.sherwood@arm.com>
31476 * calls.c (load_register_parameters): Cope with polynomial
31477 mode sizes. Require a constant size for BLKmode parameters
31478 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
31479 forces a parameter to be padded at the lsb end in order to
31480 fill a complete number of words, require the parameter size
31481 to be ordered wrt UNITS_PER_WORD.
31483 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31484 Alan Hayward <alan.hayward@arm.com>
31485 David Sherwood <david.sherwood@arm.com>
31487 * reload1.c (spill_stack_slot_width): Change element type
31488 from unsigned int to poly_uint64_pod.
31489 (alter_reg): Treat mode sizes as polynomial.
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 * reload.c (complex_word_subreg_p): New function.
31496 (reload_inner_reg_of_subreg, push_reload): Use it.
31498 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31499 Alan Hayward <alan.hayward@arm.com>
31500 David Sherwood <david.sherwood@arm.com>
31502 * lra-constraints.c (process_alt_operands): Reject matched
31503 operands whose sizes aren't ordered.
31504 (match_reload): Refer to this check here.
31506 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31507 Alan Hayward <alan.hayward@arm.com>
31508 David Sherwood <david.sherwood@arm.com>
31510 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
31511 that the mode size is in the set {1, 2, 4, 8, 16}.
31513 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31514 Alan Hayward <alan.hayward@arm.com>
31515 David Sherwood <david.sherwood@arm.com>
31517 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
31518 Use plus_constant instead of gen_rtx_PLUS.
31520 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31521 Alan Hayward <alan.hayward@arm.com>
31522 David Sherwood <david.sherwood@arm.com>
31524 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
31525 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
31526 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
31527 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
31528 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
31529 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
31530 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
31531 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
31532 * config/i386/i386.c (ix86_push_rounding): ...this new function.
31533 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
31535 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
31536 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
31537 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
31538 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
31539 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
31540 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
31541 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
31542 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
31543 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
31544 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
31546 * expr.c (emit_move_resolve_push): Treat the input and result
31547 of PUSH_ROUNDING as a poly_int64.
31548 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
31549 (emit_push_insn): Likewise.
31550 * lra-eliminations.c (mark_not_eliminable): Likewise.
31551 * recog.c (push_operand): Likewise.
31552 * reload1.c (elimination_effects): Likewise.
31553 * rtlanal.c (nonzero_bits1): Likewise.
31554 * calls.c (store_one_arg): Likewise. Require the padding to be
31555 known at compile time.
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 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
31562 Use plus_constant instead of gen_rtx_PLUS.
31564 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31565 Alan Hayward <alan.hayward@arm.com>
31566 David Sherwood <david.sherwood@arm.com>
31568 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
31569 rather than an int.
31571 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31572 Alan Hayward <alan.hayward@arm.com>
31573 David Sherwood <david.sherwood@arm.com>
31575 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
31576 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
31577 via stack temporaries. Treat the mode size as polynomial too.
31579 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31580 Alan Hayward <alan.hayward@arm.com>
31581 David Sherwood <david.sherwood@arm.com>
31583 * expr.c (expand_expr_real_2): When handling conversions involving
31584 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
31585 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
31586 as a poly_uint64 too.
31588 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31589 Alan Hayward <alan.hayward@arm.com>
31590 David Sherwood <david.sherwood@arm.com>
31592 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
31594 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31595 Alan Hayward <alan.hayward@arm.com>
31596 David Sherwood <david.sherwood@arm.com>
31598 * combine.c (can_change_dest_mode): Handle polynomial
31599 REGMODE_NATURAL_SIZE.
31600 * expmed.c (store_bit_field_1): Likewise.
31601 * expr.c (store_constructor): Likewise.
31602 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
31603 and polynomial REGMODE_NATURAL_SIZE.
31604 (gen_lowpart_common): Likewise.
31605 * reginfo.c (record_subregs_of_mode): Likewise.
31606 * rtlanal.c (read_modify_subreg_p): Likewise.
31608 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31609 Alan Hayward <alan.hayward@arm.com>
31610 David Sherwood <david.sherwood@arm.com>
31612 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
31613 numbers of elements.
31615 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31616 Alan Hayward <alan.hayward@arm.com>
31617 David Sherwood <david.sherwood@arm.com>
31619 * match.pd: Cope with polynomial numbers of vector elements.
31621 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31622 Alan Hayward <alan.hayward@arm.com>
31623 David Sherwood <david.sherwood@arm.com>
31625 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
31626 in a POINTER_PLUS_EXPR.
31628 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31629 Alan Hayward <alan.hayward@arm.com>
31630 David Sherwood <david.sherwood@arm.com>
31632 * omp-simd-clone.c (simd_clone_subparts): New function.
31633 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
31634 (ipa_simd_modify_function_body): Likewise.
31636 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31637 Alan Hayward <alan.hayward@arm.com>
31638 David Sherwood <david.sherwood@arm.com>
31640 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
31641 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
31642 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
31643 (expand_vector_condition, vector_element): Likewise.
31644 (subparts_gt): New function.
31645 (get_compute_type): Use subparts_gt.
31646 (count_type_subparts): Delete.
31647 (expand_vector_operations_1): Use subparts_gt instead of
31648 count_type_subparts.
31650 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31651 Alan Hayward <alan.hayward@arm.com>
31652 David Sherwood <david.sherwood@arm.com>
31654 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
31655 (vect_compile_time_alias): ...this new function. Do the calculation
31656 on poly_ints rather than trees.
31657 (vect_prune_runtime_alias_test_list): Update call accordingly.
31659 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31660 Alan Hayward <alan.hayward@arm.com>
31661 David Sherwood <david.sherwood@arm.com>
31663 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
31665 (vect_schedule_slp_instance): Likewise.
31667 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31668 Alan Hayward <alan.hayward@arm.com>
31669 David Sherwood <david.sherwood@arm.com>
31671 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
31672 constant and extern definitions for variable-length vectors.
31673 (vect_get_constant_vectors): Note that the number of units
31674 is known to be constant.
31676 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31677 Alan Hayward <alan.hayward@arm.com>
31678 David Sherwood <david.sherwood@arm.com>
31680 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
31681 of units as polynomial. Choose between WIDE and NARROW based
31684 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31685 Alan Hayward <alan.hayward@arm.com>
31686 David Sherwood <david.sherwood@arm.com>
31688 * tree-vect-stmts.c (simd_clone_subparts): New function.
31689 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
31691 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31692 Alan Hayward <alan.hayward@arm.com>
31693 David Sherwood <david.sherwood@arm.com>
31695 * tree-vect-stmts.c (vectorizable_call): Treat the number of
31696 vectors as polynomial. Use build_index_vector for
31697 IFN_GOMP_SIMD_LANE.
31699 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31700 Alan Hayward <alan.hayward@arm.com>
31701 David Sherwood <david.sherwood@arm.com>
31703 * tree-vect-stmts.c (get_load_store_type): Treat the number of
31704 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
31705 for variable-length vectors.
31706 (vectorizable_mask_load_store): Treat the number of units as
31707 polynomial, asserting that it is constant if the condition has
31708 already been enforced.
31709 (vectorizable_store, vectorizable_load): Likewise.
31711 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31712 Alan Hayward <alan.hayward@arm.com>
31713 David Sherwood <david.sherwood@arm.com>
31715 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
31716 of units as polynomial. Punt if we can't tell at compile time
31717 which vector contains the final result.
31719 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31720 Alan Hayward <alan.hayward@arm.com>
31721 David Sherwood <david.sherwood@arm.com>
31723 * tree-vect-loop.c (vectorizable_induction): Treat the number
31724 of units as polynomial. Punt on SLP inductions. Use an integer
31725 VEC_SERIES_EXPR for variable-length integer reductions. Use a
31726 cast of such a series for variable-length floating-point
31729 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31730 Alan Hayward <alan.hayward@arm.com>
31731 David Sherwood <david.sherwood@arm.com>
31733 * tree.h (build_index_vector): Declare.
31734 * tree.c (build_index_vector): New function.
31735 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
31736 of units as polynomial, forcibly converting it to a constant if
31737 vectorizable_reduction has already enforced the condition.
31738 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
31739 to create a {1,2,3,...} vector.
31740 (vectorizable_reduction): Treat the number of units as polynomial.
31741 Choose vectype_in based on the largest scalar element size rather
31742 than the smallest number of units. Enforce the restrictions
31745 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31746 Alan Hayward <alan.hayward@arm.com>
31747 David Sherwood <david.sherwood@arm.com>
31749 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
31750 number of units as polynomial.
31752 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31753 Alan Hayward <alan.hayward@arm.com>
31754 David Sherwood <david.sherwood@arm.com>
31756 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
31757 * target.def (autovectorize_vector_sizes): Return the vector sizes
31758 by pointer, using vector_sizes rather than a bitmask.
31759 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
31760 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
31761 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
31763 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
31764 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
31765 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
31766 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
31767 * omp-general.c (omp_max_vf): Likewise.
31768 * omp-low.c (omp_clause_aligned_alignment): Likewise.
31769 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
31770 * tree-vect-loop.c (vect_analyze_loop): Likewise.
31771 * tree-vect-slp.c (vect_slp_bb): Likewise.
31772 * doc/tm.texi: Regenerate.
31773 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
31775 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
31776 the vector size as a poly_uint64 rather than an unsigned int.
31777 (current_vector_size): Change from an unsigned int to a poly_uint64.
31778 (get_vectype_for_scalar_type): Update accordingly.
31779 * tree.h (build_truth_vector_type): Take the size and number of
31780 units as a poly_uint64 rather than an unsigned int.
31781 (build_vector_type): Add a temporary overload that takes
31782 the number of units as a poly_uint64 rather than an unsigned int.
31783 * tree.c (make_vector_type): Likewise.
31784 (build_truth_vector_type): Take the number of units as a poly_uint64
31785 rather than an unsigned int.
31787 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31788 Alan Hayward <alan.hayward@arm.com>
31789 David Sherwood <david.sherwood@arm.com>
31791 * target.def (get_mask_mode): Take the number of units and length
31792 as poly_uint64s rather than unsigned ints.
31793 * targhooks.h (default_get_mask_mode): Update accordingly.
31794 * targhooks.c (default_get_mask_mode): Likewise.
31795 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
31796 * doc/tm.texi: Regenerate.
31798 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31799 Alan Hayward <alan.hayward@arm.com>
31800 David Sherwood <david.sherwood@arm.com>
31802 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
31803 * omp-general.c (omp_max_vf): Likewise.
31804 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
31805 (expand_omp_simd): Handle polynomial safelen.
31806 * omp-low.c (omplow_simd_context): Add a default constructor.
31807 (omplow_simd_context::max_vf): Change from int to poly_uint64.
31808 (lower_rec_simd_input_clauses): Update accordingly.
31809 (lower_rec_input_clauses): Likewise.
31811 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31812 Alan Hayward <alan.hayward@arm.com>
31813 David Sherwood <david.sherwood@arm.com>
31815 * tree-vectorizer.h (vect_nunits_for_cost): New function.
31816 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
31817 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
31818 (vect_analyze_slp_cost): Likewise.
31819 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
31820 (vect_model_load_cost): Likewise.
31822 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31823 Alan Hayward <alan.hayward@arm.com>
31824 David Sherwood <david.sherwood@arm.com>
31826 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
31827 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
31828 from an unsigned int * to a poly_uint64_pod *.
31829 (calculate_unrolling_factor): New function.
31830 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
31832 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31833 Alan Hayward <alan.hayward@arm.com>
31834 David Sherwood <david.sherwood@arm.com>
31836 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
31837 from an unsigned int to a poly_uint64.
31838 (_loop_vec_info::slp_unrolling_factor): Likewise.
31839 (_loop_vec_info::vectorization_factor): Change from an int
31841 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
31842 (vect_get_num_vectors): New function.
31843 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
31844 (vect_get_num_copies): Use vect_get_num_vectors.
31845 (vect_analyze_data_ref_dependences): Change max_vf from an int *
31846 to an unsigned int *.
31847 (vect_analyze_data_refs): Change min_vf from an int * to a
31849 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
31850 than an unsigned HOST_WIDE_INT.
31851 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
31852 (vect_analyze_data_ref_dependence): Change max_vf from an int *
31853 to an unsigned int *.
31854 (vect_analyze_data_ref_dependences): Likewise.
31855 (vect_compute_data_ref_alignment): Handle polynomial vf.
31856 (vect_enhance_data_refs_alignment): Likewise.
31857 (vect_prune_runtime_alias_test_list): Likewise.
31858 (vect_shift_permute_load_chain): Likewise.
31859 (vect_supportable_dr_alignment): Likewise.
31860 (dependence_distance_ge_vf): Take the vectorization factor as a
31861 poly_uint64 rather than an unsigned HOST_WIDE_INT.
31862 (vect_analyze_data_refs): Change min_vf from an int * to a
31864 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
31865 vfm1 as a poly_uint64 rather than an int. Make the same change
31866 for the returned bound_scalar.
31867 (vect_gen_vector_loop_niters): Handle polynomial vf.
31868 (vect_do_peeling): Likewise. Update call to
31869 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
31870 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
31872 * tree-vect-loop.c (vect_determine_vectorization_factor)
31873 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
31874 (vect_get_known_peeling_cost): Likewise.
31875 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
31876 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
31877 (vect_transform_loop): Likewise. Use the lowest possible VF when
31878 updating the upper bounds of the loop.
31879 (vect_min_worthwhile_factor): Make static. Return an unsigned int
31880 rather than an int.
31881 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
31882 polynomial unroll factors.
31883 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
31884 (vect_make_slp_decision): Likewise.
31885 (vect_supported_load_permutation_p): Likewise, and polynomial
31887 (vect_analyze_slp_cost): Handle polynomial vf.
31888 (vect_slp_analyze_node_operations): Likewise.
31889 (vect_slp_analyze_bb_1): Likewise.
31890 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
31891 than an unsigned HOST_WIDE_INT.
31892 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
31893 (vectorizable_load): Handle polynomial vf.
31894 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
31896 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
31898 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31899 Alan Hayward <alan.hayward@arm.com>
31900 David Sherwood <david.sherwood@arm.com>
31902 * match.pd: Handle bit operations involving three constants
31903 and try to fold one pair.
31905 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31907 * tree-vect-loop-manip.c: Include gimple-fold.h.
31908 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
31909 niters_maybe_zero parameters. Handle other cases besides a step of 1.
31910 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
31911 Add a path that uses a step of VF instead of 1, but disable it
31913 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
31914 and niters_no_overflow parameters. Update calls to
31915 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
31916 Create a new SSA name if the latter choses to use a ste other
31917 than zero, and return it via niters_vector_mult_vf_var.
31918 * tree-vect-loop.c (vect_transform_loop): Update calls to
31919 vect_do_peeling, vect_gen_vector_loop_niters and
31920 slpeel_make_loop_iterate_ntimes.
31921 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
31922 (vect_gen_vector_loop_niters): Update declarations after above changes.
31924 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
31926 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
31927 128-bit round to integer instructions.
31928 (ceil<mode>2): Likewise.
31929 (btrunc<mode>2): Likewise.
31930 (round<mode>2): Likewise.
31932 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31934 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
31935 unaligned VSX load/store on P8/P9.
31936 (expand_block_clear): Allow the use of unaligned VSX
31937 load/store on P8/P9.
31939 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31941 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
31943 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
31944 swap associated with both a load and a store.
31946 2018-01-02 Andrew Waterman <andrew@sifive.com>
31948 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
31949 * config/riscv/riscv.md (clear_cache): Use it.
31951 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
31953 * web.c: Remove out-of-date comment.
31955 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31957 * expr.c (fixup_args_size_notes): Check that any existing
31958 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
31959 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
31960 (emit_single_push_insn): ...here.
31962 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31964 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
31965 (const_vector_encoded_nelts): New function.
31966 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
31967 (const_vector_int_elt, const_vector_elt): Declare.
31968 * emit-rtl.c (const_vector_int_elt_1): New function.
31969 (const_vector_elt): Likewise.
31970 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
31971 of CONST_VECTOR_ELT.
31973 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31975 * expr.c: Include rtx-vector-builder.h.
31976 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
31977 directly on the tree encoding.
31978 (const_vector_from_tree): Likewise.
31979 * optabs.c: Include rtx-vector-builder.h.
31980 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
31981 sequence of "u" values.
31982 * vec-perm-indices.c: Include rtx-vector-builder.h.
31983 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
31984 directly on the vec_perm_indices encoding.
31986 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31988 * doc/rtl.texi (const_vector): Describe new encoding scheme.
31989 * Makefile.in (OBJS): Add rtx-vector-builder.o.
31990 * rtx-vector-builder.h: New file.
31991 * rtx-vector-builder.c: Likewise.
31992 * rtl.h (rtx_def::u2): Add a const_vector field.
31993 (CONST_VECTOR_NPATTERNS): New macro.
31994 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
31995 (CONST_VECTOR_DUPLICATE_P): Likewise.
31996 (CONST_VECTOR_STEPPED_P): Likewise.
31997 (CONST_VECTOR_ENCODED_ELT): Likewise.
31998 (const_vec_duplicate_p): Check for a duplicated vector encoding.
31999 (unwrap_const_vec_duplicate): Likewise.
32000 (const_vec_series_p): Check for a non-duplicated vector encoding.
32001 Say that the function only returns true for integer vectors.
32002 * emit-rtl.c: Include rtx-vector-builder.h.
32003 (gen_const_vec_duplicate_1): Delete.
32004 (gen_const_vector): Call gen_const_vec_duplicate instead of
32005 gen_const_vec_duplicate_1.
32006 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
32007 (gen_const_vec_duplicate): Use rtx_vector_builder.
32008 (gen_const_vec_series): Likewise.
32009 (gen_rtx_CONST_VECTOR): Likewise.
32010 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
32011 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
32012 Build a new vector rather than modifying a CONST_VECTOR in-place.
32013 (handle_special_swappables): Update call accordingly.
32014 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
32015 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
32016 Build a new vector rather than modifying a CONST_VECTOR in-place.
32017 (handle_special_swappables): Update call accordingly.
32019 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32021 * simplify-rtx.c (simplify_const_binary_operation): Use
32022 CONST_VECTOR_ELT instead of XVECEXP.
32024 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32026 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
32027 the selector elements to be different from the data elements
32028 if the selector is a VECTOR_CST.
32029 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
32030 ssizetype for the selector.
32032 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32034 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
32035 before testing each element individually.
32036 * tree-vect-generic.c (lower_vec_perm): Likewise.
32038 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32040 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
32041 * selftest-run-tests.c (selftest::run_tests): Call it.
32042 * vector-builder.h (vector_builder::operator ==): New function.
32043 (vector_builder::operator !=): Likewise.
32044 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
32045 (vec_perm_indices::all_from_input_p): New function.
32046 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
32047 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
32048 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
32049 instead of reading the VECTOR_CST directly. Detect whether both
32050 vector inputs are the same before constructing the vec_perm_indices,
32051 and update the number of inputs argument accordingly. Use the
32052 utility functions added above. Only construct sel2 if we need to.
32054 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32056 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
32057 the broadcast of the low byte.
32058 (expand_mult_highpart): Use an explicit encoding for the permutes.
32059 * optabs-query.c (can_mult_highpart_p): Likewise.
32060 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
32061 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32062 (vectorizable_bswap): Likewise.
32063 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
32064 explicit encoding for the power-of-2 permutes.
32065 (vect_permute_store_chain): Likewise.
32066 (vect_grouped_load_supported): Likewise.
32067 (vect_permute_load_chain): Likewise.
32069 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32071 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
32072 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
32073 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
32074 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
32075 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
32076 (vect_gen_perm_mask_any): Likewise.
32078 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32080 * int-vector-builder.h: New file.
32081 * vec-perm-indices.h: Include int-vector-builder.h.
32082 (vec_perm_indices): Redefine as an int_vector_builder.
32083 (auto_vec_perm_indices): Delete.
32084 (vec_perm_builder): Redefine as a stand-alone class.
32085 (vec_perm_indices::vec_perm_indices): New function.
32086 (vec_perm_indices::clamp): Likewise.
32087 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
32088 (vec_perm_indices::new_vector): New function.
32089 (vec_perm_indices::new_expanded_vector): Update for new
32090 vec_perm_indices class.
32091 (vec_perm_indices::rotate_inputs): New function.
32092 (vec_perm_indices::all_in_range_p): Operate directly on the
32093 encoded form, without computing elided elements.
32094 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
32095 encoding. Update for new vec_perm_indices class.
32096 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
32097 the given vec_perm_builder.
32098 (expand_vec_perm_var): Update vec_perm_builder constructor.
32099 (expand_mult_highpart): Use vec_perm_builder instead of
32100 auto_vec_perm_indices.
32101 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
32102 vec_perm_indices instead of auto_vec_perm_indices. Use a single
32103 or double series encoding as appropriate.
32104 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
32105 vec_perm_indices instead of auto_vec_perm_indices.
32106 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32107 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32108 (vect_permute_store_chain): Likewise.
32109 (vect_grouped_load_supported): Likewise.
32110 (vect_permute_load_chain): Likewise.
32111 (vect_shift_permute_load_chain): Likewise.
32112 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
32113 (vect_transform_slp_perm_load): Likewise.
32114 (vect_schedule_slp_instance): Likewise.
32115 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32116 (vectorizable_mask_load_store): Likewise.
32117 (vectorizable_bswap): Likewise.
32118 (vectorizable_store): Likewise.
32119 (vectorizable_load): Likewise.
32120 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
32121 vec_perm_indices instead of auto_vec_perm_indices. Use
32122 tree_to_vec_perm_builder to read the vector from a tree.
32123 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
32124 vec_perm_builder instead of a vec_perm_indices.
32125 (have_whole_vector_shift): Use vec_perm_builder and
32126 vec_perm_indices instead of auto_vec_perm_indices. Leave the
32127 truncation to calc_vec_perm_mask_for_shift.
32128 (vect_create_epilog_for_reduction): Likewise.
32129 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
32130 from auto_vec_perm_indices to vec_perm_indices.
32131 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
32132 instead of changing individual elements.
32133 (aarch64_vectorize_vec_perm_const): Use new_vector to install
32134 the vector in d.perm.
32135 * config/arm/arm.c (expand_vec_perm_d::perm): Change
32136 from auto_vec_perm_indices to vec_perm_indices.
32137 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
32138 instead of changing individual elements.
32139 (arm_vectorize_vec_perm_const): Use new_vector to install
32140 the vector in d.perm.
32141 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
32142 Update vec_perm_builder constructor.
32143 (rs6000_expand_interleave): Likewise.
32144 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
32145 (rs6000_expand_interleave): Likewise.
32147 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32149 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
32150 to qimode could truncate the indices.
32151 * optabs.c (expand_vec_perm_var): Likewise.
32153 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32155 * Makefile.in (OBJS): Add vec-perm-indices.o.
32156 * vec-perm-indices.h: New file.
32157 * vec-perm-indices.c: Likewise.
32158 * target.h (vec_perm_indices): Replace with a forward class
32160 (auto_vec_perm_indices): Move to vec-perm-indices.h.
32161 * optabs.h: Include vec-perm-indices.h.
32162 (expand_vec_perm): Delete.
32163 (selector_fits_mode_p, expand_vec_perm_var): Declare.
32164 (expand_vec_perm_const): Declare.
32165 * target.def (vec_perm_const_ok): Replace with...
32166 (vec_perm_const): ...this new hook.
32167 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
32168 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
32169 * doc/tm.texi: Regenerate.
32170 * optabs.def (vec_perm_const): Delete.
32171 * doc/md.texi (vec_perm_const): Likewise.
32172 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
32173 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
32174 expand_vec_perm for constant permutation vectors. Assert that
32175 the mode of variable permutation vectors is the integer equivalent
32176 of the mode that is being permuted.
32177 * optabs-query.h (selector_fits_mode_p): Declare.
32178 * optabs-query.c: Include vec-perm-indices.h.
32179 (selector_fits_mode_p): New function.
32180 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
32181 is defined, instead of checking whether the vec_perm_const_optab
32182 exists. Use targetm.vectorize.vec_perm_const instead of
32183 targetm.vectorize.vec_perm_const_ok. Check whether the indices
32184 fit in the vector mode before using a variable permute.
32185 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
32186 vec_perm_indices instead of an rtx.
32187 (expand_vec_perm): Replace with...
32188 (expand_vec_perm_const): ...this new function. Take the selector
32189 as a vec_perm_indices rather than an rtx. Also take the mode of
32190 the selector. Update call to shift_amt_for_vec_perm_mask.
32191 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
32192 Use vec_perm_indices::new_expanded_vector to expand the original
32193 selector into bytes. Check whether the indices fit in the vector
32194 mode before using a variable permute.
32195 (expand_vec_perm_var): Make global.
32196 (expand_mult_highpart): Use expand_vec_perm_const.
32197 * fold-const.c: Includes vec-perm-indices.h.
32198 * tree-ssa-forwprop.c: Likewise.
32199 * tree-vect-data-refs.c: Likewise.
32200 * tree-vect-generic.c: Likewise.
32201 * tree-vect-loop.c: Likewise.
32202 * tree-vect-slp.c: Likewise.
32203 * tree-vect-stmts.c: Likewise.
32204 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
32206 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
32207 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
32208 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
32209 (aarch64_vectorize_vec_perm_const): ...this new function.
32210 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32211 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32212 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
32213 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
32214 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32215 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32216 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
32218 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
32219 check for NEON modes.
32220 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
32221 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
32222 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
32223 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
32225 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
32226 the old VEC_PERM_CONST conditions.
32227 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
32228 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
32229 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
32230 (ia64_vectorize_vec_perm_const_ok): Merge into...
32231 (ia64_vectorize_vec_perm_const): ...this new function.
32232 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
32233 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
32234 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
32235 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
32236 * config/mips/mips.c (mips_expand_vec_perm_const)
32237 (mips_vectorize_vec_perm_const_ok): Merge into...
32238 (mips_vectorize_vec_perm_const): ...this new function.
32239 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
32240 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
32241 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
32242 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
32243 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
32244 (rs6000_expand_vec_perm_const): Delete.
32245 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
32247 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32248 (altivec_expand_vec_perm_const_le): Take each operand individually.
32249 Operate on constant selectors rather than rtxes.
32250 (altivec_expand_vec_perm_const): Likewise. Update call to
32251 altivec_expand_vec_perm_const_le.
32252 (rs6000_expand_vec_perm_const): Delete.
32253 (rs6000_vectorize_vec_perm_const_ok): Delete.
32254 (rs6000_vectorize_vec_perm_const): New function.
32255 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
32256 an element count and rtx array.
32257 (rs6000_expand_extract_even): Update call accordingly.
32258 (rs6000_expand_interleave): Likewise.
32259 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
32260 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
32261 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
32262 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
32263 (rs6000_expand_vec_perm_const): Delete.
32264 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32265 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32266 (altivec_expand_vec_perm_const_le): Take each operand individually.
32267 Operate on constant selectors rather than rtxes.
32268 (altivec_expand_vec_perm_const): Likewise. Update call to
32269 altivec_expand_vec_perm_const_le.
32270 (rs6000_expand_vec_perm_const): Delete.
32271 (rs6000_vectorize_vec_perm_const_ok): Delete.
32272 (rs6000_vectorize_vec_perm_const): New function. Remove stray
32273 reference to the SPE evmerge intructions.
32274 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
32275 an element count and rtx array.
32276 (rs6000_expand_extract_even): Update call accordingly.
32277 (rs6000_expand_interleave): Likewise.
32278 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
32279 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
32281 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32283 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32285 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
32286 vector mode and that that mode matches the mode of the data
32288 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
32289 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
32290 directly using expand_vec_perm_1 when forcing selectors into
32292 (expand_vec_perm_var): New function, split out from expand_vec_perm.
32294 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32296 * optabs-query.h (can_vec_perm_p): Delete.
32297 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
32298 * optabs-query.c (can_vec_perm_p): Split into...
32299 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
32300 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
32301 particular selector is valid.
32302 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32303 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32304 (vect_grouped_load_supported): Likewise.
32305 (vect_shift_permute_load_chain): Likewise.
32306 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
32307 (vect_transform_slp_perm_load): Likewise.
32308 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32309 (vectorizable_bswap): Likewise.
32310 (vect_gen_perm_mask_checked): Likewise.
32311 * fold-const.c (fold_ternary_loc): Likewise. Don't take
32312 implementations of variable permutation vectors into account
32313 when deciding which selector to use.
32314 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
32315 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
32316 with a false third argument.
32317 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
32318 to test whether the constant selector is valid and can_vec_perm_var_p
32319 to test whether a variable selector is valid.
32321 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32323 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
32324 * optabs-query.c (can_vec_perm_p): Likewise.
32325 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
32326 instead of vec_perm_indices.
32327 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
32328 (vect_gen_perm_mask_checked): Likewise,
32329 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
32330 (vect_gen_perm_mask_checked): Likewise,
32332 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32334 * optabs-query.h (qimode_for_vec_perm): Declare.
32335 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
32336 (qimode_for_vec_perm): ...this new function.
32337 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
32339 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32341 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
32342 does not have a conditional at the top.
32344 2018-01-02 Richard Biener <rguenther@suse.de>
32346 * ipa-inline.c (big_speedup_p): Fix expression.
32348 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32351 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
32354 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32358 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
32359 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
32360 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
32361 cond_taken_branch_cost 3->4.
32363 2018-01-01 Jakub Jelinek <jakub@redhat.com>
32365 PR tree-optimization/83581
32366 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
32367 TODO_cleanup_cfg if any changes have been made.
32369 PR middle-end/83608
32370 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
32371 convert_modes if target mode has the right side, but different mode
32374 PR middle-end/83609
32375 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
32376 last argument when extracting from CONCAT. If either from_real or
32377 from_imag is NULL, use expansion through memory. If result is not
32378 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
32379 the parts directly to inner mode, if even that fails, use expansion
32382 PR middle-end/83623
32383 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
32384 check for bswap in mode rather than HImode and use that in expand_unop
32387 Copyright (C) 2018 Free Software Foundation, Inc.
32389 Copying and distribution of this file, with or without modification,
32390 are permitted in any medium without royalty provided the copyright
32391 notice and this notice are preserved.