1 2018-11-17 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/87546
4 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
5 min_precision variable, initially set it to orig_precision, only does
6 something if op_type's precision is <= min_precision and update
7 min_precision whenever calling set_op.
9 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
12 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
14 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
16 * doc/invoke.texi (Option Summary): Fix whitespace and line
17 breaks in @gccoptlist environments.
18 (Warning Options): Likewise.
19 (Optimize Options): Likewise.
20 (PowerPC SPE Options): Likewise.
21 (RS/6000 and PowerPC Options): Likewise.
23 2018-11-16 Jeff Law <law@redhat.com>
25 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
28 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
31 * doc/invoke.texi (Optimize Options): Update options enabled by
32 fprofile-generate, -fprofile-use, and -fauto-profile.
34 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
37 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
38 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
39 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
41 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
42 * targhooks.c (default_print_patchable_function_entry): Assert
43 asm_int_op does not return a NULL string.
44 * varasm.c (integer_asm_op): Return the op for a partial int type
45 when the requested size does not correspond to an integer type.
46 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
47 * doc/tm.texi: Regenerate.
49 2018-11-12 Jason Merrill <jason@redhat.com>
51 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
53 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
55 * config/rs6000/constraints.md (wF constraint): Remove power9
56 fusion documentation. Just document wF's use for power8 fusion.
57 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
58 (fusion_gpr_addis): Delete power9 fusion support. Change power8
59 fusion support to require the upper 12 bits to be all 0's or all
61 (fusion_gpr_mem_load): Add comment.
62 (fusion_addis_mem_combo_load): Remove power9 fusion support.
63 (fusion_addis_mem_combo_store): Delete.
64 (fusion_offsettable_mem_operand): Delete.
65 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
67 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
68 (POWERPC_MASKS): Delete power9 fusion.
69 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
70 (fusion_p9_p): Delete.
71 (expand_fusion_p9_load): Delete.
72 (expand_fusion_p9_store): Delete.
73 (emit_fusion_p9_load): Delete.
74 (emit_fusion_p9_store): Delete.
75 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
77 (rs6000_option_override_internal): Set power8 fusion based on
78 whether we are tuning for power8. Delete power9 fusion support.
79 (rs6000_opt_masks): Delete -mpower9-fusion switch.
80 (emit_fusion_load): Rename emit_fusion_load_store to
81 emit_fusion_load, and drop fusion store support. Update callers.
82 (emit_fusion_load_store): Likewise.
83 (emit_fusion_gpr_load): Likewise.
84 (fusion_p9_p): Delete.
85 (expand_fusion_p9_load): Delete.
86 (expand_fusion_p9_store): Delete.
87 (emit_fusion_p9_load): Delete.
88 (emit_fusion_p9_store): Delete.
89 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
92 (power9 fusion peephole2s): Delete.
93 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
94 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
95 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
96 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
97 (fusion_p9_<mode>_constant): Delete.
98 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
100 * doc/md.texi (PowerPC constraints): Update wF constraint
101 documentation for power8 fusion only.
103 2018-11-16 Jakub Jelinek <jakub@redhat.com>
105 PR rtl-optimization/87475
106 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
107 CROSSING_JUMP_P insns.
108 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
110 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
113 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
114 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
115 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
117 2018-11-16 Jakub Jelinek <jakub@redhat.com>
120 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
121 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
123 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
125 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
126 location_t parameter.
127 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
128 (switch_decision_tree::do_jump_if_equal): Likewise.
129 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
130 of switch statement to emit_case_nodes.
131 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
132 set it on the newly built GIMPLE comparison statement.
133 (switch_decision_tree::do_jump_if_equal): Likewise.
134 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
135 in calls to do_jump_if_equal as well as recursive calls.
137 2018-11-16 Andi Kleen <ak@linux.intel.com>
139 * config/i386/i386.md: Allow memory operands to ptwrite.
141 2018-11-16 Michael Matz <matz@suse.de>
143 * system.h (PRsa): New macro.
144 (SIZE_AMOUNT): Cast number to uint64_t.
145 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
146 (pool_usage::dump_footer): Likewise and also use PRIu64.
147 * bitmap.h (bitmap_usage::dump): Likewise.
148 * ggc-common.c (ggc_usage::dump): Likewise.
149 * ggc-page.c (ggc_print_statistics): Likewise.
150 * input.c (dump_line_table_statistics): Likewise.
151 * mem-stats.h (mem_usage::dump): Likewise.
152 (mem_usage::dump_footer): Likewise.
153 * rtl.c (dump_rtx_statistics): Likewise.
154 * tree-cfg.c (dump_cfg_stats): Likewise.
155 * tree-dfa.c (dump_dfa_stats): Likewise.
156 * tree-phinodes.c (phinodes_print_statistics): Likewise.
157 * tree-ssanames (ssanames_print_statistics): Likewise.
158 * vec.c (vec_usage::dump): Likewise.
159 (vec_usage::dump_footer): Likewise.
161 2018-11-16 Richard Biener <rguenther@suse.de>
163 PR tree-optimization/88011
164 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
165 replacing set_value_range_to_undefined and
166 set_value_range_to_varying with method calls.
168 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
170 * config/s390/s390.md
171 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
173 (*r<noxa>sbg_<mode>_sll): Likewise.
174 (*r<noxa>sbg_<mode>_srl): Likewise.
175 (*r<noxa>sbg_sidi_srl): New pattern.
177 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
179 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
180 Default to TARGET_VXWORKS_RTP.
181 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
182 of TARGET_VXWORKS_RTP.
183 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
184 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
185 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
187 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
190 * doc/extend.texi (Common Type Attributes): Make it explicit
191 that attribute "packed" can apply to C++ classes.
193 2018-11-15 Martin Sebor <msebor@redhat.com>
197 * tree.c (type_argument_type): New function.
198 * tree.h (type_argument_type): Declare it.
199 * gcc/doc/extend.texi (alloc_align): Update and clarify.
200 (alloc_size, nonnull, sentinel): Same.
202 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
203 Kwok Cheung Yeung <kcy@codesourcery.com>
205 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
206 int_mode_for_size fails.
207 (vectorizable_load): Likewise.
209 2018-11-15 David Malcolm <dmalcolm@redhat.com>
211 * doc/ux.texi (Group logically-related diagnostics): Move
212 discussion of auto_diagnostic_group into this new subsection.
213 Give an example of where this grouping is used.
215 2018-11-15 David Malcolm <dmalcolm@redhat.com>
218 * Makefile.in (OBJS): Move json.o to...
219 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
220 * common.opt (fdiagnostics-format=): New option.
221 (diagnostics_output_format): New enum.
222 * diagnostic-format-json.cc: New file.
223 * diagnostic.c (default_diagnostic_final_cb): New function, taken
224 from start of diagnostic_finish.
225 (diagnostic_initialize): Initialize final_cb to
226 default_diagnostic_final_cb.
227 (diagnostic_finish): Move "being treated as errors" messages to
228 default_diagnostic_final_cb. Call any final_cb.
229 (default_diagnostic_finalizer): Add diagnostic_t param.
230 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
231 diagnostic_finalizer callback.
232 * diagnostic.h (enum diagnostics_output_format): New enum.
233 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
234 (struct diagnostic_context): Add "final_cb".
235 (default_diagnostic_finalizer): Add diagnostic_t param.
236 (diagnostic_output_format_init): New decl.
237 * doc/invoke.texi (-fdiagnostics-format): New option.
238 * dwarf2out.c (gen_producer_string): Ignore
239 OPT_fdiagnostics_format_.
240 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
241 * lto-wrapper.c (append_diag_options): Ignore it.
242 * opts.c (common_handle_option): Handle it.
244 2018-11-15 David Malcolm <dmalcolm@redhat.com>
246 PR tree-optimization/88015
247 * graphite-isl-ast-to-gimple.c
248 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
250 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
252 2018-11-15 Richard Biener <rguenther@suse.de>
255 * gimple.c (gimple_call_flags): Union flags from decl, type
257 * trans-mem.c (is_tm_pure_call): Simplify.
259 2018-11-15 Richard Biener <rguenther@suse.de>
261 PR tree-optimization/88031
262 * tree-vect-loop.c (vectorizable_reduction): Move check
263 for multiple types earlier so we get the expected dump.
264 Simplify calls to vectorizable_condition.
265 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
266 * tree-vect-stmts.c (vectorizable_condition): Instead of
267 reduc_def and reduc_index take just a flag. Simplify
268 code-generation now that we can rely on the defs being set up.
269 (vectorizable_comparison): Remove unused argument.
271 2018-11-15 Richard Biener <rguenther@suse.de>
273 PR tree-optimization/88030
274 * tree-complex.c (need_eh_cleanup): New global.
275 (update_complex_assignment): Mark blocks that need EH update.
276 (expand_complex_comparison): Likewise.
277 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
278 perform EH cleanup and schedule CFG cleanup if that did anything.
280 2018-11-15 Jakub Jelinek <jakub@redhat.com>
282 PR rtl-optimization/88018
283 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
284 on the presence of fallthru edge, rather than if it is a USE or not.
286 2018-11-15 Richard Biener <rguenther@suse.de>
289 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
290 analyze_subscript_affine_affine properly.
292 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
294 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
295 (GLIBC_DYNAMIC_LINKER): Define.
296 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
298 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
300 PR tree-optimization/84648
301 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
302 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
303 by calling adjust_cond_for_loop_until_wrap.
305 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
308 * doc/extend.texi (Common Function Attributes): Clarify linker
309 restrictions on "aligned" attribute.
310 (Common Variable Attributes): Likewise. Mention that linker
311 restrictions don't apply to stack-allocated variables.
313 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
315 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
318 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
322 * doc/invoke.texi (Optimize Options): Remove bad example about
323 interaction between -flto and -O. Replace it with a note that
324 you need to compile with -O and not just link. Copy-edit -flto
325 discussion to reduce verbiage and improve flow.
327 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
330 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
331 also suppress many optimizations. Alphabetize option lists for
332 -O1, -O2, and -Os. Add list of options disabled with -Og, and
333 correct documentation for those options to say that.
334 * opts.c (default_options_table): Sort table by level and option
335 name, to make it easier to correlate to the manual.
337 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
339 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
340 printing of the names of segment registers.
342 2018-11-14 Nathan Sidwell <nathan@acm.org>
346 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
349 2018-11-14 David Malcolm <dmalcolm@redhat.com>
351 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
353 2018-11-14 Jakub Jelinek <jakub@redhat.com>
355 P1236R1 - Signed integers are two's complement
356 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
359 * hash-map.h (hash_map::iterator::reference_pair): New class.
360 (hash_map::iterator::operator*): Return it rather than std::pair.
362 2018-11-14 Jeff Law <law@redhat.com>
364 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
365 iff the operand is a constant.
367 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
369 * gimple-ssa-evrp-analyze.c
370 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
371 ignore_equivs_equal_p to equal_p.
372 * ipa-cp.c (meet_with_1): Use equal_p instead of
373 ignore_equivs_equal_p.
374 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
375 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
376 (value_range::operator==): Remove.
377 (value_range::operator!=): Remove.
378 (vrp_prop::visit_stmt): Use equal_p.
379 * tree-vrp.h (value_range): Remove operator==, operator!=,
380 ignore_equivs_equal_p.
381 * vr-values.c (update_value_range): Use equal_p.
383 2018-11-14 Michael Matz <matz@suse.de>
386 * gimplify.c (collect_fallthrough_labels): Add new argument,
387 return location via that, don't modify statements.
388 (warn_implicit_fallthrough_r): Adjust call, don't use
389 statement location directly.
391 2018-11-14 Richard Biener <rguenther@suse.de>
394 * tree-data-ref.c (split_constant_offset): Add wrapper
395 allocating a cache hash-map.
396 (split_constant_offset_1): Cache results of expanding
397 expressions from SSA def stmts.
399 2018-11-14 Richard Biener <rguenther@suse.de>
402 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
404 (lambda_vector_mult_const): Likewise.
405 (lambda_matrix_right_hermite): Use lambda_int temporaries.
407 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
408 Jackson Woodruff <jackson.woodruff@arm.com>
410 PR 71026/tree-optimization
411 * match.pd: Simplify floating point comparisons.
413 2018-11-14 Jakub Jelinek <jakub@redhat.com>
415 PR rtl-optimization/87817
416 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
417 last argument has low 8 bits clear, fold to 0.
419 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
421 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
422 MAX_OFILE_ALIGNMENT as needed.
424 2018-11-14 Richard Biener <rguenther@suse.de>
426 PR tree-optimization/88019
427 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
428 COND_EXPR arguments to non-trapping overflow.
430 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
432 * config/arc/arc.h (reg_class): Reorder registers classes, remove
433 unused register classes.
434 (REG_CLASS_NAMES): Likewise.
435 (REG_CLASS_CONTENTS): Likewise.
436 (FIXED_REGISTERS): Make lp_count fixed.
437 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
438 (PROGRAM_COUNTER_REGNO): Remove.
439 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
440 register classes, use constants for register numbers, remove
441 reg_class_contents references.
442 (arc_process_double_reg_moves): Add asserts.
443 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
444 lra_in_progress predicate.
445 (arc_init_reg_tables): Remove unused register classes.
446 (arc_register_move_cost): Likewise.
447 (arc_preferred_reload_class): Likewise.
448 (hwloop_optimize): Update rtx patterns involving lp_count register.
449 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
450 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
451 CHEAP_CORE_REGS. Former one will be used for LRA.
452 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
453 one will be used for LRA.
454 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
455 one will be used for LRA.
456 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
457 Former one will be used for LRA.
458 ("f"): Delete constraint.
461 ("l"): Change it from register constraint to constraint.
462 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
463 (movhi_insn): Likewise.
464 (movsi_insn): Update pattern.
467 (stack_tie): Remove 'b' constraint letter.
469 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
470 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
471 (R57_REG, R59_REG, PCL_REG): Likewise.
472 (ILINK1_REGNUM): Renamed to ILINK1_REG.
473 (ILINK2_REGNUM): Renamed to ILINK2_REG.
476 (Rcw): Remove unused reg classes.
477 * config/arc/predicates.md (dest_reg_operand): Just default on
478 register_operand predicate.
479 (mpy_dest_reg_operand): Likewise.
480 (move_dest_operand): Use macros instead of constants.
481 * config/arc/arc.opt (mlra): Switch to lra as default.
483 2018-11-14 Richard Biener <rguenther@suse.de>
485 PR tree-optimization/87974
486 * tree-vect-loop.c (vectorizable_reduction): When computing
487 the vectorized reduction PHI vector type ignore constant
490 2018-11-14 Jakub Jelinek <jakub@redhat.com>
492 PR tree-optimization/87977
493 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
494 stmt, build a new one and replace the old one with it. Formatting fix.
495 Call release_ssa_name (x) if !has_other_use and !delete_div.
496 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
497 verify lhs of stmt is still def.
499 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
501 PR rtl-optimization/87507
502 * lower-subreg.c (operand_for_swap_move_operator): New function.
503 (simple_move): Strip simple operators.
504 (find_pseudo_copy): Likewise.
505 (resolve_operand_for_swap_move_operator): New function.
506 (resolve_simple_move): Strip simple operators and swap operands.
508 2018-11-13 Jakub Jelinek <jakub@redhat.com>
510 PR tree-optimization/87898
511 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
512 (ipa_simd_modify_function_body): Remove debug stmts where the first
513 argument was changed into a non-decl.
515 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
517 PR rtl-optimization/87899
518 * lra-lives.c (start_living): Update white space in comment.
519 (enum point_type): New.
520 (sparseset_contains_pseudos_p): New function.
521 (update_pseudo_point): Likewise.
522 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
523 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
524 handling. Move early exit after adding conflicts.
525 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
526 if regno is already live. Remove all handling of program points.
527 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
528 after adding conflicts. Remove all handling of program points and
529 ignore_reg_for_conflicts.
530 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
531 and do not guard call to mark_pseudo_live.
532 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
533 and do not guard call to mark_pseudo_dead.
534 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
535 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
536 Use new function update_pseudo_point. Handle register copies by
537 removing the source register from the live set. Handle INOUT operands.
538 Update to the next program point using the unused_set, dead_set and
540 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
542 2018-11-13 David Malcolm <dmalcolm@redhat.com>
544 * builtins.c: Replace "source_location" with "location_t".
545 * diagnostic-show-locus.c: Likewise.
546 * diagnostic.c: Likewise.
547 * dumpfile.c: Likewise.
548 * gcc-rich-location.h: Likewise.
549 * genmatch.c: Likewise.
550 * gimple.h: Likewise.
551 * gimplify.c: Likewise.
553 * input.h: Likewise. Eliminate the typedef.
554 * omp-expand.c: Likewise.
555 * selftest.h: Likewise.
556 * substring-locations.h (get_source_location_for_substring):
558 (get_location_within_string): ...this.
559 * tree-cfg.c: Replace "source_location" with "location_t".
560 * tree-cfgcleanup.c: Likewise.
561 * tree-diagnostic.c: Likewise.
562 * tree-into-ssa.c: Likewise.
563 * tree-outof-ssa.c: Likewise.
564 * tree-parloops.c: Likewise.
565 * tree-phinodes.c: Likewise.
566 * tree-phinodes.h: Likewise.
567 * tree-ssa-loop-ivopts.c: Likewise.
568 * tree-ssa-loop-manip.c: Likewise.
569 * tree-ssa-phiopt.c: Likewise.
570 * tree-ssa-phiprop.c: Likewise.
571 * tree-ssa-threadupdate.c: Likewise.
572 * tree-ssa.c: Likewise.
573 * tree-ssa.h: Likewise.
574 * tree-vect-loop-manip.c: Likewise.
576 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
578 * doc/extend.texi: Fix typo in the weakref description.
580 2018-11-13 Richard Biener <rguenther@suse.de>
582 PR tree-optimization/86991
583 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
584 group building until we have successfully detected the SLP
586 (vect_is_simple_reduction): Remove fixup code here.
588 2018-11-13 David Malcolm <dmalcolm@redhat.com>
590 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
591 (dump_gimple_stmt): Use it.
592 (dump_gimple_stmt_loc): Likewise.
593 (dump_gimple_expr): Likewise.
594 (dump_gimple_expr_loc): Likewise.
595 (dump_generic_expr): Likewise.
596 (dump_generic_expr_loc): Likewise.
597 (dump_printf): Likewise.
598 (dump_printf_loc): Likewise.
599 (dump_dec): Likewise.
600 (dump_dec): Likewise.
601 (dump_hex): Likewise.
602 (dump_symtab_node): Likewise.
603 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
604 Guard dump call with dump_enabled_p.
605 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
606 * graphite-optimize-isl.c (optimize_isl): Likewise.
607 * graphite.c (graphite_transform_loops): Likewise.
608 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
609 * tree-parloops.c (parallelize_loops): Likewise.
610 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
611 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
612 (vect_prune_runtime_alias_test_list): Likewise.
613 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
614 (vect_estimate_min_profitable_iters): Likewise.
615 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
616 (vect_build_slp_tree_2): Likewise.
617 (vect_supported_load_permutation_p): Likewise.
618 (vect_slp_analyze_operations): Likewise.
619 (vect_slp_analyze_bb_1): Likewise.
620 (vect_slp_bb): Likewise.
621 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
622 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
623 (pass_slp_vectorize::execute): Likewise.
624 (increase_alignment): Likewise.
626 2018-11-13 David Malcolm <dmalcolm@redhat.com>
629 * ipa-inline.c (report_inline_failed_reason): Guard calls to
630 cl_target_option_print_diff and cl_optimization_print_diff with
633 2018-11-13 David Malcolm <dmalcolm@redhat.com>
635 * doc/invoke.texi (-fsave-optimization-record): Note that the
636 output is compressed.
637 * optinfo-emit-json.cc: Include <zlib.h>.
638 (optrecord_json_writer::write): Compress the output.
640 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
642 * tree-vrp.c (value_range_base::dump): Dump type.
643 Do not use INF nomenclature for 1-bit types.
644 (dump_value_range): Group all variants to common dumping code.
645 (debug): New overloaded functions for value_ranges.
646 (value_range_base::dump): Remove no argument version.
647 (value_range::dump): Same.
649 2018-11-13 Richard Biener <rguenther@suse.de>
651 PR tree-optimization/87931
652 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
653 nested cycles we support to latch computations vectorizable_reduction
656 2018-11-13 Martin Liska <mliska@suse.cz>
658 PR tree-optimization/87885
659 * cfghooks.c (account_profile_record): Rename
661 (profile_record_check_consistency): ... this.
662 Calculate missing num_mismatched_freq_in.
663 (profile_record_account_profile): New function
664 that calculates time and size of a function.
665 * cfghooks.h (struct profile_record): Remove
667 (struct cfg_hooks): Remove after_pass flag.
668 (account_profile_record): Rename to ...
669 (profile_record_check_consistency): ... this.
670 (profile_record_account_profile): New.
671 * cfgrtl.c (rtl_account_profile_record): Remove
673 * passes.c (check_profile_consistency): Do only
675 (account_profile): Calculate size and time of
677 (pass_manager::dump_profile_report): Reformat
679 (execute_one_ipa_transform_pass): Call
680 consistency check before clean upand call account_profile
681 after a clean up is done.
682 (execute_one_pass): Call check_profile_consistency and
683 account_profile instead of using after_pass flag..
684 * tree-cfg.c (gimple_account_profile_record): Likewise.
686 2018-11-13 Martin Liska <mliska@suse.cz>
689 * config/i386/i386.c (ix86_option_override_internal): Error
690 about usage -mabi=ms and -fsanitize={,kernel-}address.
692 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
694 * config/aarch64/aarch64.c
695 (aarch64_vectorize_preferred_vector_alignment): Change return type to
697 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
698 alignment being a poly int.
699 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
700 return type to poly_uint64.
701 * target.def (default_preferred_vector_alignment): Likewise.
702 * targhooks.c (default_preferred_vector_alignment): Likewise.
703 * targhooks.h (default_preferred_vector_alignment): Likewise.
704 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
705 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
707 (vect_update_misalignment_for_peel): Likewise.
708 (vect_enhance_data_refs_alignment): Likewise.
709 (vect_find_same_alignment_drs): Likewise.
710 (vect_duplicate_ssa_name_ptr_info): Likewise.
711 (vect_setup_realignment): Likewise.
712 (vect_can_force_dr_alignment_p): Change alignment parameter type to
714 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
715 mask with a compile time variable vector alignment.
716 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
718 (vect_do_peeling): Exit early if vector alignment is not constant.
719 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
721 (vectorizable_store): Likewise.
722 (vectorizable_load): Likweise.
723 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
725 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
727 (vect_can_force_dr_alignment_p): Change alignment parameter type to
730 2018-11-13 Richard Biener <rguenther@suse.de>
732 PR tree-optimization/87962
733 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
734 detect outer reduction for disqualifying in-loop uses.
736 2018-11-13 Richard Biener <rguenther@suse.de>
738 PR tree-optimization/87967
739 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
740 for constants for the scalar loop.
742 2018-11-13 Alan Modra <amodra@gmail.com>
744 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
745 integer constants when -mcmodel=medium.
747 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
749 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
750 * config/arc/elf.h (PROFILE_HOOK): Define.
751 * config/arc/linux.h (PROFILE_HOOK): Likewise.
753 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
755 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
756 jump table data in the text section.
758 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
760 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
761 to fit the eh_return pattern.
762 * config/arc/arc.md (eh_return): Define.
763 (VUNSPEC_ARC_EH_RETURN): Likewise.
764 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
766 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
769 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
771 * common/config/arc/arc-common.c (arc_option_optimization_table):
772 Millicode optimization is default on for size optimizations.
773 * config/arc/arc-protos.h (arc_check_multi): New function.
774 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
775 (ENTER_LEAVE_START_REG): Define.
776 (ENTER_LEAVE_END_REG): Likewise.
777 (arc_override_options): Disable millicode when long calls option
779 (arc_frame_info): Change it from int to bool.
780 (arc_compute_frame_size): Clean up.
781 (arc_save_restore): Remove.
782 (frame_save_reg): New function.
783 (frame_restore_reg): Likewise.
784 (arc_enter_leave_p): Likewise.
785 (arc_save_callee_saves): Likewise.
786 (arc_restore_callee_saves): Likewise.
787 (arc_save_callee_enter): Likewise.
788 (arc_restore_callee_leave): Likewise.
789 (arc_save_callee_milli): Likewise.
790 (arc_restore_callee_milli): Likewise.
791 (arc_expand_prologue): Reimplement to emit enter/leave
793 (arc_expand_epilogue): Likewise.
794 (arc_check_multi): New function.
795 * config/arc/arc.md (push_multi_fp): New pattern.
796 (push_multi_fp_blink): Likewise.
797 (pop_multi_fp): Likewise.
798 (pop_multi_fp_blink): Likewise.
799 (pop_multi_fp_ret): Likewise.
800 (pop_multi_fp_blink_ret): Likewise.
801 * config/arc/arc.opt (mmillicode): Update option.
802 (mcode-density-frame): New option.
803 * config/arc/predicates.md (push_multi_operand): New predicate.
804 (pop_multi_operand): Likewise.
805 * doc/invoke.texi (ARC): Update ARC options information.
808 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
810 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
811 * config/arc/arc.c (operands_ok_ldd_std): New function.
812 (mem_ok_for_ldd_std): Likewise.
813 (gen_operands_ldd_std): Likewise.
814 * config/arc/arc.md: Add peephole2 rules for std/ldd.
816 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
818 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
819 test on stack_usage_file.
820 (lang_dependent_init): Do not open the .su file if generating LTO.
822 2018-11-13 Jakub Jelinek <jakub@redhat.com>
824 PR rtl-optimization/87918
825 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
826 simplify_gen_relational rather than simplify_gen_binary.
828 2018-11-13 Richard Biener <rguenther@suse.de>
830 * tree-ssanames.h (set_range_info): Use value_range_base.
831 (get_range_info): Likewise.
832 * tree-ssanames.c (set_range_info): Likewise.
833 (get_range_info): Likewise.
834 * tree-vrp.c (value_range_base::union_helper): Split
835 out common parts of value_range[_base]::union_.
836 (value_range_base::union_): Update.
837 (value_range::union_): Likewise.
838 (determine_value_range_1): Use value_range_base.
839 (determine_value_range): Likewise.
840 * tree-vrp.h (value_range_base::union_helper): Move ...
841 (value_range::union_helper): ... from here.
843 2018-11-13 Alan Modra <amodra@gmail.com>
845 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
847 (rs6000_secondary_reload_gpr): Don't call find_replacement.
849 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
852 * doc/invoke.texi (Optimize Options): Clarify that the
853 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
854 apply to data cache size.
856 2018-11-13 Alan Modra <amodra@gmail.com>
858 * config/rs6000/predicates.md (logical_const_operand),
859 (logical_operand): Correct comment.
860 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
862 2018-11-13 Alan Modra <amodra@gmail.com>
864 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
866 2018-11-13 Alan Modra <amodra@gmail.com>
868 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
870 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
871 covered by alternative.
872 (movcc_internal1): Ignore h for register preference.
873 (mov<mode>_hardfloat64): Likewise.
874 (mov<mode>_softfloat): Ignore c, l, h for register preference.
876 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
878 PR preprocessor/47823
879 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
882 * doc/extend.texi (Pragmas): Note additional pragmas documented
885 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
889 * tree-scalar-evolution.c (expression_expensive_p): Make
890 BUILTIN POPCOUNT as expensive when backend does not define it.
892 2018-11-12 Fredrik Noring <noring@nocrew.org>
894 * config.gcc: Update with-llsc defaults for MIPS r5900.
896 2018-11-12 Martin Liska <mliska@suse.cz>
898 PR gcov-profile/87442
899 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
901 * doc/invoke.texi: Document them.
902 * tree-profile.c (parse_profile_filter): New.
903 (parse_profile_file_filtering): Likewise.
904 (release_profile_file_filtering): Likewise.
905 (include_source_file_for_profile): Likewise.
906 (tree_profiling): Filter source files based on the
909 2018-11-12 Martin Liska <mliska@suse.cz>
912 * doc/extend.texi: Add missing values for __builtin_cpu_is and
913 __builtin_cpu_supports for x86 target.
915 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
918 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
920 (Arithmetic): Likewise for "lo_sum".
922 2018-11-12 Renlin Li <renlin.li@arm.com>
925 * dse.c (get_stored_val): Add check for compile-time constantness
928 2018-11-12 Sudakshina Das <sudi.das@arm.com>
930 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
931 (ARMv8_5a): New fgroup.
932 (armv8.5-a): New arch.
933 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
935 * config/arm/arm-tables.opt: Regenerate.
936 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
937 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
938 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
939 Add matching rules for -march=armv8.5-a and extensions.
940 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
941 Add sb and predres to all armv8-a except armv8.5-a.
943 2018-11-12 Richard Biener <rguenther@suse.de>
945 * tree-vrp.h (value_range[_base]::set): Make public. Provide
946 overload for single value.
947 (value_range[_base]::set_nonnull): New.
948 (value_range[_base]::set_null): Likewise.
949 (value_range): Document bitmap copying behavior, mark
950 copy constructor and assignment operator deleted.
951 (value_range::move): New.
952 (value_range::set_and_canonicalize): Default bitmap to zero.
953 (set_value_range_to_nonnull): Remove.
954 (set_value_range_to_null): Likewise.
955 (set_value_range): Likewise.
956 (set_value_range_to_value): Likewise.
957 (extract_range_from_unary_expr): Work on value_range_base.
958 (extract_range_from_binary_expr_1): Likewise. Rename to...
959 (extract_range_from_binary_expr): ... this.
960 * tree-vrp.c (value_range::update): Clear equiv bitmap
962 (value_range::move): New, move equiv bitmap.
963 (value_range_base::set_undefined): Avoid assignment.
964 (value_range::set_undefined): Likewise.
965 (value_range_base::set_varying): Likewise.
966 (value_range::set_varying): Likewise.
967 (set_value_range): Remove.
968 (value_range_base::set): New overload for value.
969 (value_range::set): Likewise.
970 (set_value_range_to_nonnull): Remove.
971 (value_range_base::set_nonnull): New.
972 (value_range::set_nonnull): Likewise.
973 (set_value_range_to_null): Remove.
974 (value_range_base::set_null): New.
975 (value_range::set_null): Likewise.
976 (range_is_null): Work on value_range_base.
977 (range_is_nonnull): Likewise.
978 (ranges_from_anti_range): Likewise.
979 (extract_range_into_wide_ints): Likewise.
980 (extract_range_from_multiplicative_op): Likewise.
981 (extract_range_from_binary_expr): Likewise. Update for API changes.
982 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
984 (value_range::intersect_helper): Avoid copy and assignment.
985 (value_range::union_helper): Likewise.
986 (determine_value_range_1): Adjust.
987 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
988 Avoid assignment by using move.
989 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
990 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
992 * tree-ssanames.c (get_range_info): Likewise.
993 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
994 * vr-values.c (vr_values::get_value_range): Adjust.
995 (vr_values::update_value_range): Likewise.
996 (symbolic_range_based_on_p): Work on value_range_base.
997 (vr_values::extract_range_from_binary_expr): Use value_range_base.
998 (vr_values::extract_range_from_unary_expr): Likewise.
999 (vr_values::extract_range_from_cond_expr): Avoid assignment.
1000 (vr_values::extract_range_from_comparison): Adjust.
1001 (vr_values::check_for_binary_op_overflow): Use value_range_base.
1002 (vr_values::extract_range_basic): Adjust.
1003 (vr_values::adjust_range_with_scev): Likewise.
1004 (vr_values::vrp_visit_assignment_or_call): Likewise.
1005 (vr_values::get_vr_for_comparison): Change API to avoid
1006 assignment and copy construction.
1007 (vr_values::compare_name_with_value): Adjust accordingly.
1008 (vr_values::compare_names): Likewise.
1009 (vr_values::extract_range_from_phi_node): Avoid assignment and
1010 bogus in-place modify of equiv bitmap.
1011 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
1012 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
1013 for extract_range_from_unary_expr API change.
1014 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
1016 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
1018 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
1020 2018-11-12 Richard Biener <rguenther@suse.de>
1022 * tree-vrp.h (value_range_base::symbolic_p,
1023 value_range_base::constant_p, value_range_base::zero_p,
1024 value_range_base::singleton_p): Move from value_range.
1025 (value_range::dump): Add.
1026 * gimple-ssa-evrp-analyze.c
1027 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
1028 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
1029 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1031 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
1032 (value_range::constant_p): Likewise.
1033 (value_range::singleton_p): Likewise.
1034 (value_range_base::dump): Add.
1035 (set_value_range_to_undefined): Remove.
1036 (set_value_range_to_varying): Likewise.
1037 (range_int_cst_p): Take value_range_base argument.
1038 (range_int_cst_singleton_p): Likewise.
1039 (value_range_constant_singleton): Likewise.
1040 (vrp_set_zero_nonzero_bits): Likewise.
1041 (extract_range_from_multiplicative_op): Use set_varying.
1042 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
1043 (extract_range_from_unary_expr): Likewise.
1044 (dump_value_range_base): Change to overload of dump_value_range.
1045 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
1046 (vrp_prop::visit_stmt): Likewise.
1047 (value_range::intersect_helper): Likewise.
1048 (value_range::union_helper): Likewise.
1049 (determine_value_range_1): Likewise.
1051 2018-11-12 Richard Biener <rguenther@suse.de>
1053 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
1054 (set_value_range_to_null): Likewise.
1055 * vr-values.c (vr_values::extract_range_from_comparison):
1056 Clear equiv for constant singleton ranges.
1058 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
1060 * config/i386/sse.md: Combine VFIXUPIMM* patterns
1061 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1062 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1063 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
1064 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1065 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1066 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
1068 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1071 * doc/extend.texi (Common Type Attributes): For the align type
1072 attribute, copy language about decreasing alignment from the
1073 corresponding variable attribute.
1075 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1077 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
1080 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1083 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
1084 -frtti and -fno-rtti code.
1086 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
1089 * doc/extend.texi (Other Builtins): Document probability associated
1090 with __builtin_expect.
1092 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
1095 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
1096 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
1097 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
1098 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
1100 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1102 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
1104 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
1106 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
1108 2018-11-11 Richard Biener <rguenther@suse.de>
1110 * tree-vrp.h (class value_range_base): New base class for
1111 value_range containing all but the m_equiv member.
1112 (dump_value_range_base): Add.
1113 (range_includes_zero_p): Work on value_range_base.
1114 * tree-vrp.c (value_range_base::set): Split out base handling
1116 (value_range::set): this.
1117 (value_range::set_equiv): New.
1118 (value_range_base::value_range_base): New constructors.
1119 (value_range_base::check): Split out base handling from...
1120 (value_range::check): this.
1121 (value_range::equal_p): Refactor in terms of
1122 ignore_equivs_equal_p which is now member of the base.
1123 (value_range_base::set_undefined): New.
1124 (value_range_base::set_varying): Likewise.
1125 (value_range_base::dump):Split out base handling from...
1126 (value_range::dump): this.
1127 (value_range_base::set_and_canonicalize): Split out base handling
1129 (value_range::set_and_canonicalize): this.
1130 (value_range_base::union_): New.
1131 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
1133 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
1134 instead of value_range everywhere.
1135 (ipcp_vr_lattice::print): Use dump_value_range_base.
1136 (ipcp_vr_lattice::meet_with): Adjust.
1137 (ipcp_vr_lattice::meet_with_1): Likewise.
1138 (ipa_vr_operation_and_type_effects): Likewise.
1139 (propagate_vr_across_jump_function): Likewise.
1140 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
1141 (ipa_get_value_range): Likewise.
1142 (ipa_set_jfunc_vr): Likewise.
1143 (ipa_compute_jump_functions_for_edge): Likewise.
1145 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
1148 * doc/invoke.texi (Optimize Options): Add @opindex entries
1149 for the positive forms of -fno-xxx and -mno-xxx options
1150 that were lacking them.
1152 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
1154 * combine.c (make_more_copies): Only make an intermediate copy if the
1155 dest of a move is a pseudo.
1157 2018-11-09 Maya Rashish <coypu@sdf.org>
1160 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
1161 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
1163 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
1167 * doc/invoke.texi (Optimize Options): Clarify default behavior
1168 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
1170 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
1172 PR tree-optimization/87940
1173 * expr.c (string_constant): Don't strip NOPS in subexpressions.
1174 Fold PLUS_EXPR correctly.
1176 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
1179 * config/s390/s390.c (s390_safe_relative_long_p): New function.
1180 (annotate_constant_pool_refs): Skip insns which support
1181 relative addressing.
1182 (annotate_constant_pool_refs_1): New helper function.
1183 (find_constant_pool_ref): Skip insns which support relative
1185 (find_constant_pool_ref_1): New helper function.
1186 (replace_constant_pool_ref): Skip insns which support
1187 relative addressing.
1188 (replace_constant_pool_ref_1): New helper function.
1189 (s390_mainpool_start): Adapt to the new signature.
1190 (s390_mainpool_finish): Likewise.
1191 (s390_chunkify_start): Likewise.
1192 (s390_chunkify_finish): Likewise.
1193 (pass_s390_early_mach::execute): Likewise.
1194 (s390_prologue_plus_offset): Likewise.
1195 (s390_emit_prologue): Likewise.
1196 (s390_emit_epilogue): Likewise.
1198 2018-11-09 Jakub Jelinek <jakub@redhat.com>
1200 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
1201 but unsupported lastprivate with conditional modifier.
1203 2018-11-09 Jeff Law <law@redhat.com>
1205 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
1206 unused argument better. Add gcc_unreachable to silence warning.
1208 2018-11-09 Martin Sebor <msebor@redhat.com>
1211 * attribs.c (has_attribute): New helper function.
1212 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
1213 * attribs.h (decls_mismatched_attributes): Declare.
1214 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
1215 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
1216 * common.opt (-Wattribute-alias): Take an argument.
1217 (-Wno-attribute-alias): New option.
1218 * doc/extend.texi (Common Function Attributes): Document copy.
1219 (Common Variable Attributes): Same.
1220 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
1221 (-Wattribute-alias): Document new option argument.
1223 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
1225 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
1226 matching criteria. Remove unused array initializer.
1228 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
1229 Jinsong Ji <jji@us.ibm.com>
1231 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
1232 constraints by introducing a new temporary.
1233 (_mm_cvtss_si64): Likewise.
1235 2018-11-09 Martin Liska <mliska@suse.cz>
1237 * common.opt: Add -fipa-stack-alignment flag.
1238 * doc/invoke.texi: Document it.
1239 * final.c (rest_of_clean_state): Guard stack
1240 shrinking with flag.
1242 2018-11-09 Martin Liska <mliska@suse.cz>
1244 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
1246 (ipa_discover_variable_flags): ... this.
1247 * common.opt: Come up with new flag -fipa-reference-addressable.
1248 * doc/invoke.texi: Document it.
1249 * ipa-reference.c (propagate): Call the renamed fn.
1250 * ipa-visibility.c (whole_program_function_and_variable_visibility):
1252 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
1254 (ipa_discover_variable_flags): ... this. Discover
1255 non-addressable variables only with the newly added flag.
1256 * opts.c: Enable the newly added flag with -O1 and higher
1259 2018-11-09 David Malcolm <dmalcolm@redhat.com>
1261 * json.cc (selftest::test_writing_literals): Fix comment.
1263 2018-11-09 Martin Liska <mliska@suse.cz>
1265 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
1266 string to a stack buffer.
1267 (aarch64_parse_cpu): Likewise.
1268 (aarch64_parse_tune): Likewise.
1270 2018-11-09 Richard Biener <rguenther@suse.de>
1272 PR tree-optimization/87953
1273 * tree-vect-loop.c (vectorizable_reduction): For analysis
1274 always pass ops[0] to vectorizable_condition.
1276 2018-11-09 Stafford Horne <shorne@gmail.com>
1277 Richard Henderson <rth@twiddle.net>
1278 Joel Sherrill <joel@rtems.org>
1280 * common/config/or1k/or1k-common.c: New file.
1281 * config/or1k/*: New.
1282 * config.gcc (or1k*-*-*): New.
1283 * configure.ac (or1k*-*-*): New test for openrisc tls.
1284 * configure: Regenerated.
1285 * doc/install.texi: Document OpenRISC triplets.
1286 * doc/invoke.texi: Document OpenRISC arguments.
1287 * doc/md.texi: Document OpenRISC.
1289 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
1291 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
1292 (arm7tdmi-s): Delete CPU.
1293 (arm710t): Add aliases for arm720t and arm740t.
1294 (arm720t, arm740t): Delete CPUs.
1295 (arm920t): Add aliases for arm920, arm922t and arm940t.
1296 (arm920, arm922t, arm940t): Delete CPUs.
1297 (arm10tdmi): Add alias for arm1020t.
1298 (arm1020t): Delete CPU.
1299 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
1300 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
1301 (arm10e): Add aliases for arm1020e and arm1022e.
1302 (arm1020e, arm1022e): Delete CPU.
1303 * config/arm/arm.md (generic_sched): Remove entries that are now
1305 (generic_vfp): Likewise.
1306 * config/arm/arm1020e.md: Simplify tuning selection based on alias
1308 * config/arm/arm-tune.md: Regenerated.
1309 * config/arm/arm-tables.opt: Regenerated.
1311 2018-11-09 Richard Biener <rguenther@suse.de>
1313 PR tree-optimization/87621
1314 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
1315 op with only phi inputs.
1316 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
1317 (ch_base::copy_headers): Run CSE on copied loop headers.
1318 (pass_ch_vect::process_loop_p): Simplify.
1320 2018-11-09 Alexandre Oliva <oliva@adacore.com>
1322 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
1323 for non-w64 x86_64 biarch.
1325 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
1327 PR rtl-optimization/86438
1328 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
1329 of in_b for the compare if in_b is SET_DEST.
1332 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
1333 non-toplevel UNSPEC.
1335 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1337 * tree-vrp.c (value_range::check): Do not access internals
1339 (value_range::singleton_p): Same.
1340 (value_range::type): Same.
1341 (vrp_finalize): Use value_range API.
1343 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1345 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
1347 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1349 * vr-values.c (vr_values::get_value_range): Use value_range API
1350 instead of piecing together ranges.
1351 (vr_values::update_value_range): Same.
1353 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
1355 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
1356 of performing ad-hoc calculations.
1357 * tree-ssanames.c (set_range_info): New overloaded function
1358 accepting value_range &.
1359 (get_range_info): Same.
1360 * tree-ssanames.h (set_range_info_raw): Remove.
1361 (set_range_info): New prototype.
1362 (get_range_info): Same.
1363 * tree-vrp.h (value_range::null_p): Rename to zero_p.
1364 * tree-vrp.c (value_range::null_p): Same.
1366 2018-11-09 Jan Hubicka <jh@suse.cz>
1368 * tree.c (fld_type_variant_equal_p): Test user align flag.
1369 (flt_type_variant): Copy user align flag.
1370 (fld_incomplete_type_of): Clear it.
1372 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1374 * config/arm/neon.md (div<mode>3): New pattern.
1376 2018-11-08 Andi Kleen <ak@linux.intel.com>
1378 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
1379 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
1380 (ix86_handle_option): Handle OPT_mptwrite.
1381 * config/i386/cpuid.h (bit_PTWRITE): Add.
1382 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
1383 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
1384 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
1385 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
1386 (ix86_option_override_internal): Handle PTA_PTWRITE.
1387 (ix86_valid_target_attribute_inner_p): Define ptwrite.
1388 (def_builtin2): Force UINT64 to be 64bit only.
1389 * config/i386/i386.h (TARGET_PTWRITE): Add.
1390 (TARGET_PTWRITE_P): Add.
1392 * config/i386/i386.md: Define ptwrite.
1393 * config/i386/i386.opt: Add -mptwrite.
1394 * config/i386/immintrin.h (_ptwrite64): Add.
1396 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
1397 * doc/invoke.texi: Document -mptwrite.
1399 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
1401 PR rtl-optimization/87600
1402 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
1403 * lra-constraints.c (process_alt_operands): Skip illegal hard
1404 register usage. Prefer reloading non hard register operands.
1406 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
1409 * doc/invoke.texi (Optimize Options): Clarify default behavior
1410 for -fno-sched-interblock and -fno-sched-spec.
1412 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
1414 * collect2.c (linker_select): Add USE_LLD_LD.
1415 (ld_suffixes): Add ld.lld.
1416 (main): Handle -fuse-ld=lld.
1417 * common.opt (-fuse-ld=lld): New option.
1418 * doc/invoke.texi (-fuse-ld=lld): Document.
1419 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
1421 2018-11-08 Paul Koning <ni1d@arrl.net>
1423 * config/pdp11/constraints.md: Add "Z" series constraints for use
1424 with pre-dec and post-inc addressing.
1425 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
1426 (pdp11_expand_operands): Add int argument (word count).
1427 (pdp11_sp_frame_offset): Delete.
1428 (pdp11_cmp_length): New function.
1429 (pushpop_regeq): New function.
1430 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
1432 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
1434 (pdp11_initial_elimination_offset): Ditto.
1435 (pdp11_expand_operands): Add word count argument. Bugfixes.
1436 (output_move_multiple): Change how pointer adjustment is done.
1437 (pdp11_gen_int_label): Correct format.
1438 (output_ascii): Ditto.
1439 (pdp11_asm_output_var): Add code for DEC assembler case.
1440 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
1442 (legitimate_const_double_p): Ditto.
1443 (pdp11_register_move_cost): Adjust for new register classes.
1444 (pdp11_regno_reg_class): Ditto.
1445 (expand_block_move): Delete.
1446 (pushpop_regeq): New function.
1447 (pdp11_legitimate_address_p): Bugfix in check for constant
1449 (pdp11_sp_frame_offset): Delete.
1450 (pdp11_reg_save_size): New helper function for new frame layout.
1451 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
1452 (pdp11_expand_shift): Bugfix in check for constant shift count.
1453 (pdp11_shift_length): Ditto.
1454 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
1455 (pdp11_cmp_length): New function.
1456 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
1457 some compile options.
1458 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
1459 (CALL_USED_REGISTERS): Ditto.
1460 (ELIMINABLE_REGS): Ditto.
1461 (REGISTER_NAMES): Ditto.
1462 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
1464 (REG_CLASS_NAMES): Ditto.
1465 (REG_CLASS_CONTENTS): Ditto. Also remove
1466 HARD_FRAME_POINTER_REGNUM.
1467 (CPU_REG_CLASS): New macro.
1468 (CLASS_MAX_NREGS): Adjust for new register classes.
1469 (FUNCTION_PROFILER): Make no-op.
1470 (may_call_alloca): Remove unused declaration.
1471 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
1472 (ASM_OUTPUT_SKIP): Fix format.
1473 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
1474 (HARD_FRAME_POINTER_REGNUM): Remove.
1476 (*rts): Rename. Remove epilogue related checks.
1477 (cmpsi, cmpdi): New insn.
1478 (cbranch<mode>4): Change to apply to SI and DI modes as well.
1479 (mov<mode>): Change constraints to enforce that push/pop
1480 destination cannot use the same register as source.
1481 (*mov<mode><cc_cc>): Ditto.
1482 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
1483 at assembly output rather than as RTL expander.
1484 (zero_extendqihi2): Bugfix in check for same registers.
1485 (adddi3_nocc): Bugfix in check for constant operand.
1486 (addsi3_nocc): Ditto.
1487 (subdi3_nocc): Ditto.
1488 (subsi3_nocc): Ditto.
1489 (negdi2_nocc): Copy input to pdp11_expand_operands.
1490 (negsi2_nocc): Ditto.
1491 (bswap2_nocc): Ditto.
1492 * config/pdp11/pdp11.opt (mlra): Fix documentation.
1493 * config/pdp11/t-pdp11: Use -Os.
1495 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
1497 * config/arm/parsecpu.awk (/alias/): New parsing rule.
1498 (/begin cpu/): Check that the cpu name hasn't been previously defined.
1499 (gen_comm_data): Print out CPU alias tables.
1500 (check_cpu): Match aliases when checking the CPU name.
1501 * config/arm/arm-protos.h (cpu_alias): New structure.
1502 (cpu_option): Add entry for aliases.
1503 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
1504 strongarm1100 and strongarm1110.
1505 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
1506 (config/arm/arm-generic.md): Remove redundant references to
1507 strongarm110, strongarm1100 and strongarm1110.
1508 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
1509 Scan aliases for additional hints.
1510 (arm_parse_cpu_option_name): Also match a cpu name against the list
1512 * config/arm/arm-tables.opt: Regenerated.
1513 * config/arm/arm-tune.md: Regenerated.
1515 2018-11-08 Jakub Jelinek <jakub@redhat.com>
1517 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
1518 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
1519 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
1520 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
1521 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
1522 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
1523 * gengtype.c (open_base_files): Add omp-general.h.
1524 * gimple.c (gimple_build_omp_critical):
1525 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
1526 gimple_omp_taskgroup_set_clauses.
1527 (gimple_build_omp_atomic_load): Add mo argument, call
1528 gimple_omp_atomic_set_memory_order.
1529 (gimple_build_omp_atomic_store): Likewise.
1530 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
1531 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
1533 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
1534 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
1535 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
1536 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
1537 different value for GF_OMP_ATOMIC_NEED_VALUE.
1538 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
1540 (struct gimple_statement_omp_single_layout): And remove here.
1541 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
1542 than gimple_statement_omp_single_layout.
1543 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
1545 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
1546 (gimple_omp_subcode): Formatting fix.
1547 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
1548 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
1549 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
1550 gimple_omp_teams_host, gimple_omp_teams_set_host,
1551 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
1552 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
1553 gimple_omp_taskgroup_set_clauses): New inline functions.
1554 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
1555 (gimple_build_omp_atomic_store): Likewise.
1556 (gimple_omp_atomic_seq_cst_p): Remove.
1557 (gimple_omp_atomic_memory_order): New function.
1558 (gimple_omp_atomic_set_seq_cst): Remove.
1559 (gimple_omp_atomic_set_memory_order): New function.
1560 (gimple_build_omp_taskgroup): Add clauses argument.
1561 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
1562 (dump_gimple_omp_task): Print taskwait with depend clauses.
1563 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
1564 dump_omp_atomic_memory_order.
1565 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
1566 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
1567 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
1568 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
1569 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
1570 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
1571 ORT_UNTIED_TASKLOOP enumerators.
1572 (enum gimplify_defaultmap_kind): New.
1573 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
1574 target_map_pointers_as_0len_arrays members, add defaultmap.
1575 (new_omp_context): Initialize defaultmap member.
1576 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1577 (maybe_fold_stmt): Don't fold even in host teams regions.
1578 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
1579 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
1580 ctx->omp_firstprivatize_variable.
1581 (omp_add_variable): Don't add private/firstprivate for VLAs in
1583 (omp_default_clause): Print "taskloop" rather than "task" if
1585 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1586 Handle new defaultmap clause kinds.
1587 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
1588 iterator to be lastprivate or private. Fix up diagnostics if linear
1589 is used on collapse>1 simd iterator.
1590 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
1591 (gimplify_omp_depend): New function.
1592 (gimplify_scan_omp_clauses): Add shared clause on parallel for
1593 combined parallel master taskloop{, simd} if taskloop has
1594 firstprivate, lastprivate or reduction clause. Handle
1595 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
1596 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
1597 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
1598 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
1599 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
1601 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
1602 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
1603 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
1604 OMP_CLAUSE_{TASK,IN}_REDUCTION.
1605 (gimplify_omp_task): Handle taskwait with depend clauses.
1606 (gimplify_omp_for): Add shared clause on parallel for combined
1607 parallel master taskloop{, simd} if taskloop has firstprivate,
1608 lastprivate or reduction clause. Use ORT_TASKLOOP or
1609 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
1610 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
1611 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
1612 __for_end and __for_range temporaries on OMP_PARALLEL for
1613 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
1614 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
1615 sandwiched in between two taskloops.
1616 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
1617 instead of ctx->omp_firstprivatize_variable.
1618 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
1619 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
1620 host teams, use gimplify_and_return_first etc. for body like
1621 for target or target data constructs, and at the end call
1622 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
1623 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
1624 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
1625 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
1626 gimple_omp_atomic_set_seq_cst calls.
1627 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
1628 case, handle taskgroup clauses.
1629 * lto-streamer-out.c (hash_tree): Handle
1630 OMP_CLAUSE_{TASK,IN}_REDUCTION.
1631 * Makefile.in (GTFILES): Add omp-general.h.
1632 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
1633 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
1634 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
1635 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
1636 BUILT_IN_GOMP_LOOP_DOACROSS_START,
1637 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
1638 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
1639 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
1640 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
1641 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
1642 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
1643 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
1644 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
1645 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
1646 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
1647 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
1648 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
1649 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
1650 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
1651 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
1652 * omp-expand.c (workshare_safe_to_combine_p): Return false for
1653 non-worksharing loops.
1654 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
1655 (determine_parallel_type): Don't combine parallel with worksharing
1656 which has _reductemp_ clause.
1657 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
1658 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
1659 if there is nonmonotonic modifier or if there is no modifier and no
1660 ordered clause. For dynamic and guided schedule without monotonic
1661 and nonmonotonic modifier, default to nonmonotonic.
1662 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
1663 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
1665 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
1666 there are any reduction clauses.
1667 (expand_taskwait_call): New function.
1668 (expand_teams_call): New function.
1669 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
1670 expand_teams_call for it. Formatting fix. Handle taskwait with
1672 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
1673 of worksharing loops with task reductions.
1674 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
1675 expansion of worksharing loops with task reductions.
1676 (expand_omp_sections): Handle expansion of sections with task
1678 (expand_omp_synch): For host teams call expand_omp_taskreg.
1679 (omp_memory_order_to_memmodel): New function.
1680 (expand_omp_atomic_load, expand_omp_atomic_store,
1681 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
1682 instead of gimple_omp_atomic_seq_cst_p.
1683 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
1684 depend clauses as a standalone directive.
1685 * omp-general.c (enum omp_requires): New variable.
1686 (omp_extract_for_data): Initialize have_reductemp member. Allow
1687 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
1688 GT_EXPR loops depending on incr sign. Formatting fixes.
1689 * omp-general.h (struct omp_for_data): Add have_reductemp member.
1690 (enum omp_requires): New enum.
1691 (omp_requires_mask): Declare.
1692 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
1694 * omp-low.c (struct omp_context): Add task_reductions and
1695 task_reduction_map fields.
1696 (is_host_teams_ctx): New function.
1697 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
1698 (use_pointer_for_field): Use is_global_var instead of
1699 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
1701 (build_outer_var_ref): Ignore taskgroup outer contexts.
1702 (delete_omp_context): Release task_reductions and task_reduction_map.
1703 (scan_sharing_clauses): Don't add any fields for reduction clause on
1704 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
1705 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
1706 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
1707 Handle OMP_CLAUSE_NONTEMPORAL.
1708 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
1710 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
1711 clauses with task modifier.
1712 (scan_omp_task): Handle taskwait with depend clauses.
1713 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
1714 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
1715 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
1716 (scan_omp_for): Fix comment formatting.
1717 (scan_omp_teams): Handle host teams constructs.
1718 (check_omp_nesting_restrictions): Allow teams with no outer
1719 OpenMP context. Adjust diagnostics for teams strictly nested into
1720 some explicit OpenMP construct other than target. Allow OpenMP atomics
1721 inside of simd regions.
1722 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
1723 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
1724 taskreg_nesting_level while scanning host teams construct.
1725 (task_reduction_read): New function.
1726 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
1727 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
1728 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
1729 second argument create_tmp_var if it is NULL. Don't ignore shared
1730 clauses in is_host_teams_ctx contexts. Handle
1731 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
1733 (lower_reduction_clauses): Ignore reduction clauses with task
1734 modifier. Remove second argument create_tmp_var if it is NULL.
1735 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
1736 (lower_send_clauses): Ignore reduction clauses with task modifier.
1737 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
1738 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
1739 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
1740 rely that it is the last stmt in body so far. Ignore outer taskgroup
1742 (omp_task_reductions_find_first, omp_task_reduction_iterate,
1743 lower_omp_task_reductions): New functions.
1744 (lower_omp_sections): Handle reduction clauses with taskgroup
1745 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
1746 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
1747 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
1749 (lower_omp_taskgroup): Handle taskgroup reductions.
1750 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
1751 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
1752 (lower_depend_clauses): If there are any
1753 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
1754 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
1755 seen, assume lowering is done already and return early. Set kind
1756 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
1757 (lower_omp_taskreg): Handle reduction clauses with task modifier on
1758 parallel construct. Handle reduction clause on taskloop construct.
1759 Handle taskwait with depend clauses.
1760 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
1761 for host teams constructs.
1762 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
1763 nontemporal and _reductemp_ clause entries.
1764 (omp_clause_code_name): Likewise.
1765 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
1766 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
1767 * tree-core.h (enum omp_clause_code): Add
1768 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
1769 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
1770 (struct tree_base): Add omp_atomic_memory_order field into union.
1771 Remove OMP_ATOMIC_SEQ_CST comment.
1772 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
1773 and OMP_CLAUSE_DEPEND_DEPOBJ.
1774 (struct tree_omp_clause): Add subcode.defaultmap_kind.
1775 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
1776 OpenMP constructs with body and clauses operands.
1777 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
1778 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
1779 (OMP_TASKGROUP_CLAUSES): Define.
1780 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
1781 OMP_CLAUSE__LOOPTEMP_.
1782 (OMP_ATOMIC_SEQ_CST): Remove.
1783 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
1784 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
1785 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
1786 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
1787 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
1788 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
1789 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
1790 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
1791 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
1792 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
1794 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
1795 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
1796 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
1797 (convert_local_omp_clauses): Likewise. Remove useless test.
1798 * tree-parloops.c (create_call_for_reduction_1): Pass
1799 OMP_MEMORY_ORDER_RELAXED as new argument to
1800 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
1801 * tree-pretty-print.c (dump_omp_iterators): New function.
1802 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
1803 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
1804 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
1805 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
1806 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
1807 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
1808 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
1809 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
1810 (dump_omp_atomic_memory_order): New function.
1811 (dump_generic_node): Use it. Print taskgroup clauses. Print
1812 taskwait with depend clauses.
1813 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
1814 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
1815 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
1816 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
1817 write_ts_omp_clause_tree_pointers): Likewise.
1819 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1822 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
1823 "-missed", "-note", and "-all" sub-options.
1824 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
1826 (can_early_inline_edge_p): Likewise.
1827 (want_early_inline_function_p): Likewise.
1828 (want_inline_self_recursive_call_p): Likewise.
1829 (recursive_inlining): Likewise.
1830 (inline_small_functions): Likewise.
1831 (flatten_function): Likewise.
1832 (ipa_inline): Likewise.
1833 (inline_always_inline_functions): Likewise.
1834 (early_inline_small_functions): Likewise.
1835 (early_inliner): Likewise.
1836 * tree-inline.c (expand_call_inline): Likewise.
1838 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1840 * pretty-print.c (pp_format): Handle %f.
1841 (selftest::test_pp_format): Add test of %f.
1842 * pretty-print.h (pp_double): New macro.
1844 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1846 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
1847 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
1848 (dump_pretty_printer::decode_format): Implement "%C" for
1850 (selftest::test_capture_of_dump_calls): Rename "where" to
1851 "stmt_loc". Convert test_decl to a function decl and set its
1852 location. Add a symbol_table_test RAII instance and a
1853 cgraph_node, using it to test "%C" and dump_symtab_node.
1855 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
1858 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
1860 2018-11-08 David Malcolm <dmalcolm@redhat.com>
1862 * cgraph.c: Include "selftest.h".
1863 (saved_symtab): New variable.
1864 (selftest::symbol_table_test::symbol_table_test): New ctor.
1865 (selftest::symbol_table_test::~symbol_table_test): New dtor.
1866 (selftest::test_symbol_table_test): New test.
1867 (selftest::cgraph_c_tests): New.
1868 * cgraph.h (saved_symtab): New decl.
1869 (selftest::symbol_table_test): New class.
1870 * selftest-run-tests.c (selftest::run_tests): Call
1871 selftest::cgraph_c_tests.
1872 * selftest.h (selftest::cgraph_c_tests): New decl.
1874 2018-11-08 Richard Biener <rguenther@suse.de>
1876 * tree-data-ref.h (lambda_int): New typedef.
1877 (lambda_vector_gcd): Adjust.
1878 (lambda_vector_new): Likewise.
1879 (lambda_matrix_new): Likewise.
1880 * tree-data-ref.c (print_lambda_vector): Adjust.
1882 2018-11-08 Richard Biener <rguenther@suse.de>
1884 PR tree-optimization/87929
1885 * tree-complex.c (expand_complex_comparison): Clean EH.
1887 2018-11-08 Martin Liska <mliska@suse.cz>
1889 * doc/extend.texi: Reword.
1890 * predict.c (expr_expected_value_1): Likewise.
1892 2018-11-08 Richard Biener <rguenther@suse.de>
1894 PR tree-optimization/87913
1895 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
1896 of extreme values to ordered comparisons.
1898 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1901 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
1903 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1905 * doc/invoke.texi: Remove leading dash from @opindex entries
1906 throughout the file.
1908 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
1911 * doc/invoke.texi (Option Summary): Add -e and --entry.
1912 (Link Options): Likewise.
1914 2018-11-07 Nathan Sidwell <nathan@acm.org>
1917 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
1918 --disable-checking bootstrap.
1920 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
1922 * configure: Regenerated.
1924 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1927 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
1928 to the mode of the widest field iff the widest field has mode class
1929 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
1932 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
1934 * common.opt: Add -Wattribute-warning.
1935 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
1936 * expr.c (expand_expr_real_1): Add new attribute to warning_at
1937 call to allow user configure behavior of "warning" attribute.
1939 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1941 * target.def: Put @: after every vs., e.g., and i.e. where it is
1942 followed by whitespace.
1943 * doc/extend.texi: Ditto.
1944 * doc/fragments.texi: Ditto.
1945 * doc/gimple.texi: Ditto.
1946 * doc/implement-c.texi: Ditto.
1947 * doc/install.texi: Ditto.
1948 * doc/invoke.texi: Ditto.
1949 * doc/md.texi: Ditto.
1950 * doc/plugins.texi: Ditto.
1951 * doc/rtl.texi: Ditto.
1952 * doc/sourcebuild.texi: Ditto.
1953 * doc/tm.texi.in: Ditto.
1954 * doc/ux.texi: Ditto.
1955 * doc/tm.texi: Regenerate.
1957 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1959 * config/arm/arm-cpus.in (ares): New entry.
1960 * config/arm/arm-tables.opt: Regenerate.
1961 * config/arm/arm-tune.md: Likewise.
1962 * doc/invoke.texi (ARM Options): Document ares.
1964 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1966 * config/aarch64/aarch64-cores.def (ares): Define.
1967 * config/aarch64/aarch64-tune.md: Regenerate.
1968 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
1970 2018-11-07 Jan Hubicka <jh@suse.cz>
1972 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
1973 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
1974 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
1977 2018-11-07 Richard Biener <rguenther@suse.de>
1979 PR tree-optimization/87914
1980 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
1982 (vectorizable_reduction): Handle shifts and rotates by dispatching
1983 to vectorizable_shift.
1984 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
1985 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
1987 (vectorizable_shift): Export and handle being called as
1989 (vect_analyze_stmt): Call vectorizable_shift after
1990 vectorizable_reduction.
1991 * tree-vectorizer.h (vectorizable_shift): Declare.
1993 2018-11-07 Jan Hubicka <jh@suse.cz>
1995 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
1996 than const decls in TREE_VALUE of enum.
1997 (dump_type_inheritance_graph): Improve duplicate dumping.
1998 (free_enum_values): New.
1999 (build_type_inheritance_graph): Use it.
2000 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
2001 which are not main variants or not ODR types.
2002 (verify_type_variant): Expect variants to have no TYPE_VALUES.
2004 2018-11-07 Richard Biener <rguenther@suse.de>
2006 * ipa-inline.c (want_inline_small_function_p): Compute
2007 big_speedup_p lazily and last.
2009 2018-11-07 Jan Hubicka <jh@suse.cz>
2011 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
2012 building incomplete variant of complete type.
2013 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
2014 variant of complete type.
2016 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2018 * config/mips/mips.c: Fix typo in documentation of
2019 mips_loongson_ext2_prefetch_cookie.
2020 (mips_option_override): fix brain twister logical.
2021 * config/mips/mips.h: Fix typo in documentation of
2022 ISA_HAS_CTZ_CTO and define pattern.
2023 * config/mips/mips.md (prefetch): Hoist EXT2 above
2025 (prefetch_indexed): Hoist EXT2 above the EXT block.
2027 2018-11-07 Jan Hubicka <jh@suse.cz>
2029 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
2030 return and parameter types of function and method types.
2031 (free_lang_data_in_cgraph): Update.
2033 2018-11-07 Martin Liska <mliska@suse.cz>
2035 PR rtl-optimization/87868
2036 * postreload-gcse.c (eliminate_partially_redundant_load): Set
2037 threshold to max_count if we would overflow.
2038 * profile-count.h: Make max_count a public constant.
2040 2018-11-07 Martin Liska <mliska@suse.cz>
2042 * mem-stats.h: Fix GNU coding style.
2044 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2046 * config/mips/gs264e.md: New.
2047 * config/mips/mips-cpus.def: Define gs264e.
2048 * config/mips/mips-tables.opt: Regenerate.
2049 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
2051 (mips_issue_rate): Add support for gs264e.
2052 (mips_multipass_dfa_lookahead): Likewise.
2053 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
2054 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
2055 (MIPS_ASE_MSA_SPEC): New.
2056 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
2057 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
2058 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
2059 * config/mips/mips.md: Include gs264e.md.
2060 (processor): Add gs264e.
2061 * config/mips/mips.opt (MSA): Use Mask instead of Var.
2062 * doc/invoke.texi: Add gs264e to supported architectures.
2064 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2066 * config/mips/gs464e.md: New.
2067 * config/mips/mips-cpus.def: Define gs464e.
2068 * config/mips/mips-tables.opt: Regenerate.
2069 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
2071 (mips_issue_rate): Add support for gs464e.
2072 (mips_multipass_dfa_lookahead): Likewise.
2073 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
2074 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
2075 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
2076 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
2077 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
2078 * config/mips/mips.md: Include gs464e.md.
2079 (processor): Add gs464e.
2080 * doc/invoke.texi: Add gs464e to supported architectures.
2082 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2084 * config/mips/loongson3a.md: Rename to ...
2085 * config/mips/gs464.md: ... here.
2086 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
2087 as an alias of gs464 processor.
2088 * config/mips/mips-tables.opt: Regenerate.
2089 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
2090 instead of PROCESSOR_LOONGSON_3A.
2091 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
2093 (mips_option_override): Enable MMI and EXT for gs464.
2094 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
2095 Rename TUNE_LOONGSON_3A to TUNE_GS464.
2096 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
2097 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
2098 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
2100 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
2101 (processor): Add gs464;
2102 * doc/invoke.texi: Add gs464 to supported architectures.
2104 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2106 * config/mips/mips-protos.h
2107 (mips_loongson_ext2_prefetch_cookie): New prototype.
2108 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
2109 (mips_option_override): Enable TARGET_LOONGSON_EXT when
2110 TARGET_LOONGSON_EXT2 is true.
2111 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
2112 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
2113 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
2114 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
2115 TARGET_LOONGSON_EXT2.
2116 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
2117 (define_insn "ctz<mode>2"): New insn pattern.
2118 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
2119 (define_insn "prefetch_indexed_<mode>"): Include
2120 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
2121 * config/mips/mips.opt (-mloongson-ext2): Add option.
2122 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
2124 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2126 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
2127 __mips_loongson_ext.
2128 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
2129 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
2131 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
2132 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
2133 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
2134 instead of TARGET_LOONGSON_3A.
2135 * config/mips/mips.opt (-mloongson-ext): Add option.
2136 * gcc/doc/invoke.texi (-mloongson-ext): Document.
2138 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
2140 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
2141 * config/mips/loongson.md: Move to ...
2142 * config/mips/loongson-mmi.md: here; Adjustment.
2143 * config/mips/loongson.h: Move to ...
2144 State as deprecated. Include loongson-mmiintrin.h for back
2145 compatibility and warning.
2146 * config/mips/loongson-mmiintrin.h: ... here.
2147 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
2148 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
2149 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
2150 (mips_option_override): Make sure MMI use hard float;
2151 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
2152 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
2153 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
2154 TARGET_LOONGSON_VECTORS.
2155 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
2156 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
2157 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
2158 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
2160 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
2161 TARGET_LOONGSON_VECTORS.
2162 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
2163 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
2164 (Loongson MMI patterns): Include loongson-mmi.md instead of
2166 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
2167 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
2169 2018-11-07 Richard Biener <rguenther@suse.de>
2172 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
2173 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
2175 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
2177 PR rtl-optimization/87874
2178 * lra.c (lra_substitute_pseudo): Do not create a subreg for
2181 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
2183 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
2184 if not in indexed or indirect form.
2185 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
2186 (bswap<mode>2_store): Ditto.
2188 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
2190 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
2193 2018-11-06 Richard Biener <rguenther@suse.de>
2195 PR tree-optimization/86850
2196 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
2197 instead of src.m_vec.
2199 2018-11-06 Jan Hubicka <jh@suse.cz>
2201 * tree.c (fld_simplified_type_name): Break out form ...
2202 (free_lang_data_in_type): ... here.
2203 (fld_type_variant_equal_p): Use it.
2205 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2207 * config/default-d.c: Include memmodel.h.
2209 * config/sol2-d.c: New file.
2210 * config/t-sol2 (sol2-d.o): New rule.
2211 * config.gcc <*-*-solaris2*>: Set d_target_objs,
2212 target_has_targetdm.
2214 2018-11-06 Jan Hubicka <jh@suse.cz>
2216 * tree.c (fld_type_variant): Also copy alignment; be sure that
2217 new variant is equal.
2219 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
2222 * config/s390/s390.md: Add relative_long attribute.
2224 2018-11-06 Jan Hubicka <jh@suse.cz>
2226 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
2227 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
2228 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
2229 stream TYPE_NEEDS_CONSTRUCTING.
2230 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
2231 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
2233 2018-11-06 Richard Biener <rguenther@suse.de>
2235 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
2237 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
2239 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2241 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
2242 REG_CLASS_CONTENTS[GEN_REGS].
2243 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
2245 2018-11-06 Jan Hubicka <jh@suse.cz>
2247 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
2249 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2252 * config/sol2.h (ASAN_CC1_SPEC): Define.
2253 (LD_WHOLE_ARCHIVE_OPTION): Define.
2254 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
2255 (ASAN_REJECT_SPEC): Provide default.
2256 (LIBASAN_EARLY_SPEC): Define.
2257 (LIBTSAN_EARLY_SPEC): Define.
2258 (LIBLSAN_EARLY_SPEC): Define.
2259 * config/i386/sol2.h (CC1_SPEC): Redefine.
2260 (ASAN_REJECT_SPEC): Define.
2262 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
2263 (TARGET_ASAN_SHADOW_OFFSET): Define.
2264 (sparc_asan_shadow_offset): New function.
2265 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
2266 (ASAN_REJECT_SPEC): Define.
2268 2018-11-06 Jan Hubicka <jh@suse.cz>
2270 * tree.c (fld_type_variant): Copy canonical type.
2271 (fld_incomplete_type_of): Check that canonical types looks sane;
2272 copy canonical type.
2273 (verify_type): Accept when incomplete type has complete canonical type.
2275 2018-11-06 Jan Hubicka <jh@suse.cz>
2277 * tree.c (free_lang_data): Reset overwite_assembler_name,
2278 print_xnode, print_decl, print_type and print_identifier of
2281 2018-11-06 Richard Biener <rguenther@suse.de>
2283 PR tree-optimization/87889
2284 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
2285 Do nothing if old and new arg are the same
2287 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
2290 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
2291 attributes for operands 3 and 4.
2293 2018-11-06 Richard Biener <rguenther@suse.de>
2296 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
2297 matching bitfield insertion.
2299 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
2301 * auto-inc-dec.c: Include valtrack.h. Improve comments.
2302 (reg_next_debug_use): New.
2303 (attempt_change): Propagate adjusted expression into affected
2305 (merge_in_block): Track uses in debug insns.
2306 (pass_inc_dec::execute): Allocate and release
2309 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
2311 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
2312 (_mm512_fixupimm_round_pd): Update parameters and builtin.
2313 (_mm512_maskz_fixupimm_round_pd): Ditto.
2314 (_mm512_fixupimm_round_ps): Ditto.
2315 (_mm512_maskz_fixupimm_round_ps): Ditto.
2316 (_mm_fixupimm_round_sd): Ditto.
2317 (_mm_maskz_fixupimm_round_sd): Ditto.
2318 (_mm_fixupimm_round_ss): Ditto.
2319 (_mm_maskz_fixupimm_round_ss): Ditto.
2320 (_mm512_fixupimm_pd): Ditto.
2321 (_mm512_maskz_fixupimm_pd): Ditto.
2322 (_mm512_fixupimm_ps): Ditto.
2323 (_mm512_maskz_fixupimm_ps): Ditto.
2324 (_mm_fixupimm_sd): Ditto.
2325 (_mm_maskz_fixupimm_sd): Ditto.
2326 (_mm_fixupimm_ss): Ditto.
2327 (_mm_maskz_fixupimm_ss): Ditto.
2328 (_mm512_mask_fixupimm_round_pd): Update builtin.
2329 (_mm512_mask_fixupimm_round_ps): Ditto.
2330 (_mm_mask_fixupimm_round_sd): Ditto.
2331 (_mm_mask_fixupimm_round_ss): Ditto.
2332 (_mm512_mask_fixupimm_pd): Ditto.
2333 (_mm512_mask_fixupimm_ps): Ditto.
2334 (_mm_mask_fixupimm_sd): Ditto.
2335 (_mm_mask_fixupimm_ss): Ditto.
2336 * config/i386/avx512vlintrin.h:
2337 (_mm256_fixupimm_pd): Update parameters and builtin.
2338 (_mm256_maskz_fixupimm_pd): Ditto.
2339 (_mm256_fixupimm_ps): Ditto.
2340 (_mm256_maskz_fixupimm_ps): Ditto.
2341 (_mm_fixupimm_pd): Ditto.
2342 (_mm_maskz_fixupimm_pd): Ditto.
2343 (_mm_fixupimm_ps): Ditto.
2344 (_mm_maskz_fixupimm_ps): Ditto.
2345 (_mm256_mask_fixupimm_pd): Update builtin.
2346 (_mm256_mask_fixupimm_ps): Ditto.
2347 (_mm_mask_fixupimm_pd): Ditto.
2348 (_mm_mask_fixupimm_ps): Ditto.
2349 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
2350 * config/i386/i386-builtin.def: Update builtin definitions.
2351 * config/i386/i386.c: Handle new builtin types and remove useless ones.
2352 * config/i386/sse.md: Update VFIXUPIMM* patterns.
2353 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2354 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2355 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
2356 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2357 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2358 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
2359 * config/i386/subst.md:
2360 (round_saeonly_sd_mask_operand4): Add new subst_attr.
2361 (round_saeonly_sd_mask_op4): Ditto.
2362 (round_saeonly_expand_operand5): Ditto.
2363 (round_saeonly_expand): Update.
2365 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
2367 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
2369 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
2371 PR rtl-optimization/87871
2372 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
2374 2018-11-05 Paul Koning <ni1d@arrl.net>
2376 * doc/sourcebuild.texi (target attributes): Document new "inf"
2377 effective target keyword.
2379 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2381 * config/s390/s390.c (s390_register_move_cost): Increase costs for
2382 moves involving the CC reg.
2384 2018-11-05 Richard Biener <rguenther@suse.de>
2386 PR tree-optimization/87873
2387 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
2389 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
2390 * tree-vect-loop.c (vect_transform_loop): When splitting the
2391 loop exit also create forwarder PHIs for constants.
2392 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
2393 Handle constant to_arg, add extra checking we match up the correct
2396 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2398 * config/s390/s390.md: QImode and HImode for load on condition.
2400 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
2402 * config/s390/predicates.md: Fix typo.
2403 * config/s390/s390.md: Allow immediates for load on condition.
2405 2018-11-05 Martin Liska <mliska@suse.cz>
2407 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
2408 * bitmap.h (struct bitmap_usage): Likewise.
2409 * ggc-common.c (SCALE): Remove.
2411 (struct ggc_usage): Use SIZE_AMOUNT. And update
2413 * ggc-page.c (SCALE): Remove.
2414 (STAT_LABEL): Remove.
2415 (ggc_print_statistics): Use SIZE_AMOUNT.
2416 * gimple.h (SCALE): Remove.
2418 * input.c (ONE_K): Remove.
2421 (STAT_LABEL): Likewise.
2422 (FORMAT_AMOUNT): Likewise.
2423 (dump_line_table_statistics): Use SIZE_AMOUNT.
2424 * mem-stats.h (struct mem_usage): Likewise.
2425 * rtl.c (dump_rtx_statistics): Likewise.
2426 (rtx_alloc_counts): Change type to size_t.
2427 (rtx_alloc_sizes): Likewise.
2428 (rtx_count_cmp): New.
2429 (dump_rtx_statistics): Sort first based on counts.
2430 * tree.c (tree_nodes_cmp): New.
2431 (tree_codes_cmp): New.
2432 (dump_tree_statistics): Sort first based on counts.
2433 * system.h (ONE_K): New.
2435 (SIZE_SCALE): Likewise.
2436 (SIZE_LABEL): Likewise.
2437 (SIZE_AMOUNT): Likewise.
2438 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
2439 * tree-dfa.c (dump_dfa_stats): Likewise.
2440 * tree-phinodes.c (phinodes_print_statistics): Likewise.
2441 * tree-ssanames.c (ssanames_print_statistics): Likewise.
2442 * tree.c (dump_tree_statistics): Likewise.
2443 * vec.c (struct vec_usage): Likewise.
2444 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
2445 trigger a -Werror=format-overflow with
2446 --enable-gather-detailed-stats.
2448 2018-11-05 Martin Liska <mliska@suse.cz>
2450 * mem-stats.h (mem_alloc_description::release_instance_overhead):
2452 * vec.c (struct vec_usage): Register m_element_size.
2453 (vec_prefix::register_overhead): New arguments: elements and
2455 (vec_prefix::release_overhead): Subtract elements.
2456 * vec.h (struct vec_prefix): Change signature.
2457 (va_heap::reserve): Pass proper arguments.
2458 (va_heap::release): Likewise.
2460 2018-11-05 Martin Liska <mliska@suse.cz>
2462 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
2466 2018-11-05 Richard Biener <rguenther@suse.de>
2468 * tree-scalar-evolution.h (final_value_replacement_loop): Update
2470 * tree-scalar-evolution.c (final_value_replacement_loop): Return
2471 whether anything was done.
2472 (scev_const_prop): Remove constant propagation part, fold
2474 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
2475 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
2478 2018-11-05 Jakub Jelinek <jakub@redhat.com>
2480 PR tree-optimization/87859
2481 * gimple-ssa-store-merging.c (struct merged_store_group): Add
2482 only_constants and first_nonmergeable_order members.
2483 (merged_store_group::merged_store_group): Initialize them.
2484 (merged_store_group::do_merge): Clear only_constants member if
2485 adding something other than INTEGER_CST store.
2486 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
2487 stores with order >= first_nonmergeable_order. Use
2488 merged_store->only_constants instead of always recomputing it.
2489 Set merged_store->first_nonmergeable_order if we've skipped any
2490 stores. Attempt to merge overlapping INTEGER_CST stores that
2491 we would otherwise skip.
2494 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
2496 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
2499 * config/i386/emmintrin.h (__v16qs): New to cope with option
2501 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
2502 (_mm_cmplt_epi8): Likewise.
2503 (_mm_cmpgt_epi8): Likewise.
2505 2018-11-05 Richard Biener <rguenther@suse.de>
2507 PR rtl-optimization/87852
2508 * fwprop.c (use_killed_between): Only consider single-defs of the
2509 use whose definition statement dominates the use.
2511 2018-11-05 Martin Liska <mliska@suse.cz>
2514 * doc/invoke.texi: Remove options that are
2515 not disabled with -Os.
2517 2018-11-05 Martin Liska <mliska@suse.cz>
2520 * doc/extend.texi: Update constrain about the last argument
2521 of __builtin_expect_with_probability.
2523 2018-11-05 Martin Liska <mliska@suse.cz>
2526 * predict.c (expr_expected_value_1): Verify
2527 that last argument is a real constants and emit
2530 2018-11-05 Martin Liska <mliska@suse.cz>
2532 PR gcov-profile/77698
2533 * ipa-profile.c (ipa_profile): Adjust hotness threshold
2536 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2538 PR tree-optimization/86572
2539 * builtins.c (c_strlen): Handle negative offsets in a safe way.
2541 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2543 PR tree-optimization/87672
2544 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
2545 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
2547 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
2550 * cfgexpand.c (pass_expand::execute): Move the call to
2551 finish_eh_generation in front of the call to expand_stack_alignment.
2553 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
2555 * common/config/i386/i386-common.c (processor_alias_table): Add
2557 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
2558 (case ${target}): Add znver2.
2559 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
2560 -march=native recognize znver2 processors.
2561 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
2562 * config/i386/i386.c (m_znver2): New definition.
2563 (m_ZNVER): New definition.
2564 (m_AMD_MULTIPLE): Includes m_znver2.
2565 (processor_cost_table): Add znver2 entry.
2566 (processor_target_table): Add znver2 entry.
2567 (get_builtin_code_for_version): Set priority for
2569 (processor_model): Add M_AMDFAM17H_ZNVER2.
2570 (arch_names_table): Ditto.
2571 (ix86_reassociation_width): Include znver2.
2572 * config/i386/i386.h (TARGET_znver2): New definition.
2573 (struct ix86_size_cost): Add TARGET_ZNVER2.
2574 (enum processor_type): Add PROCESSOR_ZNVER2.
2575 * config/i386/i386.md (define_attr "cpu"): Add znver2.
2576 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
2577 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
2578 (ix86_adjust_cost): Add znver2.
2579 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
2580 * gcc/doc/extend.texi: Add details about znver2.
2581 * gcc/doc/invoke.texi: Add details about znver2.
2583 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
2587 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
2590 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2592 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
2593 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
2595 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
2597 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
2598 shrink-wrapping when -mtrack-speculation.
2600 2018-11-02 Richard Biener <rguenther@suse.de>
2602 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
2603 (pop_cost_one_pair): Do not free pair.
2604 (pop_best_coalesce): Likewise.
2605 (create_coalesce_list): Initialize obstack.
2606 (delete_coalesce_list): Free obstack.
2607 (find_coalesce_pair): Obstack-allocate coalesce pairs.
2608 (add_cost_one_coalesce): Likewise.
2609 (struct live_track): Remove bitmap pointer indirections.
2610 (new_live_track): Adjust.
2611 (delete_live_track): Likewise.
2612 (live_track_remove_partition): Likewise.
2613 (live_track_add_partition): Likewise.
2614 (live_track_live_p): Likewise.
2615 (live_track_process_def): Likewise.
2616 (live_track_clear_base_vars): Likewise.
2618 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2620 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
2621 * configure: Regenerate.
2622 * config.in: Regenerate.
2623 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
2624 HAVE_AS_REGISTER_PSEUDO_OP guard.
2625 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
2626 !HAVE_AS_REGISTER_PSEUDO_OP support.
2627 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
2630 2018-11-02 Richard Biener <rguenther@suse.de>
2632 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
2635 2018-11-02 Richard Biener <rguenther@suse.de>
2637 PR tree-optimization/87776
2638 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
2639 executable when iterating but running into rpo-vn-max-loop-depth
2640 and not eliding the iteration.
2642 2018-11-30 Jan Hubicka <jh@suse.cz>
2644 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
2647 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
2649 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
2651 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
2652 Combine with rs6000_address_for_fpconvert.
2653 (rs6000_address_for_fpconvert): Combine with
2654 rs6000_force_indexed_or_indirect_mem.
2655 (rs6000_expand_vector_init): Change function call from
2656 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
2657 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
2658 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
2659 (floatsi<mode>2_lfiwax_mem): Ditto.
2660 (floatunssi<mode>2_lfiwzx): Ditto.
2661 (floatunssi<mode>2_lfiwzx_mem): Ditto.
2662 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
2663 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
2664 (fix_trunc<mode>si2_stfiwx): Ditto.
2665 (fixuns_trunc<mode>si2_stfiwx): Ditto.
2666 (float_<mode>si2_hw): Ditto.
2667 (floatuns_<mode>si2_hw): Ditto.
2668 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
2669 (vsx_splat_<mode>): Ditto.
2671 2018-11-01 Joseph Myers <joseph@codesourcery.com>
2673 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
2674 at end of assembler input text.
2675 * configure: Regenerate.
2677 2018-11-01 Jakub Jelinek <jakub@redhat.com>
2679 PR tree-optimization/87826
2680 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
2681 negative or larger or equal to type's precision.
2683 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
2685 * opts.c (default_options_table): Do not enable
2686 OPT_fdelayed_branch at -Og.
2687 * doc/invoke.texi (-fdelayed-branch): Document it.
2689 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
2691 * optabs-libfuncs.c (build_libfunc_function_visibility):
2692 New, split out from...
2693 (build_libfunc_function): ... here.
2694 (init_one_libfunc_visibility): New, split out from ...
2695 (init_one_libfunc): ... here.
2697 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
2698 scratch register need not be early-clobber. Document the reason
2699 why we cannot use ST<OP>.
2701 2018-10-31 Joseph Myers <joseph@codesourcery.com>
2704 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
2705 line for second argument of AC_DEFINE_UNQUOTED.
2706 * doc/install.texi (Tools/packages necessary for modifying GCC):
2707 Update to autoconf 2.69 and automake 1.15.1.
2708 * aclocal.m4, config.in, configure: Regenerate.
2710 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
2712 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
2715 2018-10-31 Martin Liska <mliska@suse.cz>
2718 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
2719 Add new argument invalid_extension.
2720 (aarch64_get_all_extension_candidates): New function.
2721 (aarch64_rewrite_selected_cpu): Add NULL to function call.
2722 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
2724 (aarch64_get_all_extension_candidates): New function.
2725 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
2726 argument invalid_extension.
2727 (aarch64_parse_cpu): Likewise.
2728 (aarch64_print_hint_for_extensions): New function.
2729 (aarch64_validate_mcpu): Provide hint about invalid extension.
2730 (aarch64_validate_march): Likewise.
2731 (aarch64_handle_attr_arch): Pass new argument.
2732 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
2733 (aarch64_handle_attr_isa_flags): Likewise.
2735 2018-10-31 Richard Biener <rguenther@suse.de>
2739 * tree-outof-ssa.c (insert_backedge_copies): Restrict
2740 copy generation to useful cases. Place the copy before
2741 the definition of the backedge value when possible.
2743 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2745 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
2746 * config/arc/arc.c (arc_active_insn): New function.
2747 (check_store_cacheline_hazard): Likewise.
2748 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
2749 (arc_override_options): Disable delay slot scheduler for older
2751 (arc_store_addr_hazard_p): New implementation, old one renamed to
2753 (arc_store_addr_hazard_internal_p): Renamed.
2754 (arc_reorg): Don't combine into brcc instructions which are part
2755 of hardware hazard solution.
2756 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
2757 (tune_arc700): Likewise.
2758 * config/arc/arc.opt (arc7xx): New tune value.
2759 * config/arc/arc700.md: Improve A7 scheduler.
2761 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2763 * config/arc/arc.c (arc_override_options): Remove
2764 TARGET_COMPACT_CASESI.
2765 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
2766 (CASE_VECTOR_MODE): Likewise.
2767 (CASE_VECTOR_PC_RELATIVE): Likewise.
2768 (CASE_VECTOR_SHORTEN_MODE): Likewise.
2769 (CASE_VECTOR_SHORTEN_MODE1): Delete.
2770 (ADDR_VEC_ALIGN): Update.
2771 (ASM_OUTPUT_CASE_LABEL): Undefine.
2772 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
2773 (TARGET_BI_BIH): Define.
2774 (DEFAULT_BRANCH_INDEX): Likewise.
2775 * config/arc/arc.md (casesi): Rework to accept BI/BIH
2776 instructions, remove compact_casesi use case.
2777 (casesi_compact_jump): Remove.
2778 (casesi_dispatch): New pattern.
2779 * config/arc/arc.opt: Add mbranch-index option. Deprecate
2780 compact_casesi option.
2781 * doc/invoke.texi: Document mbranch-index option.
2783 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2785 * config/arc/arc.c (arc_get_tp): Remove function.
2786 (arc_emit_call_tls_get_addr): Likewise.
2787 (arc_call_tls_get_addr): New function.
2788 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
2789 * config/arc/arc.md (tls_load_tp_soft): Remove.
2790 (tls_gd_get_addr): Likewise.
2792 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
2794 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
2795 (mulsi3_600_lib): Remove pattern.
2796 (umulsi3_highpart_600_lib_le): Likewise.
2797 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
2798 (umulsidi3): Remove call to umulsidi3_600_lib.
2799 (umulsidi3_600_lib): Remove pattern.
2800 (peephole2): Remove peephole using the above deprecated patterns.
2802 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2805 * config/arm/arm.c (arm_option_check_internal): Disable the combined
2806 use of -mslow-flash-data and -mword-relocations.
2807 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
2808 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
2810 * doc/invoke.texi (-mword-relocations): Mention conflict with
2812 (-mslow-flash-data): Reciprocally.
2814 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
2816 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
2817 16-byte modes held in GP registers to use an even regno.
2819 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
2820 (aarch64_atomic_ldop_supported_p): Remove.
2821 (aarch64_gen_atomic_ldop): Remove.
2822 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
2823 Fully expand LSE operations here.
2824 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
2825 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
2826 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
2827 and use ATOMIC_LDOP instead; use register_operand for the input;
2828 drop the split and emit insns directly.
2829 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
2830 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
2831 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
2833 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
2834 (aarch64_gen_atomic_ldop): Don't call it.
2835 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
2836 Use aarch64_reg_or_zero.
2837 (aarch64_atomic_exchange<ALLI>): Likewise.
2838 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
2839 operand 0; use aarch64_reg_or_zero for input; merge ...
2840 (@aarch64_atomic_swp<ALLI>): ... this and remove.
2842 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
2843 (aarch64_split_compare_and_swap): Use it.
2844 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
2845 test oldval against the proper predicate.
2846 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
2847 Use nonmemory_operand for expected.
2848 (cas_short_expected_pred): New.
2849 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
2850 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
2851 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
2852 (aarch64_plushi_operand): New.
2854 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
2855 Force oldval into the rval register for TARGET_LSE; emit the compare
2856 during initial expansion so that it may be deleted if unused.
2857 (aarch64_gen_atomic_cas): Remove.
2858 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
2859 Change =&r to +r for operand 0; use match_dup for operand 2;
2860 remove is_weak and mod_f operands as unused. Drop the split
2862 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
2863 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
2864 (@aarch64_atomic_cas<GPI>): Similarly.
2866 2018-10-31 Richard Biener <rguenther@suse.de>
2868 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
2871 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
2873 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
2874 * config/aarch64/aarch64.md: Include saphira.md
2875 * config/aarch64/saphira.md: New file for pipeline description.
2877 2018-10-30 Martin Sebor <msebor@redhat.com>
2880 * gimple-ssa-sprintf.c (format_directive): Use %G to include
2882 (sprintf_dom_walker::compute_format_length):
2883 Avoid setting POSUNDER4K here.
2884 (get_destination_size): Handle null argument values.
2885 (get_user_idx_format): New function.
2886 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
2887 functions, including user-defined with attribute format printf.
2888 Use %G to include inlining context.
2889 Set POSUNDER4K here.
2891 2018-10-30 Jan Hubicka <jh@suse.cz>
2893 * params.def (lto-partitions): Bump from 32 to 128.
2895 2018-10-30 Jan Hubicka <jh@suse.cz>
2898 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
2900 (free_lang_data_in_type): Forward declare.
2901 (fld_type_variant_equal_p): New function.
2902 (fld_type_variant): New function
2903 (fld_incomplete_types): New hash.
2904 (fld_incomplete_type_of): New function
2905 (fld_simplfied-type): New function.
2906 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
2907 (free_lang_data): Allocate and free fld_incomplete_type; update call
2908 of free_lang_data_in_decl.
2910 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
2912 * gcov.c (output_lines): Remove duplicate line.
2914 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
2916 * config/rs6000/rs6000.md (bswapdi2): Force address into register
2917 if not in indexed or indirect form.
2918 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
2919 (bswapdi2_store): Ditto.
2920 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
2922 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
2923 Prototype for helper function.
2925 2018-10-30 Martin Sebor <msebor@redhat.com>
2927 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
2928 (target, pragma GCC optimize, pragma GCC target): Ditto.
2930 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
2932 * doc/extend.texi: Fix prototype and description of
2933 __builtin_expect_with_probability.
2935 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2937 * cgraph.h (clone_function_name_1): Replaced by new
2938 clone_function_name_numbered that takes name as string; for
2939 privatize_symbol_name_1 use only.
2940 (clone_function_name): Renamed to
2941 clone_function_name_numbered to be explicit about numbering.
2942 (clone_function_name): New two-argument function that does
2943 not number its output.
2944 (clone_function_name): New three-argument function that
2945 takes a number to append to its output.
2946 * cgraphclones.c (duplicate_thunk_for_node):
2947 (clone_function_name_1): Renamed.
2948 (clone_function_name_numbered): Two new functions.
2949 (clone_function_name): Improved documentation.
2950 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
2951 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
2952 * final.c (final_scan_insn_1): Use the new clone_function_name
2954 * multiple_target.c (create_dispatcher_calls): Ditto.
2955 (create_target_clone): Ditto.
2956 * omp-expand.c (grid_expand_target_grid_body): Ditto.
2957 * omp-low.c (create_omp_child_function_name): Ditto.
2958 * omp-simd-clone.c (simd_clone_create): Ditto.
2959 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
2960 new clone_function_name without numbering.
2962 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
2964 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
2965 Assert that the allocation size is not zero.
2967 2018-10-30 Richard Biener <rguenther@suse.de>
2969 PR tree-optimization/87800
2970 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
2971 non-induction or reduction PHIs.
2973 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
2975 * config/aarch64/falkor-tag-collision-avoidance.c
2976 (execute_tag_collision_avoidance): Call df_note_add_problem.
2978 2018-10-30 Martin Liska <mliska@suse.cz>
2980 * doc/extend.texi: Fix typo in documentation
2981 of __builtin_expect_with_probability.
2983 2018-10-29 David Malcolm <dmalcolm@redhat.com>
2986 * input.c (get_substring_ranges_for_loc): Detect if
2987 linemap_resolve_location gives us a NULL map, and reject
2990 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
2992 * config.gcc (xstormy16-*-elf): Set tm_d_file.
2994 2018-10-29 David Malcolm <dmalcolm@redhat.com>
2995 Martin Sebor <msebor@redhat.com>
2996 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
2998 * doc/ux.texi (Quoting): New subsection, adapted from material at
2999 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
3000 MartinSebor and ManuelLopezIbanez.
3001 (Fix-it hints): Note that fix-it hints shouldn't be marked for
3004 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
3007 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
3010 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
3012 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
3014 2018-10-29 Paul Koning <ni1d@arrl.net>
3016 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
3018 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
3020 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
3021 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
3022 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
3023 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
3024 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
3025 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
3026 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
3027 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
3028 Change 'vector' to '__vector'.
3029 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
3030 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
3031 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
3032 _mm_avg_pu16): Likewise. And, whitespace corrections.
3034 2018-10-29 Richard Biener <rguenther@suse.de>
3036 PR tree-optimization/87785
3037 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
3040 2018-10-29 Olivier Hainque <hainque@adacore.com>
3042 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
3044 2018-10-29 Olivier Hainque <hainque@adacore.com>
3046 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
3047 ports configuration macro, defaults to "gnu".
3048 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
3049 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
3051 2018-10-29 Olivier Hainque <hainque@adacore.com>
3053 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
3055 2018-10-29 Richard Biener <rguenther@suse.de>
3057 PR tree-optimization/87790
3058 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
3059 (vect_make_slp_decision): Adjust.
3060 (vect_slp_analyze_bb_1): Likewise.
3061 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
3064 2018-10-29 Richard Biener <rguenther@suse.de>
3066 PR tree-optimization/87785
3067 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
3069 (vect_build_slp_tree): Likewise.
3070 (vect_gather_slp_loads): New function.
3071 (vect_analyze_slp_instance): Gather loads separately from the
3074 2018-10-29 Martin Liska <mliska@suse.cz>
3076 * Makefile.in: Make dependency to json.o.
3077 * doc/gcov.texi: Document new JSON format, remove
3078 old intermediate format documentation.
3079 * gcov.c (struct function_info): Come up with m_name and
3081 (function_info::function_info): Initialize it.
3082 (function_info::~function_info): Release it.
3083 (main): Rename flag_intermediate_format to flag_json_format.
3084 (print_usage): Describe --json-format.
3085 (process_args): Set flag_json_format.
3086 (output_intermediate_line): Remove.
3087 (output_intermediate_json_line): Likewise.
3088 (get_gcov_intermediate_filename): Return new extension
3090 (output_intermediate_file): Implement JSON emission.
3091 (output_json_intermediate_file): Implement JSON emission.
3092 (generate_results): Use ::get_name for function name.
3093 Handle JSON output file.
3094 (read_graph_file): Use ::get_name instead of cplus_demangle.
3095 (read_count_file): Likewise.
3096 (solve_flow_graph): Likewise.
3097 (add_line_counts): Likewise.
3098 (accumulate_line_counts): Use new flag_json_format.
3099 (output_function_details): Use ::get_name instead of cplus_demangle.
3100 (output_lines): Likewise.
3101 * json.cc (test_writing_literals): Add new tests.
3102 * json.h (class literal): Add new boolean constructor.
3104 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
3106 PR rtl-optimization/87701
3107 PR rtl-optimization/87780
3108 * combine.c (make_more_copies): Rewrite.
3110 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
3112 * doc/generic.texi (ABSU_EXPR): Document.
3113 * match.pd (absu(x)*absu(x) -> x*x): Handle.
3114 (absu(absu(X)) -> absu(X)): Likewise.
3115 (absu(-X) -> absu(X)): Likewise.
3116 (absu(X) where X is nonnegative -> X): Likewise.
3118 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
3120 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
3121 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
3122 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
3123 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
3124 (s-tm-texi): Also check timestamp on d-target.def.
3125 (generated_files): Add TM_D_H and d-target-hooks-def.h.
3126 (build/genhooks.o): Also depend on D_TARGET_DEF.
3127 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
3129 * config/aarch64/aarch64-d.c: New file.
3130 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
3132 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
3134 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
3135 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
3136 * config/arm/arm-d.c: New file.
3137 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
3138 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
3139 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3140 * config/arm/t-arm (arm-d.o): New rule.
3141 * config/default-d.c: New file.
3142 * config/glibc-d.c: New file.
3143 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3144 * config/i386/i386-d.c: New file.
3145 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
3146 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
3147 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3148 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
3149 * config/i386/t-i386 (i386-d.o): New rule.
3150 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3151 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3152 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
3153 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3154 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
3155 * config/mips/mips-d.c: New file.
3156 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
3157 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
3158 * config/mips/t-mips (mips-d.o): New rule.
3159 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3160 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3161 * config/powerpcspe/powerpcspe-d.c: New file.
3162 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
3164 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3165 Support GNU D by using 0 as the language type.
3166 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
3167 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
3168 * config/riscv/riscv-d.c: New file.
3169 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
3171 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
3172 * config/riscv/t-riscv (riscv-d.o): New rule.
3173 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3174 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
3175 * config/rs6000/rs6000-d.c: New file.
3176 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
3178 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3179 Support GNU D by using 0 as the language type.
3180 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
3181 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
3182 * config/s390/s390-d.c: New file.
3183 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
3184 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
3185 * config/s390/t-s390 (s390-d.o): New rule.
3186 * config/sparc/sparc-d.c: New file.
3187 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
3189 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
3190 * config/sparc/t-sparc (sparc-d.o): New rule.
3191 * config/t-glibc (glibc-d.o): New rule.
3192 * configure: Regenerated.
3193 * configure.ac (tm_d_file): New variable.
3194 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
3195 * doc/contrib.texi (Contributors): Add self for the D frontend.
3196 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
3197 * doc/install.texi (Configuration): Mention libphobos as an option for
3198 --enable-shared. Mention d as an option for --enable-languages.
3199 (Testing): Mention check-d as a target.
3200 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
3201 name suffixes. Mention d as a -x option.
3202 * doc/sourcebuild.texi (Top Level): Mention libphobos.
3203 * doc/standards.texi (Standards): Add section on D language.
3204 * doc/tm.texi: Regenerated.
3205 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
3206 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
3207 * dwarf2out.c (is_dlang): New function.
3208 (gen_compile_unit_die): Use DW_LANG_D for D.
3209 (declare_in_namespace): Return module die for D, instead of adding
3210 extra declarations into the namespace.
3211 (gen_namespace_die): Generate DW_TAG_module for D.
3212 (gen_decl_die): Handle CONST_DECLSs for D.
3213 (dwarf2out_decl): Likewise.
3214 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
3215 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
3216 kinds of aggregates.
3217 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
3218 * genhooks.c: Include d/d-target.def.
3220 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
3223 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
3224 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
3225 (STACK_DYNAMIC_OFFSET): Likewise.
3227 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
3230 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
3233 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
3235 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
3236 return type and other typos.
3238 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
3240 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
3241 a shorter sequence with fewer branches.
3242 (emit_final_str_compare_gpr): Ditto.
3244 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3246 * config/rs6000/tmmintrin.h: New file.
3247 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
3249 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3251 * config/rs6000/mmintrin.h: Enable 32bit compilation.
3252 * config/rs6000/xmmintrin.h: Likewise.
3254 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
3256 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
3258 2018-10-26 Richard Biener <rguenther@suse.de>
3260 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
3262 (vect_mark_slp_stmts_relevant): Likewise.
3263 (vect_detect_hybrid_slp_stmts): Likewise.
3264 (vect_bb_slp_scalar_cost): Likewise.
3265 (vect_remove_slp_scalar_calls): Likewise.
3267 2018-10-26 Jan Hubicka <jh@suse.cz>
3269 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
3270 (warn_types_mismatch): Fix walk of DECL_NAME.
3271 (odr_types_equivalent_p): Fix overactive assert.
3273 2018-10-26 Richard Biener <rguenther@suse.de>
3275 PR tree-optimization/87105
3276 * tree-vectorizer.h (_slp_tree::refcnt): New member.
3277 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
3279 (vect_create_new_slp_node): Initialize refcnt to one.
3281 (scalar_stmts_set_t, bst_fail): Remove.
3282 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
3283 (vect_build_slp_tree): Add bst_map argument and lookup
3284 already created SLP nodes.
3285 (vect_print_slp_tree): Handle a SLP graph, print SLP node
3287 (vect_slp_rearrange_stmts): Handle a SLP graph.
3288 (vect_analyze_slp_instance): Adjust and free SLP nodes from
3289 the CSE map. Fix indenting.
3290 (vect_schedule_slp_instance): Add short-cut.
3292 2018-10-26 Martin Liska <mliska@suse.cz>
3295 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
3296 addr_expr and not with pointers.
3298 2018-10-26 Jan Hubicka <jh@suse.cz>
3300 * tree.c (free_lang_data_in_type): Only check main variants.
3301 * ipa-devirt.c (warn_odr): Make static.
3302 (types_same_for_odr): Drop strict variant.
3303 (types_odr_comparable): Likewise.
3304 (odr_or_derived_type_p): Look for main variants.
3305 (odr_name_hasher::equal): Cleanup comment.
3306 (odr_subtypes_equivalent): Add warn and warned arguments; check main
3308 (type_variants_equivalent_p): break out from ...
3309 (odr_types_equivalent): ... here; go for main variants where needed.
3310 (warn_odr): ... here; turn static.
3311 (warn_types_mismatch): Compare mangled names of main variants.
3312 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
3313 (type_with_linkage_p): Sanity check that we look at main variant.
3314 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
3315 * tree.h (types_same_for_odr): Drop strict argument.
3317 2018-10-26 Richard Biener <rguenther@suse.de>
3319 PR tree-optimization/87746
3320 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
3321 Simplify and fix WRT strided store groups with size not
3322 equal to step in element count.
3323 (vect_analyze_group_access_1): Dump the whole group.
3325 2018-10-25 Carl Love <cel@us.ibm.com>
3327 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
3328 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
3329 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
3330 precicion entry for each overloaded builtin.
3331 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
3332 VSCEDPUO): Rename overloaded name.
3333 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
3334 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
3335 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
3336 define_expand for xscmpexqp instruction.
3337 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
3339 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3340 Jinsong Ji <jji@us.ibm.com>
3342 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
3343 function with vec_sl.
3344 (_mm_slli_epi32): Likewise.
3345 (_mm_slli_epi64): Likewise.
3346 (_mm_srai_epi16): Replace deprecated function with vec_sra.
3347 (_mm_srai_epi32): Likewise.
3348 (_mm_srli_epi16): Replace deprecated function with vec_sr.
3349 (_mm_srli_epi32): Likewise.
3350 (_mm_srli_epi64): Likewise.
3351 (_mm_sll_epi16): Replace deprecated function with vec_sl.
3352 (_mm_sll_epi32): Likewise.
3353 (_mm_sll_epi64): Likewise.
3354 (_mm_sra_epi16): Replace deprecated function with vec_sra.
3355 (_mm_sra_epi32): Likewise.
3356 (_mm_srl_epi16): Replace deprecated function with vec_sr.
3357 (_mm_srl_epi32): Likewise.
3358 (_mm_srl_epi64): Likewise.
3360 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3361 Jinsong Ji <jji@us.ibm.com>
3363 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
3364 comparison operators with vec_cmp* for compatibility due to
3365 unfortunate history; clean up formatting and use types more
3367 (_mm_sll_epi32): Likewise.
3368 (_mm_sll_epi64): Likewise.
3369 (_mm_srl_epi16): Likewise.
3370 (_mm_srl_epi32): Likewise.
3371 (_mm_srl_epi64): Likewise.
3373 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
3374 Jinsong Ji <jji@us.ibm.com>
3376 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
3377 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
3378 __vector __bool int. Use vec_cmpgt in preference to deprecated
3379 function vec_vcmpgtfp.
3380 (_mm_max_ps): Likewise.
3382 2018-10-25 Jeff Law <law@redhat.com>
3384 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
3387 2018-10-25 Martin Sebor <msebor@redhat.com>
3389 * doc/extend.texi (aligned): Expand attribute description.
3390 (Alignment): Rename section. Discuss function arguments.
3392 2018-10-25 Jan Hubicka <jh@suse.cz>
3394 * ipa-devirt.c (main_odr_variant): Remove.
3395 (hash_odr_name, types_same_for_odr, types_odr_comparable,
3396 odr_name_hasher::equal, odr_subtypes_equivalent_p):
3397 Drop use of main_odr_variant.
3398 (add_type_duplicate): Silence confused warnings on integer types.
3399 (get_odr_type): Always look for main variant.
3400 (register_odr_type): Simplify.
3402 2018-10-25 Richard Biener <rguenther@suse.de>
3404 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
3405 Initialize ng to silence error with release checking bootstrap.
3407 2018-10-25 Richard Biener <rguenther@suse.de>
3409 * tree-if-conv.c: Include tree-ssa-sccvn.h.
3410 (tree_if_conversion): Run CSE on the if-converted loop body.
3412 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
3414 * config/s390/constraints.md (ZL): New constraint.
3415 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
3417 * config/s390/s390.md (movdi_larl): Remove.
3418 (movdi_64): Add the LARL alternative.
3420 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
3423 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
3424 (RTX_CODE_HWINT_P): New macro.
3425 (rtx_code_size): Use RTX_CODE_HWINT_P ().
3427 2018-10-25 Jan Hubicka <jh@suse.cz>
3429 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
3432 2018-10-25 Richard Biener <rguenther@suse.de>
3434 PR tree-optimization/87665
3435 PR tree-optimization/87745
3436 * tree-vectorizer.h (get_earlier_stmt): Remove.
3437 (get_later_stmt): Pick up UID from the original non-pattern stmt.
3439 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
3441 * options.texi (Deprecated): Move list to Var section.
3443 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
3444 Jinsong Ji <jji@us.ibm.com>
3446 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
3447 __vector long to __vector long long.
3448 (_mm_cvtpd_ps): Likewise.
3449 (_mm_cvttpd_epi32): Likewise.
3450 (_mm_cvtpi32_pd): Likewise.
3451 (_mm_unpackhi_epi64): Likewise.
3452 (_mm_unpacklo_epi64): Likewise.
3454 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
3456 PR rtl-optimization/87720
3457 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
3459 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
3461 * gimple-ssa-isolate-paths.c
3462 (find_implicit_erroneous_behavior): Do not change code if the
3463 pass is running for warnings only.
3464 (find_explicit_erroneous_behavior): Likewise.
3466 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
3468 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
3469 Define as rs6000_mangle_decl_assembler_name.
3470 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
3471 long double to IEEE long double, switch the names of the long
3472 double built-in functions to be <func>f128 instead of <func>l.
3474 2018-10-24 Martin Sebor <msebor@redhat.com>
3476 * doc/extend.texi (nonnull): List no-argument form. Reference
3477 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
3479 2018-10-24 Richard Biener <rguenther@suse.de>
3481 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
3483 2018-10-24 Martin Liska <mliska@suse.cz>
3485 PR tree-optimization/84436
3486 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
3488 (switch_conversion::contains_linear_function_p): New.
3489 (switch_conversion::build_one_array): Support linear
3490 transformation on input.
3491 * tree-switch-conversion.h (struct switch_conversion): Add
3492 contains_linear_function_p declaration.
3494 2018-10-24 Richard Biener <rguenther@suse.de>
3496 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
3497 if its argument is CONSTANT_CLASS_P.
3499 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
3501 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
3502 it is wrong for forward declarations.
3504 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
3506 * config/s390/s390.c (s390_check_qrst_address): Add the missing
3507 SYMBOL_REF_P () check.
3509 2018-10-24 Richard Biener <rguenther@suse.de>
3511 PR tree-optimization/87105
3512 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
3513 dump classification.
3514 (vect_analyze_data_ref_accesses): Handle duplicate loads and
3515 stores by splitting the affected group after the fact.
3516 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
3517 fail the SLP build because of size constraints.
3519 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3521 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
3522 * configure: Regenerate.
3523 * config.in: Regenerate.
3524 * varasm.c (mergeable_string_section): Use readonly_data_section
3525 if linker doesn't support SHF_MERGE with alignment > 8.
3526 (mergeable_constant_section): Likewise.
3528 2018-10-24 Richard Biener <rguenther@suse.de>
3530 PR tree-optimization/84013
3531 * tree-ssa-structalias.c (struct msdi_data): New struct for
3532 marshalling data to walk_stmt_load_store_ops.
3533 (maybe_set_dependence_info): Refactor as callback for
3534 walk_stmt_load_store_ops.
3535 (compute_dependence_clique): Set restrict info on all stmt kinds.
3537 2018-10-24 Martin Liska <mliska@suse.cz>
3539 * cgraph.c (cgraph_node::dump):
3540 Remove reduntant dumps and make tp_first_run dump more compact.
3542 2018-10-24 Richard Biener <rguenther@suse.de>
3544 PR tree-optimization/87665
3545 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
3548 2018-10-12 Jeff Law <law@redhat.com>
3550 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
3553 2018-10-23 Richard Biener <rguenther@suse.de>
3555 * tree-vrp.c (add_assert_info): Guard dump_printf with
3557 * gimple-ssa-evrp-analyze.c
3558 (evrp_range_analyzer::record_ranges_from_incoming_edge):
3559 Use value_range::ignore_equivs_equal_p.
3561 2018-10-23 Richard Biener <rguenther@suse.de>
3563 PR tree-optimization/87105
3564 PR tree-optimization/87608
3565 * passes.def (pass_all_early_optimizations): Add early phi-opt
3567 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
3568 addition to debug stmts.
3569 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
3570 and abs replacement early.
3571 * tree-cfg.c (gimple_empty_block_p): Likewise.
3573 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
3576 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
3577 specified to configure.
3578 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
3580 2018-10-23 Richard Biener <rguenther@suse.de>
3582 PR tree-optimization/87700
3583 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
3585 2018-10-23 Jakub Jelinek <jakub@redhat.com>
3588 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
3589 second argument from __mmask16 to __mmask8.
3590 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
3591 _mm_mask_packs_epi32): Likewise.
3592 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
3594 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
3596 2018-10-23 Richard Biener <rguenther@suse.de>
3598 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
3600 2018-10-23 Richard Biener <rguenther@suse.de>
3602 PR tree-optimization/86144
3603 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
3604 over simd attribute.
3606 2018-10-23 Richard Biener <rguenther@suse.de>
3608 PR tree-optimization/87693
3609 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
3610 the case we do not find the taken edge.
3612 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
3613 Jinsong Ji <jji@us.ibm.com>
3615 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
3616 (_mm_store_pd): Use unaligned vector type for pointer cast.
3617 (_mm_maskmoveu_si128): Likewise.
3618 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
3619 (_mm_store_ps): Use unaligned vector type for pointer cast.
3621 2018-10-22 Paul Koning <ni1d@arrl.net>
3623 * symtab.c (symtab_node::increase_alignment): Correct max
3626 2018-10-22 Yury Gribov <tetra2005@gmail.com>
3628 PR tree-optimization/87633
3629 * match.pd: Do not generate unordered integer comparisons.
3631 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
3633 PR rtl-optimization/87600
3634 * combine.c: Add include of expr.h.
3635 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
3636 register to a pseudo.
3637 (make_more_copies): New function, add a copy to a new pseudo after
3638 the moves from hard registers into pseudos.
3639 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
3640 later. Call make_more_copies.
3642 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
3644 * lra-constraints.c (process_alt_operands): New local array,
3645 matching_early_clobber. Check matching_early_clobber before
3646 decrementing reject, and set matching_early_clobber after.
3648 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
3651 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
3652 call output_addr_const and hope for the best.
3654 2018-10-22 Richard Biener <rguenther@suse.de>
3656 * gimple-ssa-evrp-analyze.c
3657 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
3658 smarter about what ranges to use.
3659 * tree-vrp.c (add_assert_info): Dump here.
3660 (register_edge_assert_for_2): Instead of here at multiple but
3663 * gcc.dg/tree-ssa/evrp12.c: New testcase.
3664 * gcc.dg/predict-6.c: Adjust.
3665 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
3666 * gcc.dg/tree-ssa/vrp02.c: Likewise.
3667 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
3669 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
3670 Richard Biener <rguenther@suse.de>
3672 * bitmap.h: Update data structure documentation, including a
3673 description of bitmap views as either linked-lists or splay trees.
3674 (struct bitmap_element_def): Update comments for splay tree bitmaps.
3675 (struct bitmap_head_def): Likewise.
3676 (bitmap_list_view, bitmap_tree_view): New prototypes.
3677 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
3679 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
3680 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
3681 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
3682 released bitmap element here.
3683 (bitmap_element_free): Remove.
3684 (bitmap_elt_clear_from): Work on splay tree bitmaps.
3685 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
3686 this function similar ones such that linked-list bitmap implementation
3687 functions are grouped.
3688 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
3689 and moved for grouping.
3690 (bitmap_list_insert_element_after): Renamed from
3691 bitmap_elt_insert_after, and moved for grouping.
3692 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
3693 (bitmap_tree_link_left, bitmap_tree_link_right,
3694 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
3695 bitmap_tree_link_element, bitmap_tree_unlink_element,
3696 bitmap_tree_find_element): New functions for splay-tree bitmap
3698 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
3699 Renamed and moved, see above entries.
3700 (bitmap_tree_listify_from): New function to convert part of a splay
3701 tree bitmap to a linked-list bitmap.
3702 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
3703 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
3704 (bitmap_find_bit): Remove.
3705 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
3706 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
3707 Handle splay tree bitmaps.
3708 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
3709 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
3710 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
3711 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
3712 bitmap_intersect_compl_p, bitmap_ior_and_compl,
3713 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
3714 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
3715 corresponding changes to use linked-list specific bitmap_element
3716 manipulation functions as applicable for efficiency.
3717 (bitmap_tree_to_vec): New function.
3718 (debug_bitmap_elt_file): New function split out from ...
3719 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
3720 (bitmap_print): Likewise.
3722 PR tree-optimization/63155
3723 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
3725 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
3728 2018-10-22 Martin Liska <mliska@suse.cz>
3730 PR tree-optimization/87686
3732 2018-08-29 Martin Liska <mliska@suse.cz>
3734 * tree-switch-conversion.c (switch_conversion::expand):
3735 Strenghten assumption about gswitch statements.
3737 2018-10-22 Martin Liska <mliska@suse.cz>
3739 * ipa-icf.c (sem_item::compare_attributes): Remove.
3740 (sem_item::compare_referenced_symbol_properties): Use
3741 attribute_list_equal instead.
3742 (sem_function::equals_wpa): Likewise.
3743 * ipa-icf.h: Remove compare_attributes.
3745 2018-10-22 Richard Biener <rguenther@suse.de>
3748 * mem-stats.h (mem_usage::operator==): Fix pasto.
3750 2018-10-22 Richard Biener <rguenther@suse.de>
3752 PR tree-optimization/87640
3753 * tree-vrp.c (set_value_range_with_overflow): Decompose
3755 (extract_range_from_binary_expr_1): Adjust.
3757 2018-10-22 Martin Jambor <mjambor@suse.cz>
3759 * tree-eh.h (stmt_could_throw_p): Add function parameter.
3760 (stmt_can_throw_external): Likewise.
3761 (stmt_can_throw_internal): Likewise.
3762 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
3763 (lower_eh_constructs_2): Likewise.
3764 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
3765 (stmt_can_throw_external): Likewise.
3766 (stmt_can_throw_internal): Likewise.
3767 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
3768 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
3769 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
3770 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
3771 (pass_lower_eh_dispatch::execute): Pass cfun to
3772 stmt_can_throw_external.
3773 (cleanup_empty_eh): Likewise.
3774 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
3775 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
3776 stmt_can_throw_external instead of pushing it to cfun.
3777 (symbol_table::create_edge): Likewise.
3778 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
3779 stmt_can_throw_internal.
3780 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
3781 to stmt_could_throw_p.
3782 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
3783 stmt_can_throw_internal.
3784 (pass_store_merging::execute): Likewise.
3785 * gimple-ssa-strength-reduction.c
3786 (find_candidates_dom_walker::before_dom_children): Pass cfun to
3788 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
3789 stmt_can_throw_internal.
3790 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
3791 to stmt_can_throw_external.
3792 (check_stmt): Pass cfun to stmt_could_throw_p.
3793 (check_stmt): Pass cfun to stmt_can_throw_external.
3794 (pass_nothrow::execute): Likewise.
3795 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
3796 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
3797 stmt_can_throw_internal.
3798 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
3799 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
3800 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
3801 * tree-complex.c (expand_complex_libcall): Pass cfun to
3802 stmt_could_throw_p and to stmt_can_throw_internal.
3803 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
3804 * tree-inline.c (copy_edges_for_bb): Likewise.
3805 (maybe_move_debug_stmts_to_successors): Likewise.
3806 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
3808 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
3809 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
3810 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
3811 stmt_can_throw_internal.
3812 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
3813 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
3815 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
3816 stmt_can_throw_internal.
3817 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
3818 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
3820 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
3821 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
3822 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3823 (convert_mult_to_fma_1): Likewise.
3824 (convert_to_divmod): Likewise.
3825 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
3826 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
3827 * tree-ssa-propagate.c
3828 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
3829 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
3830 (maybe_optimize_range_tests): Likewise.
3831 (linearize_expr_tree): Likewise.
3832 (reassociate_bb): Likewise.
3833 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
3834 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
3835 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
3836 (handle_char_store): Likewise.
3837 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
3838 stmt_can_throw_internal.
3839 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
3841 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
3842 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
3843 (vectorizable_simd_clone_call): Likewise.
3844 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
3845 (gimple_stringop_fixed_value): Likewise.
3847 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
3849 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
3850 literal pool references.
3851 (s390_check_qrst_address): Adapt to the new behavior of
3852 s390_loadrelative_operand_p ().
3854 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3857 * config/i386/sse.md (*andnot<mode>3_bcst): New.
3859 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3862 * config/i386/sse.md (*<code><mode>3_bcst): New.
3864 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
3867 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
3868 V4DI, V16SI and V8DI.
3869 (*sub<mode>3<mask_name>_bcst): New.
3870 (*add<mode>3<mask_name>_bcst): Likewise.
3872 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
3873 Jinsong Ji <jji@us.ibm.com>
3875 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
3876 __m64 with __vector unsigned long long for compatibility.
3877 (_mm_movemask_epi8): Likewise.
3878 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
3879 (_mm_cvttps_pi32): Likewise.
3880 (_mm_cvtpi32_ps): Likewise.
3881 (_mm_cvtps_pi16): Likewise.
3882 (_mm_loadh_pi): Likewise.
3883 (_mm_storeh_pi): Likewise.
3884 (_mm_movehl_ps): Likewise.
3885 (_mm_movelh_ps): Likewise.
3886 (_mm_loadl_pi): Likewise.
3887 (_mm_storel_pi): Likewise.
3888 (_mm_movemask_ps): Likewise.
3889 (_mm_shuffle_pi16): Likewise.
3891 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3894 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
3895 __builtin_ia32_vfnmsubpd512_mask.
3896 (_mm512_mask_fnmsub_round_pd): Likewise.
3897 (_mm512_fnmsub_pd): Likewise.
3898 (_mm512_mask_fnmsub_pd): Likewise.
3899 (_mm512_maskz_fnmsub_round_pd): Use
3900 __builtin_ia32_vfnmsubpd512_maskz.
3901 (_mm512_maskz_fnmsub_pd): Likewise.
3902 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
3903 (_mm512_mask_fnmsub_round_ps): Likewise.
3904 (_mm512_fnmsub_ps): Likewise.
3905 (_mm512_mask_fnmsub_ps): Likewise.
3906 (_mm512_maskz_fnmsub_round_ps): Use
3907 __builtin_ia32_vfnmsubps512_maskz.
3908 (_mm512_maskz_fnmsub_ps): Likewise.
3909 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
3910 __builtin_ia32_vfnmsubpd256_mask.
3911 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
3912 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
3913 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
3914 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
3915 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
3916 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
3917 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
3918 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
3919 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
3920 __builtin_ia32_vfnmsubpd.
3921 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
3922 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
3923 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
3924 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
3925 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
3926 * config/i386/i386-builtin.def: Add
3927 __builtin_ia32_vfnmsubpd256_mask,
3928 __builtin_ia32_vfnmsubpd256_maskz,
3929 __builtin_ia32_vfnmsubpd128_mask,
3930 __builtin_ia32_vfnmsubpd128_maskz,
3931 __builtin_ia32_vfnmsubps256_mask,
3932 __builtin_ia32_vfnmsubps256_maskz,
3933 __builtin_ia32_vfnmsubps128_mask,
3934 __builtin_ia32_vfnmsubps128_maskz,
3935 __builtin_ia32_vfnmsubpd512_mask,
3936 __builtin_ia32_vfnmsubpd512_maskz,
3937 __builtin_ia32_vfnmsubps512_mask,
3938 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
3939 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
3940 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
3941 __builtin_ia32_vfnmsubpd256.
3942 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
3943 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
3944 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
3946 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
3948 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
3950 (fmai_vmfnmsub_<mode><round_name>): Likewise.
3952 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3955 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
3956 __builtin_ia32_vfnmaddpd512_mask.
3957 (_mm512_mask_fnmadd_round_pd): Likewise.
3958 (_mm512_fnmadd_pd): Likewise.
3959 (_mm512_mask_fnmadd_pd): Likewise.
3960 (_mm512_maskz_fnmadd_round_pd): Use
3961 __builtin_ia32_vfnmaddpd512_maskz.
3962 (_mm512_maskz_fnmadd_pd): Likewise.
3963 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
3964 (_mm512_mask_fnmadd_round_ps): Likewise.
3965 (_mm512_fnmadd_ps): Likewise.
3966 (_mm512_mask_fnmadd_ps): Likewise.
3967 (_mm512_maskz_fnmadd_round_ps): Use
3968 __builtin_ia32_vfnmaddps512_maskz.
3969 (_mm512_maskz_fnmadd_ps): Likewise.
3970 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
3971 __builtin_ia32_vfnmaddpd256_mask.
3972 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
3973 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
3974 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
3975 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
3976 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
3977 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
3978 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
3979 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
3980 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
3981 __builtin_ia32_vfnmaddpd.
3982 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
3983 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
3984 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
3985 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
3986 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
3987 * config/i386/i386-builtin.def: Add
3988 __builtin_ia32_vfnmaddpd256_mask,
3989 __builtin_ia32_vfnmaddpd256_maskz,
3990 __builtin_ia32_vfnmaddpd128_mask,
3991 __builtin_ia32_vfnmaddpd128_maskz,
3992 __builtin_ia32_vfnmaddps256_mask,
3993 __builtin_ia32_vfnmaddps256_maskz,
3994 __builtin_ia32_vfnmaddps128_mask,
3995 __builtin_ia32_vfnmaddps128_maskz,
3996 __builtin_ia32_vfnmaddpd512_mask,
3997 __builtin_ia32_vfnmaddpd512_maskz,
3998 __builtin_ia32_vfnmaddps512_mask,
3999 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
4000 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
4001 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
4002 __builtin_ia32_vfnmaddpd256.
4003 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
4004 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
4005 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
4007 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
4009 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
4011 (fmai_vmfnmadd_<mode><round_name>): Likewise.
4013 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4016 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
4017 __builtin_ia32_vfmsubpd512_mask.
4018 (_mm512_mask_fmsub_round_pd): Likewise.
4019 (_mm512_fmsub_pd): Likewise.
4020 (_mm512_mask_fmsub_pd): Likewise.
4021 (_mm512_maskz_fmsub_round_pd): Use
4022 __builtin_ia32_vfmsubpd512_maskz.
4023 (_mm512_maskz_fmsub_pd): Likewise.
4024 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
4025 (_mm512_mask_fmsub_round_ps): Likewise.
4026 (_mm512_fmsub_ps): Likewise.
4027 (_mm512_mask_fmsub_ps): Likewise.
4028 (_mm512_maskz_fmsub_round_ps): Use
4029 __builtin_ia32_vfmsubps512_maskz.
4030 (_mm512_maskz_fmsub_ps): Likewise.
4031 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
4032 __builtin_ia32_vfmsubpd256_mask.
4033 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
4034 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
4035 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
4036 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
4037 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
4038 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
4039 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
4040 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
4041 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
4042 __builtin_ia32_vfmsubpd.
4043 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
4044 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
4045 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
4046 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
4047 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
4048 * config/i386/i386-builtin.def: Add
4049 __builtin_ia32_vfmsubpd256_mask,
4050 __builtin_ia32_vfmsubpd256_maskz,
4051 __builtin_ia32_vfmsubpd128_mask,
4052 __builtin_ia32_vfmsubpd128_maskz,
4053 __builtin_ia32_vfmsubps256_mask,
4054 __builtin_ia32_vfmsubps256_maskz,
4055 __builtin_ia32_vfmsubps128_mask,
4056 __builtin_ia32_vfmsubps128_maskz,
4057 __builtin_ia32_vfmsubpd512_mask,
4058 __builtin_ia32_vfmsubpd512_maskz,
4059 __builtin_ia32_vfmsubps512_mask,
4060 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
4061 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
4062 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
4063 __builtin_ia32_vfmsubpd256.
4064 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
4065 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
4066 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
4068 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
4070 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
4072 (fmai_vmfmsub_<mode><round_name>): Likewise.
4074 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4076 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
4077 Remove plus. Renamed to ...
4078 (*sub<mode>3<mask_name>_bcst): This.
4079 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
4080 (*add<mode>3<mask_name>_bcst): This.
4082 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4085 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
4087 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
4090 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
4091 (_mm_or_epi32): Likewise.
4092 (_mm256_xor_epi32): Likewise.
4093 (_mm_xor_epi32): Likewise.
4094 (_mm256_or_epi64): Likewise.
4095 (_mm_or_epi64): Likewise.
4096 (_mm256_xor_epi64): Likewise.
4097 (_mm_xor_epi64): Likewise.
4099 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
4102 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
4104 2018-10-20 Jakub Jelinek <jakub@redhat.com>
4107 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
4109 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
4111 * doc/ux.texi: Move @section directly after @node.
4113 2018-10-19 Jakub Jelinek <jakub@redhat.com>
4117 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
4118 depend closely nested inside of loop with ordered clause with
4121 2018-10-19 David Malcolm <dmalcolm@redhat.com>
4123 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
4124 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
4125 * doc/ux.texi: New file.
4127 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4129 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
4130 be the first CR field allocated.
4132 2018-10-19 Richard Biener <rguenther@suse.de>
4135 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
4136 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
4138 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
4141 * config/i386/sse.md
4142 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
4143 (*add<mode>3<mask_name>_bcst_2): Likewise.
4145 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
4147 * config/i386/sse.md
4148 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
4149 Replace nonimmediate_operand with register_operand.
4150 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
4152 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
4155 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
4157 PR rtl-optimization/87596
4158 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
4159 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
4160 for instructions in FROM..TO range.
4162 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
4164 * cfgexpand.c (expand_one_var): Use specific wording in error message
4165 for non-local frame variables.
4166 * stor-layout.c (layout_decl): Do not issue a warning for them.
4168 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
4170 * haifa-sched.c (priority): Add force_recompute parameter.
4171 (apply_replacement): Call priority () with force_recompute = true.
4172 (restore_pattern): Likewise.
4174 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4176 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
4177 HOST_BITS_PER_WIDE_INT.
4178 (test_vector_ops_duplicate): Likewise.
4180 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4183 * config/i386/sse.md (VF_AVX512): New.
4184 (avx512bcst): Likewise.
4185 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
4187 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
4189 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
4192 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
4194 * doc/invoke.texi (-dumpversion): Improve grammar.
4195 (-dumpfullversion): Make more consistent with -dumpversion.
4197 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
4199 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
4200 Set JUMP_LABEL to the jump insn.
4201 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
4202 Predict emitted jump and add label to jump insn.
4204 2018-10-18 David Malcolm <dmalcolm@redhat.com>
4206 PR tree-optimization/87562
4207 * input.c (get_substring_ranges_for_loc): Use
4208 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
4209 getting the linemap for the endpoint. Verify that it's either
4210 in the same linemap as the start point's spelling location, or
4211 at least in the same file.
4213 2018-10-18 Richard Biener <rguenther@suse.de>
4215 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
4216 feed width-specific load/store costs through ix86_vec_cost.
4217 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
4218 (k8_cost): Likewise.
4219 (bdver_cost): Likewise.
4220 (znver1_cost): Likewise.
4221 (btver1_cost): Likewise.
4222 (btver2_cost): Likewise.
4224 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4226 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
4227 to simplify subreg of vec_merge.
4229 2018-10-18 Richard Biener <rguenther@suse.de>
4231 * config/i386/i386.c: Fix costing of vector FMA.
4233 2018-10-18 Richard Biener <rguenther@suse.de>
4235 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
4237 (ix86_builtin_vectorization_cost): For vec_construct properly
4238 cost insertion into SSE regs.
4239 (...): Adjust calls to ix86_vec_cost.
4241 2018-10-18 Richard Biener <rguenther@suse.de>
4245 2018-02-07 Richard Biener <rguenther@suse.de>
4247 PR tree-optimization/84204
4248 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
4251 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
4254 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
4256 (test_vector_ops_duplicate): Add test for a scalar subreg of a
4257 VEC_MERGE of a VEC_DUPLICATE.
4259 2018-10-17 Joseph Myers <joseph@codesourcery.com>
4261 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
4262 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
4263 * doc/standards.texi (C Language): Document C2X.
4264 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
4265 (rl78_option_override): Handle "GNU C2X" language name.
4267 2018-10-17 Joseph Myers <joseph@codesourcery.com>
4269 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
4270 Document C17 as published in 2018.
4272 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
4275 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
4276 bail out if both sides do not have the same storage order.
4278 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4280 * bitmap.c (bitmap_head::dump): New.
4281 * bitmap.h (bitmap_head): Add dump().
4282 * gimple-ssa-evrp-analyze.c
4283 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
4284 (evrp_range_analyzer::set_ssa_range_info): Same.
4285 (evrp_range_analyzer::record_ranges_from_phis): Same.
4286 (evrp_range_analyzer::record_ranges_from_stmt): Same.
4287 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
4288 * gimple-ssa-sprintf.c (get_int_range): Same.
4289 (format_integer): Same.
4290 (sprintf_dom_walker::handle_gimple_call): Same.
4291 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
4292 (ipcp_vr_lattice::top_p): Same.
4293 (ipcp_vr_lattice::bottom_p): Same.
4294 (ipcp_vr_lattice::set_to_bottom): Same.
4295 (ipa_vr_operation_and_type_effects): Same.
4296 (propagate_vr_across_jump_function): Same.
4297 (ipcp_store_vr_results): Same.
4298 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
4299 (ipa_print_node_jump_functions_for_edge): Same.
4300 (ipa_get_value_range): Same.
4301 (ipa_compute_jump_functions_for_edge): Same.
4302 (ipa_write_jump_function): Same.
4303 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
4304 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
4306 * vr-values.c (set_value_range_to_nonnegative): Same.
4307 (set_value_range_to_truthvalue): Same.
4308 (vr_values::get_value_range): Same.
4309 (vr_values::set_defs_to_varying): Same.
4310 (vr_values::update_value_range): Same.
4311 (symbolic_range_based_on_p): Same.
4312 (vr_values::op_with_boolean_value_range_p): Same.
4313 (vr_values::extract_range_for_var_from_comparison_expr): Same.
4314 (vr_values::extract_range_from_ssa_name): Same.
4315 (vr_values::extract_range_from_binary_expr): Same.
4316 (vr_values::extract_range_from_unary_expr): Same.
4317 (vr_values::extract_range_from_cond_expr): Same.
4318 (vr_values::extract_range_from_comparison): Same.
4319 (vr_values::check_for_binary_op_overflow): Same.
4320 (vr_values::extract_range_basic): Same.
4321 (vr_values::extract_range_from_assignment): Same.
4322 (compare_ranges): Same.
4323 (compare_range_with_value): Same.
4324 (vr_values::adjust_range_with_scev): Same.
4325 (vrp_valueize): Same.
4326 (vrp_valueize_1): Same.
4327 (vr_values::get_vr_for_comparison): Same.
4328 (vr_values::compare_name_with_value): Same.
4329 (vr_values::compare_names): Same.
4330 (vr_values::vrp_evaluate_conditional): Same.
4331 (find_case_label_ranges): Same.
4332 (vr_values::vrp_visit_switch_stmt): Same.
4333 (vr_values::extract_range_from_phi_node): Same.
4334 (vr_values::simplify_div_or_mod_using_ranges): Same.
4335 (vr_values::simplify_bit_ops_using_ranges): Same.
4336 (test_for_singularity): Same.
4337 (range_fits_type_p): Same.
4338 (vr_values::simplify_cond_using_ranges_1): Same.
4339 (vr_values::simplify_switch_using_ranges): Same.
4340 (vr_values::simplify_float_conversion_using_ranges): Same.
4341 (vr_values::two_valued_val_range_p): Same.
4342 (vr_values::add_equivalence): Move to value_range::equiv_add.
4343 * vr-values.h (vr_values::add_equivalence): Remove.
4344 (VR_INITIALIZER): Remove.
4345 * tree-vrp.c (value_range::set): New.
4346 (value_range::equiv_add): New.
4347 (value_range::value_range): New.
4348 (value_range::deep_copy): New.
4349 (value_range::check): New.
4350 (value_range::equal_p): New.
4351 (value_range::ignore_equivs_equal_p): New.
4352 (value_range::operator==): New.
4353 (value_range::operator!=): New.
4354 (value_range::symbolic_p): New.
4355 (value_range::numeric_p): New.
4356 (value_range::set_undefined): New.
4357 (value_range::set_varying): New.
4358 (value_range::may_contain_p): New.
4359 (value_range::equiv_clear): New.
4360 (value_range::singleton_p): New.
4361 (value_range::intersect): New.
4362 (value_range::dump): New.
4363 (value_range::set_and_canonicalize): New.
4364 (set_value_range): Adjust for value_range API.
4365 (set_value_range_to_undefined): Same.
4366 (set_value_range_to_varying): Same.
4367 (set_and_canonicalize_value_range): Same.
4368 (set_value_range_to_nonnull): Same.
4369 (set_value_range_to_null): Same.
4370 (range_is_null): Same.
4371 (range_is_nonnull): Same.
4372 (range_int_cst_p): Same.
4373 (range_int_cst_singleton_p): Same.
4374 (symbolic_range_p): Same.
4375 (range_includes_zero_p): Same.
4376 (value_range_constant_singleton): Same.
4377 (vrp_set_zero_nonzero_bits): Same.
4378 (ranges_from_anti_range): Same.
4379 (extract_range_into_wide_ints): Same.
4380 (extract_range_from_multiplicative_op): Same.
4381 (set_value_range_with_overflow): Same.
4382 (extract_range_from_binary_expr_1): Same.
4383 (extract_range_from_unary_expr): Same.
4384 (dump_value_range): Same.
4385 (debug_value_range): Same.
4386 (vrp_prop::check_array_ref): Same.
4387 (vrp_prop::check_mem_ref): Same.
4388 (vrp_prop::vrp_initialize): Same.
4389 (vrp_prop::visit_stmt): Same.
4390 (intersect_ranges): Same.
4391 (vrp_prop::visit_phi): Same.
4392 (vrp_prop::vrp_finalize): Same.
4393 (determine_value_range_1): Same.
4394 (determine_value_range): Same.
4395 (vrp_intersect_ranges_1): Rename to...
4396 (vrp_intersect_1): this.
4397 (vrp_intersect_ranges): Rename to...
4398 (value_range::intersect_helper): ...this.
4399 (vrp_meet_1): Rename to...
4400 (value_range::union_helper): ...this.
4401 (vrp_meet): Rename to...
4402 (value_range::union_): ...this.
4403 (copy_value_range): Remove.
4404 * tree-vrp.h (struct value_range): Rewrite into a proper class.
4405 (value_range::vrtype): New.
4406 (value_range::type): New.
4407 (value_range::equiv): New.
4408 (value_range::min): New.
4409 (value_range::max): New.
4410 (value_range::varying_p): New.
4411 (value_range::undefined_p): New.
4412 (value_range::null_p): New.
4413 (value_range::equiv_add): New.
4414 (copy_value_range): Remove.
4416 2018-10-17 David Malcolm <dmalcolm@redhat.com>
4418 * Makefile.in (SELFTEST_TARGETS): New.
4419 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
4420 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
4421 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
4423 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
4424 (selftest-c++-gdb, selftest-c++-valgrind): Move to
4426 * configure: Regenerate.
4427 * configure.ac (selftest_languages): New.
4429 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4431 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
4432 overflow wraps argument.
4433 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
4434 wide_int_range_multiplicative_op.
4435 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
4436 overflow wraps argument.
4437 (wide_int_range_multiplicative_op): Same.
4438 (wide_int_range_lshift): Same.
4439 (wide_int_range_div): Same.
4440 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
4441 (wide_int_range_lshift): Same.
4442 (wide_int_range_div): Same.
4444 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
4446 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
4447 use sign as argument.
4448 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
4449 wide_int_range_shift_undefined_p.
4451 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4453 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
4455 (@despeculate_copy<ALLI_TI:mode>): ... This.
4456 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
4459 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4461 * config.gcc: Obsolete *-*-solaris2.10*.
4462 * doc/install.texi (Specific, *-*-solaris2*): Document it.
4464 2018-10-12 Jeff Law <law@redhat.com>
4466 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
4467 reg + sym +- const_int addressing modes.
4469 2018-10-15 David Malcolm <dmalcolm@redhat.com>
4471 * common.opt (fdiagnostics-minimum-margin-width=): New option.
4472 * diagnostic-show-locus.c (layout::layout): Apply the minimum
4474 (layout::start_annotation_line): Only print up to 3 of the
4475 margin character, to avoid touching the left-hand side.
4476 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
4477 minimum margin width, as set by test_diagnostic_context's ctor.
4478 (selftest::test_fixit_insert_containing_newline): Likewise.
4479 (selftest::test_fixit_insert_containing_newline_2): Likewise.
4480 (selftest::test_line_numbers_multiline_range): Clear
4481 dc.min_margin_width.
4482 * diagnostic.c (diagnostic_initialize): Initialize
4484 * diagnostic.h (struct diagnostic_context): Add field
4486 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
4487 * opts.c (common_handle_option): Handle
4488 OPT_fdiagnostics_minimum_margin_width_.
4489 * selftest-diagnostic.c
4490 (selftest::test_diagnostic_context::test_diagnostic_context):
4491 Initialize min_margin_width to 6.
4492 * toplev.c (general_init): Initialize global_dc->min_margin_width.
4494 2018-10-15 David Malcolm <dmalcolm@redhat.com>
4496 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
4497 Fix usage of "error_at_rich_loc" in the comment.
4499 2018-10-15 Renlin Li <renlin.li@arm.com>
4502 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
4503 if-conversioned loop when it contains ifn with types not
4504 supported by backend.
4505 * internal-fn.c (expand_direct_optab_fn): Add an assert.
4506 (direct_internal_fn_supported_p): New helper function.
4507 * internal-fn.h (direct_internal_fn_supported_p): Declare.
4509 2018-10-15 Jakub Jelinek <jakub@redhat.com>
4512 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4513 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4514 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4516 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
4518 PR tree-optimization/87022
4519 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
4520 bits in dist vector rather than the first one.
4522 2018-10-15 Richard Biener <rguenther@suse.de>
4525 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
4526 (visit_loadstore): When a used restrict tag escaped verify that
4527 the points-to solution of "other" pointers do not include
4529 (compute_dependence_clique): If a used restrict tag escaped
4530 communicated that down to visit_loadstore.
4532 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
4534 * config/s390/s390.c (s390_expand_vec_init): Force vector element
4535 into reg if it isn't a general operand.
4537 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4540 * config/i386/sse.md (*vec_dupv2di): Add register source to
4543 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
4546 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
4547 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
4548 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
4549 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
4551 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
4553 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
4554 (notice_args_size): Set it in the current trace if no insn that can
4555 throw internally has been seen yet.
4556 (connect_traces): When connecting args_size between traces, allow the
4557 incoming values not to match if there is an insn setting it before the
4558 first insn that can throw internally; in that case, force the creation
4559 of a CFI note on this latter insn.
4561 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
4563 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
4564 the base class, because using the injected-class-name was not clearly
4565 specified until DR 176.
4567 2018-10-12 Paul Koning <ni1d@arrl.net>
4569 * config/pdp11/pdp11.md (doloop_end): New expander.
4570 (doloop_end_insn): renamed from "doloop_end".
4571 (addqi3): New pattern.
4572 (subqi3): New pattern.
4573 * config/pdp11/predicates.md (incdec_operand): New predicate.
4575 2018-10-12 Yury Gribov <tetra2005@gmail.com>
4578 * real.c (format_helper::can_represent_integral_type_p): New function
4579 * real.h (format_helper::can_represent_integral_type_p): Ditto.
4580 * match.pd: New pattern.
4582 2018-10-12 Alexandre Oliva <oliva@adacore.com>
4584 * configure.ac: Introduce --enable-large-address-aware
4585 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
4586 * doc/install.texi: Document it.
4587 * configure, config.in: Rebuilt.
4588 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
4589 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
4590 (LINK_SPEC): Insert it.
4591 * config/i386/mingw-w64.h: Likewise.
4593 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
4595 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
4597 PR rtl-optimization/87600
4598 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
4600 2018-10-12 Paul Koning <ni1d@arrl.net>
4602 * doc/md.texi (doloop_end): Document that the pattern code may
4603 need to check operand mode.
4605 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
4607 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
4608 to zero-extend between int and floating-point registers.
4609 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
4610 ldp into floating-point registers. Add type and arch attributes.
4611 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
4612 Use f_loads for type attribute.
4614 2018-10-11 Martin Sebor <msebor@redhat.com>
4616 * doc/extend.texi (attribute packed): Correct typos.
4618 2018-10-11 Martin Sebor <msebor@redhat.com>
4620 * doc/extend.texi (attribute flatten): Mention interaction with
4623 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
4626 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
4627 Set new_decl virtual flag to zero.
4629 2018-10-11 Martin Sebor <msebor@redhat.com>
4632 * doc/extend.texi (attribute format_arg): Discuss using multiple
4633 attributes on a single function.
4635 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
4637 PR tree-optimization/86829
4638 * match.pd (sin (atan (x))): New simplification rules.
4639 (cos (atan (x))): Likewise.
4640 * real.c (build_sinatan_real): New function.
4641 * real.h (build_sinatan_real): Prototype.
4643 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4645 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
4647 (fold_mergeeo_helper): New helper function.
4648 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
4649 intrinsics. Correct some whitespace indentation issues.
4651 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
4654 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
4655 Use HOST_WIDE_INT_1U for shift.
4657 2018-10-11 Doug Rupp <rupp@adacore.com>
4658 Olivier Hainque <hainque@adacore.com>
4660 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
4661 Pass --relax to the linker for RTPs.
4662 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
4664 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
4665 Jan Hubicka <jh@suse.cz>
4666 Martin Jambor <mjambor@suse.cz>
4668 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
4669 the same elements are repeated rather than printing all of them.
4670 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
4671 "repeated" elements.
4672 * read-rtl-function.c (test_loading_repeat): New function.
4673 (read_rtl_function_c_tests): Call test_loading_repeat.
4674 * rtl-tests.c (test_dumping_repeat): New function.
4675 (rtl_tests_c_tests): Call test_dumping_repeat.
4677 2018-10-11 Richard Biener <rguenther@suse.de>
4679 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
4680 bdver?_cost): Unify to ...
4681 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
4682 * config/i386/i386.c (processor_cost_table): Adjust.
4684 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
4687 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
4688 the thunk when expanding to GIMPLE.
4690 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
4692 * varasm.c (mergeable_string_section): Don't try to move zero-length
4693 strings to the merge section.
4695 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
4698 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
4700 2018-10-10 Jakub Jelinek <jakub@redhat.com>
4703 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
4704 to special_args set.
4706 2018-10-10 Richard Biener <rguenther@suse.de>
4708 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
4709 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
4710 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
4711 and recursing and pattern terminating the recursion on SSE
4712 vector width using ix86_expand_reduc.
4713 (reduc_sminmax_scal_<mode>): Split into part reducing to half
4714 width and recursing and SSE2 vector variant doing the final
4715 reduction with ix86_expand_reduc.
4716 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
4717 with terminating the recursion at AVX level, splitting that
4720 2018-10-09 David Malcolm <dmalcolm@redhat.com>
4722 * genmatch.c (error_cb): Rename to...
4723 (diagnostic_cb): ...this, converting int params to enums.
4724 (fatal_at): Update for renaming.
4725 (warning_at): Likewise.
4727 * input.c (selftest::ebcdic_execution_charset::apply):
4728 Update for renaming of...
4729 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
4731 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
4732 converting level and reason to enums.
4733 (class selftest::lexer_error_sink): Rename to...
4734 (class selftest::lexer_test_options): ...this, renaming field
4735 "m_errors" to "m_diagnostics".
4736 (selftest::lexer_test_options::apply): Update for renaming of...
4737 (selftest::lexer_test_options::on_error): ...this, renaming to...
4738 (selftest::lexer_test_options::on_diagnostic): ...this
4739 converting level and reason to enums.
4740 (selftest::test_lexer_string_locations_raw_string_unterminated):
4741 Update for renamings.
4742 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
4745 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
4747 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
4748 * config/rs6000/pmmintrin.h: New file.
4750 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
4752 PR tree-optimization/86659
4753 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
4755 2018-10-09 Richard Biener <rguenther@suse.de>
4757 PR tree-optimization/63155
4758 * tree-ssa-structalias.c: Include tree-ssa.h.
4759 (get_constraint_for_ssa_var): For undefs return nothing_id.
4760 (find_func_aliases): Cleanup PHI handling.
4762 2018-10-09 Richard Biener <rguenther@suse.de>
4764 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
4767 2018-10-09 Martin Liska <mliska@suse.cz>
4769 * asan.c (asan_emit_stack_protection): If a stack variable
4770 is located in a same file as current function, then emit
4771 line info into variable definition string.
4773 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4775 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
4778 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4780 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
4783 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
4785 PR tree-optimization/86659
4786 * gimple-match.h (struct gimple_match_op): Add reverse field.
4787 (gimple_match_op::set_op): New overloaded method.
4788 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
4789 the REF_REVERSE_STORAGE_ORDER flag on the value.
4790 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
4791 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
4793 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
4796 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
4797 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
4799 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
4802 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
4803 Defined with __builtin_ia32_vfmaddsubpd512_mask.
4805 2018-10-08 Richard Biener <rguenther@suse.de>
4807 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
4808 cost the same as AVX128 ones.
4810 2018-10-08 Paul Koning <ni1d@arrl.net>
4812 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
4813 (expand_block_move): New function.
4814 * config/pdp11/pdp11.c (output_block_move): Remove.
4815 (expand_block_move): New function.
4816 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
4817 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
4818 (*movmemhi1): Remove.
4820 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
4822 * config/s390/2827.md: Increase latencies for some FP instructions.
4824 2018-10-08 Richard Biener <rguenther@suse.de>
4826 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
4828 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
4829 * tree-vectorizer.h (dump_stmt_cost): Adjust.
4830 (add_stmt_cost): Dump return value of the hook.
4832 2018-10-08 Richard Biener <rguenther@suse.de>
4834 PR tree-optimization/63155
4835 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
4836 (ssa_propagation_engine::ssa_propagate): Remove redundant
4837 bitmap bit clearing.
4839 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
4841 PR rtl-optimization/86939
4842 PR rtl-optimization/87479
4843 * ira.h (non_conflicting_reg_copy_p): New prototype.
4844 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
4845 (make_hard_regno_dead): Don't add conflicts for register
4846 ignore_reg_for_conflicts.
4847 (make_object_dead): Likewise.
4848 (non_conflicting_reg_copy_p): New function.
4849 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
4850 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
4851 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
4852 (make_hard_regno_dead): Don't add conflicts for register
4853 ignore_reg_for_conflicts. Remove special conflict handling of
4854 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
4855 check_pic_pseudo_p and update callers.
4856 (mark_pseudo_dead): Don't add conflicts for register
4857 ignore_reg_for_conflicts.
4858 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
4860 2018-10-05 Andrew Waterman <andrew@sifive.com>
4861 Jim Wilson <jimw@sifive.com>
4863 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
4864 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
4865 new pattern using HONOR_SNANS that emits one extra instruction.
4867 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4869 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
4870 patterns): Merge SI and DI patterns to a GPR pattern.
4871 (unnamed define_insn and define_split for record form of that): Merge
4872 to a single define_insn_and_split pattern.
4874 2018-10-05 David Malcolm <dmalcolm@redhat.com>
4877 * input.c (expand_location_to_spelling_point): Add param "aspect"
4878 and use rather than hardcoding LOCATION_ASPECT_CARET.
4879 (get_substring_ranges_for_loc): Handle the case of a single token
4880 within a macro expansion.
4881 * input.h (expand_location_to_spelling_point): Add "aspect" param,
4882 defaulting to LOCATION_ASPECT_CARET.
4884 2018-10-05 Paul Koning <ni1d@arrl.net>
4886 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
4887 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
4888 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
4889 (pdp11_guard_type): New function.
4891 2018-10-05 Paul Koning <ni1d@arrl.net>
4893 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
4894 * config/pdp11/pdp11.opt (mfloat32): Remove.
4896 * doc/invoke.texi (pdp11 -mfloat32): Remove:
4897 (pdp11 -mfloat64): Remove.
4899 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4901 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
4902 (*cmp<mode>_cc_i387): Ditto.
4903 (*cmpu<mode>_cc_i387): Ditto.
4904 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4905 * config/i386/i386.c (ix86_expand_fp_compare): Remove
4907 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
4908 Emit x86_sahf_1 pattern.
4909 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
4910 (ix86_expand_carry_flag_compare): Ditto.
4912 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4914 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
4915 to reg_or_0_operand. Add "C" constraint.
4916 (*cmpxf_cc_i387): Ditto.
4917 (*cmp<mode>_i387): Change operand 2 predicate
4918 to nonimm_or_0_operand. Add "C" constraint.
4919 (*cmp<mode>_cc_i387): Ditto.
4920 (*cmp<mode>_0_i387): Remove insn pattern.
4921 (*cmp<mode>_0_cc_i387): Ditto.
4923 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
4925 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
4926 * config/i386/predicates.md (nonimm_or_0_operand): Rename
4927 from vector_move_operand. Update all uses.
4929 2018-10-05 Martin Sebor <msebor@redhat.com>
4931 PR tree-optimization/87490
4932 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
4935 2018-10-05 Richard Biener <rguenther@suse.de>
4937 PR tree-optimization/63155
4938 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
4939 vertical space in dumpfiles.
4940 * tree-ssa-propagate.h
4941 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
4942 * tree-ssa-propagate.c (cfg_blocks_back): New global.
4943 (ssa_edge_worklist_back): Likewise.
4944 (curr_order): Likewise.
4945 (cfg_blocks_get): Remove abstraction.
4946 (cfg_blocks_add): Likewise.
4947 (cfg_blocks_empty_p): Likewise.
4948 (add_ssa_edge): Add to current or next worklist based on
4950 (add_control_edge): Likewise.
4951 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
4953 (ssa_propagation_engine::ssa_propagate): ... here. Unify
4954 iteration from CFG and SSA edge worklist so we process
4955 everything in RPO order, prioritizing forward progress
4957 (ssa_prop_init): Allocate new worklists, do not dump
4959 (ssa_prop_fini): Free new worklists.
4961 2018-10-05 Richard Biener <rguenther@suse.de>
4963 * tree-core.h (tree_block::abstract_flag): Remove.
4964 (tree_block::block_num): Make full 32bits.
4965 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
4966 * tree.h (BLOCK_ABSTRACT): Remove.
4967 * dwarf2out.c (gen_lexical_block_die): Remove dead code
4968 resulting from BLOCK_ABSTRACT being always false.
4969 (gen_inlined_subroutine_die): Likewise.
4970 (gen_block_die): Likewise.
4971 * tree.c (block_ultimate_origin): Likewise.
4972 * tree-pretty-print.c (dump_block_node): Remove code dealing
4973 with BLOCK_ABSTRACT.
4974 * tree-ssa-live.c (dump_scope_block): Likewise.
4975 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
4976 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
4978 2018-10-05 Richard Biener <rguenther@suse.de>
4980 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
4981 is asked for initialize mode to the component mode of the
4984 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
4987 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
4988 assembler for -mavx.
4989 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
4991 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4994 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
4996 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
4997 to be DImode. When using mffscrn, force the operand to a register.
4999 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
5001 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
5002 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
5003 X87MODEF mode iterator.
5004 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
5005 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
5006 X87MODEF mode iterator.
5008 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
5010 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
5011 -Wno-prio-ctor-dtor.
5013 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5015 * Makefile.in (OBJS): Add opt-problem.o.
5016 * dump-context.h: Include "selftest.h.
5017 (selftest::temp_dump_context): New forward decl.
5018 (class dump_context): Make friend of class
5019 selftest::temp_dump_context.
5020 (dump_context::dump_loc_immediate): New decl.
5021 (class dump_pretty_printer): Move here from dumpfile.c.
5022 (class temp_dump_context): Move to namespace selftest.
5023 (temp_dump_context::temp_dump_context): Add param
5024 "forcibly_enable_dumping".
5025 (selftest::verify_dumped_text):
5026 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
5027 (selftest::verify_item):
5028 (ASSERT_IS_TEXT): Move here from dumpfile.c.
5029 (ASSERT_IS_TREE): Likewise.
5030 (ASSERT_IS_GIMPLE): Likewise.
5031 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
5033 (dump_context::dump_loc_immediate): ...this new function.
5034 (class dump_pretty_printer): Move to dump-context.h.
5035 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
5036 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
5037 (temp_dump_context::temp_dump_context): Move to "selftest"
5038 namespace. Add param "forcibly_enable_dumping", and use it to
5039 conditionalize the use of m_pp;
5040 (selftest::verify_dumped_text): Make non-static.
5041 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
5042 (selftest::verify_item): Make non-static.
5043 (ASSERT_IS_TEXT): Move to dump-context.h.
5044 (ASSERT_IS_TREE): Likewise.
5045 (ASSERT_IS_GIMPLE): Likewise.
5046 (selftest::test_capture_of_dump_calls): Pass "true" for new
5047 param of temp_dump_context.
5048 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
5049 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
5051 * opt-problem.cc: New file.
5052 * opt-problem.h: New file.
5053 * optinfo-emit-json.cc
5054 (selftest::test_building_json_from_dump_calls): Pass "true" for
5055 new param of temp_dump_context.
5056 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
5057 (optinfo::emit_for_opt_problem): New function.
5058 (optinfo::emit): Clarity which emit_item is used.
5059 * optinfo.h (optinfo::get_dump_location): New accessor.
5060 (optinfo::emit_for_opt_problem): New decl.
5061 (optinfo::emit): Make const.
5062 * selftest-run-tests.c (selftest::run_tests): Call
5063 selftest::opt_problem_cc_tests.
5064 * selftest.h (selftest::opt_problem_cc_tests): New decl.
5065 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
5066 bool to opt_result, converting fprintf messages to
5067 opt_result::failure_at calls. Add "stmt" param for use by the
5069 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
5070 (runtime_alias_check_p): Convert return type from bool to
5071 opt_result, converting dump_printf calls to
5072 opt_result::failure_at, using the statement DDR_A for their
5074 (find_data_references_in_stmt): Convert return type from bool to
5075 opt_result, converting "return false" to opt_result::failure_at
5077 * tree-data-ref.h: Include "opt-problem.h".
5078 (dr_analyze_innermost): Convert return type from bool to opt_result,
5079 and add a const gimple * param.
5080 (find_data_references_in_stmt): Convert return type from bool to
5082 (runtime_alias_check_p): Likewise.
5083 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
5084 dr_analyze_innermost.
5085 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
5086 Convert return type from bool to opt_result, adding a message for
5087 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
5088 (vect_analyze_data_ref_dependence): Convert return type from bool
5089 to opt_result. Change sense of return type from "false"
5090 effectively meaning "no problems" to "false" meaning a problem,
5091 so that "return false" becomes "return opt_result::success".
5092 Convert "return true" calls to opt_result::failure_at, using
5093 the location of statement A rather than vect_location.
5094 (vect_analyze_data_ref_dependences): Convert return type from bool
5096 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
5097 calls to opt_result::failure_at, using the stmt location rather
5099 (vect_verify_datarefs_alignment): Convert return type from bool
5101 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
5102 into multiple more-tightly-scoped copies.
5103 (vect_analyze_data_refs_alignment): Convert return type from bool
5105 (vect_analyze_data_ref_accesses): Likewise, converting a
5106 "return false" to a "return opt_result::failure_at", adding a
5108 (vect_prune_runtime_alias_test_list): Convert return type from
5109 bool to opt_result, converting dump_printf_loc to
5110 opt_result::failure_at. Add a %G to show the pertinent statement,
5111 and use the stmt's location rather than vect_location.
5112 (vect_find_stmt_data_reference): Convert return type from
5113 bool to opt_result, converting dump_printf_loc to
5114 opt_result::failure_at, using stmt's location.
5115 (vect_analyze_data_refs): Convert return type from bool to
5116 opt_result. Convert "return false" to "return
5117 opt_result::failure_at", adding messages as needed.
5118 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
5119 type from bool to opt_result.
5120 (vect_determine_vf_for_stmt): Likewise.
5121 (vect_determine_vectorization_factor): Likewise, converting
5122 dump_printf_loc to opt_result::failure_at, using location of phi
5123 rather than vect_location.
5124 (vect_analyze_loop_form_1): Convert return type from bool to
5125 opt_result, converting dump_printf_loc calls, retaining the use of
5127 (vect_analyze_loop_form): Convert return type from loop_vec_info
5128 to opt_loop_vec_info.
5129 (vect_analyze_loop_operations): Convert return type from bool to
5130 opt_result, converting dump_printf_loc calls, using the location
5131 of phi/stmt rather than vect_location where available. Convert
5132 various "return false" to "return opt_result::failure_at" with
5133 "unsupported phi" messages.
5134 (vect_get_datarefs_in_loop): Convert return type from bool to
5135 opt_result. Add a message for the
5136 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
5137 (vect_analyze_loop_2): Convert return type from bool to
5138 opt_result. Ensure "ok" is set to a opt_result::failure_at before
5139 each "goto again;", adding new messages where needed.
5140 Add "unsupported grouped {store|load}" messages.
5141 (vect_analyze_loop): Convert return type from loop_vec_info to
5143 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
5145 * tree-vect-stmts.c (process_use): Likewise, converting
5146 dump_printf_loc call and using stmt location, rather than
5148 (vect_mark_stmts_to_be_vectorized): Likeise.
5149 (vect_analyze_stmt): Likewise, adding a %G.
5150 (vect_get_vector_types_for_stmt): Convert return type from bool to
5151 opt_result, converting dump_printf_loc calls and using stmt
5152 location, rather than vect_location.
5153 (vect_get_mask_type_for_stmt): Convert return type from tree to
5154 opt_tree, converting dump_printf_loc calls and using stmt location.
5155 * tree-vectorizer.c: Include "opt-problem.h.
5156 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
5157 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
5158 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
5159 enabled, use it to report at the top level "couldn't vectorize
5160 loop" followed by the problem.
5161 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
5162 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
5164 (vect_analyze_stmt): Likewise.
5165 (vect_get_vector_types_for_stmt): Likewise.
5166 (tree vect_get_mask_type_for_stmt): Likewise.
5167 (vect_analyze_data_ref_dependences): Likewise.
5168 (vect_enhance_data_refs_alignment): Likewise.
5169 (vect_analyze_data_refs_alignment): Likewise.
5170 (vect_verify_datarefs_alignment): Likewise.
5171 (vect_analyze_data_ref_accesses): Likewise.
5172 (vect_prune_runtime_alias_test_list): Likewise.
5173 (vect_find_stmt_data_reference): Likewise.
5174 (vect_analyze_data_refs): Likewise.
5175 (vect_analyze_loop): Convert return type from loop_vec_info to
5177 (vect_analyze_loop_form): Likewise.
5178 (vect_analyze_slp): Convert return type from bool to opt_result.
5180 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5182 * doc/invoke.texi (-fopt-info): Document new "internals"
5184 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
5185 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
5187 (optinfo_verbosity_options): Add "internals".
5188 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
5189 (dump_context::apply_dump_filter_p): New member function.
5190 (dump_context::dump_loc): Use apply_dump_filter_p rather than
5191 explicitly masking the dump_kind.
5192 (dump_context::begin_scope): Increment the scope depth first. Use
5193 apply_dump_filter_p rather than explicitly masking the dump_kind.
5194 (dump_context::emit_item): Use apply_dump_filter_p rather than
5195 explicitly masking the dump_kind.
5196 (dump_dec): Likewise.
5197 (dump_hex): Likewise.
5198 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
5199 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
5200 (opt_info_switch_p): Update handling of default
5201 MSG_OPTIMIZED_LOCATIONS to cope with default of
5202 MSG_PRIORITY_USER_FACING.
5203 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
5204 masking the dump_kind.
5205 (selftest::test_capture_of_dump_calls): Update test_dump_context
5206 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
5207 than MSG_ALL. Generalize scope test to be run at all four
5208 combinations of with/without MSG_PRIORITY_USER_FACING and
5209 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
5210 for each of the two values.
5211 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
5212 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
5213 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
5214 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
5215 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
5216 with MSG_PRIORITY_*.
5217 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
5218 dump messages as MSG_PRIORITY_USER_FACING.
5219 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
5220 about the interaction with MSG_PRIORITY_*.
5222 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5224 * varasm.c (output_constant): Add new parameter merge_strings.
5225 Make strings properly zero terminated in merge string sections.
5226 (mergeable_string_section): Don't fail if the last char is non-zero.
5227 (assemble_variable_contents): Handle merge string sections.
5228 (assemble_variable): Likewise.
5229 (assemble_constant_contents): Likewise.
5230 (output_constant_def_contents): Likewise.
5231 (output_constructor_array_range,
5232 output_constructor_regular_field): Adjust call to output_constant.
5233 (output_object_block): Adjust call to assemble_constant_contents
5234 and assemble_variable_contents.
5236 2018-10-04 Martin Liska <mliska@suse.cz>
5239 * cgraphunit.c (process_function_and_variable_attributes):
5240 Warn about a function with alias attribute and a body.
5242 2018-10-04 Martin Liska <mliska@suse.cz>
5245 * multiple_target.c (redirect_to_specific_clone): New function.
5246 (ipa_target_clone): Use it.
5247 * tree-inline.c: Fix comment.
5249 2018-10-04 David Malcolm <dmalcolm@redhat.com>
5251 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
5253 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
5254 (gcc::dump_manager::register_pass): New member function, adapted
5255 from loop body in gcc::pass_manager::register_pass, adding a
5256 call to update_dfi_for_opt_info.
5257 (gcc::dump_manager::opt_info_enable_passes): Store the
5258 -fopt-info options into the new fields. Move the loop
5260 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
5262 * dumpfile.h (struct opt_pass): New forward decl.
5263 (gcc::dump_manager::register_pass): New decl.
5264 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
5265 (class gcc::dump_manager): Add fields "m_optgroup_flags",
5266 "m_optinfo_flags", and "m_optinfo_filename".
5267 * passes.c (gcc::pass_manager::register_pass): Move all of the
5268 dump-handling code to gcc::dump_manager::register_pass.
5270 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
5272 PR rtl-optimization/87466
5273 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
5274 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
5275 * doc/tm.texi: Regenerate.
5276 * ira-lives.c (process_bb_node_lives): Use the new target hook.
5277 * lra-lives.c (process_bb_lives): Likewise.
5278 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
5281 2018-10-04 Tamar Christina <tamar.christina@arm.com>
5283 * params.c (add_params): Fix initialization.
5285 2018-10-04 Martin Liska <mliska@suse.cz>
5287 PR gcov-profile/84107
5288 * tree-profile.c (init_ic_make_global_vars):
5289 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
5290 Come up with new ic_tuple* variables. Emit
5291 __gcov_indirect_call{,_topn} variables.
5292 (gimple_gen_ic_profiler): Access the variable
5294 (gimple_gen_ic_func_profiler): Access
5295 __gcov_indirect_call.callee field.
5296 (gimple_init_gcov_profiler): Use ptr_type_node.
5297 * value-prof.c (gimple_ic): Use ptr_type_node.
5299 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5301 PR tree-optimization/85787
5302 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
5303 malloc_candidate_p into this function and add support for
5304 detecting multiple phis.
5305 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
5307 2018-10-04 Martin Liska <mliska@suse.cz>
5310 * ipa-inline.c (inline_to_all_callers_1):
5311 Call ultimate_alias_target for node being inlined.
5313 2018-10-03 Jeff Law <law@redhat.com>
5315 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
5317 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
5318 * tree.h (get_typenode_from_name): Prototype.
5320 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
5322 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5323 Change operand 2 predicate to nonimmediate_operand.
5324 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5326 2018-10-03 Martin Sebor <msebor@redhat.com>
5327 Jeff Law <law@redhat.com>
5329 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
5331 (get_string_length): Detect unterminated arrays.
5332 (format_string): Same.
5333 (format_directive): Warn about unterminated arrays.
5334 (handle_gimple_call): Mark statements with no_warning as needed.
5336 2018-10-03 Jim Wilson <jimw@sifive.com>
5338 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
5339 also define __riscv_abi_rve. Delete trailing white space.
5341 2018-10-03 Paul Koning <ni1d@arrl.net>
5343 Enable LRA register allocator for PDP11.
5344 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
5347 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
5348 * config/pdp11/pdp11.opt (-mlra): New option.
5349 * doc/invoke.texi (PDP-11 Options): Document -mlra.
5351 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
5353 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
5354 (*<absneg:code>extend<mode>xf2): Ditto.
5356 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
5358 PR tree-optimization/87415
5359 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
5362 2018-10-02 Jeff Law <law@redhat.com>
5364 * gimple-fold.c (get_range_strlen): Only set *nonstr when
5365 an unterminated string is discovered. Bubble up range
5366 even for unterminated strings.
5367 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
5368 indicates the string was not terminated via NONSTR.
5370 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
5372 * tree-vrp.c (extract_range_from_unary_expr): Special case all
5373 pointer conversions.
5374 Do not do anything special for anti-ranges.
5376 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
5378 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
5379 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
5381 2018-10-03 Martin Liska <mliska@suse.cz>
5383 PR gcov-profile/86109
5384 * coverage.c (coverage_begin_function): Do not
5385 mark lambdas as artificial.
5386 * tree-core.h (struct GTY): Remove tm_clone_flag
5387 and introduce new lambda_function.
5388 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
5390 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
5393 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
5394 P8_VECTOR and VSX are enabled.
5396 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
5398 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5399 0x3907 as CPU model number.
5401 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
5403 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
5404 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
5405 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
5406 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
5407 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
5408 * config/s390/s390.md: Likewise. Rename also the cpu attribute
5409 value from arch12 to z14.
5411 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
5413 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
5415 (isinf<mode>2): Ditto.
5417 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
5419 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
5420 before emitting fxam. Perform calculations in XFmode.
5422 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
5424 * match.pd (((X /[ex] A) +- B) * A): New transformation.
5426 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
5429 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
5430 * tree.c (signed_or_unsigned_type_for): Handle complex.
5432 2018-10-02 Jeff Law <law@redhat.com>
5434 * gimple-fold.c (get_range_strlen): Remove dead code.
5436 2018-10-02 Martin Sebor <msebor@redhat.com>
5437 Jeff Law <law@redhat.com>
5439 * builtins.c (unterminated_array): Add new arguments.
5440 If argument is not terminated, bubble up size and exact
5442 (expand_builtin_strnlen): Detect, avoid expanding
5443 and diagnose unterminated arrays.
5444 (c_strlen): Fill in offset of start of unterminated strings.
5445 * builtins.h (unterminated_array): Update prototype.
5447 2018-10-02 Richard Biener <rguenther@suse.de>
5449 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
5450 of haddv4df, first reduce to SSE width and exploit the fact
5451 that we only need element zero with the reduction result.
5452 (reduc_plus_scal_v2df): Likewise.
5454 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
5456 * dojump.h (do_jump): Delete.
5457 (do_jump_1): Likewise.
5458 (split_comparison): Move around.
5459 * dojump.c (do_jump): Make static.
5460 (do_jump_1): Likewise.
5461 (jumpifnot): Move around.
5462 (jumpifnot_1): Likewise.
5464 (jumpif_1): Likewise.
5465 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
5467 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
5469 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
5470 insns in the delay slot and add_insn_after for the jump insn.
5472 2018-10-02 Richard Biener <rguenther@suse.de>
5474 * tree-inline.c (expand_call_inline): Use the location of
5475 the callee declaration for the inline-entry marker.
5476 * final.c (notice_source_line): Remove special-casing of
5477 NOTE_INSN_INLINE_ENTRY.
5479 2018-10-01 Carl Love <cel@us.ibm.com>
5482 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
5483 (__builtin_mtfsb0): New.
5484 (__builtin_mtfsb1): New.
5485 ( __builtin_set_fpscr_rn): New.
5486 (__builtin_set_fpscr_drn): New.
5487 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
5488 (rs6000_expand_set_fpscr_rn_builtin): Add.
5489 (rs6000_expand_set_fpscr_drn_builtin): Add.
5490 (rs6000_expand_builtin): Add case statement entries for
5491 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
5492 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
5493 RS6000_BUILTIN_MFFSL.
5494 (rs6000_init_builtins): Add ftype initialization and def_builtin
5495 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
5496 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
5497 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
5498 rs6000_mffscdrn): Add define_insn.
5499 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
5500 * doc/extend.texi: Add documentation for the builtins.
5502 2018-10-01 Richard Biener <rguenther@suse.de>
5504 PR tree-optimization/87465
5505 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
5506 causing branch miscounts.
5508 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5510 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
5511 aarch64_option_default_param): New.
5512 (params.h): Include.
5513 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
5514 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
5515 stack-clash protection validation code.
5517 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5519 * params.c (validate_param): New.
5520 (add_params): Use it.
5521 (set_param_value): Refactor param validation into validate_param.
5522 (diagnostic.h): Include.
5523 * diagnostic.h (diagnostic_ready_p): New.
5525 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5527 * params.c (set_param_value):
5528 Add index of parameter being validated.
5529 * common/common-target.def (option_validate_param): New.
5530 * common/common-targhooks.h (default_option_validate_param): New.
5531 * common/common-targhooks.c (default_option_validate_param): New.
5532 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
5533 * doc/tm.texi: Regenerate.
5535 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5538 * config/aarch64/aarch64.c (aarch64_override_options_internal):
5539 Add validation for stack-clash parameters and set defaults.
5541 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5544 * configure.ac: Add stack-clash-protection-guard-size.
5545 * doc/install.texi: Document it.
5546 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
5547 * params.def: Update comment for guard-size.
5548 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
5549 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
5550 * configure: Regenerate.
5552 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5555 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
5556 STACK_DYNAMIC_OFFSET): New.
5557 * config/aarch64/aarch64.c (aarch64_layout_frame):
5558 Update outgoing args size.
5559 (aarch64_stack_clash_protection_alloca_probe_range,
5560 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5562 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5565 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
5567 * target.def (stack_clash_protection_alloca_probe_range): New.
5568 (stack_clash_protection_final_dynamic_probe): Remove.
5569 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
5570 (default_stack_clash_protection_final_dynamic_probe): Remove.
5571 * targhooks.c: Likewise.
5572 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
5573 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
5574 * doc/tm.texi: Regenerate.
5576 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5579 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
5580 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
5581 aarch64_clamp_to_uimm12_shift): New.
5582 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
5583 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
5585 2018-10-01 Tamar Christina <tamar.christina@arm.com>
5588 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
5590 2018-10-01 Jeff Law <law@redhat.com>
5591 Richard Sandiford <richard.sandiford@linaro.org>
5592 Tamar Christina <tamar.christina@arm.com>
5595 * config/aarch64/aarch64.md
5596 (probe_stack_range): Add k (SP) constraint.
5597 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
5598 STACK_CLASH_MAX_UNROLL_PAGES): New.
5599 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
5600 stack probes for stack clash.
5601 (aarch64_allocate_and_probe_stack_space): New.
5602 (aarch64_expand_prologue): Use it.
5603 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
5604 (aarch64_sub_sp): Add emit_move_imm optional param.
5606 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
5608 PR tree-optimization/87261
5609 * match.pd: Remove trailing whitespace.
5610 Add (x & y) | ~(x | y) -> ~(x ^ y),
5611 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
5613 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
5615 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
5617 (ashlsi3_insn): Update instruction constraints.
5618 (ashrsi3_insn): Likewise.
5619 (rotrsi3): Likewise.
5620 (add_shift): Likewise.
5621 * config/arc/constraints.md (Csz): New 32 bit constraint. It
5622 avoids placing in the limm field small constants which, otherwise,
5623 could end into a small instruction.
5625 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
5627 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
5628 destination register is not odd-even.
5629 (umaddsidi4_split): Likewise.
5631 2018-10-01 Richard Biener <rguenther@suse.de>
5633 * tree-inline.c (expand_call_inline): Store origin of fn
5634 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
5635 * tree.c (block_ultimate_origin): Simplify and do some
5638 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
5640 * config/i386/mmx.md (EMMS): New int iterator.
5641 (emms): New int attribute.
5642 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
5643 EMMS int iterator. Explicitly declare clobbers.
5644 (mmx_emms): Remove expander.
5646 * config/i386/predicates.md (emms_operation): Remove predicate.
5647 (vzeroall_pattern): New predicate.
5648 (vzeroupper_pattern): Rename from vzeroupper_operation.
5649 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
5650 vzeroupper_pattern and vzeroall_pattern predicates.
5652 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
5654 PR rtl-optimization/86939
5655 * ira-lives.c (make_hard_regno_born): Rename from this...
5656 (make_hard_regno_live): ... to this. Remove update to conflict
5657 information. Update function comment.
5658 (make_hard_regno_dead): Add conflict information update. Update
5660 (make_object_born): Rename from this...
5661 (make_object_live): ... to this. Remove update to conflict information.
5662 Update function comment.
5663 (make_object_dead): Add conflict information update. Update function
5665 (mark_pseudo_regno_live): Call make_object_live.
5666 (mark_pseudo_regno_subword_live): Likewise.
5667 (mark_hard_reg_dead): Update function comment.
5668 (mark_hard_reg_live): Call make_hard_regno_live.
5669 (process_bb_node_lives): Likewise.
5670 * lra-lives.c (make_hard_regno_born): Rename from this...
5671 (make_hard_regno_live): ... to this. Remove update to conflict
5672 information. Remove now uneeded check_pic_pseudo_p argument.
5673 Update function comment.
5674 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
5675 to conflict information. Update function comment.
5676 (mark_pseudo_live): Remove update to conflict information. Update
5678 (mark_pseudo_dead): Add conflict information update.
5679 (mark_regno_live): Call make_hard_regno_live.
5680 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
5681 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
5683 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
5686 * config/i386/i386.c (construct_container): Use TImode for
5687 BLKmode values in 2 integer registers.
5689 2018-09-29 Jeff Law <law@redhat.com>
5691 * builtins.c (unterminated_array): Pass in c_strlen_data * to
5692 c_strlen rather than just a tree *.
5693 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
5694 Update recursive calls appropriately. If caller did not provide a
5695 suitable data pointer, create a local one. When a non-terminated
5696 string is discovered, bubble up information about the string via the
5697 c_strlen_data object.
5698 * builtins.h (c_strlen): Update prototype.
5699 (c_strlen_data): New structure.
5700 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
5701 For a type 2 call, if c_strlen indicates a non-terminated string
5702 use the length of the non-terminated string.
5703 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
5705 2018-09-29 Jakub Jelinek <jakub@redhat.com>
5708 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
5709 __m512d type for __A argument rather than __m512.
5711 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
5713 * match.pd (simple_comparison): Don't optimize if either operand is
5714 a function pointer when target needs function pointer canonicalization.
5716 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5718 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
5719 power5 .. power9 to remove indirection.
5720 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
5721 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
5722 ASM_CPU_476_SPEC): Delete.
5723 (ASM_CPU_SPEC): Adjust.
5724 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
5725 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
5727 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5729 * config.in: Delete HAVE_AS_DCI.
5730 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
5731 * config/rs6000/rs6000.h: Ditto.
5732 * configure.ac: Delete HAVE_AS_DCI.
5733 * configure: Regenerate.
5735 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
5737 * config.in (HAVE_AS_LWSYNC): Delete.
5738 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
5739 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
5741 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
5742 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
5744 * configure.ac: Delete HAVE_AS_LWSYNC.
5745 * configure: Regenerate.
5747 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
5748 Pierre-Marie de Rodat <derodat@adacore.com>
5750 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
5751 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
5752 (cgraph_node::create_thunk): Add indirect_offset parameter.
5753 (thunk_adjust): Likewise.
5754 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
5755 and initialize the corresponding field with it.
5756 (cgraph_node::dump): Dump indirect_offset field.
5757 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
5758 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
5759 (thunk_adjust): Add indirect_offset parameter and deal with it.
5760 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
5761 pass it to thunk_adjust. Do not call the target hook if it's non-zero
5762 or if the thunk is external or local. Fix formatting. Do not chain
5763 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
5764 if any, in the GIMPLE representation.
5765 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
5766 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
5767 (input_node): Read indirect_offset field.
5768 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
5769 call to thunk_adjust.
5770 * tree-nested.c (struct nesting_info): Add thunk_p field.
5771 (create_nesting_tree): Set it.
5772 (convert_all_function_calls): Copy static chain from targets to thunks.
5773 (finalize_nesting_tree_1): Return early for thunks.
5774 (unnest_nesting_tree_1): Do not finalize thunks.
5775 (gimplify_all_functions): Do not gimplify thunks.
5777 2018-09-28 David Malcolm <dmalcolm@redhat.com>
5779 * opt-suggestions.c (option_proposer::build_option_suggestions):
5780 Release "option_values".
5782 2018-09-28 David Malcolm <dmalcolm@redhat.com>
5784 * coverage.c (get_coverage_counts): Convert problem-reporting dump
5785 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
5786 * dumpfile.c (kind_as_string): New function.
5787 (dump_loc): Rather than a hardcoded prefix of "note: ", use
5788 kind_as_string to vary the prefix based on dump_kind.
5789 (selftest::test_capture_of_dump_calls): Update for above.
5791 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
5793 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
5794 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
5796 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
5798 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
5799 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
5800 INVALID_REGNUM instead of FPSR_REG.
5801 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
5802 * config/i386/i386.md: Update comment of FP compares.
5803 (fldenv): Do not clobber FPSR_REG.
5805 2018-09-28 Richard Biener <rguenther@suse.de>
5807 * tree.h (BLOCK_ORIGIN): New.
5808 * omp-expand.c (grid_expand_target_grid_body): Assign
5809 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
5810 * tree-inline.c (remap_block): Likewise.
5811 * auto-profile.c (get_function_decl_from_block): Simplify
5812 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
5813 * langhooks.c (lhd_print_error_function): Likewise.
5814 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
5816 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
5817 * tree.c (block_nonartificial_location): Likewise.
5818 (block_ultimate_origin): Likewise.
5819 * tree-pretty-print.c (percent_K_format): Likewise. Remove
5820 no longer needed LTO case.
5822 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
5823 Jan Hubicka <jh@suse.cz>
5824 Martin Jambor <mjambor@suse.cz>
5826 * simplify-rtx.c (simplify_merge_mask): New function.
5827 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
5828 same masks are used in op1 or op2.
5829 (test_vec_merge): New function.
5830 (test_vector_ops): Call test_vec_merge.
5832 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
5834 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
5835 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
5836 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
5837 (cypress_costs): Set it.
5838 (supersparc_costs): Likewise.
5839 (hypersparc_costs): Likewise.
5840 (leon_cost): Likewise.
5841 (leon3_costs): Likewise.
5842 (sparclet_costs): Likewise.
5843 (ultrasparc_costs): Likewise.
5844 (ultrasparc_costs): Likewise.
5845 (niagara_costs): Likewise.
5846 (niagara2_costs): Likewise.
5847 (niagara3_costs): Likewise.
5848 (niagara4_costs): Likewise.
5849 (niagara7_costs): Likewise.
5850 (m8_costs): Likewise.
5851 (TARGET_CAN_FOLLOW_JUMP): Define.
5852 (pass_work_around_errata::gate): Minor tweak.
5853 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
5854 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
5855 Automaitcally clear MASK_FSMULD mask for V7 processors.
5856 (sparc_can_follow_jump): New static function.
5857 (output_ubranch): Deal with CROSSING_JUMP_P.
5858 (sparc_use_sched_lookahead): Rewrite using switch statement.
5859 (sparc_issue_rate): Reorder.
5860 (sparc_branch_cost): New function.
5862 2018-09-27 Martin Sebor <msebor@redhat.com>
5864 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
5865 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
5866 (int_fits_type_p): Same.
5868 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
5870 * config/i386/i386.md (FPCR_REG): Remove.
5871 (UNSPEC_FLDCW): Remove.
5872 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
5873 (x86_fldcw_1): Remove insn pattern.
5874 (fnstenv): Do not clobber FPCR_REG.
5876 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
5877 (CALL_USED_REGISTERS): Ditto.
5878 (REG_ALLOC_ORDER): Ditto.
5879 (REG_CLASS_CONTENTS): Ditto.
5880 (HI_REGISTER_NAMES): Ditto.
5881 (ADDITIONAL_REGISTER_NAMES): Use defines instead
5882 of numerical constants.
5883 * config/i386/i386.c (regclass_map): Remove fpsr register.
5884 (dbx_register_map): Ditto.
5885 (dbx64_register_map): Ditto.
5886 (svr4_dbx_register_map): Ditto.
5887 (print_reg): Do not handle FPCR_REG.
5889 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
5892 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
5893 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
5894 Delete, always treat as true.
5895 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
5896 Ditto. Simplify remaining code.
5897 * config/powerpcspe/powerpcspe.h: Ditto.
5898 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
5899 Simplify remaining code.
5900 (rs6000_expand_builtin): Ditto.
5901 * config/rs6000/rs6000.h: Ditto.
5902 * configure.ac: Ditto.
5903 * configure: Regenerate.
5905 2018-09-27 Martin Liska <mliska@suse.cz>
5907 * coverage.c (get_coverage_counts): Revert the formatting
5908 of missing profile opt info.
5910 2018-09-27 Richard Biener <rguenther@suse.de>
5914 * dwarf2out.c (set_block_origin_self): Do not mark outermost
5915 block as we do not output that.
5916 (gen_inlined_subroutine_die): Elide the originally outermost
5917 block, matching what we do for concrete instances.
5918 (decls_for_scope): Add parameter specifying whether to recurse
5921 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
5922 Tom de Vries <tom@codesourcery.com>
5926 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
5927 STORE_FLAG_VALUE == 1.
5929 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
5931 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
5932 constant definitions.
5933 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
5934 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
5935 ("speculation_barrier"): New expander definition.
5937 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
5939 PR gcov-profile/86957
5940 * common.opt: New warning option -Wmissing-profile.
5941 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
5942 * doc/invoke.texi: Document -Wmissing-profile.
5944 2018-09-26 Jim Wilson <jimw@sifive.com>
5946 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
5947 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
5949 2018-09-26 Martin Sebor <msebor@redhat.com>
5951 * tree.c (zerop): Change return type to bool.
5952 (integer_zerop, integer_onep, integer_each_onep): Same.
5953 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
5954 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
5955 (real_onep, real_minus_onep, chain_index): Same.
5956 (print_type_hash_statistics, type_list_equal): Same.
5957 * tree.h (zerop): Same.
5958 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
5959 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
5960 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
5961 (real_onep, real_minus_onep, chain_index): Same.
5962 (print_type_hash_statistics, type_list_equal): Same.
5964 2018-09-26 Jim Wilson <jimw@sifive.com>
5966 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
5968 2018-09-26 Jakub Jelinek <jakub@redhat.com>
5971 * config/i386/i386.c: Include debug.h and dwarf2out.h.
5972 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
5975 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
5977 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
5979 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
5981 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
5982 and FP_SECOND_SSE_REGS.
5983 (REG_CLASS_NAMES): Ditto.
5984 (REG_CLASS_CONTENTS): Ditto.
5985 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
5986 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
5987 (ix86_preferred_output_reload_class): Ditto.
5988 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
5989 clobber constraint to "=&f".
5990 (fix_truncdi_i387): Ditto.
5991 (lrintxfdi2): Ditto.
5992 (fistdi2_<rounding>): Ditto.
5993 (fpremxf4_i387): Change "=u" constraint to "=f".
5994 (fprem1xf4_i387): Ditto.
5996 (fptanxf4_i387): Ditto.
5997 (fxtractxf3_i387): Ditto.
5998 (fscalexf4_i387): Ditto.
5999 (atan2xf3): Change "u" constraint to "f".
6000 (fyl2xxf3_i387): Ditto.
6001 (fyl2xp1xf3_i387): Ditto.
6003 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
6006 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
6007 for removed I387_MASK_PM entity.
6010 2018-09-26 Jeff Law <law@redhat.com>
6012 2018-09-26 Alexey Neyman <stilor@att.net>
6014 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
6015 headers are no longer pulled in by <isl/val.h>.
6017 2018-09-26 Richard Biener <rguenther@suse.de>
6020 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
6021 or concrete instance DIE to the tree. Create abstract origin
6022 attributes also for concrete instances.
6024 2018-09-26 Alexey Neyman <stilor@att.net>
6026 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
6027 headers are no longer pulled in by <isl/val.h>.
6029 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
6031 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
6032 Use new helper functions.
6033 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
6034 Use new helper functions.
6035 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
6036 aarch_mm_needs_release): New declarations.
6037 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
6038 aarch_mm_needs_release): New.
6040 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
6042 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
6043 (arm32_output_mi_thunk): Deal with long calls.
6045 2018-09-26 Richard Biener <rguenther@suse.de>
6049 * tree-inline.c (expand_call_inline): When the location
6050 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
6051 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
6052 the inserted BLOCK to make inlined_function_outer_scope_p
6054 * dwarf2out.c (add_call_src_coords_attributes): Do not add
6055 coords for reserved locations.
6057 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6059 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
6060 (*call_indirect_nonlocal_sysv<mode>): Ditto.
6061 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
6062 (*sibcall_nonlocal_sysv<mode>): Ditto.
6063 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
6064 (<bd>_<mode>): Ditto.
6065 (<bd>tf_<mode>): Ditto.
6067 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6069 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
6070 control string as a list of templates instead of as C code.
6071 (*altivec_movti): Ditto.
6072 * config/rs6000/darwin.md (movdf_low_di): Ditto.
6074 2018-09-25 Jim Wilson <jimw@sifive.com>
6076 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
6077 when target symbol is weak.
6079 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
6082 * builtins.c (unterminated_array): Simplify.
6083 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
6084 where pointer arithmetic is safe.
6086 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
6089 * config/rs6000/altivec.md (altivec_vspltb): Use
6090 const_0_to_15_operand instead of u5bit_cint_operand.
6091 (*altivec_vspltb_internal): Ditto.
6092 (altivec_vspltb_direct): Ditto.
6093 (altivec_vsplth): Use const_0_to_7_operand instead of
6095 (*altivec_vsplth_internal): Ditto.
6096 (altivec_vsplth_direct): Ditto.
6097 (altivec_vspltw): Use const_0_to_3_operand instead of
6099 (*altivec_vspltw_internal): Ditto.
6100 (altivec_vspltw_direct): Ditto.
6101 (altivec_vspltsf): Ditto.
6102 (*altivec_vspltsf_internal): Ditto.
6103 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
6104 various splats with the proper size immediate. Reorder the various
6105 cases by ascending size of immediate, and put all such together.
6107 2018-09-25 Richard Biener <rguenther@suse.de>
6110 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
6112 (maybe_create_die_with_external_ref): Do not create
6113 DW_TAG_imported_unit here.
6114 (add_abstract_origin_attribute): Handle external BLOCK refs.
6115 (dwarf2out_abstract_function): Simplify LTO case.
6116 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
6117 rather than using maybe_create_die_with_external_ref.
6119 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
6122 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
6123 (frndintxf2_mask_pm_i387): Ditto.
6124 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
6125 Enable for !flag_trapping_math.
6126 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
6127 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
6128 Change operand 1 predicate to nonimmediate_operand.
6129 (attr "i387_cw"): Remove mask_pm.
6130 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
6131 (enum ix86_entity): Remove I387_MASK_PM.
6132 * config/i386/i386.c (ix86_i387_mode_needed): Do not
6133 handle I387_MASK_PM.
6134 (ix86_mode_needed): Ditto.
6135 (ix86_mode_after): Ditto.
6136 (ix86_mode_entry): Ditto.
6137 (ix86_mode_exit): Ditto.
6138 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
6140 2018-09-25 Jakub Jelinek <jakub@redhat.com>
6142 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
6143 to_update_switch_stmts to vNULL instead of calling create on them
6146 2018-09-25 Richard Biener <rguenther@suse.de>
6148 PR tree-optimization/87402
6149 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
6150 (visit_phi): Re-instantiate handling of supposed to be VARYING
6151 but non-VARYING backedge value.
6153 2018-09-25 Richard Biener <rguenther@suse.de>
6156 * dwarf2out.c (struct sym_off_pair): New.
6157 (external_die_map): New global.
6158 (lookup_decl_die): When in LTO create DIEs lazily from the
6160 (lookup_block_die): New function, create DIEs lazily in LTO.
6161 (equate_block_to_die): New function.
6162 (dwarf2out_die_ref_for_decl): During WPA get the association
6163 from the external DIE map.
6164 (dwarf2out_register_external_die): Record mapping into the
6166 (maybe_create_die_with_external_ref): New function split out from
6167 DIE generation part of old dwarf2out_register_external_die.
6168 (add_abstract_origin_attribute): Do not return the DIE. When
6169 in LTO reference externals directly.
6170 (dwarf2out_abstract_function): When in LTO ignore calls for
6171 decls with external DIEs (already present abstract instances).
6172 (gen_call_site_die): Adjust.
6173 (add_high_low_attributes): Likewise.
6174 (gen_lexical_block_die): Likewise.
6175 (gen_inlined_subroutine_die): Likewie.
6176 (gen_block_die): Likewise.
6177 (dwarf2out_inline_entry): Likewise.
6178 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
6181 2018-09-25 Martin Liska <mliska@suse.cz>
6183 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
6184 integers and not by a float value.
6186 2018-09-25 Martin Liska <mliska@suse.cz>
6189 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
6191 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
6192 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
6193 * godump.c (go_format_type): Remove extra parenthesis.
6195 2018-09-25 Martin Liska <mliska@suse.cz>
6197 * alias.c (set_dest_equal_p): Remove unused function.
6198 * config/i386/i386.c (def_builtin_pure2): Likewise.
6199 * diagnostic-show-locus.c (class layout): Remove
6201 (layout::layout): Likewise here.
6202 * dump-context.h (class temp_dump_context): Likewise.
6203 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
6204 (add_AT_loclistsptr): Likewise.
6205 (add_AT_offset): Likewise.
6206 (get_AT_hi_pc): Likewise.
6207 (is_comdat_die): Likewise.
6208 (type_is_enum): Likewise.
6209 (ceiling): Likewise.
6210 (add_AT_vms_delta): Likewise.
6211 (is_class_die): Likewise.
6212 * edit-context.c (class line_event): Remove unused field.
6213 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
6215 * ipa-cp.c (ipa_get_vr_lat): Likewise.
6216 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
6217 (ok_for_base_p_nonstrict): Likewise.
6218 * tree-chrec.c (is_not_constant_evolution): Likewise.
6219 (chrec_fold_poly_cst): Likewise.
6220 * tree-if-conv.c (has_pred_critical_p): Likewise.
6221 * tree-ssa-coalesce.c (print_exprs): Likewise.
6222 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
6223 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
6224 * tree-vrp.c (value_ranges_intersect_p): Likewise.
6225 (value_range_nonnegative_p): Likewise.
6227 2018-09-25 Martin Liska <mliska@suse.cz>
6229 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
6230 Do not handle "GNU Pascal".
6231 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
6233 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
6234 from documentation. Likewise.
6235 * dbxout.c (dbxout_range_type): Likewise.
6236 * doc/cpp.texi: Likewise.
6237 * doc/extend.texi: Likewise.
6238 * doc/frontends.texi: Likewise.
6239 * doc/invoke.texi: Remove Pascal entry.
6240 * tree.def (CLEANUP_POINT_EXPR): Likewise.
6241 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
6243 2018-09-25 Martin Liska <mliska@suse.cz>
6246 * doc/invoke.texi: Document all parameters and remove default
6249 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
6252 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
6253 contains HOST_WIDE_INTs when computing its size.
6255 2018-09-24 Jim Wilson <jimw@sifive.com>
6258 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
6260 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
6262 2018-09-24 Andrew Pinski <apinski@marvell.com>
6264 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
6265 access prev before checking it for NULLness in the
6266 AARCH64_FUSE_CMP_BRANCH case.
6268 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
6271 * config/i386/i386.c (rest_of_insert_endbranch): Set
6272 endbr_queued_at_entrance to true and don't insert ENDBR if
6273 x86_function_profiler will be called.
6274 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
6276 * config/i386/i386.h (machine_function): Add
6277 endbr_queued_at_entrance.
6279 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
6281 * genattrtab.c (mk_attr_alt): Use alternative_mask.
6282 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
6284 (check_attr_test): Use alternative_mask.
6285 (get_attr_value): Likewise.
6286 (compute_alternative_mask): Use alternative_mask and XWINT.
6287 (make_alternative_compare): Use alternative_mask.
6288 (attr_alt_subset_p): Use XWINT.
6289 (attr_alt_subset_of_compl_p): Likewise.
6290 (attr_alt_intersection): Use alternative_mask and XWINT.
6291 (attr_alt_union): Likewise.
6292 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
6293 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
6294 (simplify_test_exp): Use alternative_mask and XWINT.
6295 (write_test_expr): Use alternative_mask and XWINT, adjust bit
6296 number calculation to support 64 bits. Generate code that
6297 checks 64-bit masks.
6298 (main): Use alternative_mask.
6299 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
6301 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
6304 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
6305 RETURN+USE when returning via %r14.
6307 2018-09-24 Martin Liska <mliska@suse.cz>
6309 * gcov.c (output_lines): Print colorization legend
6310 for both flag_use_colors and flag_use_hotness_colors.
6313 2018-09-24 Martin Liska <mliska@suse.cz>
6315 * coverage.c (get_coverage_counts): Use warning_at
6316 with current_function_decl location. Use %qD in warning
6319 2018-09-24 Martin Liska <mliska@suse.cz>
6321 * memory-block.h (memory_block_pool::release): Annotate with
6322 valgrind that the memory is not accessible.
6324 2018-09-24 Martin Liska <mliska@suse.cz>
6327 * asan.c: Make asan_handled_variables extern.
6329 * cfgexpand.c (expand_stack_vars): Make sure
6330 a representative is unpoison if another
6331 variable in the partition is handled by
6332 use-after-scope sanitization.
6334 2018-09-24 Richard Biener <rguenther@suse.de>
6336 PR tree-optimization/63155
6337 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
6338 the worklist when the edge of the respective argument isn't
6341 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6343 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
6344 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
6345 (MASK_CLASS_P): Update for rename.
6346 (MAYBE_MASK_CLASS_P): Ditto.
6347 (REG_CLASS_NAMES): Update.
6348 (REG_CLASS_CONTENT): Update.
6349 * config/i386/i386.c (regclass_map): Update for MASK_REG
6350 and ALL_MASK_REGS rename.
6351 * config/i386/constraints.md (Yk): Update for rename.
6354 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6356 * config/i386/i386.h (enum reg_class): Remove
6357 EVEX_SSE_REGS and MOD4_SSE_REGS.
6358 (REG_CLASS_NAMES): Update.
6359 (REG_CLASS_CONTENT): Update.
6360 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
6361 registers as ALL_SSE_REGS.
6362 (ix86_additional_allocno_class_p): Remove.
6363 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
6364 (ix86_register_priority): Lower priority of EVEX SSE registers.
6365 Use IN_RANGE macro where appropriate.
6366 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
6367 AVX-5124VNNIW checks.
6368 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
6369 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
6370 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
6371 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
6372 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
6373 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
6374 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
6375 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
6376 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
6377 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
6378 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
6379 * config/i386/constraints.md (Yh): Remove.
6381 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
6383 * config/i386/i386.c (regclass_map): Declare integer REX registers
6386 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
6388 * doc/service.texi (Service): Switch the fsf.org link to https.
6390 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
6393 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
6394 Define to speculation_safe_value_not_needed.
6396 2018-09-21 Florian Weimer <fweimer@redhat.com>
6399 * doc/extend.texi (Common Function Attributes): Mention that
6400 noreturn suppresses tail call optimization.
6402 2018-09-21 Jeff Law <law@redhat.com>
6404 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
6405 vr_values::cleanup_edges_and_switches.
6406 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
6408 (identify_jump_threads): Remove EDGE_IGNORE handling.
6409 (execute_vrp): Move handling of to_remove_edges and
6410 to_update_switch_stmts into vr_values class member functions.
6411 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
6412 (to_update_switch_stmts): Likewise.
6413 * vr-values.c: Include cfghooks.h.
6414 (vr_values::vr_values): Initialize to_remove_edges and
6415 to_update_switch_stmts.
6416 (vr_values::~vr_values): Verify to_remove_edges and
6417 to_update_switch_stmts are empty.
6418 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
6419 (vr_values::cleanup_edges_and_switches): New member function.
6420 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
6421 function. Add new data members.
6423 2018-09-21 David Malcolm <dmalcolm@redhat.com>
6425 PR tree-optimization/87309
6426 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
6427 calls with pflags and alt_flags.
6428 (selftest::test_capture_of_dump_calls): Add test of interaction of
6429 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
6431 2018-09-21 Olivier Hainque <hainque@adacore.com>
6433 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
6435 2018-09-21 Olivier Hainque <hainque@adacore.com>
6437 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
6439 2018-09-21 Olivier Hainque <hainque@adacore.com>
6441 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
6442 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
6443 (VXWORKS_LIBS_RTP): Minor reordering.
6445 2018-09-21 Olivier Hainque <hainque@adacore.com>
6447 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
6448 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
6450 2018-09-21 Olivier Hainque <hainque@adacore.com>
6452 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
6453 (PTRDIFF_TYPE): Likewise.
6455 2018-09-21 Olivier Hainque <hainque@adacore.com>
6457 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
6458 triplet, similar to support for VxWorks7.
6459 * config/vxworks-dummy.h: Provide a default definition
6460 of TARGET_VXWORKS64 to 0.
6462 2018-09-21 Olivier Hainque <hainque@adacore.com>
6464 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
6465 * config/vxworks-dummy.h: here.
6467 2018-09-21 Olivier Hainque <hainque@adacore.com>
6469 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
6471 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
6472 Bo Zhou <zbo.zhou@hisilicon.com>
6474 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
6475 * config/aarch64/aarch64-tune.md: Regenerated.
6476 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
6477 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
6478 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
6480 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
6481 Julian Brown <julian@codesourcery.com>
6483 * builtins.c (get_builtin_sync_mem): Handle address spaces.
6485 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6487 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
6488 if the call takes a static chain.
6490 2018-09-21 Martin Liska <mliska@suse.cz>
6492 * auto-profile.c (autofdo_source_profile::read): Do not
6494 (read_profile): Do not add working sets.
6495 (read_autofdo_file): Remove sum_all.
6496 (afdo_callsite_hot_enough_for_early_inline): Remove const
6498 * coverage.c (struct counts_entry): Remove gcov_summary.
6499 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
6500 do not support GCOV_TAG_PROGRAM_SUMMARY.
6501 (get_coverage_counts): Remove summary and expected
6503 * coverage.h (get_coverage_counts): Likewise.
6504 * doc/gcov-dump.texi: Remove -w option.
6505 * gcov-dump.c (dump_working_sets): Remove.
6506 (main): Do not support '-w' option.
6507 (print_usage): Likewise.
6508 (tag_summary): Likewise.
6509 * gcov-io.c (gcov_write_summary): Do not dump
6511 (gcov_read_summary): Likewise.
6512 (gcov_histo_index): Remove.
6513 (gcov_histogram_merge): Likewise.
6514 (compute_working_sets): Likewise.
6515 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
6517 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
6518 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
6519 (GCOV_HISTOGRAM_SIZE): Remove.
6520 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
6521 (struct gcov_summary): Simplify rapidly just
6522 to runs and sum_max fields.
6523 (gcov_histo_index): Remove.
6524 (NUM_GCOV_WORKING_SETS): Likewise.
6525 (compute_working_sets): Likewise.
6526 * gcov-tool.c (print_overlap_usage_message): Remove
6527 trailing empty line.
6528 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
6529 (output_lines): Remove program related line.
6530 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
6531 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
6533 (input_profile_summary): Do not read it.
6534 (merge_profile_summaries): And do not merge it.
6535 (input_symtab): Do not call removed function.
6536 * modulo-sched.c (sms_schedule): Do not print sum_max.
6537 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
6538 removed when histogram method was invented.
6539 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
6541 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
6543 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
6544 and dump selected value.
6545 * profile.c (add_working_set): Remove.
6546 (get_working_sets): Likewise.
6547 (find_working_set): Likewise.
6548 (get_exec_counts): Do not work with working sets.
6549 (read_profile_edge_counts): Do not inform as sum_max is removed.
6550 (compute_branch_probabilities): Likewise.
6551 (compute_value_histograms): Remove argument for call of
6552 get_coverage_counts.
6553 * profile.h: Do not make gcov_summary const.
6555 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
6557 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
6559 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
6561 PR tree-optimization/86990
6562 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
6563 Check that the entire merged store group is made of constants only for
6566 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
6568 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
6569 (VTABLE_VERIFICATION_SPEC): Likewise.
6570 (SANITIZER_EARLY_SPEC): Likewise.
6571 (SANITIZER_SPEC): Likewise.
6572 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
6573 * doc/invoke.texi (Link Options): Document -r.
6575 2018-09-20 Richard Biener <rguenther@suse.de>
6578 * gimplify.c (gimplify_expr): Retain alignment of
6579 addressable lvalue in dereference.
6581 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
6584 * configure.ac: Check for .loc is_stmt support.
6585 * configure, config.in: Rebuilt.
6586 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
6589 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6591 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
6592 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
6595 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6597 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
6599 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
6602 2018-09-20 Marek Polacek <polacek@redhat.com>
6604 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
6606 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
6608 PR tree-optimization/87288
6609 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
6610 into account when determining PEELING_FOR_NITERS.
6612 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
6614 PR tree-optimization/86877
6615 * tree-vect-loop.c (vect_analyze_loop_2): Call
6616 vect_verify_datarefs_alignment.
6618 2018-09-19 Marek Polacek <polacek@redhat.com>
6620 * doc/invoke.texi: Document -Wclass-conversion.
6622 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
6624 * config/pa/pa.c (pa_adjust_priority): Delete.
6625 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
6627 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
6628 (atomic_storehi): Likewise.
6629 (atomic_storesi): Likewise.
6630 (atomic_loaddi): Restore compare and swap exchange loop code.
6632 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6634 PR rtl-optimization/86902
6635 * combine.c (try_combine): When changing the CC mode used, don't change
6636 an unrelated mode in other_insn to that new CC mode.
6638 2018-09-19 David Malcolm <dmalcolm@redhat.com>
6640 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
6641 with %T in place of calls to dump_generic_expr.
6642 (prune_runtime_alias_test_list): Likewise.
6643 (create_runtime_alias_checks): Likewise.
6644 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
6645 (vect_analyze_data_ref_dependence): Likewise.
6646 (vect_slp_analyze_data_ref_dependence): Likewise.
6647 (vect_record_base_alignment): Likewise. Use %G in place of call
6648 to dump_gimple_stmt.
6649 (vect_compute_data_ref_alignment): Likewise.
6650 (verify_data_ref_alignment): Likewise.
6651 (vect_find_same_alignment_drs): Likewise.
6652 (vect_analyze_group_access_1): Likewise.
6653 (vect_analyze_data_ref_accesses): Likewise.
6654 (dependence_distance_ge_vf): Likewise.
6655 (dump_lower_bound): Likewise.
6656 (vect_prune_runtime_alias_test_list): Likewise.
6657 (vect_find_stmt_data_reference): Likewise.
6658 (vect_analyze_data_refs): Likewise.
6659 (vect_create_addr_base_for_vector_ref): Likewise.
6660 (vect_create_data_ref_ptr): Likewise.
6661 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
6662 (vect_can_advance_ivs_p): Likewise.
6663 (vect_update_ivs_after_vectorizer): Likewise.
6664 (vect_gen_prolog_loop_niters): Likewise.
6665 (vect_prepare_for_masked_peels): Likewise.
6666 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
6667 (vect_determine_vectorization_factor): Likewise.
6668 (vect_is_simple_iv_evolution): Likewise.
6669 (vect_analyze_scalar_cycles_1): Likewise.
6670 (vect_analyze_loop_operations): Likewise.
6671 (report_vect_op): Likewise.
6672 (vect_is_slp_reduction): Likewise.
6673 (check_reduction_path): Likewise.
6674 (vect_is_simple_reduction): Likewise.
6675 (vect_create_epilog_for_reduction): Likewise.
6676 (vect_finalize_reduction:): Likewise.
6677 (vectorizable_induction): Likewise.
6678 (vect_transform_loop_stmt): Likewise.
6679 (vect_transform_loop): Likewise.
6680 (optimize_mask_stores): Likewise.
6681 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
6682 (vect_split_statement): Likewise.
6683 (vect_recog_over_widening_pattern): Likewise.
6684 (vect_recog_average_pattern): Likewise.
6685 (vect_determine_min_output_precision_1): Likewise.
6686 (vect_determine_precisions_from_range): Likewise.
6687 (vect_determine_precisions_from_users): Likewise.
6688 (vect_mark_pattern_stmts): Likewise.
6689 (vect_pattern_recog_1): Likewise.
6690 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
6691 (vect_record_max_nunits): Likewise.
6692 (vect_build_slp_tree_1): Likewise.
6693 (vect_build_slp_tree_2): Likewise.
6694 (vect_print_slp_tree): Likewise.
6695 (vect_analyze_slp_instance): Likewise.
6696 (vect_detect_hybrid_slp_stmts): Likewise.
6697 (vect_detect_hybrid_slp_1): Likewise.
6698 (vect_slp_analyze_operations): Likewise.
6699 (vect_slp_analyze_bb_1): Likewise.
6700 (vect_transform_slp_perm_load): Likewise.
6701 (vect_schedule_slp_instance): Likewise.
6702 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
6703 (vect_mark_stmts_to_be_vectorized): Likewise.
6704 (vect_init_vector_1): Likewise.
6705 (vect_get_vec_def_for_operand): Likewise.
6706 (vect_finish_stmt_generation_1): Likewise.
6707 (vect_check_load_store_mask): Likewise.
6708 (vectorizable_call): Likewise.
6709 (vectorizable_conversion): Likewise.
6710 (vectorizable_operation): Likewise.
6711 (vectorizable_load): Likewise.
6712 (vect_analyze_stmt): Likewise.
6713 (vect_is_simple_use): Likewise.
6714 (vect_get_vector_types_for_stmt): Likewise.
6715 (vect_get_mask_type_for_stmt): Likewise.
6716 * tree-vectorizer.c (increase_alignment): Likewise.
6718 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
6720 * doc/rtl.texi: Adjust vec_select description.
6721 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
6722 non-constant selectors.
6724 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
6726 * config/aarch64/aarch64-protos.h
6727 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
6728 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
6729 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
6730 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
6731 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
6732 (AARCH64_FL_PROFILE): Move index so flags are ordered.
6733 (AARCH64_ISA_RCPC8_4): New flag.
6734 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
6735 to aarch64_offset_9bit_signed_unscaled_p.
6736 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
6738 * config/aarch64/constraints.md (Ust): New constraint.
6739 * config/aarch64/predicates.md.
6740 (aarch64_9bit_offset_memory_operand): New predicate.
6741 (aarch64_rcpc_memory_operand): New predicate.
6743 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
6745 PR rtl-optimization/87361
6746 * rtlanal.c (nonzero_bits1): Revert accidental change.
6748 2018-09-19 Richard Biener <rguenther@suse.de>
6750 PR tree-optimization/87349
6751 PR tree-optimization/87342
6752 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
6754 2018-09-18 Marek Polacek <polacek@redhat.com>
6756 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
6757 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
6759 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
6761 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
6763 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
6765 PR rtl-optimization/86882
6766 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
6768 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
6770 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
6771 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
6773 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
6776 * doc/invoke.texi (Link Options): Fix formatting and grammar.
6778 2018-09-18 Richard Biener <rguenther@suse.de>
6781 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
6782 (compute_samebase_partition_bases): Likewise.
6783 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
6784 (gimple_can_coalesce_p): Simplify.
6786 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
6788 Handle a library implementation of ffs calling __builtin_ffs.
6789 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
6790 (mmix_init_libfuncs): New function: make __builtin_ffs expand
6793 2018-09-17 David Malcolm <dmalcolm@redhat.com>
6795 * diagnostic-show-locus.c (class layout_range): Add field
6797 (layout_range::layout_range): Add "original_idx" param and use it
6798 to initialize new field.
6799 (make_range): Use 0 for original_idx.
6800 (layout::layout): Pass in index to calls to
6801 maybe_add_location_range.
6802 (layout::maybe_add_location_range): Add param "original_idx" and
6803 pass it on to layout_range.
6804 (layout::print_any_labels): Pass on range->m_original_idx to
6806 (gcc_rich_location::add_location_if_nearby): Use 0 for
6808 * gcc-rich-location.h (text_range_label::get_text): Update for new
6810 (range_label_for_type_mismatch::get_text): Likewise.
6812 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
6814 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
6816 2018-09-17 David Malcolm <dmalcolm@redhat.com>
6818 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
6819 format_string_diagnostic_t.
6820 (fmtwarn_n): Likewise.
6821 * substring-locations.c
6822 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
6823 (format_warning_n_va): Convert to...
6824 (format_string_diagnostic_t::emit_warning_n_va): ...this.
6825 (format_warning_va): Convert to...
6826 (format_string_diagnostic_t::emit_warning_va): ...this.
6827 (format_warning_at_substring): Convert to...
6828 (format_string_diagnostic_t::emit_warning): ...this.
6829 (format_warning_at_substring_n): Convert to...
6830 (format_string_diagnostic_t::emit_warning_n): ...this.
6831 * substring-locations.h (class format_string_diagnostic_t): New
6833 (format_warning_va): Convert to
6834 format_string_diagnostic_t::emit_warning_va.
6835 (format_warning_n_va): Convert to
6836 format_string_diagnostic_t::emit_warning_n_va.
6837 (format_warning_at_substring): Convert to
6838 format_string_diagnostic_t::emit_warning.
6839 (format_warning_at_substring_n): Convert to
6840 format_string_diagnostic_t::emit_warning_n.
6842 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
6843 Bernd Schmidt <bernds_cb1@t-online.de>
6845 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
6848 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
6850 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
6851 operand 0 predicate to nonimmediate operand.
6852 (rint<mode>2_frndint): Remove insn pattern.
6853 (rint<mode>2): Change operand 1 predicate to general_operand.
6854 Extend operand 1 to XFmode and generate rintxf2 insn.
6855 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
6856 Do not use X87MODEF mode macro.
6857 (frndintxf2_<rounding>_i387): Rename from
6858 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
6859 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
6860 to XFmode and generate significandxf3 insn.
6862 2018-09-17 Richard Biener <rguenther@suse.de>
6864 PR tree-optimization/87328
6865 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
6866 visiting unexecutable backedges when not iterating.
6867 (do_rpo_vn): Mark all edges not executable even when not
6870 2018-09-17 Martin Jambor <mjambor@suse.cz>
6873 * doc/invoke.texi (Warning Options): Likewise.
6875 2018-09-17 Richard Biener <rguenther@suse.de>
6877 PR tree-optimization/87301
6878 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
6879 clean EH info from leftover copy assignments.
6881 2018-09-17 Martin Liska <mliska@suse.cz>
6883 PR gcov-profile/85871
6884 * gcov.c (output_intermediate_file): Fix out of bounds
6887 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
6889 * config/arc/arc.c: Object attributes for core4 not reflected
6891 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
6894 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
6896 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
6898 2018-09-17 Martin Liska <mliska@suse.cz>
6900 * doc/gcov.texi: Document new option --use-hotness-colors.
6901 * gcov.c (struct source_info): Declare new field.
6902 (source_info::source_info): Set default for maximum_count.
6903 (print_usage): Add new -q option.
6904 (process_args): Process it.
6905 (accumulate_line_info): Save src->maximum_count.
6906 (output_line_beginning): Make color line number if
6907 flag_use_hotness_colors is set.
6908 (output_line_details): Pass default argument value.
6909 (output_lines): Pass src->maximum_count.
6911 2018-09-17 Martin Liska <mliska@suse.cz>
6913 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
6914 Use processor_names table.
6915 * config/i386/i386.c (ix86_default_align): Use
6916 processor_cost_table for alignment values.
6917 (ix86_option_override_internal): Use processor_names.
6918 (ix86_function_specific_print): Likewise.
6919 * config/i386/i386.h (struct processor_costs):
6920 Add alignment values.
6921 (struct ptt): Remove and replace with const char *.
6922 * config/i386/x86-tune-costs.h (struct processor_costs):
6923 Declare default alignments for all costs.
6925 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
6927 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
6928 symbolics or VR_VARYING ranges for ABS_EXPR.
6929 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
6930 when range will wrap.
6932 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
6935 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
6936 before and after a JUMP_TABLE_DATA.
6938 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
6941 * dojump.c (do_compare_and_jump): Canonicalize function pointers
6942 when one operand is a function pointer. Use POINTER_TYPE_P and
6943 FUNC_OR_METHOD_TYPE_P.
6944 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
6945 * fold-const.c (build_range_check): Likewise.
6946 * match.pd (simple_comparison): Likewise.
6948 2018-09-14 David Malcolm <dmalcolm@redhat.com>
6951 * spellcheck.c (get_edit_distance_cutoff): New function.
6952 (selftest::test_edit_distance_unit_test_oneway): Rename to...
6953 (selftest::test_get_edit_distance_one_way): ...this.
6954 (selftest::test_get_edit_distance_unit): Rename to...
6955 (selftest::test_get_edit_distance_both_ways): ...this.
6956 (selftest::test_edit_distances): Move tests to this new function,
6957 and test some more pairs of strings. Update for above renaming.
6958 (selftest::get_old_cutoff): New function.
6959 (selftest::test_get_edit_distance_cutoff): New function.
6960 (selftest::assert_suggested_for): New function.
6961 (ASSERT_SUGGESTED_FOR): New macro.
6962 (selftest::assert_not_suggested_for): New function.
6963 (ASSERT_NOT_SUGGESTED_FOR): New macro.
6964 (selftest::test_suggestions): New function.
6965 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
6966 tests to selftest::test_edit_distances and call it. Add calls to
6967 selftest::test_get_edit_distance_cutoff and
6968 selftest::test_suggestions.
6969 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
6970 (best_match::consider): Replace hard-coded cutoff calculation with
6972 (best_match::get_cutoff): New declaration.
6973 (best_match::get_best_meaningful_candidate): Likewise.
6975 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
6977 * builtins.c (fold_builtin_strlen): Remove TODO comment.
6979 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
6982 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
6984 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
6985 terminated string literal.
6987 2018-09-14 Martin Sebor <msebor@redhat.com>
6989 * builtins.c (unterminated_array): Handle ARRAY_REF.
6990 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
6991 * builtins.h (unterminated_array): Declare extern.
6992 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
6994 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
6997 2018-09-14 Martin Sebor <msebor@redhat.com>
6998 Jeff Law <law@redhat.com>
7000 * builtins.c (unterminated_array): New.
7001 (expand_builtin_strcpy): Adjust.
7002 (expand_builtin_strcpy_args): Detect unterminated arrays.
7003 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
7004 unterminated arrays.
7005 * gimple-fold.h (get_maxval_strlen): Add argument.
7006 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
7008 * gimple-fold.c (get_range_strlen): Add argument.
7009 (get_maxval_strlen): Adjust.
7010 * gimple-fold.h (get_range_strlen): Add argument.
7012 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
7014 * config/i386/movdirintrin.h: Fix copyright year.
7016 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
7018 * reg-stack.c: Include regs.h.
7019 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
7020 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
7021 FIRST_STACK_REG, not DFmode.
7022 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
7023 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
7024 (change stack): Default register mode to the reg_raw_mode of
7025 FIRST_STACK_REG, not DFmode.
7026 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
7027 (*swapxf): Rename from swapxf.
7029 2018-09-14 Carl Love <cel@us.ibm.com>
7031 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
7032 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
7034 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
7037 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
7040 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
7043 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
7045 2018-09-14 Jason Merrill <jason@redhat.com>
7047 Fix --enable-gather-detailed-mem-stats.
7048 * hash-table.c (hash_table_usage): Change from variable to function.
7049 * hash-table.h: Adjust.
7050 * Makefile.in: Add missing dependencies on hash-table.h.
7052 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7054 PR tree-optimization/87259
7056 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
7057 execute_cse_reciprocals_1 has tried transforming.
7059 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
7061 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
7062 VR_VARYING for PLUS/MINUS_EXPR.
7064 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
7066 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
7069 2018-09-14 Richard Biener <rguenther@suse.de>
7072 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
7073 bits for the merged partition.
7075 2018-09-13 Martin Sebor <msebor@redhat.com>
7076 Bernd Edlinger <bernd.edlinger@hotmail.de>
7078 * builtins.h (c_srlen): Add argument.
7079 * builtins.c (warn_string_no_nul): New function.
7080 (c_strlen): Add argument and use it. Update recursive calls.
7081 Pass DECL argument to string_constant to get info on non
7082 terminated strings. Update *NONSTR as needed.
7083 (fold_builtin_strlen): Add argument to calls to c_strlen.
7084 Warn for unterminated arrays.
7085 (warn_string_no_null): Add prototype.
7086 * expr.c (string_constant): Update arguments. Update recursive
7087 calls appropriately. Detect missing NUL terminator and outermost
7088 declaration its missing in.
7089 Improve checks for arrays with nonzero lower bound or elements
7090 that are not a single byte. Simplify offset computation.
7091 Simplify checks for non-NUL terminated strings.
7092 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
7093 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
7095 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
7097 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
7099 * fold-const.c (c_getstr): Fix function comment. Remove unused third
7100 argument. Fix range checks.
7101 * fold-const.h (c_getstr): Adjust protoype.
7102 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
7103 string is constant but contains no NUL byte.
7105 * expr.c (string_constant): Adjust function comment.
7106 Remove bogus check for zero termination.
7108 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
7110 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
7111 (get_constant_size): Don't make STRING_CSTs larger than they are.
7112 (check_string_literal): New check function for STRING_CSTs.
7113 (output_constant): Use it.
7115 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
7118 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
7120 2018-09-13 Richard Biener <rguenther@suse.de>
7122 PR tree-optimization/87263
7123 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
7124 (struct unwind_state): Add max_rpo field.
7125 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
7126 Compute max_rpo, the max RPO number a block can be backwards reached
7127 from. Re-write non-iterating mode to a RPO ordered worklist approach,
7128 separating it from the iterating mode.
7130 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
7132 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
7133 (rfs_decision): New scheduling decision.
7135 2018-09-13 Richard Biener <rguenther@suse.de>
7138 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
7139 (vn_nary_op_insert_pieces_predicated): Do not write useless
7140 valid_dominated_by_p entry outside of the allocated storage.
7142 2018-09-13 Omar Sandoval <osandov@osandov.com>
7143 Tom de Vries <tdevries@suse.de>
7146 * dwarf2out.c (is_c): New function.
7147 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
7149 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
7152 * config/aarch64/aarch64.md (*aarch64_bfxil):
7154 * config/aarch64/constraints.md (Ulc): Define.
7155 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
7157 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
7160 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
7162 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
7163 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
7164 aarch64_layout_frame call.
7165 (aarch64_expand_epilogue): Likewise.
7166 (aarch64_initial_elimination_offset): Likewise.
7167 (aarch64_get_separate_components): Likewise.
7168 (aarch64_use_return_insn_p): Likewise.
7169 (aarch64_layout_frame): Remove unneeded check.
7171 2018-09-13 Jakub Jelinek <jakub@redhat.com>
7173 * configure.ac: Only append
7174 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
7175 gcc_config_arguments if it was never reconfigured or last reconfigure
7176 was with different arguments.
7177 * configure: Regenerated.
7179 2018-09-13 Jakub Jelinek <jakub@redhat.com>
7180 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7183 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
7184 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
7186 2018-09-13 Jakub Jelinek <jakub@redhat.com>
7188 PR tree-optimization/87287
7189 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
7190 X % C == 0 to X % (unsigned) C == 0 optimization to ...
7191 * match.pd (X % C == 0): ... here. New optimization.
7193 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7196 * expr.h (maybe_optimize_mod_cmp): Declare.
7197 * expr.c (mod_inv): New function.
7198 (maybe_optimize_mod_cmp): New function.
7199 (do_store_flag): Use it.
7200 * cfgexpand.c (expand_gimple_cond): Likewise.
7202 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
7203 Julian Brown <julian@codesourcery.com>
7206 * gimplify.c (gimplify_scan_omp_clauses): Set
7207 target_firstprivatize_array_bases in OpenACC parallel and kernels
7208 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
7209 OpenACC data regions.
7211 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
7213 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
7214 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
7216 2018-09-12 Richard Biener <rguenther@suse.de>
7218 PR tree-optimization/87280
7219 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
7220 edge but unreachable target.
7221 (do_rpo_vn): For conservatively handling a PHI only mark
7222 the backedge executable but not the block reachable.
7224 2018-09-12 Richard Biener <rguenther@suse.de>
7226 PR tree-optimization/87266
7227 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
7230 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
7232 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
7234 ("trunc<BFP:mode><DFP_ALL:mode>2")
7235 ("trunc<DFP_ALL:mode><BFP:mode>2")
7236 ("extend<BFP:mode><DFP_ALL:mode>2")
7237 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
7238 according to the target operand type.
7240 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7241 Andreas Krebbel <krebbel@linux.ibm.com>
7243 PR tree-optimization/86844
7244 * gimple-ssa-store-merging.c
7245 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
7246 there are any overlapping stores in between them, make sure they are
7247 also coalesced or we give up completely.
7249 2018-09-12 Jakub Jelinek <jakub@redhat.com>
7252 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
7253 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
7255 2018-09-12 Tom de Vries <tdevries@suse.de>
7257 * common.opt (gdescribe-dies): Add option.
7258 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
7259 attribute for artifical and nameless decls.
7260 (dwarf2out_register_external_die): Add description attribute to
7261 external reference die.
7262 (add_desc_attribute): New functions.
7263 (gen_subprogram_die): Add description attribute to
7264 DW_TAG_call_site_parameter.
7265 * tree-pretty-print.c (print_generic_expr_to_str): New function.
7266 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
7267 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
7269 (@item -gdescribe-dies): Add.
7271 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
7273 * tree-vrp.c (vrp_shift_undefined_p): Remove.
7274 (extract_range_from_binary_expr_1: Call
7275 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
7276 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
7279 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
7281 * gimple-ssa-warn-alloca.c
7282 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
7283 field for ALLOCA_BOUND_*_LARGE.
7285 2018-09-11 Nathan Sidwell <nathan@acm.org>
7287 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
7289 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7291 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
7292 for clobbers. Remove obsolete comment.
7294 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7296 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
7297 mpxchk, mpxld and mpxst types.
7298 (define_attr length_immediate): Remove all processing of mpx types.
7299 (define_attr prefix_0f): Ditto.
7300 (define_attr memory): Ditto.
7302 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
7304 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
7305 (log<mode>2): Change operand 1 predicate to general_operand.
7306 Extend operand 1 to XFmode and generate logxf3 insn.
7307 (log10<mode>2): Change operand 1 predicate to general_operand.
7308 Extend operand 1 to XFmode and generate log10xf3 insn.
7309 (log2<mode>2): Change operand 1 predicate to general_operand.
7310 Extend operand 1 to XFmode and generate log2xf3 insn.
7311 (fyl2xp1_extend<mode>xf3_i387): Remove.
7312 (log1p<mode>2): Change operand 1 predicate to general_operand.
7313 Extend operand 1 to XFmode and generate log1pxf3 insn.
7314 (fxtract_extend<mode>xf3_i387): Remove.
7315 (logb<mode>2): Change operand 1 predicate to general_operand.
7316 Extend operand 1 to XFmode and generate logbxf3 insn.
7317 (ilogb<mode>2): Change operand 1 predicate to general_operand.
7318 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
7319 (significand<mode>2): Change operand 1 predicate to general_operand.
7320 Extend operand 1 to XFmode and generate significandxf3 insn.
7322 2018-09-11 Nathan Sidwell <nathan@acm.org>
7324 * gcc.c (perror_with_name, pfatal_with_name): Delete.
7325 (load_specs): Use fatal_error.
7326 (DELETE_IF_ORDINARY, process_command): Use error.
7327 (execute, run_attempt): Use fatal_error.
7329 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
7331 * diagnostic-core.h (sorry_at): New prototype.
7332 * diagnostic.c (sorry_at): New function.
7334 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
7336 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
7337 by zero as VR_UNDEFINED.
7339 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
7341 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
7342 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
7343 (<sincos>mode2): New expander.
7344 (sincos_extend<mode>xf3_i387): Remove insn pattern.
7345 (sincos -> sin, cos splitters): Remove splitter patterns.
7346 (sincos<mode>3): Change operand 2 predicate to general_operand.
7347 Extend operand 2 to XFmode and generate sincosxf3 insn.
7348 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
7349 Change operand 3 predicate to const1_operand.
7350 (fptan_extend<mode>xf4_i387): Remove insn pattern.
7351 (tanxf2): Update operands in the call to fptanxf4_i387.
7352 (tan<mode>2): Change operand 1 predicate to general_operand.
7353 Extend operand 1 to XFmode and generate tanxf3 insn.
7354 (atan2xf3): Rename from *fpatanxf3_i387.
7355 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
7356 (atan2xf3): Remove expander.
7357 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
7358 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
7359 (atan<mode>2): Change operand 1 predicate to general_operand.
7360 Extend operand 1 to XFmode and generate atanxf3 insn.
7362 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
7364 * config/i386/i386.md (x87/SSE constant load splitter): Use
7365 memory_operand instead of nonimmediate_operand for input operand
7368 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
7370 * config/i386/i386.md (float partial SSE register stall splitter): Move
7371 splitter near its instruction pattern.
7372 (float_extend partial SSE register stall splitter): Ditto.
7373 (float_truncate partial SSE register stall splitter): Ditto.
7375 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
7378 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7379 to speculation_safe_value_not_needed.
7382 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
7383 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
7384 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
7385 leaf_function_p, instead use has_hard_reg_initial_val.
7387 2018-09-09 Nathan Sidwell <nathan@acm.org>
7389 * gcc.h (pfatal_with_name): Don't declare here.
7390 * gcc.c (pfatal_with_name): Make static.
7392 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
7394 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
7397 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
7399 PR rtl-optimization/85458
7400 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
7401 priority hook to reduce the priority of EXPR.
7403 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
7405 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
7406 DImode for x87 on 32bit targets. Conditionally disable x87 modes
7407 with X87_ENABLE_FLOAT. Remove preparation code.
7408 (*float<SWI48:mode><MODEF:mode>2): Rename from
7409 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
7410 math using "enabled" attribute.
7411 (*floatdi<MODEF:mode>2_i387): Rename from
7412 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
7413 enable for 32bit targets only.
7414 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
7416 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
7417 as operand 1 predicate. Rewrite as define_insn_and_split.
7418 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
7420 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
7422 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
7423 to fallthru to FLOAT case.
7425 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7428 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
7429 around folding of vec_sl to handle out of range shift values.
7431 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
7433 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
7434 Update callers to gen_fix_trunc<mode>_i387_fisttp
7435 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
7436 nonimmediate_operand.
7437 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
7438 and corresponding splitters.
7439 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
7440 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
7441 (fix_truncdi_i387_with_temp): Remove insn pattern
7442 and corresponding splitters.
7443 (fix_trunc<mode>_i387): Change operand 0 predicate to
7444 nonimmediate_operand.
7445 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
7446 and corresponding splitters.
7447 (*fistdi2_1): Remove.
7449 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
7450 (lrintxfdi2): Remove expander. Reimplement as define_insn.
7451 (*fist<mode>2_1): Remove.
7452 (fist<mode>2): Ditto.
7453 (fist<mode>2_with_temp): Remove insn pattern and corresponding
7455 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
7456 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
7457 (fistdi2_<rounding>): Change operand 0 predicate to
7458 nonimmediate_operand.
7459 (fistdi2_<rounding>_with_temp): Remove insn pattern
7460 and corresponding splitters.
7461 (fist<mode>2_<rounding>): Change operand 0 predicate to
7462 nonimmediate_operand.
7463 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
7464 and corresponding splitters.
7466 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
7468 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7470 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
7473 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7475 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7476 early gimple folding of vec_splat().
7477 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
7478 * gimple-fold.h: Add an extern define for tree_vec_extract().
7480 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
7482 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
7483 wrappers around TREE_TYPE comparisons.
7485 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7488 * config/s390/predicates.md: Add nonsym_memory_operand.
7489 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
7490 contains a SYMBOL_REF, load it into an intermediate pseudo.
7491 (s390_emit_compare_and_swap): Legitimize operand.
7492 * config/s390/s390.md: Use the new nonsym_memory_operand
7493 with UNSPECV_CAS patterns.
7495 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7498 * config/s390/s390-passes.def: New file.
7499 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
7501 (make_pass_s390_early_mach): Add declaration.
7502 * config/s390/s390.c (make_pass_s390_early_mach):
7503 (s390_option_override): Remove dynamic registration.
7504 * config/s390/t-s390: Add s390-passes.def.
7506 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
7508 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7509 Remove UNSPEC_LTREL_BASE check.
7510 (annotate_constant_pool_refs): Likewise.
7511 (find_constant_pool_ref): Likewise.
7512 (find_ltrel_base): Removed.
7513 (replace_ltrel_base): Removed.
7514 (s390_mainpool_finish): Remove replace_ltrel_base call.
7515 (s390_chunkify_start): Remove pending LTREL_BASE logic.
7516 (s390_chunkify_finish): Remove replace_ltrel_base call.
7517 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
7519 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
7522 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
7523 to speculation_safe_value_not_needed.
7525 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
7526 Bernd Schmidt <bernds_cb1@t-online.de>
7528 * config/nvptx/nvptx-opts.h: New file.
7529 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
7530 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
7532 (TARGET_SM35): New macro.
7533 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
7535 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
7537 (misa=): New option.
7538 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
7540 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
7542 * config/i386/i386.md (truncdfsf2): Remove expander.
7543 (truncdfsf2_with_temp): Ditto.
7544 (truncxf<mode>2): Ditto.
7545 (*truncdfsf_fast_mixed): Remove insn pattern.
7546 (*truncdfsf_fast_i387): Ditto.
7547 (*truncdfsf_mixed): Ditto.
7548 (*truncdfsf_i387): Ditto.
7549 (*truncdfsf2_i387_1): Ditto.
7550 (*truncxfsf2_mixed): Ditto.
7551 (*truncxfdf2_mixed): Ditto.
7552 (*truncxf<mode>2_i387_noop): Ditto. Update callers
7553 to call gen_truncxf<mode>2 instead.
7554 (*truncxf<mode>2_i387): Remove.
7555 (reg->reg splitters): Remove splitter pattern.
7556 (reg->mem splitters): Ditto.
7558 (truncdfsf2): New insn pattern.
7559 (truncxf<mode>2): Ditto.
7561 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7563 * tree-ssa-math-opts.c (is_mult_by): New function.
7564 (is_square_of): Use the above.
7565 (optimize_recip_sqrt): New function.
7566 (pass_cse_reciprocals::execute): Use the above.
7568 2018-09-05 Richard Biener <rguenther@suse.de>
7571 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
7572 to zero-init the emplaced vec.
7574 2018-09-05 Martin Liska <mliska@suse.cz>
7576 PR tree-optimization/87205
7577 * tree-switch-conversion.c (pass_lower_switch::execute):
7578 Group cases for switch statements.
7580 2018-09-05 Richard Biener <rguenther@suse.de>
7582 PR tree-optimization/87217
7583 * tree-ssa-sccvn.c (vuse_valueize): New.
7584 (vn_reference_lookup_pieces): Use it.
7585 (vn_reference_lookup): Likewise.
7587 2018-09-05 Nathan Sidwell <nathan@acm.org>
7590 * stor-layout.c (place_field): Scan forwards to check last
7591 bitfield when ms_bitfield_placement is in effect.
7593 2018-09-05 Richard Biener <rguenther@suse.de>
7596 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
7599 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
7600 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
7602 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
7603 * config.gcc (extra_objs): Build it.
7604 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
7606 * config/aarch64/aarch64-passes.def
7607 (pass_tag_collision_avoidance): New pass.
7608 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
7609 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
7610 (aarch64_classify_address): Remove static qualifier.
7611 (aarch64_address_info, aarch64_address_type): Move to...
7612 * config/aarch64/aarch64-protos.h: ... here.
7613 (make_pass_tag_collision_avoidance): New function.
7614 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
7617 2018-09-05 Martin Liska <mliska@suse.cz>
7619 * doc/gcov.texi: Update documentation of humar
7621 * gcov.c (format_count): Print one decimal place, it provides
7622 more fine number of situations like '1G' vs. '1.4G'.
7624 2018-09-05 Martin Liska <mliska@suse.cz>
7627 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
7628 * optc-gen.awk: Allow 'Var' for Deprecated options in order
7629 to generate a MASK value.
7631 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
7634 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
7635 if frame pointer isn't used.
7636 (compute_frame_pointer_to_fb_displacement): Likewise.
7638 2018-09-04 Jakub Jelinek <jakub@redhat.com>
7641 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
7642 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
7643 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
7644 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
7645 and OPTION_MASK_ISA_XSAVEC_UNSET.
7647 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
7649 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
7650 XOR operations in NAND case.
7652 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
7654 * wide-int-range.cc (wide_int_range_convert): New.
7655 * wide-int-range.h (wide_int_range_convert): New.
7656 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
7657 code into wide_int_range_convert.
7658 (extract_range_into_wide_ints): Do not munge anti range constants
7659 into the entire domain. Just return the range back.
7661 2018-09-04 Martin Liska <mliska@suse.cz>
7663 * genmatch.c (output_line_directive): Add new argument
7665 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
7667 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
7669 * doc/invoke.texi (Option Summary): Add whitespace.
7671 * doc/invoke.texi (Option Summary): Add -Waligned-new.
7673 2018-09-04 Richard Biener <rguenther@suse.de>
7675 PR tree-optimization/87211
7676 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
7677 backedge value we're supposed to treat as VARYING also number
7678 the PHI to VARYING in case it got a different value-number already.
7680 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
7682 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
7683 (extract_range_from_binary_expr_1): Do not call
7684 vrp_can_optimize_bit_op.
7685 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
7687 (wide_int_range_get_mask_and_bounds): New.
7688 (wide_int_range_optimize_bit_op): New.
7689 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
7690 (wide_int_range_bit_and): Same.
7691 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
7692 (wide_int_range_optimize_bit_op): New.
7693 (wide_int_range_get_mask_and_bounds): New.
7695 2018-09-04 Richard Biener <rguenther@suse.de>
7697 PR tree-optimization/87176
7698 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
7699 variable. When value-numbering a virtual PHI node make sure
7700 to not value-number to the backedge value.
7702 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
7704 * doc/extend.texi (Long Long, Hex Floats): Document support for
7705 long long and hex floats in more recent versions of ISO C++.
7707 2018-09-03 Richard Biener <rguenther@suse.de>
7709 PR tree-optimization/87177
7710 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
7713 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7715 * bb-reorder.c (edge_order): Convert to C-qsort-style
7716 tri-state comparator.
7717 (reorder_basic_blocks_simple): Change std::stable_sort to
7720 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7722 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
7723 tri-state comparator.
7724 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
7726 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7728 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
7729 (mergesort): ... here as maximum count for using netsort.
7730 (gcc_qsort): Set nlim to 3 if stable sort is requested.
7731 (gcc_stablesort): New.
7732 * system.h (gcc_stablesort): Declare.
7734 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
7736 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
7737 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
7738 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
7740 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
7742 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
7743 lxsdx and stxsdx alternatives.
7744 (*mov<mode>_hardfloat64): Ditto.
7745 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
7747 2018-09-03 Richard Biener <rguenther@suse.de>
7749 PR tree-optimization/87200
7750 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
7753 2018-09-03 Martin Liska <mliska@suse.cz>
7755 PR tree-optimization/87201
7756 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
7757 Fix parenthesis in an expression.
7759 2018-09-03 Richard Biener <rguenther@suse.de>
7761 PR tree-optimization/87197
7762 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
7763 visited. CSE the VN_INFO hashtable lookup.
7765 PR tree-optimization/87169
7766 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
7767 iterating make sure there's no extra backedges from irreducible
7768 regions feeding the header. Mark the destination block
7771 2018-09-03 Martin Liska <mliska@suse.cz>
7774 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
7775 * common/common-targhooks.c (default_get_valid_option_values):
7777 * common/common-targhooks.h (default_get_valid_option_values):
7779 * common/config/i386/i386-common.c: Move processor_target_table
7781 (ix86_get_valid_option_values): New function.
7782 (TARGET_GET_VALID_OPTION_VALUES): New macro.
7783 * config/i386/i386.c (struct ptt): Move to i386-common.c.
7784 (PTA_*): Move all defined masks into i386-common.c.
7785 (ix86_function_specific_restore): Use new processor_cost_table.
7786 * config/i386/i386.h (struct ptt): Moved from i386.c.
7787 (struct pta): Likewise.
7788 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
7789 * doc/tm.texi.in: Likewise.
7790 * opt-suggestions.c (option_proposer::suggest_option):
7791 Pass prefix to build_option_suggestions.
7792 (option_proposer::get_completions): Likewise.
7793 (option_proposer::build_option_suggestions): Use the new target
7795 * opts.c (struct option_help_tuple): New struct.
7796 (print_filtered_help): Use the new target hook.
7798 2018-09-03 Martin Liska <mliska@suse.cz>
7801 * predict.c (set_even_probabilities): Add likely_edges
7802 argument and handle cases where we have precisely one
7804 (combine_predictions_for_bb): Catch also likely_edges.
7805 (tree_predict_by_opcode): Handle gswitch statements.
7806 * tree-cfg.h (find_case_label_for_value): New declaration.
7807 (find_taken_edge_switch_expr): Likewise.
7808 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
7809 Find pivot in decision tree based on probabily, not by number of
7812 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
7814 * doc/standards.texi (Standards): Update Objective-C reference.
7816 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7818 * doc/install.texi (Prerequisites): Update link for MPC.
7820 2018-09-01 Michael Matz <matz@suse.de>
7822 PR tree-optimization/87074
7823 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
7824 PHIs for outer-loop uses.
7826 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7828 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
7829 * doc/invoke.texi (C Dialect Options): Ditto.
7831 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
7833 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
7835 2018-08-31 Richard Biener <rguenther@suse.de>
7837 PR tree-optimization/87168
7838 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
7839 (rpo_elim::eliminate_avail): When OP was not visited it must
7842 2018-08-31 David Malcolm <dmalcolm@redhat.com>
7844 * tree-vrp.c (copy_value_range): Convert param "from" from
7845 "value_range *" to "const value_range *".
7846 (range_is_null): Likewise for param "vr".
7847 (range_int_cst_p): Likewise.
7848 (range_int_cst_singleton_p): Likewise.
7849 (symbolic_range_p): Likewise.
7850 (value_ranges_intersect_p): Likewise for both params.
7851 (value_range_nonnegative_p): Likewise for param "vr".
7852 (value_range_constant_singleton): Likewise.
7853 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
7854 (extract_range_into_wide_ints): Likewise for param "vr".
7855 (extract_range_from_multiplicative_op): Likewise for params "vr0"
7857 (vrp_can_optimize_bit_op): Likewise.
7858 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
7860 (extract_range_from_unary_expr): Likewise.
7861 (debug_value_range): Likewise for param "vr".
7862 (value_range::dump): Add "const" qualifier.
7863 (vrp_prop::check_array_ref): Convert local "vr" from
7864 "value_range *" to "const value_range *".
7865 (vrp_prop::check_mem_ref): Likewise.
7866 (vrp_prop::visit_stmt): Likewise for local "old_vr".
7867 (vrp_intersect_ranges_1): Likewise for param "vr_1".
7868 (vrp_intersect_ranges): Likewise.
7869 (simplify_stmt_for_jump_threading): Likewise for local "vr".
7870 (vrp_prop::vrp_finalize): Likewise.
7871 * tree-vrp.h (value_range::dump): Add "const" qualifier.
7872 (vrp_intersect_ranges): Add "const" qualifier to params as above.
7873 (extract_range_from_unary_expr): Likewise.
7874 (value_range_constant_singleton): Likewise.
7875 (symbolic_range_p): Likewise.
7876 (copy_value_range): Likewise.
7877 (extract_range_from_binary_expr_1): Likewise.
7878 (range_int_cst_p): Likewise.
7879 (vrp_set_zero_nonzero_bits): Likewise.
7880 (range_int_cst_singleton_p): Likewise.
7882 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
7884 * config/aarch64/arm_neon.h (vabsd_s64): New.
7885 (vnegd_s64): Likewise.
7887 2018-08-31 Martin Jambor <mjambor@suse.cz>
7889 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
7891 2018-08-31 Martin Liska <mliska@suse.cz>
7893 * ipa-icf.c (sem_item::add_type): Use
7894 sem_item::m_type_hash_cache.
7895 * ipa-icf.h: Move the cache from sem_item_optimizer
7898 2018-08-31 Nathan Sidwell <nathan@acm.org>
7900 * doc/extend.texi (Backwards Compatibility): Remove implicit
7901 extern C leeway of () being (...).
7903 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7905 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
7907 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
7911 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
7913 2018-08-31 Jakub Jelinek <jakub@redhat.com>
7916 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
7917 gen_int_mode. Formatting fixes.
7919 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
7921 * target.def (custom_function_descriptors): Improve documentation.
7922 * doc/tm.texi.in (Trampolines): Expand discussion of function
7923 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
7924 beginning of the section.
7925 * doc/tm.texi: Regenerated.
7927 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
7929 * cfg.h (class auto_edge_flag): Spell out the template-id of the
7930 base class in the initializer list. This is a workaround for
7931 building with older GCC.
7932 (class auto_bb_flag): Likewise.
7934 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
7936 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
7937 (altivec_vcmpequ<VI_char>_p): Remove star.
7938 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
7940 (expand_strncmp_vec_sequence): New function.
7941 (emit_final_str_compare_vec): New function.
7942 (expand_strn_compare): Add support for vector strncmp.
7943 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
7944 length specification to bytes.
7945 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
7946 (vcmpnezb_p): New pattern.
7947 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
7948 for option -mstring-compare-inline-limit.
7950 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
7952 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
7953 (PTA_SKYLAKE): Add PTA_AES.
7954 (PTA_GOLDMONT): Likewise.
7956 2018-08-29 Jan Hubicka <jh@suse.cz>
7959 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
7960 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
7962 2018-08-29 Jan Hubicka <jh@suse.cz>
7964 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
7966 (hash_tree): Do not visit TYPE_STUB_DECL.
7967 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
7968 stream TYPE_STUB_DECL.
7969 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
7970 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
7971 after free_lang_data.
7972 (type_in_anonymous_namespace_p): Likewise.
7974 2018-08-29 Jan Hubicka <jh@suse.cz>
7976 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
7977 comment that it has to be even number.
7978 (class sreal): Change m_sig type to int32_t.
7979 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
7980 int64_t for temporary calculations.
7981 (sreal_verify_basics): Drop one bit from minimum and maximum.
7983 2018-08-30 Richard Biener <rguenther@suse.de>
7985 PR tree-optimization/87147
7986 * tree-ssa-sccvn.c (SSA_VISITED): New function.
7987 (visit_phi): When the degenerate result is from the backedge and
7988 we didn't visit its definition yet drop to VARYING.
7989 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
7991 2018-08-29 Jan Hubicka <jh@suse.cz>
7993 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
7995 (hash_tree): Likewise.
7997 2018-08-29 Jan Hubicka <jh@suse.cz>
7999 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
8000 and TYPE_NEXT_REF_TO.
8002 2018-08-29 Jan Hubicka <jh@suse.cz>
8004 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
8005 comment that it has to be even number.
8006 (class sreal): Change m_sig type to int32_t.
8007 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
8008 int64_t for temporary calculations.
8009 (sreal_verify_basics): Drop one bit from minimum and maximum.
8011 2018-08-30 Tamar Christina <tamar.christina@arm.com>
8013 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
8015 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
8018 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
8019 if to_add is negative.
8021 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
8024 * builtins.c (c_strlen): Improve range checks.
8026 2018-08-29 Martin Sebor <msebor@redhat.com>
8027 Jeff Law <law@redhat.com>
8029 PR tree-optimization/86714
8030 PR tree-optimization/86711
8031 * builtins.c (c_strlen): Add arguments to call to string_constant.
8032 * expr.c (string_constant): Add argument. Detect missing nul
8033 terminator and outermost declaration it's missing in.
8034 * expr.h (string_constant): Add argument.
8035 * fold-const.c (read_from_constant_string): Add arguments to call to
8037 (c_getstr): Likewise.
8038 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
8040 * tree-ssa-strlen.c (get_stridx): Likewise.
8042 2018-08-29 Jan Hubicka <jh@suse.cz>
8044 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
8045 Do not stream DECL_VINDEX.
8046 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
8047 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
8048 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
8050 2018-08-29 Richard Biener <rguenther@suse.de>
8052 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
8053 virtual operands that are not default defs to honor region
8055 (rpo_vn_valueize): Remove ineffective code here.
8057 2018-08-29 Richard Biener <rguenther@suse.de>
8059 PR tree-optimization/87132
8060 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
8061 when skipping defs reachable over backedges.
8063 2018-08-29 Richard Biener <rguenther@suse.de>
8065 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
8066 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
8067 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
8068 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
8069 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
8070 (vn_reference_lookup_3): Remove use of const_parms.
8071 (free_rpo_vn): Do not free const_parms.
8072 (do_rpo_vn): Do not call init_const_parms.
8073 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
8074 SSA_NAME_POINTS_TO_READONLY_MEMORY.
8075 (call_may_clobber_ref_p_1): Likewise.
8077 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
8080 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
8082 (-ftree-scev-cprop): Document.
8084 2018-08-29 Jan Hubicka <jh@suse.cz>
8086 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
8088 (sreal constructor): Update.
8089 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
8090 sreal:operator/): Update.
8092 2018-08-29 Martin Liska <mliska@suse.cz>
8094 * tree-switch-conversion.c (switch_conversion::expand):
8095 Strenghten assumption about gswitch statements.
8097 2018-08-29 Richard Biener <rguenther@suse.de>
8099 PR tree-optimization/87117
8100 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
8101 re-value-number released SSA VDEFs.
8103 2018-08-29 Richard Biener <rguenther@suse.de>
8105 PR tree-optimization/87126
8106 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
8108 2018-08-28 Jim Wilson <jimw@sifive.com>
8110 * config/riscv/pic.md: Rewrite.
8111 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
8113 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
8114 (SOFTF, default_load, softload, softstore): New.
8116 2018-08-28 Jeff Law <law@redhat.com>
8118 * fold-const.c (fold_binary_loc): Remove recently added assert.
8120 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
8122 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
8123 to OP parmeter of generated function.
8125 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
8127 PR tree-optimization/87009
8128 * match.pd: Add boolean optimizations.
8130 2018-08-28 Martin Sebor <msebor@redhat.com>
8133 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
8134 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
8135 (pass_walloca::gate): Use it.
8136 (alloca_call_type): Same.
8137 (pass_walloca::execute): Same.
8138 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
8140 2018-08-28 David Malcolm <dmalcolm@redhat.com>
8142 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
8143 GCC_VERSION for usage of "__gcc_dump_printf__" format from
8146 2018-08-28 Richard Biener <rguenther@suse.de>
8148 PR tree-optimization/87124
8149 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
8150 constants before looking up avail.
8152 2018-08-28 Jakub Jelinek <jakub@redhat.com>
8155 * calls.c (maybe_warn_nonstring_arg): Punt early if
8156 warn_stringop_overflow is zero. Don't call get_range_strlen
8157 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
8158 Swap comparison operands to have constants on rhs. Only use
8159 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
8160 increment lenrng[0].
8162 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
8164 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
8165 use of tree_to_shwi. Remove duplicated test for the size being
8166 a whole number of bytes.
8168 2018-08-28 Richard Biener <rguenther@suse.de>
8170 PR tree-optimization/87117
8171 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
8172 Handle removed stmt without LHS (GIMPLE_NOP).
8174 2018-08-28 Richard Biener <rguenther@suse.de>
8176 PR tree-optimization/87117
8177 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
8178 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
8180 2018-08-28 Richard Biener <rguenther@suse.de>
8182 PR tree-optimization/87117
8183 * tree-ssa-pre.c (compute_avail): Do not make expressions
8184 with predicated values available.
8185 (get_expr_value_id): Assert we do not run into predicated value
8188 2018-08-28 Richard Biener <rguenther@suse.de>
8190 PR tree-optimization/87117
8191 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
8192 get virtual operands.
8193 (get_expr_operands): Handle STRING_CST like other decls.
8195 2018-08-28 Martin Liska <mliska@suse.cz>
8197 * tree.h: Update documentation of fndecl_built_in_p
8201 2018-08-27 Jeff Law <law@redhat.com>
8202 PR tree-optimization/87110
8203 * tree-ssa-dse.c (compute_trims): Handle non-constant
8206 2018-08-27 Martin Sebor <msebor@redhat.com>
8208 PR tree-optimization/86914
8209 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
8211 2018-08-27 Martin Sebor <msebor@redhat.com>
8213 PR tree-optimization/87112
8214 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
8215 the type of the bound argument.
8217 2018-08-27 Jeff Law <law@redhat.com>
8219 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
8220 type does not have a TYPE_SIZE_UNIT.
8222 2018-08-27 Steve Ellcey <sellcey@cavium.com>
8224 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
8225 with include of backend.h.
8227 2018-08-27 Richard Biener <rguenther@suse.de>
8229 PR tree-optimization/86927
8230 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
8231 use const cond reduction code.
8233 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
8235 PR tree-optimization/85758
8236 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
8238 2018-08-27 David Malcolm <dmalcolm@redhat.com>
8241 * diagnostic-show-locus.c (class layout_range): Update for
8242 conversion of show_caret_p to a tri-state.
8243 (layout_range::layout_range): Likewise.
8244 (make_range): Likewise.
8245 (layout::maybe_add_location_range): Likewise.
8246 (layout::should_print_annotation_line_p): Don't show annotation
8247 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
8248 (layout::get_state_at_point): Update for conversion of
8249 show_caret_p to a tri-state. Bail out early for
8250 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
8251 underlining or source colorization.
8252 (gcc_rich_location::add_location_if_nearby): Update for conversion
8253 of show_caret_p to a tri-state.
8254 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
8255 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
8257 (selftest::test_one_liner_labels): Likewise.
8258 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
8259 conversion of show_caret_p to a tri-state.
8260 * pretty-print.c (text_info::set_location): Likewise.
8261 * pretty-print.h (text_info::set_location): Likewise.
8262 * substring-locations.c (format_warning_n_va): Likewise.
8263 * tree-diagnostic.c (default_tree_printer): Likewise.
8264 * tree-pretty-print.c (newline_and_indent): Likewise.
8266 2018-08-27 David Malcolm <dmalcolm@redhat.com>
8269 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
8270 line above for line-insertion fix-it hints.
8271 (selftest::test_fixit_insert_containing_newline): Update the
8272 expected results, and add a test with line-numbering enabled.
8274 2018-08-27 Martin Liska <mliska@suse.cz>
8277 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
8278 params with DECL_HAS_VALUE_EXPR_P.
8280 2018-08-27 Martin Liska <mliska@suse.cz>
8282 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
8283 selected expansion strategy.
8285 2018-08-27 Martin Liska <mliska@suse.cz>
8287 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
8288 * builtins.c (is_builtin_fn): Likewise.
8289 * attribs.c (diag_attr_exclusions): Use new function
8290 fndecl_built_in_p and remove check for FUNCTION_DECL if
8292 (builtin_mathfn_code): Likewise.
8293 (fold_builtin_expect): Likewise.
8294 (fold_call_expr): Likewise.
8295 (fold_builtin_call_array): Likewise.
8296 (fold_call_stmt): Likewise.
8297 (set_builtin_user_assembler_name): Likewise.
8298 (is_simple_builtin): Likewise.
8299 * calls.c (gimple_alloca_call_p): Likewise.
8300 (maybe_warn_nonstring_arg): Likewise.
8301 * cfgexpand.c (expand_call_stmt): Likewise.
8302 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
8303 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
8304 (cgraph_node::verify_node): Likewise.
8305 * cgraphclones.c (build_function_decl_skip_args): Likewise.
8306 (cgraph_node::create_clone): Likewise.
8307 * config/arm/arm.c (arm_insert_attributes): Likewise.
8308 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
8309 * dse.c (scan_insn): Likewise.
8310 * expr.c (expand_expr_real_1): Likewise.
8311 * fold-const.c (operand_equal_p): Likewise.
8312 (fold_binary_loc): Likewise.
8313 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8314 * gimple-low.c (lower_stmt): Likewise.
8315 * gimple-pretty-print.c (dump_gimple_call): Likewise.
8316 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
8318 * gimple.c (gimple_build_call_from_tree): Likewise.
8319 (gimple_call_builtin_p): Likewise.
8320 (gimple_call_combined_fn): Likewise.
8321 * gimplify.c (gimplify_call_expr): Likewise.
8322 (gimple_boolify): Likewise.
8323 (gimplify_modify_expr): Likewise.
8324 (gimplify_addr_expr): Likewise.
8325 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
8326 * ipa-cp.c (determine_versionability): Likewise.
8327 * ipa-fnsummary.c (compute_fn_summary): Likewise.
8328 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
8329 * ipa-split.c (visit_bb): Likewise.
8330 (split_function): Likewise.
8331 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
8332 * lto-cgraph.c (input_node): Likewise.
8333 * lto-streamer-out.c (write_symbol): Likewise.
8334 * omp-low.c (setjmp_or_longjmp_p): Likewise.
8335 (lower_omp_1): Likewise.
8336 * predict.c (strip_predict_hints): Likewise.
8337 * print-tree.c (print_node): Likewise.
8338 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
8339 * trans-mem.c (is_tm_irrevocable): Likewise.
8340 (is_tm_load): Likewise.
8341 (is_tm_simple_load): Likewise.
8342 (is_tm_store): Likewise.
8343 (is_tm_simple_store): Likewise.
8344 (is_tm_abort): Likewise.
8345 (tm_region_init_1): Likewise.
8346 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
8347 * tree-cfg.c (verify_gimple_call): Likewise.
8348 (move_stmt_r): Likewise.
8349 (stmt_can_terminate_bb_p): Likewise.
8350 * tree-eh.c (lower_eh_constructs_2): Likewise.
8351 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
8352 * tree-inline.c (remap_gimple_stmt): Likewise.
8353 (copy_bb): Likewise.
8354 (estimate_num_insns): Likewise.
8355 (fold_marked_statements): Likewise.
8356 * tree-sra.c (scan_function): Likewise.
8357 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
8358 (optimize_stack_restore): Likewise.
8359 (pass_fold_builtins::execute): Likewise.
8360 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
8361 (mark_all_reaching_defs_necessary_1): Likewise.
8362 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
8363 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
8364 (pass_forwprop::execute): Likewise.
8365 * tree-ssa-loop-im.c (stmt_cost): Likewise.
8366 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8367 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
8368 * tree-ssa-strlen.c (get_string_length): Likewise.
8369 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
8370 (find_func_aliases_for_call): Likewise.
8371 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
8372 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
8373 * tree-tailcall.c (find_tail_calls): Likewise.
8374 * tree.c (need_assembler_name_p): Likewise.
8375 (free_lang_data_in_decl): Likewise.
8376 (get_call_combined_fn): Likewise.
8377 * ubsan.c (is_ubsan_builtin_p): Likewise.
8378 * varasm.c (incorporeal_function_p): Likewise.
8379 * tree.h (DECL_BUILT_IN): Remove and replace with
8381 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
8382 (fndecl_built_in_p): New.
8384 2018-08-27 Martin Liska <mliska@suse.cz>
8386 PR tree-optimization/86847
8387 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
8388 Dump also subtree probability.
8389 (switch_decision_tree::do_jump_if_equal): New function.
8390 (switch_decision_tree::emit_case_nodes): Handle special
8391 situations in balanced tree that can be emitted much simpler.
8392 Fix calculation of probabilities that happen in tree expansion.
8393 * tree-switch-conversion.h (struct cluster): Add
8395 (struct simple_cluster): Likewise.
8396 (struct case_tree_node): Add new function has_child.
8397 (do_jump_if_equal): New.
8399 2018-08-27 Martin Liska <mliska@suse.cz>
8401 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
8402 Add new argument to bit_test_cluster constructor.
8403 (bit_test_cluster::emit): Set bits really number of values
8405 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
8406 probability argument.
8407 * tree-switch-conversion.h (struct bit_test_cluster):
8408 Add m_handles_entire_switch.
8410 2018-08-27 Martin Liska <mliska@suse.cz>
8412 PR tree-optimization/86702
8413 * tree-switch-conversion.c (jump_table_cluster::emit):
8414 Make probabilities even for values in jump table
8415 according to number of cases handled.
8416 (switch_decision_tree::compute_cases_per_edge): Pass
8417 argument to reset_out_edges_aux function.
8418 (switch_decision_tree::analyze_switch_statement): Likewise.
8419 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
8422 2018-08-27 Martin Liska <mliska@suse.cz>
8424 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
8425 cfun argument explicitly.
8426 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
8427 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
8428 function gimple_switch_default_bb.
8429 (convert_switch_statements):
8431 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
8432 * stmt.c (label_to_block_fn): Use label_to_block and pass
8433 cfun argument explicitly and use gimple_switch_label_bb.
8434 (expand_case): Likewise.
8435 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
8436 cfun argument explicitly. Likewise.
8437 (make_edges_bb): Likewise.
8438 (make_cond_expr_edges): Likewise.
8439 (get_cases_for_edge): Likewise.
8440 (make_gimple_switch_edges): Likewise.
8441 (label_to_block_fn): Likewise.
8442 (label_to_block): Likewise.
8443 (make_goto_expr_edges): Likewise.
8444 (make_gimple_asm_edges): Likewise.
8445 (main_block_label): Likewise.
8446 (group_case_labels_stmt): Likewise.
8447 (find_taken_edge_computed_goto): Likewise.
8448 (find_taken_edge_switch_expr): Likewise.
8449 (gimple_verify_flow_info): Likewise.
8450 (gimple_redirect_edge_and_branch): Likewise.
8451 (gimple_switch_label_bb): New function.
8452 (gimple_switch_default_bb): Likewise.
8453 (gimple_switch_edge): Likewise.
8454 (gimple_switch_default_edge): Likewise.
8455 * tree-cfg.h (label_to_block_fn): Remove and replace ...
8456 (label_to_block): ... with this.
8457 (gimple_switch_label_bb): New.
8458 (gimple_switch_default_bb): Likewise.
8459 (gimple_switch_edge): Likewise.
8460 (gimple_switch_default_edge): Likewise.
8461 * tree-cfgcleanup.c (convert_single_case_switch): Use
8462 new gimple functions and pass new argument to label_to_block.
8463 (cleanup_control_flow_bb):
8464 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
8465 cfun argument explicitly.
8466 (make_eh_edges): Likewise.
8467 (redirect_eh_dispatch_edge): Likewise.
8468 (lower_resx): Likewise.
8469 (lower_eh_dispatch): Likewise.
8470 (maybe_remove_unreachable_handlers): Likewise.
8471 (unsplit_eh): Likewise.
8472 (cleanup_empty_eh): Likewise.
8473 (verify_eh_edges): Likewise.
8474 (verify_eh_dispatch_edge): Likewise.
8475 * tree-ssa-dom.c (record_edge_info): Likewise.
8476 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
8477 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
8478 (thread_through_normal_block): Likewise.
8479 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
8480 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
8481 * tree-switch-conversion.c (switch_conversion::collect): Use new
8483 (switch_conversion::check_final_bb): Likewise.
8484 (switch_conversion::gather_default_values): Pass new argument
8486 (switch_conversion::build_constructors): Likewise.
8487 (switch_decision_tree::compute_cases_per_edge): Use new
8488 gimple_switch_edge function.
8489 (switch_decision_tree::analyze_switch_statement): Pass new argument
8491 (switch_decision_tree::try_switch_expansion): Use
8492 gimple_switch_default_edge.
8493 * tree-vrp.c (find_switch_asserts): Pass new argument
8495 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
8496 (vr_values::simplify_switch_using_ranges): Likewise.
8498 2018-08-27 Richard Biener <rguenther@suse.de>
8500 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
8501 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
8503 * tree-ssa-sccvn.h (struct vn_pval): New structure.
8504 (struct vn_nary_op_s): Add unwind_to member. Add
8505 predicated_values flag and put result into a union together
8506 with a linked list of vn_pval.
8507 (struct vn_ssa_aux): Add name member to make maintaining
8508 a map of SSA name to vn_ssa_aux possible. Remove no longer
8509 needed info, dfsnum, low, visited, on_sccstack, use_processed
8510 and range_info_anti_range_p members.
8511 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
8512 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
8514 (vn_valueize): New global.
8515 (vn_context_bb): Likewise.
8516 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
8517 VN_INFO_PTR_INFO): Remove.
8518 * tree-ssa-sccvn.c: ... (rewrite)
8519 (pass_fre::execute): For -O2+ initialize loops and run
8520 RPO VN in optimistic mode (iterating). For -O1 and -Og
8521 run RPO VN in non-optimistic mode.
8522 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
8523 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
8524 * doc/invoke.texi (sccvn-max-scc-size): Remove.
8525 (rpo-vn-max-loop-depth): Document.
8526 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
8527 when valuezing the VUSE signals we walked out of the region.
8528 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
8529 (phi_translate): Set VN context block to use for availability
8531 (compute_avail): Likewise.
8532 (pre_valueize): New function.
8533 (pass_pre::execute): Adjust to the RPO VN API.
8535 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
8536 (propagate_constants_for_unrolling): Remove.
8537 (tree_unroll_loops_completely): Perform value-numbering
8538 on the unrolled bodies loop parent.
8540 2018-08-27 Richard Biener <rguenther@suse.de>
8542 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
8543 for partial antic compute.
8545 2018-08-27 Jakub Jelinek <jakub@redhat.com>
8547 PR rtl-optimization/87065
8548 * combine.c (simplify_if_then_else): Formatting fix.
8549 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
8551 (known_cond): Don't return const_true_rtx for vector modes. Use
8552 CONST0_RTX instead of const0_rtx. Formatting fixes.
8554 2018-08-27 Martin Liska <mliska@suse.cz>
8556 PR gcov-profile/87069
8557 * gcov.c (process_file): Record files already processed
8558 and warn about a file being processed multiple times.
8560 2018-08-27 Martin Liska <mliska@suse.cz>
8563 * config/aarch64/aarch64.c (aarch64_override_options_internal):
8564 Set default values for x_aarch64_*_string strings.
8565 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
8566 prefix. For -mabi do not print '=ABI' in help and use
8567 <option_value> format for -msve-vector-bits and -moverride
8570 2018-08-26 Jeff Law <law@redhat.com>
8572 * config/mips/frame-header-opt.c: Include "backend.h" rather than
8575 2018-08-26 Marek Polacek <polacek@redhat.com>
8577 PR c++/87029, Implement -Wredundant-move.
8578 * doc/invoke.texi: Document -Wredundant-move.
8580 2018-08-25 Martin Sebor <msebor@redhat.com>
8582 PR tree-optimization/87059
8583 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
8584 to the same type as the other.
8585 * fold-const.c (fold_binary_loc): Assert expectation.
8587 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
8589 * config/darwin.c (machopic_legitimize_pic_address): Clean up
8590 extraneous parentheses, dead code section and formatting.
8592 2018-08-24 David Malcolm <dmalcolm@redhat.com>
8595 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
8596 wide enough for jumps in the line-numbering to be visible.
8597 (layout::print_gap_in_line_numbering): New member function.
8598 (layout::calculate_line_spans): When using line numbering, merge
8599 line spans that are only 1 line apart.
8600 (diagnostic_show_locus): When printing line numbers, show gaps in
8601 line numbering directly, rather than printing headers.
8602 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
8603 line-numbering with multiple line spans.
8604 (selftest::test_fixit_insert_containing_newline_2): Add test of
8605 line-numbering, in which the spans are close enough to be merged.
8607 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
8609 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
8610 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
8611 * tree-vrp.c (range_is_nonnull): Remove.
8612 (range_includes_zero_p): Accept value_range instead of min/max.
8613 (extract_range_from_binary_expr_1): Do not early bail on
8615 Use range_includes_zero_p instead of range_is_nonnull.
8616 (extract_range_from_unary_expr): Use range_includes_zero_p instead
8617 of range_is_nonnull.
8618 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
8619 special case VR_ANTI_RANGE.
8620 (vrp_finalize): Same.
8621 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
8623 (range_is_nonnull): Remove.
8624 * vr-values.c (vrp_stmt_computes_nonzero): Use
8625 range_includes_zero_p instead of range_is_nonnull.
8626 (extract_range_basic): Pass value_range to range_includes_zero_p
8627 instead of range_is_nonnull.
8629 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
8631 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
8632 * emit-rtl.h (rtl_data): Remove return_bnd.
8633 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
8634 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
8635 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
8636 (POINTER_BOUNDS_MODE): Remove definition.
8637 (make_pointer_bounds_mode): Remove.
8638 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
8639 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
8640 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
8641 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
8642 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
8643 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
8644 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
8646 * config/i386/i386-modes.def (BND32, BND64): Remove.
8647 * config/i386/i386.c (dbx_register_map): Remove bound registers.
8648 (dbx64_register_map): Ditto.
8649 (svr4_dbx_register_map): Ditto.
8650 (indirect_thunk_bnd_needed): Remove.
8651 (indirect_thunks_bnd_used): Ditto.
8652 (indirect_return_bnd_needed): Ditto.
8653 (indirect_return_via_cx_bnd): Ditto.
8654 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
8655 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
8656 (output_indirect_thunk): Ditto. Remove need_prefix argument.
8657 (output_indirect_thunk_function): Remove handling of
8658 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
8659 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
8660 (ix86_save_reg): Remove handling of crtl->return_bnd.
8661 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
8662 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
8663 and UNSPEC_BNDLX_ADDR.
8664 (ix86_output_indirect_branch_via_reg): Remove handling of
8665 indirect_thunk_prefix_bnd.
8666 (ix86_output_indirect_branch_via_push): Ditto.
8667 (ix86_output_function_return): Ditto.
8668 (ix86_output_indirect_function_return): Ditto.
8669 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
8670 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
8671 (CALL_USED_REGISTERS): Ditto.
8672 (REG_ALLOC_ORDER): Update for removal of bound registers.
8673 (HI_REGISTER_NAMES): Ditto.
8674 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
8675 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
8676 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
8677 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
8678 (FIRST_PSEUDO_REG): Update.
8679 (BND): Remove mode iterator.
8680 * config/i386/predicates.md (bnd_mem_operator): Remove.
8682 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
8684 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
8687 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
8689 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
8690 the case in which the permute needs only a single element and
8691 repeats for every vector of the result. Extend that case to
8692 handle variable-length vectors.
8693 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8695 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
8698 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
8699 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
8701 2018-08-24 Richard Biener <rguenther@suse.de>
8703 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
8704 bb_flags_allocated members.
8705 (auto_flag): New RAII class for allocating flags.
8706 (auto_edge_flag): New RAII class for allocating edge flags.
8707 (auto_bb_flag): New RAII class for allocating bb flags.
8708 * cfgloop.c (verify_loop_structure): Allocate temporary edge
8710 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
8711 in favor of temporarily allocated BB flag.
8712 * hsa-brig.c: Re-order includes.
8713 * hsa-dump.c: Likewise.
8714 * hsa-regalloc.c: Likewise.
8715 * print-rtl.c: Likewise.
8716 * profile-count.c: Likewise.
8718 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
8721 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
8724 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
8727 * wide-int-range.cc (wide_int_range_div): Do not ignore result
8728 from wide_int_range_multiplicative_op.
8730 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8732 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
8735 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
8737 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
8740 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
8742 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
8743 full GENERIC expression used for replacement.
8745 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
8747 * tree-vrp.c (abs_extent_range): Remove.
8748 (extract_range_into_wide_ints): Pass wide ints by reference.
8749 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
8750 Pass wide ints by reference in all calls to
8751 extract_range_into_wide_ints.
8752 * wide-int-range.cc (wide_int_range_div): New.
8753 * wide-int-range.h (wide_int_range_div): New.
8754 (wide_int_range_includes_zero_p): New.
8755 (wide_int_range_zero_p): New.
8757 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
8759 * config/aarch64/aarch64.md (arches): New enum.
8760 (arch): New enum attr.
8761 (arch_enabled): New attr.
8762 (enabled): Now uses arch_enabled only.
8763 (simd, sve, fp16): Removed attribute.
8764 (fp): Attr now defined in terms of 'arch'.
8765 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
8766 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
8767 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
8768 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
8769 attributes into 'arch'.
8770 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
8771 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
8772 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
8773 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
8774 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
8775 'simd' attribute into 'arch'.
8776 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
8777 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
8778 Convert use of 'fp' attribute to 'arch'.
8779 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
8780 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
8782 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
8783 (different modes) Merge 'fp' and 'simd' into 'arch'.
8784 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
8787 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
8789 PR rtl-optimization/87026
8790 * expmed.c (canonicalize_comparison): If we can no longer create
8791 pseudoregisters, don't.
8793 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
8796 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
8798 * config/arm/arm.c (speculation_barrier_libfunc): New static
8800 (arm_init_libfuncs): Initialize it.
8801 (arm_emit_speculation_barrier): New function.
8802 * config/arm/arm.md (speculation_barrier): Call
8803 arm_emit_speculation_barrier for architectures that do not have
8805 (speculation_barrier_insn): Only match on Armv7 or later.
8807 2018-08-23 Richard Biener <rguenther@suse.de>
8810 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
8813 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
8815 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
8816 of single-vector TBLs.
8817 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
8820 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
8823 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
8824 aarch64_evpc_tbl guard.
8826 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
8828 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
8831 2018-08-22 Martin Sebor <msebor@redhat.com>
8834 * tree-pretty-print.c (pretty_print_string): Add argument.
8835 (dump_generic_node): Call to pretty_print_string with string size.
8837 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
8839 PR rtl-optimization/86771
8840 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
8841 of two SETs into those two SETs, one to be placed at i2, if that SETs
8842 destination is modified between i2 and i3.
8844 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8846 PR tree-optimization/86725
8847 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
8849 (vect_analyze_scalar_cycles_1): Check it.
8851 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8853 PR tree-optimization/86725
8854 * tree-vect-loop.c (vect_is_simple_reduction): When treating
8855 an outer loop phi as a double reduction, make sure that the
8856 single user of the phi result is an inner loop phi.
8858 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8860 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
8861 grouped stores with gaps to a strided group.
8863 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
8865 * tree-vect-stmts.c (get_group_load_store_type)
8866 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
8867 first statement in a group.
8869 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8871 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
8872 the sequence used in gcc/gcc.c.
8874 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8877 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
8880 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8882 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
8883 Darwin10-specific unwinder-shim.
8884 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
8885 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
8886 New to cater for Darwin10 Rosetta.
8888 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8890 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
8891 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
8893 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
8898 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
8899 (dwarf2out_switch_text_section): Generate a local label for the
8900 second function sub-section and apply it as the second FDE start
8902 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
8903 second sub-section start.
8905 2018-08-22 Richard Biener <rguenther@suse.de>
8907 PR tree-optimization/86988
8908 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
8910 2018-08-22 Richard Biener <rguenther@suse.de>
8912 PR tree-optimization/86945
8913 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
8915 2018-08-22 Alexandre Oliva <oliva@adacore.com>
8917 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
8918 a comment about how uses of r2 for .sdata2 come about.
8920 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
8922 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
8924 2018-08-21 Marek Polacek <polacek@redhat.com>
8926 PR c++/86981, Implement -Wpessimizing-move.
8927 * doc/invoke.texi: Document -Wpessimizing-move.
8929 2018-08-21 Jan Hubicka <jh@suse.cz>
8931 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
8932 * tree.h (is_redundant_typedef): Remove.
8933 * dwarf2out.c (is_redundant_typedef): Turn into static function.
8935 2018-08-21 Jan Hubicka <jh@suse.cz>
8937 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
8940 2018-08-21 Tamar Christina <tamar.christina@arm.com>
8942 * expmed.c (extract_low_bits): Reject invalid subregs early.
8944 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
8947 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
8950 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8952 * config/vxworks.h: Guard vxworks_asm_out_constructor and
8953 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
8954 * config/vxworks.c: Likewise.
8956 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8958 * config/vxworks.c: Set targetm.have_ctors_dtors
8959 if HAVE_INITFINI_ARRAY_SUPPORT.
8960 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
8961 if HAVE_INITFINI_ARRAY_SUPPORT.
8963 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
8965 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
8966 default search path for VxWorks < 7.
8968 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
8970 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
8971 (get_format_string): Refer to c_getstr.
8973 2018-08-21 Tom de Vries <tdevries@suse.de>
8975 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
8976 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
8977 (debuginfo_early_stop): Declare.
8978 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
8979 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
8980 (debuginfo_early_stop): New function.
8981 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
8982 and debuginfo_early_stop.
8983 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
8984 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
8985 (general_init): Call debuginfo_early_init.
8986 (finalize): Call debuginfo_fini.
8987 (do_compile): Call debuginfo_init.
8988 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
8990 (@item -fdump-debug, @item -fdump-earlydebug): Add.
8992 2018-08-21 Tom de Vries <tdevries@suse.de>
8994 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
8995 flag_dump_noaddr and flag_dump_unnumbered.
8997 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
8999 * wide-int-range.cc (wide_int_range_abs): New.
9000 (wide_int_range_order_set): Rename from wide_int_range_min_max.
9001 * wide-int-range.h (wide_int_range_abs): New.
9002 (wide_int_range_min_max): New.
9003 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
9004 case to call wide_int_range_abs.
9005 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
9006 (extract_range_from_abs_expr): Delete.
9008 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
9011 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
9012 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
9015 2018-08-20 Nathan Sidwell <nathan@acm.org>
9016 Jeff Law <law@redhat.com>
9018 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
9019 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
9021 2018-08-20 David Malcolm <dmalcolm@redhat.com>
9024 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
9025 (decl_attributes): Likewise.
9026 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
9028 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
9029 * diagnostic-core.h (class auto_diagnostic_group): New class.
9030 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
9031 (diagnostic_report_diagnostic): Handle the first diagnostics within
9033 (emit_diagnostic): Add auto_diagnostic_group instance.
9035 (inform_n): Likewise.
9036 (warning): Likewise.
9037 (warning_at): Likewise.
9038 (warning_n): Likewise.
9039 (pedwarn): Likewise.
9040 (permerror): Likewise.
9042 (error_n): Likewise.
9043 (error_at): Likewise.
9045 (fatal_error): Likewise.
9046 (internal_error): Likewise.
9047 (internal_error_no_backtrace): Likewise.
9048 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
9049 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
9050 * diagnostic.h (struct diagnostic_context): Add fields
9051 "diagnostic_group_nesting_depth",
9052 "diagnostic_group_emission_count", "begin_group_cb",
9054 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
9055 Add auto_diagnostic_group instance(s).
9056 (find_explicit_erroneous_behavior): Likewise.
9057 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
9058 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
9059 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
9060 (gimplify_va_arg_expr): Likewise.
9061 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
9062 (HSA_SORRY_AT): Likewise.
9063 * ipa-devirt.c (compare_virtual_tables): Likewise.
9064 (warn_odr): Likewise.
9065 * multiple_target.c (expand_target_clones): Likewise.
9066 * opts-common.c (cmdline_handle_error): Likewise.
9067 * reginfo.c (globalize_reg): Likewise.
9068 * substring-locations.c (format_warning_n_va): Likewise.
9069 * tree-inline.c (expand_call_inline): Likewise.
9070 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
9071 * tree-ssa-loop-niter.c
9072 (do_warn_aggressive_loop_optimizations): Likewise.
9073 * tree-ssa-uninit.c (warn_uninit): Likewise.
9074 * tree.c (warn_deprecated_use): Likewise.
9076 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
9079 * config/i386/i386.md (eh_return): Always update EH return
9080 address in word_mode.
9082 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
9084 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
9085 TARGET_SPLIT_COMPLEX_ARG is defined.
9087 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9089 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
9091 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9094 * expr.c (expand_assignment): Assert that bitpos is positive.
9095 (store_field): Likewise
9096 (expand_expr_real_1): Make sure that bitpos is positive.
9097 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
9100 2018-08-20 Nathan Sidwell <nathan@acm.org>
9102 * Makefile.in (CPP_ID_DATA_H): Delete.
9103 (CPP_INTERNAL_H): Don't add it.
9104 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
9105 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
9107 2018-08-20 Richard Biener <rguenther@suse.de>
9109 PR tree-optimization/78655
9110 * tree-vrp.c (extract_range_from_binary_expr_1): Make
9111 pointer + offset nonnull if either operand is nonnull work.
9113 2018-08-20 Tom de Vries <tdevries@suse.de>
9115 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
9116 unless the referenced die describes the added property using
9117 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
9118 Otherwise, add a DW_AT_location to the referenced die.
9120 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
9123 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
9124 register_operand when calling ix86_set_reg_reg_cost.
9125 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
9126 Set *total to 0 for operands that satisfy x86_64_immediate_operand
9127 predicate and to 1 otherwise.
9129 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
9131 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
9132 emit a diagnostic that it is not supported and reset the option.
9133 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
9134 supported and consume the option. (ASM_FINAL_SPEC): New.
9136 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
9138 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
9141 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
9143 C-SKY port: Documentation
9145 * doc/extend.texi (C-SKY Function Attributes): New section.
9146 * doc/invoke.texi (Option Summary): Add C-SKY options.
9147 (C-SKY Options): New section.
9148 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
9150 2018-08-17 Jojo <jijie_rong@c-sky.com>
9151 Huibin Wang <huibin_wang@c-sky.com>
9152 Sandra Loosemore <sandra@codesourcery.com>
9153 Chung-Lin Tang <cltang@codesourcery.com>
9155 C-SKY port: Backend implementation
9157 * config/csky/*: New.
9158 * common/config/csky/*: New.
9160 2018-08-17 Jojo <jijie_rong@c-sky.com>
9161 Huibin Wang <huibin_wang@c-sky.com>
9162 Sandra Loosemore <sandra@codesourcery.com>
9163 Chung-Lin Tang <cltang@codesourcery.com>
9164 Andrew Jenner <andrew@codesourcery.com>
9166 C-SKY port: Configury
9168 * config.gcc (csky-*-*): New.
9169 * configure.ac: Add csky to targets for dwarf2 debug_line support.
9170 * configure: Regenerated.
9172 2018-08-17 David Malcolm <dmalcolm@redhat.com>
9174 * dump-context.h: Include "dumpfile.h".
9175 (dump_context::dump_printf_va): Convert final param from va_list
9176 to va_list *. Convert from ATTRIBUTE_PRINTF to
9177 ATTRIBUTE_GCC_DUMP_PRINTF.
9178 (dump_context::dump_printf_loc_va): Likewise.
9179 * dumpfile.c: Include "stringpool.h".
9180 (make_item_for_dump_printf_va): Delete.
9181 (make_item_for_dump_printf): Delete.
9182 (class dump_pretty_printer): New class.
9183 (dump_pretty_printer::dump_pretty_printer): New ctor.
9184 (dump_pretty_printer::emit_items): New member function.
9185 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
9187 (dump_pretty_printer::emit_item): New member function.
9188 (dump_pretty_printer::stash_item): New member function.
9189 (dump_pretty_printer::format_decoder_cb): New member function.
9190 (dump_pretty_printer::decode_format): New member function.
9191 (dump_context::dump_printf_va): Reimplement in terms of
9192 dump_pretty_printer.
9193 (dump_context::dump_printf_loc_va): Convert final param from va_list
9195 (dump_context::begin_scope): Reimplement call to
9196 make_item_for_dump_printf.
9197 (dump_printf): Update for change to dump_printf_va.
9198 (dump_printf_loc): Likewise.
9199 (selftest::test_capture_of_dump_calls): Convert "stmt" from
9200 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
9201 with %T, %E, and %G.
9202 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
9203 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
9204 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
9205 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
9206 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
9207 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
9208 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
9209 within a dump_printf_loc call to "%wu".
9210 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
9211 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
9212 missing space after "=".
9213 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
9214 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
9215 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
9216 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
9217 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
9218 duplicate "vectorized" from message.
9220 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
9222 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
9223 polyNxK_t element's TYPE_STRING_FLAG.
9225 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
9227 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
9228 (they were unnamed before). Fix comments.
9230 2018-08-17 Nathan Sidwell <nathan@acm.org>
9232 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
9234 2018-08-17 Richard Biener <rguenther@suse.de>
9236 PR tree-optimization/86841
9237 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
9239 2018-08-17 Martin Liska <mliska@suse.cz>
9241 * common.opt: Remove Warn, Init and Report for options with
9242 Ignore/Deprecated flag. Warning is done automatically for
9244 * config/i386/i386.opt: Likewise.
9245 * config/ia64/ia64.opt: Likewise.
9246 * config/rs6000/rs6000.opt: Likewise.
9247 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
9248 Remove usage of flag_check_pointer_bounds.
9249 * lto-wrapper.c (merge_and_complain): Do not handle
9250 OPT_fcheck_pointer_bounds.
9251 (append_compiler_options): Likewise.
9252 * opt-functions.awk: Do not handle Deprecated.
9253 * optc-gen.awk: Check that Var, Report and Init are not
9254 used for an option with Ignore/Deprecated flag.
9255 * opts-common.c (decode_cmdline_option): Do not report
9257 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
9259 * opts.h (struct cl_option): Remove cl_deprecated flag.
9260 (CL_ERR_DEPRECATED): Remove error enum value.
9262 2018-08-17 Richard Biener <rguenther@suse.de>
9265 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
9266 across a va-arg-pack using call adjust its return value accordingly.
9268 2018-08-16 Martin Sebor <msebor@redhat.com>
9270 PR tree-optimization/86853
9271 * gimple-ssa-sprintf.c (struct format_result): Rename member.
9272 (struct fmtresult): Add member and initialize it in ctors.
9273 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
9274 (format_string): Handle %S the same as %ls. Set MAYFAIL.
9275 (format_directive): Set POSUNDER4K when MAYFAIL is set.
9276 (parse_directive): Handle %C same as %c.
9277 (sprintf_dom_walker::compute_format_length): Adjust.
9278 (is_call_safe): Adjust.
9280 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
9282 * builtins.c (c_strlen): Add new parameter eltsize. Use it
9283 for determining how to count the elements.
9284 * builtins.h (c_strlen): Adjust prototype.
9285 * expr.c (string_constant): Add new parameter mem_size.
9286 Set *mem_size appropriately.
9287 * expr.h (string_constant): Adjust protoype.
9288 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
9289 * gimple-fold.h (get_range_strlen): Adjust prototype.
9290 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
9291 (format_string): Call get_string_length with eltsize.
9293 2018-08-16 David Malcolm <dmalcolm@redhat.com>
9295 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
9296 to emit the span, rather than setting it as the prefix.
9298 2018-08-16 David Malcolm <dmalcolm@redhat.com>
9300 * diagnostic-show-locus.c (layout::start_annotation_line): Add
9301 "margin_char" parameter, defaulting to space. Use it in place
9302 of pp_space for the initial part of the margin.
9303 (layout::print_leading_fixits): Use '+' when filling the margin
9304 of line-insertion fix-it hints.
9306 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9308 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
9311 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9313 * config/rs6000/altivec.md: Don't set length attribute to the default
9315 * config/rs6000/darwin.md: Ditto.
9316 * config/rs6000/dfp.md: Ditto.
9317 * config/rs6000/htm.md: Ditto.
9318 * config/rs6000/rs6000.md: Ditto.
9319 * config/rs6000/sync.md: Ditto.
9320 * config/rs6000/vsx.md: Ditto.
9322 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9324 * config/rs6000/altivec.md: Don't set length attribute to the default
9325 value, for branch instructions.
9326 * config/rs6000/darwin.md: Ditto.
9327 * config/rs6000/rs6000.md: Ditto.
9329 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
9331 * config/rs6000/rs6000.md (length): Always define as const_int 4.
9332 (unnamed conditional branch define_insn): Set length to 4 or 8
9333 depending on offset.
9334 (<bd>_<mode>): Similar, for alternative 0.
9335 (<bd>tf_<mode>): Ditto.
9337 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9339 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
9341 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
9343 * doc/rtl.texi: Replace old RTX class names with new names.
9346 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
9348 * expmed.h (canonicalize_comparison): New declaration.
9349 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
9350 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
9351 * optabs.c (prepare_cmp_insn): Likewise.
9352 * rtl.h (unsigned_condition_p): New function which checks if a
9353 comparison operator is unsigned.
9355 2018-08-16 Nathan Sidwell <nathan@acm.org>
9357 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
9358 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
9360 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9363 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
9364 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
9365 (mov<mov>): ..this and enable unconditionally.
9367 2018-08-16 Tamar Christina <tamar.christina@arm.com>
9369 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
9371 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
9373 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
9374 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
9375 "Common" with "Target".
9377 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
9379 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
9380 * doc/invoke.texi (mmitigate-rop): Remove.
9381 * config/i386/i386.c: Do not include "regrename.h".
9382 (ix86_rop_should_change_byte_p, reg_encoded_number)
9383 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
9385 (ix86_reorg): Remove call to ix86_mitigate_rop.
9386 * config/i386/i386.md (attr "modrm_class"): Remove.
9387 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
9388 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
9389 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
9391 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
9393 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
9394 allow folding of mergeh() and mergel() for the float and double types.
9395 (fold_mergehl_helper): Rework to handle building a permute tree
9398 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
9400 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
9403 2018-08-15 David Malcolm <dmalcolm@redhat.com>
9405 * common.opt (fdiagnostics-show-labels): New option.
9406 * diagnostic-show-locus.c (class layout_range): Add field
9408 (class layout): Add field "m_show_labels_p".
9409 (layout_range::layout_range): Add param "label" and use it to
9411 (make_range): Pass in NULL for new "label" param of layout_range's
9413 (layout::layout): Initialize m_show_labels_p.
9414 (layout::maybe_add_location_range): Pass in loc_range->m_label
9415 when constructing layout_range instances.
9416 (struct line_label): New struct.
9417 (layout::print_any_labels): New member function.
9418 (layout::print_line): Call it if label-printing is enabled.
9419 (selftest::test_one_liner_labels): New test.
9420 (selftest::test_diagnostic_show_locus_one_liner): Call it.
9421 * diagnostic.c (diagnostic_initialize): Initialize
9422 context->show_labels_p.
9423 * diagnostic.h (struct diagnostic_context): Add field
9425 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
9426 -fno-diagnostics-show-labels.
9427 * dwarf2out.c (gen_producer_string): Add
9428 OPT_fdiagnostics_show_labels to the ignored options.
9429 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
9431 (gcc_rich_location::maybe_add_expr): Likewise.
9432 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
9433 label" param, defaulting to NULL.
9434 (gcc_rich_location::add_expr): Add "label" param.
9435 (gcc_rich_location::maybe_add_expr): Likewise.
9436 (class text_range_label): New class.
9437 (class range_label_for_type_mismatch): New class.
9438 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
9439 of format_warning_va.
9440 (fmtwarn_n): Likewise for new params of format_warning_n_va.
9441 * lto-wrapper.c (merge_and_complain): Add
9442 OPT_fdiagnostics_show_labels to the "pick one setting" options.
9443 (append_compiler_options): Likewise to the dropped options.
9444 (append_diag_options): Likewise to the passed-on options.
9445 * opts.c (common_handle_option): Handle the new option.
9446 * selftest-diagnostic.c
9447 (test_diagnostic_context::test_diagnostic_context): Enable
9449 * substring-locations.c: Include "gcc-rich-location.h".
9450 (format_warning_n_va): Add "fmt_label" and "param_label" params
9451 and use them as appropriate.
9452 (format_warning_va): Add "fmt_label" and "param_label" params,
9453 passing them on to format_warning_n_va.
9454 (format_warning_at_substring): Likewise.
9455 (format_warning_at_substring_n): Likewise.
9456 * substring-locations.h (format_warning_va): Add "fmt_label" and
9457 "param_label" params.
9458 (format_warning_n_va): Likewise.
9459 (format_warning_at_substring): Likewise.
9460 (format_warning_at_substring_n): Likewise.
9461 * toplev.c (general_init): Initialize global_dc->show_labels_p.
9463 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
9466 * builtins.c (expand_builtin_memcmp): Do not expand the call
9467 when overflow is detected.
9469 2018-08-15 Martin Sebor <msebor@redhat.com>
9471 PR tree-optimization/71625
9472 * config/aarch64/aarch64-builtins.c
9473 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
9475 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
9477 * config/s390/s390.c (s390_reorg): Remove loop.
9479 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
9482 (darwin_function_switched_text_sections): Delete.
9483 * gcc/config/darwin.h
9484 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
9486 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
9489 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
9490 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
9491 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
9493 2018-08-15 Martin Liska <mliska@suse.cz>
9495 PR tree-optimization/86925
9496 * predict.c (expr_expected_value_1): When taking
9497 later predictor, assign also probability.
9498 Use fold_build2_initializer_loc in order to fold
9499 the expression in -frounding-math.
9501 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
9503 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
9505 (expand_vec_perm_1): Try the new method.
9507 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
9510 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
9511 Check whether lra_live_max_point is 0 before dividing.
9513 2018-08-14 Martin Sebor <msebor@redhat.com>
9515 PR tree-optimization/86650
9516 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
9517 (vrp_prop::check_mem_ref): Same.
9519 2018-08-13 Liu Hao <lh_mouse@126.com>
9521 * pretty-print.c (eat_esc_sequence): Swap the foreground and
9522 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
9523 and clear it thereafter, as it only works for DBCS.
9525 2018-08-13 Liu Hao <lh_mouse@126.com>
9527 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
9528 handle returned by _get_osf_handle().
9530 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9532 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
9533 for folding vec_perm.
9535 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9537 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
9538 Add support for gimple-folding of vec_pack() and vec_unpack()
9541 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
9543 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
9544 vec_xst variants to the list.
9545 (rs6000_gimple_fold_builtin): Add support for folding unaligned
9546 vector loads and stores.
9548 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
9550 * config.gcc (rs6000-ibm-aix4.x): Delete.
9551 (rs6000-ibm-aix5.1): Delete.
9552 (rs6000-ibm-aix5.2): Delete.
9553 (rs6000-ibm-aix5.3): Delete.
9554 * config/rs6000/aix43.h: Delete.
9555 * config/rs6000/aix51.h: Delete.
9556 * config/rs6000/aix52.h: Delete.
9557 * config/rs6000/t-aix43: Delete.
9559 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
9561 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9563 (s390_decompose_address): Factor out constant pool ref
9566 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9568 * config/nds32/nds32-predicates.c
9569 (nds32_can_use_bclr_p): Change return type as bool.
9570 (nds32_can_use_bset_p): Ditto.
9571 (nds32_can_use_btgl_p): Ditto.
9572 (nds32_can_use_bitci_p): Ditto.
9573 * config/nds32/nds32-protos.h
9574 (nds32_can_use_bclr_p): Change declaration.
9575 (nds32_can_use_bset_p): Ditto.
9576 (nds32_can_use_btgl_p): Ditto.
9577 (nds32_can_use_bitci_p): Ditto.
9579 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9581 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
9582 Support -msched-prolog-epilog option.
9583 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
9585 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9587 * common/config/nds32/nds32-common.c
9588 (nds32_option_optimization_table): Enalbe -malways-align.
9590 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
9592 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
9594 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
9595 OPT_misr_secure_ case.
9596 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
9597 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
9598 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
9600 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
9601 (nds32_isr_info): New field security_level.
9602 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
9603 * config/nds32/nds32.md (return_internal): Consider critical attribute.
9604 * config/nds32/nds32.opt (misr-secure): New option.
9605 * config/nds32/nds32_init.inc: New file.
9606 * config/nds32/nds32_isr.h: New file.
9608 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
9610 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
9611 Update comment for atomic instructions.
9612 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
9613 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
9615 (atomic_loaddi): Revise fence expansion to only emit fence prior to
9616 load for __ATOMIC_SEQ_CST model.
9617 (atomic_loaddi_1): Remove float register target.
9618 (atomic_storedi): Handle CONST_INT values.
9619 (atomic_storedi_1): Remove float register source. Add special case
9621 (memory_barrier): New expander and insn.
9623 2018-08-11 Jakub Jelinek <jakub@redhat.com>
9625 PR tree-optimization/86835
9626 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
9627 new_stmt after def_gsi, make sure to insert new_square_stmt after
9628 that stmt, not 2 stmts before it.
9630 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
9633 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
9636 2018-08-10 Martin Liska <mliska@suse.cz>
9639 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
9641 * builtins.c (expand_builtin_expect_with_probability):
9643 (expand_builtin_expect_with_probability): New function.
9644 (build_builtin_expect_predicate): Add new argumnet probability
9645 for BUILT_IN_EXPECT_WITH_PROBABILITY.
9646 (fold_builtin_expect):
9649 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
9650 * builtins.h (fold_builtin_expect): Set new argument.
9651 * doc/extend.texi: Document __builtin_expect_with_probability.
9652 * doc/invoke.texi: Likewise.
9653 * gimple-fold.c (gimple_fold_call): Pass new argument.
9654 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
9655 also BUILT_IN_EXPECT_WITH_PROBABILITY.
9656 * predict.c (get_predictor_value): New function.
9657 (expr_expected_value): Add new argument probability. Assume
9658 that predictor and probability are always non-null.
9659 (expr_expected_value_1): Likewise. For __builtin_expect and
9660 __builtin_expect_with_probability set probability. Handle
9661 combination in binary expressions.
9662 (tree_predict_by_opcode): Simplify code by simply calling
9663 get_predictor_value.
9664 (pass_strip_predict_hints::execute): Add handling of
9665 BUILT_IN_EXPECT_WITH_PROBABILITY.
9666 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
9668 * tree.h (DECL_BUILT_IN_P): New function.
9670 2018-08-10 Martin Liska <mliska@suse.cz>
9672 PR tree-optimization/85799
9673 * passes.def: Add argument for pass_strip_predict_hints.
9674 * predict.c (class pass_strip_predict_hints): Add new argument
9676 (strip_predictor_early): New function.
9677 (pass_strip_predict_hints::execute): Call the function to
9679 (strip_predict_hints): New function.
9680 * predict.def: Fix comment.
9682 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
9684 * Makefile.in: Clarify which tm.texi to copy over to assert the
9685 right to grant a GFDL license for all.
9687 2018-08-09 Jeff Law <law@redhat.com>
9689 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
9692 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
9694 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
9697 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9699 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
9700 reductions for variable-length vectors.
9702 2018-08-09 David Malcolm <dmalcolm@redhat.com>
9705 * common.opt (fdiagnostics-show-line-numbers): New option.
9706 * diagnostic-show-locus.c (class layout): Add fields
9707 "m_show_line_numbers_p" and "m_linenum_width";
9708 (num_digits): New function.
9709 (test_num_digits): New function.
9710 (layout::layout): Initialize new fields. Update m_x_offset
9711 logic to handle any left margin.
9712 (layout::print_source_line): Print line number when requested.
9713 (layout::start_annotation_line): New member function.
9714 (layout::print_annotation_line): Call it.
9715 (layout::print_leading_fixits): Likewise.
9716 (layout::print_trailing_fixits): Likewise. Update calls to
9717 move_to_column for new parameter.
9718 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
9719 it to potentially call start_annotation_line.
9720 (layout::show_ruler): Call start_annotation_line.
9721 (selftest::test_line_numbers_multiline_range): New selftest.
9722 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
9723 and selftest::test_line_numbers_multiline_range.
9724 * diagnostic.c (diagnostic_initialize): Initialize
9725 show_line_numbers_p.
9726 * diagnostic.h (struct diagnostic_context): Add field
9727 "show_line_numbers_p".
9728 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
9729 -fno-diagnostics-show-line-numbers.
9730 * dwarf2out.c (gen_producer_string): Add
9731 OPT_fdiagnostics_show_line_numbers to the ignored options.
9732 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
9733 one setting" options.
9734 (append_compiler_options): Likewise to the dropped options.
9735 (append_diag_options): Likewise to the passed-on options.
9736 * opts.c (common_handle_option): Handle the new option.
9737 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
9739 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
9741 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
9742 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
9743 third argument of type "const signed char" to descriptions of
9744 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
9745 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
9746 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
9747 __builtin_bcdsub_ov functions.
9749 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9751 PR tree-optimization/86858
9752 * tree-vect-loop.c (vect_is_simple_reduction): Restore
9753 flow_bb_inside_loop_p calls.
9755 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
9757 PR tree-optimization/86871
9758 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
9759 instead of gimple_assign_lhs.
9761 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
9764 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
9765 register constraint to operand 0.
9766 (add<mode>3_carryinC): Likewise.
9767 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
9769 2018-08-09 Martin Liska <mliska@suse.cz>
9772 * common.opt: Remove extra line.
9774 2018-08-09 Martin Liska <mliska@suse.cz>
9776 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
9777 at the end of a line, make first letter capital and end up
9778 a sentence with a dot.
9779 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
9780 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
9781 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
9782 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
9783 (PARAM_MAX_ISL_OPERATIONS): Likewise.
9784 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
9785 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
9786 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
9787 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
9788 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
9789 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
9790 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
9791 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
9792 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
9793 (PARAM_TREE_REASSOC_WIDTH): Likewise.
9794 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
9795 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
9796 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
9798 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
9801 * config/s390/s390.c (s390_option_override_internal): Reduce the
9802 stack-clash-protection-probe-interval param if it would be too big
9805 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
9808 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
9809 * config/m68k/m68k.c (handle_move_double): Don't call
9810 m68k_final_prescan_insn.
9811 (m68k_adjust_decorated_operand): Renamed from
9812 m68k_final_prescan_insn, remove first and third operand and
9814 (print_operand): Call it.
9815 (print_operand_address): Call it.
9817 2018-08-08 Nathan Sidwell <nathan@acm.org>
9819 * diagnostic.c (diagnostic_report_current_module): Use
9820 linemap_included_from & linemap_included_from_linemap.
9822 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
9824 * config/aarch64/aarch64-cores.def: Add phecda core.
9825 * config/aarch64/aarch64-tune.md: Regenerate.
9826 * doc/invoke.texi: Add phecda core.
9828 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
9831 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
9833 * config/s390/s390.md ("movti"): Add more alternatives for
9834 constant to GPR copies.
9836 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
9838 * config/s390/s390.c: Fix whitespace damage throughout the file.
9839 * config/s390/s390.h: Likewise.
9840 * config/s390/tpf.h: Likewise.
9842 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
9844 * config/s390/s390.c (s390_loadrelative_operand_p):
9845 Remove TARGET_CPU_ZARCH usages.
9846 (s390_rtx_costs): Likewise.
9847 (s390_legitimate_constant_p): Likewise.
9848 (s390_cannot_force_const_mem): Likewise.
9849 (legitimate_reload_constant_p): Likewise.
9850 (s390_preferred_reload_class): Likewise.
9851 (legitimize_pic_address): Likewise.
9852 (legitimize_tls_address): Likewise.
9853 (s390_split_branches): Removed.
9854 (s390_add_execute): Removed.
9855 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
9856 (s390_mainpool_start): Likewise.
9857 (s390_mainpool_finish): Likewise.
9858 (s390_mainpool_cancel): Removed.
9859 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
9860 (s390_chunkify_cancel): Likewise.
9861 (s390_return_addr_rtx): Likewise.
9862 (s390_register_info): Remove split_branches_pending_p uages.
9863 (s390_optimize_register_info): Likewise.
9864 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
9865 split_branches_pending_p usages.
9866 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
9867 (s390_load_got): Likewise.
9868 (s390_expand_split_stack_prologue): Likewise.
9869 (output_asm_nops): Likewise.
9870 (s390_function_profiler): Likewise.
9871 (s390_emit_call): Likewise.
9872 (s390_conditional_register_usage): Likewise.
9873 (s390_optimize_prologue): Likewise.
9874 (s390_reorg): Remove TARGET_CPU_ZARCH and
9875 split_branches_pending_p usages.
9876 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
9878 (s390_output_indirect_thunk_function): Likewise.
9879 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
9880 (TARGET_CPU_ZARCH_P): Removed.
9881 (struct machine_function): Remove split_branches_pending_p.
9882 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
9884 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
9886 * common/config/s390/s390-common.c (processor_flags_table):
9888 * config.gcc: Remove with_arch/with_tune support.
9889 * config/s390/2064.md: Remove cpu attribute comparisons.
9890 * config/s390/driver-native.c (s390_host_detect_local_cpu):
9892 * config/s390/linux.h (ASM_SPEC):
9893 Remove -march support.
9894 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
9895 Use a table to get an arch level.
9896 * config/s390/s390-opts.h (enum processor_type):
9898 * config/s390/s390.c
9899 (processor_table): Remove entries, add arch_level values.
9900 (s390_issue_rate): Remove cases.
9901 (s390_option_override): Adjust
9902 s390_option_override_internal() call.
9903 (s390_option_override_internal): Remove deprecation warning.
9904 (s390_valid_target_attribute_tree): Adjust
9905 s390_option_override_internal() call.
9906 * config/s390/s390.h (struct s390_processor):
9907 Share with s390-c.c, add arch_level field.
9908 * config/s390/s390.md:
9909 Remove occurrences in cpu attribute.
9910 * config/s390/s390.opt: Remove -march/-mtune support.
9911 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
9912 * doc/invoke.texi: Remove deprecation warning.
9914 2018-08-08 Luis Machado <luis.machado@linaro.org>
9916 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
9918 (qdf24xx_tunings): Set vector cost structure to
9919 qdf24xx_vector_cost.
9921 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
9922 <register_sextend>: Set to 3.
9924 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
9927 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
9928 * config/aarch64/aarch64-simd.md
9929 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
9930 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
9931 (aarch64_frecpx<mode>): ...this new pattern.
9932 * config/aarch64/aarch64-simd-builtins.def: Remove comment
9933 about aarch64_frecp<FRECP:frecp_suffix><mode>.
9935 2018-08-07 Martin Liska <mliska@suse.cz>
9938 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
9939 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
9940 * predict.def (PRED_MALLOC_NONNULL): New predictor.
9941 * doc/extend.texi: Document that malloc attribute adds
9944 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
9947 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9948 Define to speculation_safe_value_not_needed.
9950 2018-08-06 Jeff Law <law@redhat.com>
9952 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
9953 the vr_values instance to cprop_into_stmt.
9954 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
9955 (cprop_operand): Also query EVRP to determine if OP is a constant.
9957 2018-08-06 Nathan Sidwell <nathan@acm.org>
9959 * diagnostic.c (diagnostic_report_current_module): Reroll
9960 included-at loop. Translate text.
9962 2018-08-06 David Malcolm <dmalcolm@redhat.com>
9964 * function-tests.c (selftest::test_expansion_to_rtl): Call
9965 free_after_compilation.
9967 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9969 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
9971 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
9973 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
9974 loops with memory block operations from getting unrolled.
9976 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
9979 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9980 Define to speculation_safe_value_not_needed.
9982 2018-08-06 Jeff Law <law@redhat.com>
9984 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
9987 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9990 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
9991 with all enabled __intN types.
9993 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
9995 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9997 * alias.c (record_set): Check for clobber high.
9998 * cfgexpand.c (expand_gimple_stmt): Likewise.
9999 * combine-stack-adj.c (single_set_for_csa): Likewise.
10000 * combine.c (find_single_use_1): Likewise.
10001 (set_nonzero_bits_and_sign_copies): Likewise.
10002 (get_combine_src_dest): Likewise.
10003 (is_parallel_of_n_reg_sets): Likewise.
10004 (try_combine): Likewise.
10005 (record_dead_and_set_regs_1): Likewise.
10006 (reg_dead_at_p_1): Likewise.
10007 (reg_dead_at_p): Likewise.
10008 * dce.c (deletable_insn_p): Likewise.
10009 (mark_nonreg_stores_1): Likewise.
10010 (mark_nonreg_stores_2): Likewise.
10011 * df-scan.c (df_find_hard_reg_defs): Likewise.
10012 (df_uses_record): Likewise.
10013 (df_get_call_refs): Likewise.
10014 * dwarf2out.c (mem_loc_descriptor): Likewise.
10015 * haifa-sched.c (haifa_classify_rtx): Likewise.
10016 * ira-build.c (create_insn_allocnos): Likewise.
10017 * ira-costs.c (scan_one_insn): Likewise.
10018 * ira.c (equiv_init_movable_p): Likewise.
10019 (rtx_moveable_p): Likewise.
10020 (interesting_dest_for_shprep): Likewise.
10021 * jump.c (mark_jump_label_1): Likewise.
10022 * postreload-gcse.c (record_opr_changes): Likewise.
10023 * postreload.c (reload_cse_simplify): Likewise.
10024 (struct reg_use): Add source expr.
10025 (reload_combine): Check for clobber high.
10026 (reload_combine_note_use): Likewise.
10027 (reload_cse_move2add): Likewise.
10028 (move2add_note_store): Likewise.
10029 * print-rtl.c (print_pattern): Likewise.
10030 * recog.c (decode_asm_operands): Likewise.
10031 (store_data_bypass_p): Likewise.
10032 (if_test_bypass_p): Likewise.
10033 * regcprop.c (kill_clobbered_value): Likewise.
10034 (kill_set_value): Likewise.
10035 * reginfo.c (reg_scan_mark_refs): Likewise.
10036 * reload1.c (maybe_fix_stack_asms): Likewise.
10037 (eliminate_regs_1): Likewise.
10038 (elimination_effects): Likewise.
10039 (mark_not_eliminable): Likewise.
10040 (scan_paradoxical_subregs): Likewise.
10041 (forget_old_reloads_1): Likewise.
10042 * reorg.c (find_end_label): Likewise.
10043 (try_merge_delay_insns): Likewise.
10044 (redundant_insn): Likewise.
10045 (own_thread_p): Likewise.
10046 (fill_simple_delay_slots): Likewise.
10047 (fill_slots_from_thread): Likewise.
10048 (dbr_schedule): Likewise.
10049 * resource.c (update_live_status): Likewise.
10050 (mark_referenced_resources): Likewise.
10051 (mark_set_resources): Likewise.
10052 * rtl.c (copy_rtx): Likewise.
10053 * rtlanal.c (reg_referenced_p): Likewise.
10054 (single_set_2): Likewise.
10055 (noop_move_p): Likewise.
10056 (note_stores): Likewise.
10057 * sched-deps.c (sched_analyze_reg): Likewise.
10058 (sched_analyze_insn): Likewise.
10060 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10062 * cse.c (invalidate_reg): New function extracted from...
10063 (invalidate): ...here.
10064 (canonicalize_insn): Check for clobber high.
10065 (invalidate_from_clobbers): invalidate clobber highs.
10066 (invalidate_from_sets_and_clobbers): Likewise.
10067 (count_reg_usage): Check for clobber high.
10068 (insn_live_p): Likewise.
10069 * cselib.c (cselib_expand_value_rtx_1):Likewise.
10070 (cselib_invalidate_regno): Check for clobber in setter.
10071 (cselib_invalidate_rtx): Pass through setter.
10072 (cselib_invalidate_rtx_note_stores):
10073 (cselib_process_insn): Check for clobber high.
10074 * cselib.h (cselib_invalidate_rtx): Add operand.
10076 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10078 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
10079 (mark_not_eliminable): Likewise.
10080 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
10081 * lra-lives.c (process_bb_lives): Check for clobber high.
10082 * lra.c (new_insn_reg): Remember clobber highs.
10083 (collect_non_operand_hard_regs): Check for clobber high.
10084 (lra_set_insn_recog_data): Likewise.
10085 (add_regs_to_insn_regno_info): Likewise.
10086 (lra_update_insn_regno_info): Likewise.
10088 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10090 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
10091 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
10093 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10095 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
10096 (copy_insn_1): Likewise.
10097 (gen_hard_reg_clobber_high): New gen function.
10098 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
10099 * genemit.c (gen_exp): Likewise.
10100 (gen_emit_seq): Pass through info.
10101 (gen_insn): Check for CLOBBER_HIGH.
10102 (gen_expand): Pass through info.
10103 (gen_split): Likewise.
10104 (output_add_clobbers): Likewise.
10105 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
10106 (remove_clobbers): Likewise.
10107 * rtl.h (gen_hard_reg_clobber_high): New declaration.
10109 2018-08-06 Alan Hayward <alan.hayward@arm.com>
10111 * doc/rtl.texi (clobber_high): Add.
10112 (parallel): Add in clobber high
10113 * rtl.c (rtl_check_failed_code3): Add function.
10114 * rtl.def (CLOBBER_HIGH): Add expression.
10115 * rtl.h (RTL_CHECKC3): Add macro.
10116 (rtl_check_failed_code3): Add declaration.
10117 (XC3EXP): Add macro.
10119 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
10122 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
10123 cfun->machine->max_used_stack_alignment if needed.
10125 2018-08-04 Martin Sebor <msebor@redhat.com>
10127 PR tree-optimization/86571
10128 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
10131 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
10134 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10137 2018-08-03 Jeff Law <law@redhat.com>
10140 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10141 Define to speculation_safe_value_not_needed.
10143 2018-08-03 David Malcolm <dmalcolm@redhat.com>
10145 * doc/gcov.texi (-x): Remove duplicate "to".
10146 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
10147 (-Wif-not-aligned): Remove duplicate "is".
10148 (-flto): Remove duplicate "the".
10149 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
10150 duplicate "v5.00.b".
10151 (MSP430 Options): Remove duplicate "and" from the description
10152 of "-mgprel-sec=regexp".
10153 (x86 Options): Remove duplicate copies of "vmldLog102" and
10154 vmlsLog104 from description of "-mveclibabi=type".
10156 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
10158 * internal-fn.h (first_commutative_argument): Declare.
10159 * internal-fn.c (first_commutative_argument): New function.
10160 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
10161 restrictions for pattern statements. Use first_commutative_argument
10162 to look for commutative operands in calls to internal functions.
10164 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
10166 * Makefile.in (wide-int-range.o): New.
10167 * tree-vrp.c: Move all the wide_int_* functions to...
10168 * wide-int-range.cc: ...here.
10169 * tree-vrp.h: Move all the wide_int_* prototypes to...
10170 * wide-int-range.h: ...here.
10172 2018-08-03 Tom de Vries <tdevries@suse.de>
10174 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
10176 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
10177 * except.c (output_function_exception_table): Do early exit if
10178 targetm_common.except_unwind_info (&global_options) == UI_NONE.
10180 2018-08-03 Martin Liska <mliska@suse.cz>
10182 * predict.c (dump_prediction): Change to 2 digits
10185 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
10187 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
10189 (falkor_am_1_gtov_gtov): ... a new insn reservation.
10191 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
10193 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
10194 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
10195 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
10196 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
10197 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
10199 2018-08-02 David Malcolm <dmalcolm@redhat.com>
10201 * diagnostic-show-locus.c (diagnostic_show_locus): Use
10202 pp_take_prefix when saving the existing prefix.
10203 * diagnostic.c (diagnostic_append_note): Likewise.
10204 * langhooks.c (lhd_print_error_function): Likewise.
10205 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
10206 param's type. Free the existing prefix.
10207 (pp_take_prefix): New function.
10208 (pretty_printer::pretty_printer): Drop the prefix parameter.
10209 Rename the length parameter to match the comment.
10210 (pretty_printer::~pretty_printer): Free the prefix.
10211 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
10213 (struct pretty_printer): Drop the "const" from "prefix" field's
10214 type and clarify memory management.
10215 (pp_set_prefix): Drop the "const" from the 2nd param.
10216 (pp_take_prefix): New decl.
10218 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
10220 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
10221 for word_mode_ok here instead of passing as argument.
10222 (expand_block_compare): Change select_block_compare_mode() call.
10223 (expand_strncmp_gpr_sequence): New function.
10224 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
10226 2018-08-02 Jeff Law <law@redhat.com>
10229 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10230 Define to speculation_safe_value_not_needed.
10233 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10234 Define to speculation_safe_value_not_needed.
10236 2018-08-02 Tom de Vries <tdevries@suse.de>
10239 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
10240 function. Return UI_TARGET unconditionally.
10241 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
10242 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
10244 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10246 * genemit.c (print_overload_test): Fix typo.
10248 2018-08-02 Richard Biener <rguenther@suse.de>
10250 PR tree-optimization/86816
10251 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
10252 which checks for value availability before querying it.
10253 (gvn_uses_equal): Use it.
10254 (same_succ_hash): Likewise.
10255 (gimple_equal_p): Likewise.
10257 2018-08-02 Nick Clifton <nickc@redhat.com>
10260 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10261 Define to speculation_safe_value_not_needed.
10264 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10265 Define to speculation_safe_value_not_needed.
10268 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10269 Define to speculation_safe_value_not_needed.
10272 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10273 Define to speculation_safe_value_not_needed.
10276 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10277 Define to speculation_safe_value_not_needed.
10280 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10281 Define to speculation_safe_value_not_needed.
10284 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10285 Define to speculation_safe_value_not_needed.
10288 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
10289 Define to speculation_safe_value_not_needed.
10292 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10293 speculation_safe_value_not_needed.
10296 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
10297 to speculation_safe_value_not_needed.
10299 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10301 * doc/md.texi: Expand the documentation of instruction names
10302 to mention port-local uses. Document '@' in pattern names.
10303 * read-md.h (overloaded_instance, overloaded_name): New structs.
10304 (mapping): Declare.
10305 (md_reader::handle_overloaded_name): New member function.
10306 (md_reader::get_overloads): Likewise.
10307 (md_reader::m_first_overload): New member variable.
10308 (md_reader::m_next_overload_ptr): Likewise.
10309 (md_reader::m_overloads_htab): Likewise.
10310 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
10311 m_next_overload_ptr and m_overloads_htab.
10312 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
10313 (get_mode_token, get_code_token, get_int_token): New functions.
10314 (map_attr_string): Add an optional argument that passes back
10315 the associated iterator.
10316 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
10317 (md_reader::handle_overloaded_name, add_overload_instance): New
10319 (apply_iterators): Handle '@' names. Report an error if '@'
10320 is used without iterators.
10321 (initialize_iterators): Initialize the new iterator_group fields.
10322 * genopinit.c (handle_overloaded_code_for)
10323 (handle_overloaded_gen): New functions.
10324 (main): Use them to print declarations of maybe_code_for_* and
10325 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
10326 * genemit.c (print_overload_arguments, print_overload_test)
10327 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
10328 (main): Use it to print definitions of maybe_code_for_* and
10329 maybe_gen_* functions.
10330 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
10331 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
10332 instead of explicit mode checks.
10333 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
10334 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
10335 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
10336 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
10337 (aarch64_expand_compare_and_swap): Likewise
10338 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
10339 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
10340 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
10341 (aarch64_constant_pool_reload_icode): Delete.
10342 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
10343 instead of aarch64_constant_pool_reload_icode. Use
10344 code_for_aarch64_reload_mov instead of explicit mode checks.
10345 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
10346 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
10347 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
10348 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
10349 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
10350 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
10351 (aarch64_atomic_load_op_code): Delete.
10352 (aarch64_emit_atomic_load_op): Likewise.
10353 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
10354 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
10355 instead of aarch64_emit_atomic_load_op.
10356 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
10357 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
10358 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
10359 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
10360 character before the pattern name.
10361 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
10362 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
10363 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
10364 (aarch64_frecps<mode>): Likewise.
10365 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
10366 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
10367 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
10368 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
10369 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
10371 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
10373 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
10374 Allow HFmode constants if TARGET_FP_F16INST.
10376 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
10379 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
10380 No longer check last store for clobber of address register.
10382 2018-08-02 Martin Liska <mliska@suse.cz>
10384 PR gcov-profile/86817
10385 * gcov.c (process_all_functions): New function.
10387 (process_file): Move functions processing to
10388 process_all_functions.
10390 2018-08-02 David Malcolm <dmalcolm@redhat.com>
10392 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
10393 "const" to the "gimple *" and "rtx_insn *" parameters.
10394 * dumpfile.h (dump_user_location_t::dump_user_location_t):
10396 (dump_location_t::dump_location_t): Likewise.
10398 2018-08-01 Martin Sebor <msebor@redhat.com>
10400 PR tree-optimization/86650
10401 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
10402 rather than a "gcall *". Directly pass the data of interest
10403 to percent_K_format, rather than building a temporary CALL_EXPR
10405 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
10406 (gimple_fold_builtin_strncat): Adjust.
10407 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
10408 gcall* argument with gimple*.
10409 * gimple-ssa-warn-restrict.c (check_call): Same.
10410 (wrestrict_dom_walker::before_dom_children): Same.
10411 (builtin_access::builtin_access): Same.
10412 (check_bounds_or_overlap): Same
10413 (maybe_diag_overlap): Same.
10414 (maybe_diag_offset_bounds): Same.
10415 * tree-diagnostic.c (default_tree_printer): Move usage of
10416 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
10418 * tree-pretty-print.c (percent_K_format): Add argument.
10419 * tree-pretty-print.h: Add argument.
10420 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
10421 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
10422 (maybe_diag_stxncpy_trunc): Same.
10423 (handle_builtin_stxncpy): Same.
10424 (handle_builtin_strcat): Same.
10426 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10428 * match.pd: Optimise pointer range checks.
10430 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10432 PR tree-optimization/86758
10433 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
10434 to remove pattern statements.
10436 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10438 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
10439 result of dfs_enumerate_from when constructing stmt_vec_infos,
10440 instead of additionally calling get_loop_body.
10442 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10444 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
10446 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
10447 When creating an iv, assert that the step is not known to be zero.
10448 (vect_setup_realignment): Update call accordingly.
10449 * tree-vect-stmts.c (vectorizable_store): Likewise.
10450 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
10452 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10454 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
10455 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
10456 (vectorizable_reduction): Likewise.
10457 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10458 (vect_detect_hybrid_slp_stmts): Likewise.
10459 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
10461 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
10463 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
10464 (wide_int_set_zero_nonzero_bits): ...this.
10465 (zero_nonzero_bits_from_vr): Rename to...
10466 (vrp_set_zero_nonzero_bits): ...this.
10467 (extract_range_from_multiplicative_op_1): Abstract wide int
10469 (wide_int_range_multiplicative_op): ...here.
10470 (extract_range_from_binary_expr_1): Extract wide int binary
10471 operations into their own functions.
10472 (wide_int_range_lshift): New.
10473 (wide_int_range_can_optimize_bit_op): New.
10474 (wide_int_range_shift_undefined_p): New.
10475 (wide_int_range_bit_xor): New.
10476 (wide_int_range_bit_ior): New.
10477 (wide_int_range_bit_and): New.
10478 (wide_int_range_trunc_mod): New.
10479 (extract_range_into_wide_ints): New.
10480 (vrp_shift_undefined_p): New.
10481 (extract_range_from_multiplicative_op): New.
10482 (vrp_can_optimize_bit_op): New.
10483 * tree-vrp.h (value_range::dump): New.
10484 (wide_int_range_multiplicative_op): New.
10485 (wide_int_range_lshift):New.
10486 (wide_int_range_shift_undefined_p): New.
10487 (wide_int_range_bit_xor): New.
10488 (wide_int_range_bit_ior): New.
10489 (wide_int_range_bit_and): New.
10490 (wide_int_range_trunc_mod): New.
10491 (zero_nonzero_bits_from_bounds): Rename to...
10492 (wide_int_set_zero_nonzero_bits): ...this.
10493 (zero_nonzero_bits_from_vr): Rename to...
10494 (vrp_set_zero_nonzero_bits): ...this.
10495 (range_easy_mask_min_max): Rename to...
10496 (wide_int_range_can_optimize_bit_op): this.
10497 * vr-values.c (simplify_bit_ops_using_ranges): Rename
10498 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
10500 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10502 * tree-vectorizer.h (vect_orig_stmt): New function.
10503 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
10504 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
10505 (vect_create_epilog_for_reduction): Likewise.
10506 (vectorizable_live_operation): Likewise.
10507 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
10508 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
10509 * tree-vect-stmts.c (vectorizable_call): Likewise.
10510 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
10512 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10514 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
10516 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
10517 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
10518 (vect_transform_loop): Likewise.
10519 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10521 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10523 * tree-vectorizer.h (vect_schedule_slp): Return void.
10524 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10525 (vect_schedule_slp): Likewise.
10527 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10529 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
10531 (vect_transform_loop): Update calls accordingly. Schedule SLP
10532 instances before the main loop, if any exist.
10534 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
10536 PR tree-optimization/86749
10537 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
10538 If the lhs is used in a COND_EXPR, check that it is being used
10539 as the "then" or "else" value.
10541 2018-08-01 Tom de Vries <tdevries@suse.de>
10544 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
10545 speculation_safe_value_not_needed.
10547 2018-08-01 Richard Biener <rguenther@suse.de>
10549 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
10550 as base and offset.
10552 2018-08-01 Martin Liska <mliska@suse.cz>
10554 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
10555 format how successful transformation is dumped.
10556 (gimple_mod_pow2_value_transform): Likewise.
10557 (gimple_mod_subtract_transform): Likewise.
10558 (gimple_stringops_transform): Likewise.
10560 2018-08-01 Martin Liska <mliska@suse.cz>
10562 PR value-prof/35543
10563 * value-prof.c (interesting_stringop_to_profile_p):
10564 Simplify the code and add BUILT_IN_MEMMOVE.
10565 (gimple_stringops_transform): Likewise.
10567 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
10569 * config/aarch64/aarch64-simd.md
10570 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
10571 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
10572 use GPI iterator instead of SI mode.
10574 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10576 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
10577 rs6000_speculation_barrier.
10578 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
10579 new barrier pattern name.
10581 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
10583 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
10584 (speculation_barrier): New insn.
10586 2018-08-01 Richard Biener <rguenther@suse.de>
10589 * graphite.h: Include isl/id.h and isl/space.h to allow build
10592 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
10595 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
10596 mode for COFF targets.
10597 * defaults.h (TARGET_COFF): Define.
10598 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
10599 TARGET_COFF): Define.
10600 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
10601 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
10602 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
10604 2018-07-31 Alexandre Oliva <oliva@adacore.com>
10606 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
10607 * gimple-streamer-out.c (output_bb): Save it.
10608 * lto-streamer-in.c (input_struct_function_base): Restore
10609 instance discriminator if available. Create map on demand.
10610 * lto-streamer-out.c (output_struct_function_base): Save it if
10612 * final.c (decl_to_instance_map): Document LTO strategy.
10614 2018-07-31 Alexandre Oliva <oliva@adacore.com>
10615 Olivier Hainque <hainque@adacore.com>
10617 * debug.h (decl_to_instance_map_t): New type.
10618 (decl_to_instance_map): Declare.
10619 (maybe_create_decl_to_instance_map): New inline function.
10620 * final.c (bb_discriminator, last_bb_discriminator): New statics,
10621 to track basic block discriminators.
10622 (final_start_function_1): Initialize them.
10623 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
10625 (decl_to_instance_map): New variable.
10626 (map_decl_to_instance, maybe_set_discriminator): New functions.
10627 (notice_source_line): Set discriminator.
10629 2018-07-31 Ian Lance Taylor <iant@golang.org>
10631 * targhooks.c (default_have_speculation_safe_value): Add
10634 2018-07-31 David Malcolm <dmalcolm@redhat.com>
10636 * dump-context.h: Include "pretty-print.h".
10637 (dump_context::refresh_dumps_are_enabled): New decl.
10638 (dump_context::emit_item): New decl.
10639 (class dump_context): Add fields "m_test_pp" and
10641 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
10642 (temp_dump_context::get_dumped_text): New decl.
10643 (class temp_dump_context): Add field "m_pp".
10644 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
10645 (dump_context::refresh_dumps_are_enabled): ...and add a test for
10647 (set_dump_file): Update for above change.
10648 (set_alt_dump_file): Likewise.
10649 (dump_loc): New overload, taking a pretty_printer *.
10650 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
10651 to any test pretty-printer.
10652 (make_item_for_dump_gimple_stmt): New function, adapted from
10653 optinfo::add_gimple_stmt.
10654 (dump_context::dump_gimple_stmt): Call it, and use the result,
10655 eliminating the direct usage of dump_file and alt_dump_file in
10656 favor of indirectly using them via emit_item.
10657 (make_item_for_dump_gimple_expr): New function, adapted from
10658 optinfo::add_gimple_expr.
10659 (dump_context::dump_gimple_expr): Call it, and use the result,
10660 eliminating the direct usage of dump_file and alt_dump_file in
10661 favor of indirectly using them via emit_item.
10662 (make_item_for_dump_generic_expr): New function, adapted from
10664 (dump_context::dump_generic_expr): Call it, and use the result,
10665 eliminating the direct usage of dump_file and alt_dump_file in
10666 favor of indirectly using them via emit_item.
10667 (make_item_for_dump_printf_va): New function, adapted from
10668 optinfo::add_printf_va.
10669 (make_item_for_dump_printf): New function.
10670 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
10671 and use the result, eliminating the direct usage of dump_file and
10672 alt_dump_file in favor of indirectly using them via emit_item.
10673 (make_item_for_dump_dec): New function.
10674 (dump_context::dump_dec): Call it, and use the result,
10675 eliminating the direct usage of dump_file and alt_dump_file in
10676 favor of indirectly using them via emit_item.
10677 (make_item_for_dump_symtab_node): New function, adapted from
10678 optinfo::add_symtab_node.
10679 (dump_context::dump_symtab_node): Call it, and use the result,
10680 eliminating the direct usage of dump_file and alt_dump_file in
10681 favor of indirectly using them via emit_item.
10682 (dump_context::begin_scope): Reimplement, avoiding direct usage
10683 of dump_file and alt_dump_file in favor of indirectly using them
10685 (dump_context::emit_item): New member function.
10686 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
10687 Set up test pretty-printer on the underlying context. Call
10688 refresh_dumps_are_enabled.
10689 (temp_dump_context::~temp_dump_context): Call
10690 refresh_dumps_are_enabled.
10691 (temp_dump_context::get_dumped_text): New member function.
10692 (selftest::verify_dumped_text): New function.
10693 (ASSERT_DUMPED_TEXT_EQ): New macro.
10694 (selftest::test_capture_of_dump_calls): Run all tests twice, with
10695 and then without optinfo enabled. Add uses of
10696 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
10697 * dumpfile.h: Update comment for the dump_* API.
10698 * optinfo-emit-json.cc
10699 (selftest::test_building_json_from_dump_calls): Update for new
10700 param for temp_dump_context ctor.
10701 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
10702 and "m_owned" field.
10703 (optinfo_item::~optinfo_item): Likewise.
10704 (optinfo::add_item): New member function.
10705 (optinfo::emit): Update comment.
10706 (optinfo::add_string): Delete.
10707 (optinfo::add_printf): Delete.
10708 (optinfo::add_printf_va): Delete.
10709 (optinfo::add_gimple_stmt): Delete.
10710 (optinfo::add_gimple_expr): Delete.
10711 (optinfo::add_tree): Delete.
10712 (optinfo::add_symtab_node): Delete.
10713 (optinfo::add_dec): Delete.
10714 * optinfo.h (class dump_context): New forward decl.
10715 (optinfo::add_item): New decl.
10716 (optinfo::add_string): Delete.
10717 (optinfo::add_printf): Delete.
10718 (optinfo::add_printf_va): Delete.
10719 (optinfo::add_gimple_stmt): Delete.
10720 (optinfo::add_gimple_expr): Delete.
10721 (optinfo::add_tree): Delete.
10722 (optinfo::add_symtab_node): Delete.
10723 (optinfo::add_dec): Delete.
10724 (optinfo::add_poly_int): Delete.
10725 (optinfo_item::optinfo_item): Remove "owned" param.
10726 (class optinfo_item): Remove field "m_owned".
10728 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10730 PR middle-end/86705
10731 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
10732 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
10733 requested variable alignment.
10734 (expand_one_ssa_partition): Likewise.
10735 (expand_one_var): Likewise.
10737 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10739 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
10740 to speculation_safe_value_not_needed.
10742 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10744 * targhooks.h (speculation_safe_value_not_needed): New prototype.
10745 * targhooks.c (speculation_safe_value_not_needed): New function.
10746 * target.def (have_speculation_safe_value): Update documentation.
10747 * doc/tm.texi: Regenerated.
10749 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10751 * config/aarch64/iterators.md (ALLI_TI): New iterator.
10752 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
10754 (despeculate_copy<ALLI:mode>_insn): New insn.
10755 (despeculate_copyti_insn): New insn.
10756 (despeculate_simple<ALLI:mode>): New insn
10757 (despeculate_simpleti): New insn.
10758 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
10760 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
10761 aarch64_speculation_safe_value.
10762 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
10764 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10766 * config/aarch64/aarch64-speculation.cc: New file.
10767 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
10768 before pass_reorder_blocks.
10769 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
10771 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
10772 X14 and X15 when tracking speculation.
10773 * config/aarch64/aarch64.md (register name constants): Add
10774 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
10775 (unspec): Add UNSPEC_SPECULATION_TRACKER.
10776 (speculation_barrier): New insn attribute.
10777 (cmp<mode>): Allow SP in comparisons.
10778 (speculation_tracker): New insn.
10779 (speculation_barrier): Add speculation_barrier attribute.
10780 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
10781 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
10782 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
10784 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10786 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
10787 aarch64_track_speculation is true.
10788 (tb<optab><mode>1): Likewise.
10789 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
10790 generate CB[N]Z when tracking speculation.
10791 (aarch64_split_compare_and_swap): Likewise.
10792 (aarch64_split_atomic_op): Likewise.
10794 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10796 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
10798 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10800 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
10801 (speculation_barrier): New insn.
10803 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10805 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
10806 * config/arm/arm.md (speculation_barrier): New expand.
10807 (speculation_barrier_insn): New pattern.
10809 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
10811 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
10812 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
10813 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
10814 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
10816 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
10817 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
10818 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
10819 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
10820 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
10821 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
10822 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
10823 * builtins.c (expand_speculation_safe_value): New function.
10824 (expand_builtin): Call it.
10825 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
10826 * doc/extend.texi: Document __builtin_speculation_safe_value.
10827 * doc/md.texi: Document "speculation_barrier" pattern.
10828 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
10829 TARGET_HAVE_SPECULATION_SAFE_VALUE.
10830 * doc/tm.texi: Regenerated.
10831 * target.def (have_speculation_safe_value, speculation_safe_value): New
10833 * targhooks.c (default_have_speculation_safe_value): New function.
10834 (default_speculation_safe_value): New function.
10835 * targhooks.h (default_have_speculation_safe_value): Add prototype.
10836 (default_speculation_safe_value): Add prototype.
10838 2018-07-31 David Malcolm <dmalcolm@redhat.com>
10840 * dump-context.h (dump_context::dump_loc): New decl.
10841 * dumpfile.c (dump_context::dump_loc): New member function.
10842 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
10843 and dump_gimple_stmt.
10844 (dump_context::dump_gimple_expr_loc): Likewise, using
10846 (dump_context::dump_generic_expr_loc): Likewise, using
10848 (dump_context::dump_printf_loc_va): Likewise, using
10850 (dump_context::begin_scope): Explicitly using the global function
10851 "dump_loc", rather than the member function.
10853 2018-07-31 Martin Sebor <msebor@redhat.com>
10855 PR tree-optimization/86741
10856 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
10858 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
10860 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
10861 depend on whether prefetch instructions will be emitted or not.
10862 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
10863 will be emitted or not.
10864 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
10865 (TARGET_SETMEM_PFD): New macros.
10867 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10869 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
10870 (NULL_STMT_VEC_INFO): Delete.
10871 (stmt_vec_info::operator*): Likewise.
10872 (stmt_vec_info::operator gimple *): Likewise.
10873 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
10874 of NULL_STMT_VEC_INFO.
10875 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10876 (vect_reassociating_reduction_p): Likewise.
10877 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
10878 (vectorizable_store): Likewise.
10879 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
10880 (vec_info::free_stmt_vec_infos): Likewise.
10882 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10884 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
10885 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
10887 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10889 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
10890 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
10891 (vec_info::free_stmt_vec_info): New private member functions.
10892 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
10893 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
10894 * tree-parloops.c (gather_scalar_reductions): Remove calls to
10895 set_stmt_vec_info_vec and free_stmt_vec_infos.
10896 * tree-vect-loop.c (_loop_vec_info): Remove call to
10897 set_stmt_vec_info_vec.
10898 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
10899 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
10900 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
10901 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
10902 (vec_info::free_stmt_vec_info): ...these new functions. Remove
10903 assignments in {vec_info::,}new_stmt_vec_info that are redundant
10904 with the clearing in the xcalloc.
10905 (stmt_vec_info_vec): Delete.
10906 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
10907 (vectorize_loops): Likewise.
10908 (vec_info::~vec_info): Remove argument from call to
10909 free_stmt_vec_infos.
10910 (vec_info::add_stmt): Remove vinfo argument from call to
10913 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10915 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
10916 rather than a gimple stmt.
10917 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
10918 information for pattern statements when passed the original
10919 statement; instead wait to be passed the pattern statement itself.
10920 Don't call set_vinfo_for_stmt here.
10921 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
10922 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
10923 stmt_vec_infos here.
10924 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
10925 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
10926 stmt_vec_infos entry.
10928 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10930 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
10931 * tree-vectorizer.c (vec_info::replace_stmt): New function.
10932 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
10933 * tree-vect-stmts.c (vectorizable_call): Likewise.
10934 (vectorizable_simd_clone_call): Likewise.
10936 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10938 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
10939 * tree-vectorizer.c (vec_info::remove_stmt): New function.
10940 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
10941 * tree-vect-loop.c (vect_transform_loop): Likewise.
10942 * tree-vect-slp.c (vect_schedule_slp): Likewise.
10943 * tree-vect-stmts.c (vect_remove_stores): Likewise.
10945 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10947 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
10948 (vect_dr_stmt): Delete.
10949 * tree-vectorizer.c (vec_info::lookup_dr): New function.
10950 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
10952 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
10953 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
10954 (vect_verify_datarefs_alignment, vect_peeling_supportable)
10955 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
10956 (vect_analyze_data_refs): Likewise.
10957 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
10959 (vect_find_same_alignment_drs): Likewise.
10960 (vect_slp_analyze_node_dependences): Update calls accordingly.
10961 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
10962 instead of DR_VECT_AUX.
10963 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
10964 of a vector data references. Use vec_info::lookup_dr instead of
10966 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
10967 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
10968 instead of DR_VECT_AUX.
10970 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10972 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
10974 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
10976 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10977 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10978 (vect_gen_prolog_loop_niters): Likewise.
10980 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10982 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
10983 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
10984 (vect_known_alignment_in_bytes, vect_dr_behavior)
10985 (vect_get_scalar_dr_size): Take references as dr_vec_infos
10986 instead of data_references. Update calls to other routines for
10987 which the same change has been made.
10988 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
10989 dr_vec_infos instead of stmt_vec_infos.
10990 (vect_analyze_data_ref_dependence): Update call accordingly.
10991 (vect_slp_analyze_data_ref_dependence)
10992 (vect_record_base_alignments): Use DR_VECT_AUX.
10993 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
10994 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
10995 (vector_alignment_reachable_p, vect_get_data_access_cost)
10996 (vect_peeling_supportable, vect_analyze_group_access_1)
10997 (vect_analyze_group_access, vect_analyze_data_ref_access)
10998 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
10999 (vect_compile_time_alias, vect_small_gap_p)
11000 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
11001 (vect_supportable_dr_alignment): Take references as dr_vec_infos
11002 instead of data_references. Update calls to other routines for
11003 which the same change has been made.
11004 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
11005 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
11006 (vect_slp_analyze_and_verify_node_alignment)
11007 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
11008 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
11009 (vect_setup_realignment): Use dr_vec_infos. Update calls after
11011 (_vect_peel_info::dr): Replace with...
11012 (_vect_peel_info::dr_info): ...this new field.
11013 (vect_peeling_hash_get_most_frequent)
11014 (vect_peeling_hash_choose_best_peeling): Update accordingly.
11015 (vect_peeling_hash_get_lowest_cost):
11016 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
11017 routines for which the same change has been made.
11018 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
11020 * tree-vect-loop-manip.c (get_misalign_in_elems)
11021 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
11023 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11024 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
11025 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
11026 (get_group_load_store_type, get_negative_load_store_type)
11027 (vect_get_data_ptr_increment, vectorizable_store)
11028 (vectorizable_load): Likewise.
11029 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
11030 Update calls to other routines for which the same change has been made.
11032 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11034 * tree-vectorizer.h (vec_info::move_dr): New member function.
11035 (dataref_aux): Rename to...
11036 (dr_vec_info): ...this and add "dr" and "stmt" fields.
11037 (_stmt_vec_info::dr_aux): Update accordingly.
11038 (_stmt_vec_info::data_ref_info): Delete.
11039 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
11040 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
11041 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
11042 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
11044 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
11045 (STMT_VINFO_DR_INFO): New macro.
11046 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
11047 (set_dr_misalignment): Update after rename of dataref_aux.
11048 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
11049 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
11050 initialization of STMT_VINFO_DATA_REF.
11051 * tree-vectorizer.c (vec_info::move_dr): New function.
11052 * tree-vect-patterns.c (vect_recog_bool_pattern)
11053 (vect_recog_mask_conversion_pattern)
11054 (vect_recog_gather_scatter_pattern): Use it.
11055 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
11056 the "dr" and "stmt" fields of dr_vec_info instead of
11057 STMT_VINFO_DATA_REF.
11059 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11061 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
11062 (is_pattern_stmt_p): Use it.
11063 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
11064 on pattern statements.
11066 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11068 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
11069 original stmt as a stmt_vec_info rather than a gimple stmt.
11070 (vect_pattern_recog_1): Take the statement directly as a
11071 stmt_vec_info, rather than via a gimple_stmt_iterator.
11072 Update call to vect_mark_pattern_stmts.
11073 (vect_pattern_recog): Update calls accordingly.
11075 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11077 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
11078 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
11079 a vect_def_type for the first argument.
11080 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
11081 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
11082 operand if it isn't defined by a vectorized statement.
11083 (vect_build_gather_load_calls): Remove the mask_dt argument and
11084 update calls to vect_get_vec_def_for_stmt_copy.
11085 (vectorizable_bswap): Likewise the dt argument.
11086 (vectorizable_call): Update calls to vectorizable_bswap and
11087 vect_get_vec_def_for_stmt_copy.
11088 (vectorizable_simd_clone_call, vectorizable_assignment)
11089 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
11090 (vectorizable_comparison): Update calls to
11091 vect_get_vec_def_for_stmt_copy.
11092 (vectorizable_store): Likewise. Remove now-unnecessary calls to
11093 vect_is_simple_use.
11094 (vect_get_loop_based_defs): Remove dt argument and update call
11095 to vect_get_vec_def_for_stmt_copy.
11096 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
11097 and vect_get_vec_def_for_stmt_copy.
11098 (vectorizable_load): Update calls to vect_build_gather_load_calls
11099 and vect_get_vec_def_for_stmt_copy.
11100 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11101 (vectorizable_reduction, vectorizable_live_operation): Update calls
11102 to vect_get_vec_def_for_stmt_copy.
11104 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11106 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
11107 and gimple stmt arguments with a stmt_vec_info.
11108 (vect_record_base_alignments): Update calls accordingly.
11109 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
11110 and gimple stmt arguments with a stmt_vec_info.
11111 (vect_build_slp_tree_1): Remove vinfo argument and update call
11112 to vect_record_max_nunits.
11113 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
11114 and vect_record_max_nunits.
11116 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11118 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
11119 file and take a stmt_vec_info instead of a gimple stmt.
11120 (supportable_widening_operation, vect_finish_replace_stmt)
11121 (vect_finish_stmt_generation, vect_get_store_rhs)
11122 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
11123 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
11124 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
11125 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
11126 (vect_create_data_ref_ptr, bump_vector_ptr)
11127 (vect_permute_store_chain, vect_setup_realignment)
11128 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
11129 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
11130 (vectorizable_reduction, vectorizable_induction)
11131 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
11132 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
11133 than gimple stmts as arguments.
11134 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
11135 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
11136 (can_group_stmts_p, vect_check_gather_scatter)
11137 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
11138 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
11139 (vect_permute_load_chain, vect_shift_permute_load_chain)
11140 (vect_transform_grouped_load)
11141 (vect_record_grouped_load_vectors): Likewise.
11142 * tree-vect-loop.c (vect_fixup_reduc_chain)
11143 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
11144 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
11145 (vectorizable_reduction, vectorizable_induction)
11146 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
11147 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
11148 (vect_get_load_store_mask): Likewise.
11149 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
11150 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
11151 * tree-vect-stmts.c (vect_mark_relevant)
11152 (is_simple_and_all_uses_invariant)
11153 (exist_non_indexing_operands_for_use_p, process_use)
11154 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
11155 (vect_get_vec_def_for_operand, vect_get_vec_defs)
11156 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
11157 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
11158 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
11159 (get_negative_load_store_type, get_load_store_type)
11160 (vect_check_load_store_mask, vect_check_store_rhs)
11161 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
11162 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11163 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11164 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11165 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
11166 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
11167 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
11168 (supportable_widening_operation): Likewise.
11170 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11172 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
11173 a stmt_vec_info instead of a gcall.
11174 (vect_check_gather_scatter): Update call accordingly.
11175 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
11177 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
11178 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
11179 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
11180 instead of a gimple stmt.
11181 (vect_transform_loop): Update calls accordingly.
11182 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
11183 stmt_vec_infos instead of gimple stmts.
11184 (vect_analyze_slp_instance): Update use accordingly.
11185 * tree-vect-stmts.c (read_vector_array, write_vector_array)
11186 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
11187 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
11188 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
11189 (vect_gen_widened_results_half, vect_get_loop_based_defs)
11190 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
11191 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
11192 down to subroutines.
11194 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11196 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
11197 of the worklist from a vector of gimple stmts to a vector of
11199 * tree-vect-stmts.c (vect_mark_relevant, process_use)
11200 (vect_mark_stmts_to_be_vectorized): Likewise
11202 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11204 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
11205 statement before passing it to vect_analyze_stmt.
11206 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
11207 the chain of phi vector definitions. Track the exit phi via its
11209 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
11210 STMT_VINFO_REDUC_DEF.
11211 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
11212 stmt_vec_infos to handle the statement chains.
11213 (vect_get_slp_defs): Record the first statement in the node
11214 using a stmt_vec_info.
11215 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
11216 statements here and pass their stmt_vec_info down to subroutines.
11217 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
11218 down to vect_finish_stmt_generation.
11219 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
11220 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
11221 stmt_vec_infos to subroutines.
11222 (vect_remove_stores): Use stmt_vec_infos to handle the statement
11225 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11227 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
11228 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
11229 (vect_permute_store_chain, vect_setup_realignment)
11230 (vect_permute_load_chain, vect_shift_permute_load_chain)
11231 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
11232 stmts internally, and when passing values to other vectorizer routines.
11233 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
11234 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
11235 (vect_analyze_loop_operations, get_initial_def_for_reduction)
11236 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
11237 (vectorizable_reduction, vectorizable_induction)
11238 (vectorizable_live_operation, vect_transform_loop_stmt)
11239 (vect_transform_loop): Likewise.
11240 * tree-vect-patterns.c (vect_reassociating_reduction_p)
11241 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
11242 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
11243 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11244 (vect_slp_analyze_node_operations_1): Likewise.
11245 * tree-vect-stmts.c (vect_mark_relevant, process_use)
11246 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
11247 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
11248 (vect_finish_stmt_generation_1, get_group_load_store_type)
11249 (get_load_store_type, vect_build_gather_load_calls)
11250 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11251 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11252 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11253 (vectorizable_store, vectorizable_load, vectorizable_condition)
11254 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
11255 (supportable_widening_operation): Likewise.
11256 (vect_get_vector_types_for_stmt): Likewise.
11257 * tree-vectorizer.h (vect_dr_behavior): Likewise.
11259 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11261 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11262 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
11263 (vect_permute_store_chain, vect_permute_load_chain)
11264 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
11265 repeated stmt_vec_info lookups.
11266 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
11267 (vect_update_ivs_after_vectorizer): Likewise.
11268 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
11269 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
11270 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
11271 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11272 (vect_bb_slp_scalar_cost): Likewise.
11273 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
11275 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11277 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
11278 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
11279 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
11280 to get gassigns and gcalls, rather than operating on generc gimple
11282 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
11283 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
11284 (vectorizable_load, vect_analyze_stmt): Likewise.
11285 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
11287 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11289 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
11290 return stmt_vec_infos rather than gimple stmts. Do not accept
11292 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
11294 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
11295 Update use of get_later_stmt.
11296 (vect_get_constant_vectors): Update call accordingly.
11297 (vect_schedule_slp_instance): Likewise
11298 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
11299 (vect_slp_analyze_instance_dependence): Likewise.
11300 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
11302 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11304 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
11305 (stmt_info_for_cost::stmt_info): ...this new field.
11306 (add_stmt_costs): Update accordingly.
11307 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11308 (vect_get_known_peeling_cost): Likewise.
11309 (vect_estimate_min_profitable_iters): Likewise.
11310 * tree-vect-stmts.c (record_stmt_cost): Likewise.
11312 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11314 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
11315 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11316 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
11318 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
11320 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11322 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
11323 a gimple stmt to a stmt_vec_info.
11324 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
11326 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11328 * tree-vectorizer.h (vec_info::grouped_stores): Change from
11329 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11330 (_loop_vec_info::reduction_chains): Likewise.
11331 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
11333 * tree-vect-slp.c (vect_analyze_slp): Likewise.
11335 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11337 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
11338 a gimple stmt to a stmt_vec_info.
11339 (_stmt_vec_info::next_element): Likewise.
11340 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
11341 (vect_slp_analyze_and_verify_node_alignment)
11342 (vect_analyze_group_access_1, vect_analyze_group_access)
11343 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
11344 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
11345 (vect_supportable_dr_alignment): Update accordingly.
11346 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
11347 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
11348 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
11349 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11350 * tree-vect-slp.c (vect_build_slp_tree_1)
11351 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
11352 (vect_split_slp_store_group, vect_analyze_slp_instance)
11353 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
11354 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
11355 (get_group_load_store_type, get_load_store_type)
11356 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
11357 (vect_transform_stmt, vect_remove_stores): Likewise.
11359 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11361 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
11362 than a gimple stmt.
11363 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
11364 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
11365 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
11366 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
11367 (vector_alignment_reachable_p, vect_get_data_access_cost)
11368 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
11369 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
11370 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
11371 (vect_analyze_group_access_1, vect_analyze_group_access)
11372 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
11373 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
11374 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
11375 result of vect_dr_stmt and use the stmt_vec_info instead of
11376 the associated gimple stmt.
11377 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
11378 (vect_gen_prolog_loop_niters): Likewise.
11379 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
11381 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11383 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
11384 to a vec<stmt_vec_info>.
11385 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
11386 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
11387 vec<stmt_vec_info>.
11388 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
11389 to a vec<stmt_vec_info>.
11390 (bst_traits::value_type, bst_traits::value_type): Likewise.
11391 (bst_traits::hash): Update accordingly.
11392 (vect_get_and_check_slp_defs): Change the stmts parameter from
11393 a vec<gimple *> to a vec<stmt_vec_info>.
11394 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
11395 (vect_build_slp_tree): Likewise.
11396 (vect_build_slp_tree_2): Likewise. Update uses of
11397 SLP_TREE_SCALAR_STMTS.
11398 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
11399 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
11400 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
11401 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
11402 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
11403 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
11404 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
11405 (vect_get_constant_vectors, vect_get_slp_defs)
11406 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
11407 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
11408 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
11409 instead of gimple stmts.
11410 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
11411 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
11412 (vect_slp_analyze_instance_dependence): Update uses of
11413 SLP_TREE_SCALAR_STMTS.
11414 (vect_slp_analyze_and_verify_node_alignment): Likewise.
11415 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
11416 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
11417 (get_initial_defs_for_reduction): Likewise.
11418 (vect_create_epilog_for_reduction): Likewise.
11419 (vectorize_fold_left_reduction): Likewise.
11420 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
11421 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
11422 (can_vectorize_live_stmts): Likewise.
11424 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11426 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
11427 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
11428 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
11430 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
11432 (gather_scalar_reductions): Update after above interface changes.
11433 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
11434 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
11436 (vect_force_simple_reduction): Likewise.
11437 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
11438 LOOP_VINFO_REDUCTIONS.
11439 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11441 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11443 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
11444 a gimple stmt to a stmt_vec_info.
11445 * tree-vect-loop.c (vect_active_double_reduction_p)
11446 (vect_force_simple_reduction, vectorizable_reduction): Update
11449 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11451 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
11452 vec<gimple *> to a vec<stmt_vec_info>.
11453 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
11454 the reduction_phis argument from a vec<gimple *> to a
11455 vec<stmt_vec_info>.
11456 (vectorizable_reduction): Likewise the phis local variable that
11457 is passed to vect_create_epilog_for_reduction. Update for new type
11458 of SLP_TREE_VEC_STMTS.
11459 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
11460 (vectorizable_live_operation): Likewise.
11461 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
11462 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
11464 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11466 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
11467 a gimple stmt to a stmt_vec_info.
11468 (vectorizable_condition, vectorizable_live_operation)
11469 (vectorizable_reduction, vectorizable_induction): Pass back the
11470 vectorized statement as a stmt_vec_info.
11471 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
11472 use of STMT_VINFO_VEC_STMT.
11473 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
11474 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
11475 as stmt_vec_infos rather than gimple stmts.
11476 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
11477 to a stmt_vec_info.
11478 (vectorizable_live_operation): Likewise.
11479 (vectorizable_reduction, vectorizable_induction): Likewise,
11480 updating use of STMT_VINFO_VEC_STMT.
11481 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
11482 of STMT_VINFO_VEC_STMT.
11483 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
11484 (vectorizable_simd_clone_call, vectorizable_conversion)
11485 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11486 (vectorizable_store, vectorizable_load, vectorizable_condition)
11487 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
11488 from a gimple stmt to a stmt_vec_info.
11489 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
11490 pointer to a stmt_vec_info to the vectorizable_* routines.
11492 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11494 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
11495 a gimple stmt to a stmt_vec_info.
11496 (is_pattern_stmt_p): Update accordingly.
11497 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
11498 (vect_record_grouped_load_vectors): Likewise.
11499 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
11500 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
11501 (vect_model_reduction_cost): Likewise.
11502 (vect_create_epilog_for_reduction): Likewise.
11503 (vectorizable_reduction, vectorizable_induction): Likewise.
11504 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11505 Return the stmt_vec_info for the pattern statement.
11506 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
11507 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
11508 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
11509 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
11510 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
11511 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
11512 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
11513 (free_stmt_vec_info, vect_is_simple_use): Likewise.
11515 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11517 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
11518 (vect_finish_stmt_generation): Likewise.
11519 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
11520 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
11521 (vect_build_gather_load_calls): Use the return value of the above
11522 functions instead of a separate call to vinfo_for_stmt. Use narrow
11523 scopes for the input gimple stmt and wider scopes for the associated
11524 stmt_vec_info. Use vec_info::lookup_def when setting these
11525 stmt_vec_infos from an SSA_NAME definition.
11526 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
11527 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
11528 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
11529 (vectorizable_store, vectorizable_load, vectorizable_condition)
11530 (vectorizable_comparison): Likewise.
11531 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
11532 (vectorizable_reduction): Likewise.
11534 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11536 * tree-vectorizer.h (vect_is_simple_use): Add an optional
11537 stmt_vec_info * parameter before the optional gimple **.
11538 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11539 (process_use, vect_get_vec_def_for_operand_1): Update callers.
11540 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
11541 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11542 (vectorizable_live_operation): Likewise.
11543 * tree-vect-patterns.c (type_conversion_p): Likewise.
11544 (vect_look_through_possible_promotion): Likewise.
11545 (vect_recog_rotate_pattern): Likewise.
11546 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11548 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11550 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
11551 a typedef to a wrapper class.
11552 (NULL_STMT_VEC_INFO): New macro.
11553 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
11554 (stmt_vec_info::operator*): New function.
11555 (stmt_vec_info::operator gimple *): Likewise.
11556 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
11557 (add_stmt_costs): Likewise.
11558 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
11559 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
11560 (vect_get_known_peeling_cost): Likewise.
11561 (vect_estimate_min_profitable_iters): Likewise.
11562 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11563 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
11564 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
11565 (vectorizable_store, free_stmt_vec_infos): Likewise.
11566 (new_stmt_vec_info): Change return type of xcalloc to
11569 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11571 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
11572 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
11573 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
11574 a single_imm_use-based sequence.
11575 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
11577 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11579 * tree-vectorizer.h (vec_info::lookup_def): Declare.
11580 * tree-vectorizer.c (vec_info::lookup_def): New function.
11581 * tree-vect-patterns.c (vect_get_internal_def): Use it.
11582 (vect_widened_op_tree): Likewise.
11583 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11584 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
11585 (vectorizable_reduction): Likewise.
11586 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
11588 (vect_is_slp_reduction): Update calls accordingly. Use
11589 vec_info::lookup_def.
11590 (vect_is_simple_reduction): Likewise
11591 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
11593 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11595 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
11596 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
11597 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
11599 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
11600 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
11601 (vect_update_vf_for_slp, vect_analyze_loop_operations)
11602 (vect_is_slp_reduction, vectorizable_induction)
11603 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
11604 * tree-vect-patterns.c (vect_init_pattern_stmt):
11605 (vect_determine_min_output_precision_1, vect_determine_precisions)
11606 (vect_pattern_recog): Likewise.
11607 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
11608 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
11609 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
11610 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
11611 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
11612 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
11613 info field from a loop to a loop_vec_info.
11615 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11617 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
11618 (vec_info::add_stmt): Declare.
11619 * tree-vectorizer.c (vec_info::add_stmt): New function.
11620 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
11621 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
11622 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
11623 (vectorizable_induction): Likewise.
11624 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
11625 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
11626 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
11627 (vectorizable_load): Likewise.
11628 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
11629 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
11630 (vect_recog_gather_scatter_pattern): Likewise.
11631 (append_pattern_def_seq): Likewise. Remove a check that is
11632 performed by add_stmt itself.
11634 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11636 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
11637 which make_ssa_name was called with new_stmt before new_stmt
11640 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11642 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
11644 (vect_is_slp_reduction): ...here...
11645 (vect_is_simple_reduction): ...and here. Remove repetition of tests
11646 that are already known to be false.
11648 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11650 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
11651 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
11652 STMT_VINFO_NUM_SLP_USES when it's true.
11653 (vect_free_slp_instance): Add a final_p parameter and pass it to
11654 vect_free_slp_tree.
11655 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
11656 (vect_analyze_slp_instance): Likewise.
11657 (vect_slp_analyze_operations): Likewise.
11658 (vect_slp_analyze_bb_1): Likewise.
11659 * tree-vectorizer.c (vec_info): Likewise.
11660 * tree-vect-loop.c (vect_transform_loop): Likewise.
11662 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11664 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11665 function is not called for second and subsequent members of
11668 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
11670 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
11671 cases for nested loops from here to ...
11672 (vect_create_epilog_for_reduction): ...here. Only call
11673 vect_is_simple_use for inner-loop reductions.
11675 2018-07-31 Martin Liska <mliska@suse.cz>
11677 PR gcov-profile/85338
11678 PR gcov-profile/85350
11679 PR gcov-profile/85372
11680 * profile.c (struct location_triplet): New.
11681 (struct location_triplet_hash): Likewise.
11682 (output_location): Do not output a BB that
11683 is already recorded for a line.
11684 (branch_prob): Use streamed_locations.
11686 2018-07-31 Martin Liska <mliska@suse.cz>
11688 PR gcov-profile/85370
11689 * coverage.c (coverage_begin_function): Do not mark target
11690 clones as artificial functions.
11692 2018-07-31 Martin Liska <mliska@suse.cz>
11694 PR gcov-profile/83813
11695 PR gcov-profile/84758
11696 PR gcov-profile/85217
11697 PR gcov-profile/85332
11698 * profile.c (branch_prob): Do not record GOTO expressions
11699 for GIMPLE statements which locations are already streamed.
11701 2018-07-31 Olivier Hainque <hainque@adacore.com>
11703 * gcc.c (handle_spec_function): Accept a soft_matched_part
11704 argument, as do_spec_1. Pass it down to ...
11705 (eval_spec_function): Accept a soft_matched_part argument,
11706 and pass it down to ...
11707 (do_spec_2): Accept a soft_matched_part argument, and pass
11708 it down to do_spec_1.
11709 (do_spec_1): Pass soft_matched_part to handle_spec_function.
11710 (handle_braces): Update call to handle_spec_function.
11711 (driver::set_up_specs): Update calls to do_spec_2.
11712 (compare_debug_dump_opt_spec_function): Likewise.
11713 (compare_debug_self_opt_spec_function): Likewise.
11715 2018-07-31 Olivier Hainque <hainque@adacore.com>
11717 * common.opt (nolibc): New option.
11718 * doc/invoke.texi (Link Options): Document it.
11719 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
11720 * config/alpha/linux.h: Likewise.
11721 * config/arc/elf.h: Likewise.
11722 * config/arm/uclinux-elf.h: Likewise.
11723 * config/arm/unknown-elf.h: Likewise.
11724 * config/avr/avrlibc.h: Likewise.
11725 * config/bfin/bfin.h: Likewise.
11726 * config/bfin/linux.h: Likewise.
11727 * config/bfin/uclinux.h: Likewise.
11728 * config/darwin.h: Likewise.
11729 * config/darwin10.h: Likewise.
11730 * config/darwin12.h: Likewise.
11731 * config/gnu-user.h: Likewise.
11732 * config/lm32/uclinux-elf.h: Likewise.
11733 * config/pa/pa-hpux11.h: Likewise.
11734 * config/pa/pa64-hpux.h: Likewise.
11735 * config/sparc/sparc.h: Likewise.
11737 2018-07-31 Olivier Hainque <hainque@adacore.com>
11739 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
11740 undefined variables.
11742 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
11745 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
11746 instead of GEN_INT.
11748 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11750 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
11751 terminated string literal.
11753 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
11755 PR rtl-optimization/85160
11756 * combine.c (is_just_move): New function.
11757 (try_combine): Allow combining two instructions into two if neither of
11758 the original instructions was a move.
11760 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
11763 * doc/extend.texi (Global Register Variables): Discourage use of type
11765 (Local Register Variables): Likewise.
11767 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
11769 PR tree-optimization/86506
11770 * hwint.h (ceil_log2): Resync with hwint.c implementation.
11772 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
11775 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
11776 hard_regno, make sure no insn between `from` and `to` clobbers it.
11778 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
11779 Tom de Vries <tdevries@suse.de>
11781 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
11782 (PTX_DEFAULT_RUNTIME_DIM): ... this.
11783 (nvptx_goacc_validate_dims): Set default worker and gang dims to
11784 PTX_DEFAULT_RUNTIME_DIM.
11785 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
11787 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
11789 * config/pa/pa.c (pa_output_addr_vec): Align address table.
11790 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
11791 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
11793 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
11795 * config/rs6000/constraints.md (wG constraint): Delete, no longer
11797 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
11798 predicate to reflect toc fusion has been deleted.
11799 (toc_fusion_mem_raw): Delete, no longer used.
11800 (toc_fusion_mem_wrapped): Likewise.
11801 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
11803 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
11804 Delete, no longer used.
11805 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
11806 meant to be used for toc fusion.
11807 (rs6000_debug_print_mode): Delete toc fusion debugging.
11808 (rs6000_debug_reg_global): Likewise.
11809 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
11810 fusion and secondary reload support that were never used.
11811 (rs6000_option_override_internal): Delete TOC fusion, that was only
11812 partially defined, and it did not work unless you also used the
11814 (rs6000_legitimate_address_p): Delete TOC fusion support.
11815 (rs6000_opt_masks): Likewise.
11816 (fusion_wrap_memory_address): Delete function, no longer used.
11817 (fusion_split_address); Delete TOC fusion support.
11818 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
11819 longer used with toc fusion being deleted.
11820 (TARGET_TOC_FUSION_FP): Likewise.
11821 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
11823 (toc fusion spliter): Delete TOC fusion support.
11824 (toc_fusionload_<mode>): Likewise.
11825 (toc_fusionload_di): Likewise.
11826 (fusion_gpr_load_<mode>): Delete generator function, this insn no
11827 longer needs to be named. Rename predicate to delete TOC fusion.
11828 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
11829 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
11830 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
11831 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
11832 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
11834 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11836 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
11837 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
11838 __int128 in built-in function prototypes.
11839 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
11840 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
11842 2018-07-27 Martin Sebor <msebor@redhat.com>
11844 PR tree-optimization/86696
11845 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
11846 types, including enums.
11847 (handle_char_store): Be prepared for the above function to fail.
11849 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
11851 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
11852 when optimization level is lower than 2 or optimize for size.
11854 2018-07-26 Martin Sebor <msebor@redhat.com>
11856 PR tree-optimization/86043
11857 PR tree-optimization/86042
11858 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
11859 (get_string_cst_length): Rename...
11860 (get_min_string_length): ...to this. Add argument.
11861 (handle_char_store): Extend to handle multi-character stores by
11863 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
11864 * tree.h (initializer_zerop): Add argument.
11866 2018-07-26 Jakub Jelinek <jakub@redhat.com>
11868 PR middle-end/86660
11869 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
11870 declare target to variables if they have always,{to,from,tofrom} map
11873 2018-07-26 Martin Liska <mliska@suse.cz>
11876 * lto-wrapper.c: Add linker_output as prefix
11877 for ltrans_output_file.
11879 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
11881 PR rtl-optimization/85805
11882 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
11883 value for hard registers if that was written in the same mode.
11885 2018-07-26 Martin Liska <mliska@suse.cz>
11887 PR gcov-profile/86536
11888 * gcov.c (format_gcov): Use printf format %.*f directly
11889 and do not handle special values.
11891 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11893 * common/config/arc/arc-common.c (arc_option_optimization_table):
11894 Update default optimizations for size.
11896 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11898 * config/arc/arc.md (movsf_insn): Add short instruction selection.
11899 * config/arc/constraints.md (CfZ): New constraint.
11900 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
11901 (subsf3_fpu): Likewise.
11902 (cmpsf_fpu): Likewise.
11903 (cmpsf_fpu_uneq): Likewise.
11905 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11907 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
11909 (arc_is_uncached_mem_p): uncached applies to both the variable and
11912 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
11914 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
11917 2018-07-25 David Malcolm <dmalcolm@redhat.com>
11919 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
11920 field "m_scopes" from vec to auto_vec.
11922 2018-07-25 Martin Liska <mliska@suse.cz>
11924 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
11927 2018-07-25 Richard Biener <rguenther@suse.de>
11930 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
11931 special wrt context_die late.
11932 (gen_subprogram_die): Re-use DIEs in local scope.
11934 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
11936 PR tree-optimization/86644
11937 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
11939 2018-07-25 Martin Liska <mliska@suse.cz>
11941 PR middle-end/86645
11942 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
11943 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
11945 2018-07-25 Martin Liska <mliska@suse.cz>
11948 * params.def: Explain ASan abbreviation and provide
11949 a documentation link.
11951 2018-07-24 Martin Sebor <msebor@redhat.com>
11953 PR tree-optimization/86622
11954 PR tree-optimization/86532
11955 * builtins.h (string_length): Declare.
11956 * builtins.c (c_strlen): Correct handling of non-constant offsets.
11957 (check_access): Be prepared for non-constant length ranges.
11958 (string_length): Make extern.
11959 * expr.c (string_constant): Only handle the minor non-constant
11960 array index. Use string_constant to compute the length of
11961 a generic string constant.
11963 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
11965 PR tree-optimization/86618
11966 * tree-vect-stmts.c (vectorizable_call): Don't take the address
11967 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
11969 2018-07-24 David Malcolm <dmalcolm@redhat.com>
11971 PR tree-optimization/86636
11972 * json.cc (json::object::set): Fix comment. Add assertions.
11973 (json::array::append): Move here from json.h. Add comment and an
11975 (json::string::string): Likewise.
11976 * json.h (json::array::append): Move to json.cc.
11977 (json::string::string): Likewise.
11978 * optinfo-emit-json.cc
11979 (optrecord_json_writer::impl_location_to_json): Assert that we
11980 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
11981 wrapper around it. Expand the location once, rather than three
11983 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
11984 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
11986 (optrecord_json_writer::optinfo_to_json): Likewise, in four
11987 places. Fix some overlong lines.
11989 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
11991 * config/aarch64/aarch64-simd.md
11992 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
11993 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
11994 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
11995 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
11996 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
11997 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
11998 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
11999 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
12000 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
12002 2018-07-24 Jakub Jelinek <jakub@redhat.com>
12004 PR middle-end/86627
12005 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
12006 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
12007 and abs_d == d, do the power of two handling if profitable.
12009 2018-07-24 Richard Biener <rguenther@suse.de>
12011 * match.pd: Add BIT_FIELD_REF canonicalizations.
12013 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12016 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
12018 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12020 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
12021 terminated STRING_CST object.
12023 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
12025 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
12027 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
12029 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
12030 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
12031 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
12032 the elements into a register.
12033 (rs6000_split_v4si_init_di_reg): Delete.
12034 (rs6000_split_v4si_init): Delete.
12035 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
12036 (vsx_init_v4si): Rewrite as a define_expand.
12038 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
12040 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
12041 zero_extend argument from memory): New.
12043 2018-07-22 Martin Sebor <msebor@redhat.com>
12046 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
12047 diagnosing calls with unknown arguments unless -Walloca-larger-than
12048 is restricted to less than PTRDIFF_MAX bytes.
12050 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
12052 * doc/gcov.texi (Invoking Gcov): Editorial changes.
12054 2018-07-20 David Malcolm <dmalcolm@redhat.com>
12056 * pretty-print.c (text_info::set_location): Remove redundant
12057 "line_table" parameter from call to rich_location::set_range.
12059 2018-07-20 Martin Sebor <msebor@redhat.com>
12061 PR middle-end/82063
12062 * builtins.c (expand_builtin_alloca): Adjust.
12063 * calls.c (alloc_max_size): Simplify.
12064 * cgraphunit.c (cgraph_node::expand): Adjust.
12065 * common.opt (larger_than_size, warn_frame_larger_than): Remove
12067 (frame_larger_than_size): Same.
12068 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
12069 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
12071 * doc/invoke.texi (GCC Command Options): Document option arguments.
12072 Explain byte-size arguments and suffixes.
12073 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
12074 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
12075 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
12076 * doc/options.texi (UInteger): Expand.
12077 (Host_Wide_Int, ByteSize): Document new properties.
12078 * final.c (final_start_function_1): Include sizes in an error message.
12079 * function.c (frame_offset_overflow): Same.
12080 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
12081 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
12082 Diagnose unbounded alloca calls only for limits of less than
12084 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
12085 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
12087 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
12088 only for limits of less than PTRDIFF_MAX.
12089 * langhooks-def.h (lhd_handle_option): Change function argument
12091 * langhooks.c (lhd_handle_option): Same.
12092 * langhooks.h (handle_option): Same.
12093 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
12095 (var_type, var_type_struct): Same.
12096 (var_set): Handle ByteSize flag.
12097 * optc-gen.awk: Add comments to output to ease debugging. Make
12098 use of HOST_WIDE_INT where appropriate.
12099 * opts-gen-save.awk: Use %lx to format unsigned long.
12100 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
12101 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
12102 arguments. Parse bytes-size suffixes.
12103 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
12104 (enum_value_to_arg): Same.
12105 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
12106 (handle_option): Adjust.
12107 (generate_option): Change function argument to HOST_WIDE_INT.
12108 (cmdline_handle_error): Adjust.
12109 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
12110 (set_option): Change function argument to HOST_WIDE_INT.
12111 (option_enabled): Handle cl_host_wide_int.
12112 (get_option_state): Handle CLVC_SIZE.
12113 (control_warning_option): Same.
12114 * opts.c (common_handle_option): Change function argument to
12115 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
12116 OPT_Wvla_larger_than_.
12117 * opts.h (enum cl_var_type): Add an enumerator.
12118 * stor-layout.c (layout_decl): Print a more meaningful warning.
12119 * toplev.c (output_stack_usage): Adjust.
12121 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
12123 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
12124 call to inline_expand_builtin_string_cmp.
12125 (expand_builtin_strcmp): Likewise.
12126 (expand_builtin_strncmp): Likewise.
12127 (inline_string_cmp): Delete the last parameter, change char_type_node
12128 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
12130 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
12131 the inlining expansion on target where the type of the call has same or
12132 narrower precision than unsigned char.
12134 2018-07-20 David Malcolm <dmalcolm@redhat.com>
12136 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
12137 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
12138 * common.opt (fsave-optimization-record): New option.
12139 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
12140 * doc/invoke.texi (-fsave-optimization-record): New option.
12141 * dumpfile.c: Include "optinfo-emit-json.h".
12142 (struct kv_pair): Move to coretypes.h.
12143 (optgroup_options): Make non-static.
12144 (dump_context::end_scope): Call
12145 optimization_records_maybe_pop_dump_scope.
12146 * dumpfile.h (optgroup_options): New decl.
12147 * json.cc: New file.
12148 * json.h: New file.
12149 * optinfo-emit-json.cc: New file.
12150 * optinfo-emit-json.h: New file.
12151 * optinfo.cc: Include "optinfo-emit-json.h".
12152 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
12153 (optinfo_enabled_p): Check optimization_records_enabled_p.
12154 (optinfo_wants_inlining_info_p): Likewise.
12155 * optinfo.h: Update comment.
12156 * profile-count.c (profile_quality_as_string): New function.
12157 * profile-count.h (profile_quality_as_string): New decl.
12158 (profile_count::quality): New accessor.
12159 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
12160 and optinfo_emit_json_cc_tests.
12161 * selftest.h (selftest::json_cc_tests): New decl.
12162 (selftest::optinfo_emit_json_cc_tests): New decl.
12163 * toplev.c: Include "optinfo-emit-json.h".
12164 (compile_file): Call optimization_records_finish.
12165 (do_compile): Call optimization_records_start.
12166 * tree-ssa-live.c: Include optinfo.h.
12167 (remove_unused_scope_block_p): Retain inlining information if
12168 optinfo_wants_inlining_info_p returns true.
12170 2018-07-20 Richard Biener <rguenther@suse.de>
12173 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
12174 to cover -flto-partition=none.
12176 2018-07-20 Martin Liska <mliska@suse.cz>
12178 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
12179 (get_decl_source_range): Remove unused function.
12181 2018-07-20 Richard Biener <rguenther@suse.de>
12183 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
12184 (struct vn_phi_s): Likewise.
12185 (struct vn_reference_s): Likewise.
12186 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
12187 for searching the slot of an entry known to be in the hash itself.
12188 (vn_phi_hasher::equal): Likewise.
12189 (vn_reference_hasher::equal): Likewise.
12190 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
12192 (optimistic_info, current_info): Remove, keeping only valid_info.
12193 (vn_reference_lookup_1): Remove fallback lookup.
12194 (vn_reference_lookup_2): Likewise.
12195 (vn_nary_op_lookup_1): Likewise.
12196 (vn_phi_lookup): Likewise.
12197 (vn_nary_build_or_lookup_1): Make sure to not chain the built
12199 (vn_reference_insert): Adjust, chain the inserted hash element
12200 at last_inserted_ref.
12201 (vn_reference_insert_pieces): Likewise.
12202 (visit_reference_op_call): Likewise.
12203 (vn_nary_op_insert_into): Chain the inserted hash element at
12204 last_inserted_nary.
12205 (vn_nary_op_insert_pieces): Adjust.
12206 (vn_nary_op_insert): Likewise.
12207 (vn_nary_op_insert_stmt): Likewise.
12208 (vn_phi_insert): Adjust, chain the inserted hash element at
12210 (process_scc): Remove clearing and copying the optimistic
12211 table. Instead remove elements inserted during an optimistic
12212 iteration from the single table we maintain.
12213 (init_scc_vn): Adjust.
12214 (free_scc_vn): Likewise.
12215 (sccvn_dom_walker::record_cond): Likewise.
12216 (sccvn_dom_walker::after_dom_children): Likewise.
12218 2018-07-19 Martin Sebor <msebor@redhat.com>
12220 PR tree-optimization/84047
12221 PR tree-optimization/83776
12222 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
12223 (check_array_bounds): Call it.
12225 2018-07-19 Martin Sebor <msebor@redhat.com>
12227 * align.h (align_flags): Use member initialization.
12229 2018-07-19 David Malcolm <dmalcolm@redhat.com>
12231 * Makefile.in (OBJS): Add optinfo.o.
12232 * coretypes.h (class symtab_node): New forward decl.
12233 (struct cgraph_node): New forward decl.
12234 (class varpool_node): New forward decl.
12235 * dump-context.h: New file.
12236 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
12238 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
12239 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
12240 (set_alt_dump_file): Likewise.
12241 (dump_context::~dump_context): New dtor.
12242 (dump_gimple_stmt): Move implementation to...
12243 (dump_context::dump_gimple_stmt): ...this new member function.
12244 Add the stmt to any pending optinfo, creating one if need be.
12245 (dump_gimple_stmt_loc): Move implementation to...
12246 (dump_context::dump_gimple_stmt_loc): ...this new member function.
12247 Start a new optinfo and add the stmt to it.
12248 (dump_gimple_expr): Move implementation to...
12249 (dump_context::dump_gimple_expr): ...this new member function.
12250 Add the stmt to any pending optinfo, creating one if need be.
12251 (dump_gimple_expr_loc): Move implementation to...
12252 (dump_context::dump_gimple_expr_loc): ...this new member function.
12253 Start a new optinfo and add the stmt to it.
12254 (dump_generic_expr): Move implementation to...
12255 (dump_context::dump_generic_expr): ...this new member function.
12256 Add the tree to any pending optinfo, creating one if need be.
12257 (dump_generic_expr_loc): Move implementation to...
12258 (dump_context::dump_generic_expr_loc): ...this new member
12259 function. Add the tree to any pending optinfo, creating one if
12261 (dump_printf): Move implementation to...
12262 (dump_context::dump_printf_va): ...this new member function. Add
12263 the text to any pending optinfo, creating one if need be.
12264 (dump_printf_loc): Move implementation to...
12265 (dump_context::dump_printf_loc_va): ...this new member function.
12266 Start a new optinfo and add the stmt to it.
12267 (dump_dec): Move implementation to...
12268 (dump_context::dump_dec): ...this new member function. Add the
12269 value to any pending optinfo, creating one if need be.
12270 (dump_context::dump_symtab_node): New member function.
12271 (dump_context::get_scope_depth): New member function.
12272 (dump_context::begin_scope): New member function.
12273 (dump_context::end_scope): New member function.
12274 (dump_context::ensure_pending_optinfo): New member function.
12275 (dump_context::begin_next_optinfo): New member function.
12276 (dump_context::end_any_optinfo): New member function.
12277 (dump_context::s_current): New global.
12278 (dump_context::s_default): New global.
12279 (dump_scope_depth): Delete global.
12280 (dumpfile_ensure_any_optinfo_are_flushed): New function.
12281 (dump_symtab_node): New function.
12282 (get_dump_scope_depth): Reimplement in terms of dump_context.
12283 (dump_begin_scope): Likewise.
12284 (dump_end_scope): Likewise.
12285 (selftest::temp_dump_context::temp_dump_context): New ctor.
12286 (selftest::temp_dump_context::~temp_dump_context): New dtor.
12287 (selftest::verify_item): New function.
12288 (ASSERT_IS_TEXT): New macro.
12289 (ASSERT_IS_TREE): New macro.
12290 (ASSERT_IS_GIMPLE): New macro.
12291 (selftest::test_capture_of_dump_calls): New test.
12292 (selftest::dumpfile_c_tests): Call it.
12293 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
12294 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
12295 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
12296 descriptive comment.
12297 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
12298 (dump_node, dump_bb): Move these unrelated decls.
12299 (class dump_manager): Add leading comment.
12300 * optinfo.cc: New file.
12301 * optinfo.h: New file.
12303 2018-07-19 Michael Collison <michael.collison@arm.com>
12304 Richard Henderson <rth@redhat.com>
12306 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
12307 (subti): Handle op1 zero.
12308 (subvti4, usub4ti4): New.
12309 (*sub<GPI>3_compare1_imm): New.
12310 (sub<GPI>3_carryinCV): New.
12311 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
12312 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
12314 2018-07-19 Michael Collison <michael.collison@arm.com>
12315 Richard Henderson <rth@redhat.com>
12317 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
12318 (addti3): Create simpler code if low part is already known to be 0.
12319 (addvti4, uaddvti4): New.
12320 (*add<GPI>3_compareC_cconly_imm): New.
12321 (*add<GPI>3_compareC_cconly): New.
12322 (*add<GPI>3_compareC_imm): New.
12323 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
12324 handle constants within this pattern..
12325 (*add<GPI>3_compareV_cconly_imm): New.
12326 (*add<GPI>3_compareV_cconly): New.
12327 (*add<GPI>3_compareV_imm): New.
12328 (add<GPI>3_compareV): New.
12329 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
12330 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
12331 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
12332 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
12333 with 'comparison' operator.
12334 (*add<GPI>3_compareV_cconly_imm): Ditto.
12335 (*add<GPI>3_compareV_cconly): Ditto.
12336 (*add<GPI>3_compareV_imm): Ditto.
12337 (add<GPI>3_compareV): Ditto.
12338 (add<mode>3_carryinC): Ditto.
12339 (*add<mode>3_carryinC_zero): Ditto.
12340 (*add<mode>3_carryinC): Ditto.
12341 (add<mode>3_carryinV): Ditto.
12342 (*add<mode>3_carryinV_zero): Ditto.
12343 (*add<mode>3_carryinV): Ditto.
12345 2018-07-19 Michael Collison <michael.collison@arm.com>
12346 Richard Henderson <rth@redhat.com>
12348 * config/aarch64/aarch64-modes.def (CC_V): New.
12349 * config/aarch64/aarch64-protos.h
12350 (aarch64_addti_scratch_regs): Declare
12351 (aarch64_subvti_scratch_regs): Declare.
12352 (aarch64_expand_subvti): Declare.
12353 (aarch64_gen_unlikely_cbranch): Declare
12354 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
12355 for signed overflow using CC_Vmode.
12356 (aarch64_get_condition_code_1): Handle CC_Vmode.
12357 (aarch64_gen_unlikely_cbranch): New function.
12358 (aarch64_addti_scratch_regs): New function.
12359 (aarch64_subvti_scratch_regs): New function.
12360 (aarch64_expand_subvti): New function.
12362 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12364 * config/aarch64/aarch64-option-extensions.def: New entry for profile
12366 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
12367 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
12370 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12373 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
12374 address check not strict.
12376 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
12378 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
12380 (store_pair_lanes<mode>): Likewise.
12381 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
12382 enum value 'ADDR_QUERY_LDP_STP_N'.
12383 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
12384 (aarch64_print_address_internal): Add declaration.
12385 (aarch64_print_ldpstp_address): Remove.
12386 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
12387 (aarch64_print_operand): Change printing of 'y'.
12388 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
12389 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
12390 'true' rather than '1'.
12391 * config/aarch64/constraints.md (Uml): Likewise.
12392 (Uml): Rename to Umn.
12395 2018-07-19 Richard Biener <rguenther@suse.de>
12397 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
12399 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
12400 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
12401 (vn_reference_hasher): Likewise.
12402 (struct vn_tables_s): Remove obstack and alloc-pool members.
12403 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
12404 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
12405 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
12406 (vn_reference_insert_pieces): Likewise.
12407 (alloc_vn_nary_op_noinit): Adjust.
12408 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
12409 (vn_phi_eq): Adjust.
12410 (shared_lookup_phiargs): Remove.
12411 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
12412 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
12413 (visit_reference_op_call): Likewise.
12414 (copy_nary, copy_phi, copy_reference): Remove.
12415 (process_scc): Rewind the obstack when iterating. Do not
12416 copy the elements to valid_info but just move them from one
12417 hashtable to the other.
12418 (allocate_vn_table): Adjust.
12419 (free_vn_table): Likewise.
12420 (init_scc_vn): Likewise.
12421 (free_scc_vn): Likewise.
12423 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
12426 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
12427 indirect_return as function type attribute.
12428 (ix86_attribute_table): Change indirect_return to function
12430 * doc/extend.texi: Update indirect_return attribute.
12432 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
12434 * wide-int.h (widest2_int): New.
12435 * gimple-fold.c (arith_overflowed_p): Use it.
12436 * tree.h (widest2_int_cst): New.
12437 * tree-vrp.c (wide_int_binop_overflow): Rename from
12438 vrp_int_const_binop.
12439 Rewrite to work on trees.
12440 (extract_range_from_multiplicative_op_1): Abstract code to...
12441 (wide_int_range_min_max): ...here.
12442 (wide_int_range_cross_product): ...and here.
12443 (extract_range_from_binary_expr_1): Abstract overflow code to...
12444 (wide_int_range_mult_wrapping): ...here.
12445 * tree-vrp.h (wide_int_range_cross_product): New.
12446 (wide_int_range_mult_wrapping): New.
12448 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
12449 Julia Koval <julia.koval@intel.com>
12451 * config/i386/x86-tune-costs.h (skylake_memcpy,
12452 skylake_memset): Replace rep_prefix with unrolling for size 512.
12454 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
12456 PR middle-end/86544
12457 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
12458 comparision with EQ_EXPR in last stmt.
12460 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
12462 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
12463 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
12464 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
12465 previously known as "PowerPC AltiVec Built-in Functions". Move
12466 some material to new subsubsections "PowerPC AltiVec Built-in
12467 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
12469 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
12470 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
12471 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
12472 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
12474 2018-07-18 Richard Biener <rguenther@suse.de>
12476 PR tree-optimization/86557
12477 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
12480 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
12482 * config/s390/s390.c (s390_function_profiler): Generate CFI.
12484 2018-07-17 Jeff Law <law@redhat.com>
12486 * config/arm/arm.c (get_label_padding): Update for recent
12487 changes to label_to_alignment.
12489 PR tree-optimization/86010
12490 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
12492 * config/mips/mips.c (vr4130_align_insns): Update for recent
12493 changes to label_to_alignment.
12495 * config/frv/frv.c (frv_label_align): Update for recent changes
12496 to label_to_alignment.
12498 * config/nios2/nios2.c (nios2_label_align): Update for recent
12499 changes which dropped ALIGN_LABELS_LOG.
12501 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
12503 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
12504 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
12506 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12508 * config/arc/arc.c (arc_label_align): Use align_labels instead of
12509 deprecated align_labels_log.
12511 2018-07-17 Richard Biener <rguenther@suse.de>
12514 * dwarf2out.c (init_sections_and_labels): Always generate
12515 a debug_line_str_section for early LTO debug.
12516 (dwarf2out_finish): Reset debug_line_str_hash output early.
12517 Bump counter for extra dwarf5 .debug_loc labels to not conflict
12519 (dwarf2out_early_finish): Output debug_line_str.
12521 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
12523 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
12524 index register on z196 or later.
12526 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
12528 * config/s390/s390.c (s390_default_align): Set default function
12530 (s390_override_options_after_change): Call s390_default align.
12531 (s390_option_override_internal): Call s390_default_align.
12532 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
12534 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12536 PR middle-end/86542
12537 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
12538 to _looptemp_ clauses, other than the first two.
12540 2018-07-17 Martin Liska <mliska@suse.cz>
12542 * opts.c: Do not enable OPT_falign_* for -Os.
12544 2018-07-17 Martin Liska <mliska@suse.cz>
12546 * align.h (MAX_CODE_ALIGN): New.
12547 (MAX_CODE_ALIGN_VALUE): New.
12548 * common/config/i386/i386-common.c (ix86_handle_option):
12549 (MAX_CODE_ALIGN): Moved to align.h.
12550 * final.c (MAX_CODE_ALIGN): Likewise.
12551 * opts.c (parse_and_check_align_values):
12552 (MAX_CODE_ALIGN): Likewise.
12553 (MAX_CODE_ALIGN_VALUE): Likewise.
12555 2018-07-17 Martin Liska <mliska@suse.cz>
12557 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
12558 in order to fulfil coding style.
12559 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
12560 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12561 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
12562 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
12563 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
12564 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
12565 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
12566 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
12567 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12569 2018-07-17 Martin Liska <mliska@suse.cz>
12571 * align.h: New file.
12572 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
12574 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
12575 align_flags of label_to_alignment.
12576 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
12578 * config/m68k/m68k.c: Do not use removed align_labels_value and
12580 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
12581 (LOOP_ALIGN): Likewise.
12582 (LABEL_ALIGN): Likewise.
12583 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12584 Remove not used macro.
12585 (rs6000_loop_align): Change return type to align_flags.
12586 (rs6000_loop_align_max_skip): Remove.
12587 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
12588 Change return type to align_flags.
12589 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
12590 Remove not used macro.
12591 (rs6000_loop_align): Change return type to align_flags.
12592 (rs6000_loop_align_max_skip): Remove.
12593 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
12594 * config/rx/rx-protos.h (rx_align_for_label): Make it
12596 * config/rx/rx.c (rx_align_for_label): Change return type
12598 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
12600 into align_flags class.
12601 (LABEL_ALIGN): Likewise.
12602 (LOOP_ALIGN): Likewise.
12603 * config/s390/s390.c (s390_label_align): Use align_flags
12605 (s390_asm_output_function_label): Likewise.
12606 * config/sh/sh.c (sh_override_options_after_change):
12607 Use align_flags class directly without macros.
12608 (find_barrier): Likewise.
12609 (barrier_align): Likewise.
12610 (sh_loop_align): Likewise.
12611 * config/spu/spu.c (spu_option_override):
12612 Use align_flags_tuple::get_value instead of removed macros.
12613 (spu_sched_init): Likewise.
12614 * config/spu/spu.h (GTY): Likewise.
12615 * config/visium/visium.c (visium_option_override):
12616 Set "8" as default secondary alignment.
12617 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
12618 in order to guarantee secondary alignment of 8.
12619 * coretypes.h: Include align.h header file.
12620 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
12621 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
12622 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
12623 * doc/tm.texi.in: Likewise.
12624 * final.c (struct label_alignment): Remove not used structure.
12625 (LABEL_ALIGN): Change type to align_flags.
12626 (LOOP_ALIGN): Likewise.
12627 (JUMP_ALIGN): Likewise.
12628 (default_loop_align_max_skip): Remove.
12629 (default_label_align_max_skip): Likewise.
12630 (default_jump_align_max_skip): Likewise.
12631 (default_label_align_after_barrier_max_skip):
12632 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
12633 (LABEL_TO_MAX_SKIP): Remove.
12634 (label_to_alignment): Return align_flags type instead of integer.
12635 (label_to_max_skip): Remove.
12636 (align_fuzz): Use align_flags type.
12637 (compute_alignments): Use align_flags type and use align_flags::max
12638 to combine multiple alignments.
12639 (grow_label_align): Grow vec instead of C array.
12640 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
12641 (shorten_branches): Use align_flags type and use align_flags::max
12642 to combine multiple alignments.
12643 (final_scan_insn_1): Remove usage of secondary alignment that comes
12644 from label alignment, but instead use proper secondary alignment
12645 which is computed in grow_label_align.
12646 * flags.h (struct align_flags_tuple): Move to align.h.
12647 (struct align_flags): Likewise.
12648 (state_align_loops): Rename to align_loops.
12649 (state_align_jumps): Rename to align_jumps.
12650 (state_align_labels): Rename to align_labels.
12651 (state_align_functions): Rename to align_functions.
12652 (align_loops_log): Remove.
12653 (align_jumps_log): Remove.
12654 (align_labels_log): Remove.
12655 (align_functions_log): Remove.
12656 (align_loops_max_skip): Remove.
12657 (align_jumps_max_skip): Remove.
12658 (align_labels_max_skip): Remove.
12659 (align_functions_max_skip): Remove.
12660 (align_loops_value): Remove.
12661 (align_jumps_value): Remove.
12662 (align_labels_value): Remove.
12663 (align_functions_value): Remove.
12664 * output.h (label_to_alignment): Change return type to align_flags.
12665 (label_to_max_skip): Remove.
12666 * target.def: Remove loop_align_max_skip, label_align_max_skip,
12667 jump_align_max_skip macros.
12668 * targhooks.h (default_loop_align_max_skip): Remove.
12669 (default_label_align_max_skip): Likewise.
12670 (default_jump_align_max_skip): Likewise.
12671 (default_label_align_after_barrier_max_skip): Remove.
12672 * toplev.c (read_log_maxskip): Use ::normalize function.
12673 (parse_N_M): Remove not used argument and also call ::normalize.
12674 (parse_alignment_opts): Do not pass unused arguments.
12675 * varasm.c (assemble_start_function): Use directly align_functions
12676 instead of removed macros.
12677 * system.h: Do not poison removed macros.
12679 2018-07-17 Jakub Jelinek <jakub@redhat.com>
12681 PR middle-end/86539
12682 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
12683 and cond temporaries don't have reference type if iterator has
12684 pointer type. For init use &for_pre_body instead of pre_p if
12685 for_pre_body is non-empty.
12687 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12689 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
12690 double-double modes to SFmode directly directly.
12691 (trunc<mode>sf2_fprs): Delete.
12693 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12695 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
12696 for conversions between IFmode and the decimal floating point modes.
12697 (init_float128_ieee): Use the correct names for conversions between
12698 KFmode and the decimal floating point modes.
12700 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
12702 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
12703 for the conversions between TDmode and IFmode.
12704 (init_float128_ieee): Use more correct names for the conversions
12705 between TDmode and KFmode.
12707 2018-07-16 Jakub Jelinek <jakub@redhat.com>
12709 PR tree-optimization/86526
12710 * builtins.c (expand_builtin_memcmp): Formatting fixes.
12711 (inline_expand_builtin_string_cmp): Likewise.
12712 (inline_string_cmp): Likewise. Use c_readstr instead of
12713 builtin_memcpy_read_str. Add unit_mode temporary.
12715 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
12717 PR middle-end/86528
12718 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
12719 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
12721 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
12723 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
12724 Alphabetize prototypes of built-in functions, separating out
12725 built-in functions that are listed in this section but should be
12726 described elsewhere.
12728 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
12731 * expmed.c (emit_store_flag): Do not emit setcc followed by a
12732 conditional move when trapping comparison was split to a
12733 non-trapping one (and vice versa).
12735 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12737 * config/s390/s390.c (s390_function_profiler): Generate nops
12738 instead of profiler call sequences.
12739 * config/s390/s390.opt: Add the new option.
12741 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12743 * config/s390/s390.c (s390_function_profiler): Generate
12744 __mcount_loc section.
12745 * config/s390/s390.opt: Add the new option.
12747 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
12749 * common.opt: Add the new warning.
12750 * config/s390/s390.c (s390_function_profiler): Emit "brasl
12751 %r0,__fentry__" when -mfentry is specified.
12752 (s390_option_override_internal): Disallow -mfentry for 31-bit
12754 * config/s390/s390.opt: Add the new option.
12756 2018-07-16 Richard Biener <rguenther@suse.de>
12759 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
12760 for function-local FUNCTION_DECL and RESULT_DECL immediately.
12762 2018-07-16 Martin Liska <mliska@suse.cz>
12765 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
12768 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
12770 * config/arc/arcHS.md: Update ARCHS scheduling rules.
12772 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
12774 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
12776 * config/arc/arc-cpus.def (hs4x): New cpu.
12778 * config/arc/arc-tables.opt: Regenerate.
12779 * config/arc/arc.c (arc_sched_issue_rate): New function.
12780 (TARGET_SCHED_ISSUE_RATE): Define.
12781 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
12782 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
12784 (attr tune): Add ARCHS4x tune values.
12785 (attr tune_dspmpy): Define.
12786 (*tst): Correct instruction type.
12787 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
12788 * config/arc/arcHS4x.md: New file.
12789 * config/arc/fpu.md: Update instruction type attributes.
12790 * config/arc/t-multilib: Regenerate.
12792 2018-07-16 Tom de Vries <tdevries@suse.de>
12795 * var-tracking.c (vt_initialize): Fix pre_dec handling.
12797 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12799 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
12802 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
12804 PR tree-optimization/86514
12805 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
12806 conversion to a boolean type from a type with greater precision.
12808 2018-07-16 Tom de Vries <tdevries@suse.de>
12810 * var-tracking.c (vt_initialize): Print adjusted insn slim if
12811 dump_flags request TDF_SLIM.
12813 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
12815 * fold-const.c (int_const_binop_1): Abstract...
12816 (wide_int_binop): ...wide int code here.
12817 (poly_int_binop): ...poly int code here.
12818 Abstract the rest of int_const_binop_1 into int_const_binop.
12819 * fold-const.h (wide_int_binop): New.
12820 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
12821 Remove useless PLUS/MINUS_EXPR case.
12822 (zero_nonzero_bits_from_vr): Move wide int code...
12823 (zero_nonzero_bits_from_bounds): ...here.
12824 (extract_range_from_binary_expr_1): Move mask optimization code...
12825 (range_easy_mask_min_max): ...here.
12826 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
12827 (range_easy_mask_min_max): New.
12829 2018-07-15 Jeff Law <law@redhat.com>
12832 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
12835 2018-07-14 Jim Wilson <jimw@sifive.com>
12837 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
12839 2018-07-14 Paul Koning <ni1d@arrl.net>
12841 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
12843 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
12845 * lto-streamer-out.c (copy_function_or_variable): Dump info about
12848 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
12849 Steve Munroe <munroesj52@gmail.com>
12851 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
12852 (_mm_andnot_si128): Likewise.
12853 (_mm_or_si128): Likewise.
12854 (_mm_xor_si128): Likewise.
12856 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
12858 PR middle-end/78809
12859 * builtins.c (expand_builtin_memcmp): Inline the calls first
12860 when result_eq is false.
12861 (expand_builtin_strcmp): Inline the calls first.
12862 (expand_builtin_strncmp): Likewise.
12863 (inline_string_cmp): New routine. Expand a string compare
12864 call by using a sequence of char comparison.
12865 (inline_expand_builtin_string_cmp): New routine. Inline expansion
12866 a call to str(n)cmp/memcmp.
12867 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
12869 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
12871 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
12873 * config/arm/driver-arm.c: Include arm-native.h.
12874 (host_detect_local_cpu): Use auto-generated data tables.
12875 (vendors, arm_cpu_table): Delete. Move part information to ...
12876 * config/arm/arm-cpus.in: ... here.
12877 * config/arm/parsecpu.awk (gen_native): New function.
12878 (vendor, part): New CPU fields.
12879 (END): Add support for building the native CPU detection tables.
12880 * config/arm/t-arm (arm-native.h): Add build rule.
12881 (driver-arm.o): Add dependency on arm-native.h.
12883 2018-07-13 Richard Biener <rguenther@suse.de>
12885 PR middle-end/85974
12886 * match.pd (addr1 - addr2): Allow either of the operand to
12889 2018-07-13 Tom de Vries <tdevries@suse.de>
12891 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
12894 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
12896 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
12897 arrays instead of numbered variables.
12899 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
12901 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
12902 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
12904 2018-07-13 Richard Biener <rguenther@suse.de>
12907 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
12908 instead of get_context_die.
12910 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
12911 Richard Biener <rguenther@suse.de>
12913 PR middle-end/86489
12914 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
12915 that the loop latch destination where phi is defined.
12917 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
12919 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
12920 (riscv_expand_epilogue): Add assertion to check interrupt mode.
12921 (riscv_set_current_function): Extract getting interrupt type to new
12923 (riscv_get_interrupt_type): New function.
12924 (riscv_merge_decl_attributes): New function, checking interrupt type is
12926 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
12928 2018-07-12 Paul Koning <ni1d@arrl.net>
12930 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
12933 2018-07-12 Paul Koning <ni1d@arrl.net>
12935 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
12936 zero reference, add doloop_end instead.
12937 * doc/md.texi (decrement_and_branch_until_zero): Remove.
12938 (Looping patterns): Remove decrement_and_branch_until_zero. Add
12939 detail for doloop_end.
12941 2018-07-12 Martin Sebor <msebor@redhat.com>
12944 * attribs.c (decl_attributes): Reject conflicting attributes before
12945 calling attribute handlers.
12947 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
12949 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
12951 (gcc::dump_manager::get_dump_file_name): likewise.
12952 (dump_begin): Likewise.
12953 * dumpfile.h (dump_begin): Update prototype.
12954 (gcc::dump_manager::get_dump_file_name,
12955 gcc::dump_manager::get_dump_file_name): Update prototype.
12957 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12959 * internal-fn.h (vectorizable_internal_fn_p): New function.
12960 * tree-vect-slp.c (compatible_calls_p): Likewise.
12961 (vect_build_slp_tree_1): Remove nops argument. Handle calls
12962 to internal functions.
12963 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
12965 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12967 * fold-const.h (inverse_conditions_p): Declare.
12968 * fold-const.c (inverse_conditions_p): New function.
12969 * match.pd: Use inverse_conditions_p. Add folds of view_converts
12970 that test the inverse condition of a conditional internal function.
12971 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
12972 * internal-fn.c (internal_fn_mask_index): Handle conditional
12973 internal functions.
12974 (vectorized_internal_fn_supported_p): New function.
12975 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
12976 (any_pred_load_store): Replace with...
12977 (need_to_predicate): ...this new variable.
12978 (redundant_ssa_names): New variable.
12979 (ifcvt_can_use_mask_load_store): Move initial checks to...
12980 (ifcvt_can_predicate): ...this new function. Handle tree codes
12981 for which a conditional internal function exists.
12982 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
12983 instead of ifcvt_can_use_mask_load_store. Update after variable
12985 (predicate_load_or_store): New function, split out from
12986 predicate_mem_writes.
12987 (check_redundant_cond_expr): New function.
12988 (value_available_p): Likewise.
12989 (predicate_rhs_code): Likewise.
12990 (predicate_mem_writes): Rename to...
12991 (predicate_statements): ...this. Use predicate_load_or_store
12992 and predicate_rhs_code.
12993 (combine_blocks, tree_if_conversion): Update after above name changes.
12994 (ifcvt_local_dce): Handle redundant_ssa_names.
12995 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
12996 general conditional functions.
12997 * tree-vect-stmts.c (vectorizable_call): Likewise.
12999 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13000 Alan Hayward <alan.hayward@arm.com>
13001 David Sherwood <david.sherwood@arm.com>
13003 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
13004 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
13005 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
13006 plus and minus and convert them into IFN_COND_FMA-based sequences.
13007 (convert_mult_to_fma): Handle conditional plus and minus.
13009 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13011 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
13012 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
13013 (cond_fnms_optab): New optabs.
13014 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
13015 internal functions.
13016 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
13017 * internal-fn.h (get_conditional_internal_fn): Declare.
13018 (get_unconditional_internal_fn): Likewise.
13019 * internal-fn.c (cond_ternary_direct): New macro.
13020 (expand_cond_ternary_optab_fn): Likewise.
13021 (direct_cond_ternary_optab_supported_p): Likewise.
13022 (FOR_EACH_COND_FN_PAIR): Likewise.
13023 (get_conditional_internal_fn): New function.
13024 (get_unconditional_internal_fn): Likewise.
13025 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
13026 (gimple_match_op::gimple_match_op): Add a new overload for 5
13028 (gimple_match_op::set_op): Likewise.
13029 (gimple_resimplify5): Declare.
13030 * genmatch.c (decision_tree::gen): Generate simplifications for
13032 * gimple-match-head.c (gimple_simplify): Define an overload for
13033 5 operands. Handle calls with 5 arguments in the top-level overload.
13034 (convert_conditional_op): Handle conversions from unconditional
13035 internal functions to conditional ones.
13036 (gimple_resimplify5): New function.
13037 (build_call_internal): Pass a fifth operand.
13038 (maybe_push_res_to_seq): Likewise.
13039 (try_conditional_simplification): Try converting conditional
13040 internal functions to unconditional internal functions.
13041 Handle 3-operand unconditional forms.
13042 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
13043 Define ternary equivalents of the current rules for binary conditional
13044 internal functions.
13045 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
13046 ternary operations.
13047 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
13048 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
13049 (optab): Handle them.
13050 (SVE_COND_FP_TERNARY): New int iterator.
13051 (sve_fmla_op, sve_fmad_op): New int attributes.
13052 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
13053 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
13054 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
13056 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
13058 * target.def (preferred_else_value): New target hook.
13059 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
13060 * doc/tm.texi: Regenerate.
13061 * targhooks.h (default_preferred_else_value): Declare.
13062 * targhooks.c (default_preferred_else_value): New function.
13063 * internal-fn.h (conditional_internal_fn_code): Declare.
13064 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
13065 (get_conditional_internal_fn): Use it.
13066 (conditional_internal_fn_code): New function.
13067 * gimple-match.h (gimple_match_cond): New struct.
13068 (gimple_match_op): Add a cond member function.
13069 (gimple_match_op::gimple_match_op): Update all forms to take a
13071 * genmatch.c (expr::gen_transform): Use the same condition as res_op
13072 for the suboperation, but don't specify a particular else_value.
13073 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
13074 (visit_nary_op, visit_reference_op_load): Pass
13075 gimple_match_cond::UNCOND to the gimple_match_op constructor.
13076 * gimple-match-head.c: Include tree-eh.h
13077 (convert_conditional_op): New function.
13078 (maybe_resimplify_conditional_op): Likewise.
13079 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
13080 (gimple_resimplify2): Likewise.
13081 (gimple_resimplify3): Likewise.
13082 (gimple_resimplify4): Likewise.
13083 (maybe_push_res_to_seq): Return null for conditional operations.
13084 (try_conditional_simplification): New function.
13085 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
13087 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
13089 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
13091 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
13093 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
13095 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
13097 (hash_tree): Do not hash DECL_FCONTEXT
13098 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
13099 Do not stream DECL_FCONTEXT.
13100 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
13101 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
13103 2018-07-12 Richard Biener <rguenther@suse.de>
13106 * dwarf2out.c (gen_block_die): Only output blocks when they have
13107 at least one !DECL_IGNORED_P variable.
13109 2018-07-12 Richard Biener <rguenther@suse.de>
13112 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
13113 Remove -mieee-fp handling.
13115 2018-07-12 Richard Biener <rguenther@suse.de>
13117 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
13118 left-over from last patch.
13120 2018-07-12 Jakub Jelinek <jakub@redhat.com>
13122 PR tree-optimization/86492
13123 * gimple-ssa-store-merging.c
13124 (imm_store_chain_info::coalesce_immediate_stores): Call
13125 check_no_overlap even for the merge_overlapping case. Formatting fix.
13127 2018-07-12 Richard Biener <rguenther@suse.de>
13129 PR middle-end/86479
13130 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
13131 move possibly trapping operations into the conditional.
13133 2018-07-12 Richard Biener <rguenther@suse.de>
13135 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
13136 (vn_lookup_simplify_result): Remove recursion limit applied
13138 (vn_nary_build_or_lookup_1): Adjust.
13139 (try_to_simplify): Likewise.
13140 * gimple-match-head.c (gimple_resimplify1): Instead apply one
13142 (gimple_resimplify2): Likewise.
13143 (gimple_resimplify3): Likewise.
13144 (gimple_resimplify4): Likewise.
13146 2018-07-11 Jakub Jelinek <jakub@redhat.com>
13148 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
13149 Use __mmask64 type instead of __mmask8 for __M argument.
13150 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
13151 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
13153 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
13154 __mmask16 for __M argument.
13155 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
13156 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
13157 to __mmask16 instead of __mmask8.
13158 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
13159 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
13160 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
13161 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
13162 instead of __mmask16 for __U argument.
13163 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
13164 __mmask16 instead of __mmask8 for __U argument.
13165 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
13167 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
13169 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
13171 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
13173 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
13175 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
13177 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
13179 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
13181 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
13183 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
13185 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
13186 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
13187 return type as well as __M argument type and all casts from __mmask8
13189 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
13190 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
13191 return type as well as __M argument type and all casts from __mmask8
13193 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
13194 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
13195 return type as well as __M argument type and all casts from __mmask8
13197 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
13198 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
13199 return type as well as __M argument type and all casts from __mmask8
13201 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
13202 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
13205 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
13207 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
13208 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
13211 2018-07-11 Paul Koning <ni1d@arrl.net>
13213 * doc/md.texi (define_subst): Document how multiple occurrences of
13214 the same argument in the replacement pattern are handled.
13216 2018-07-11 Paul Koning <ni1d@arrl.net>
13218 * doc/extend.texi (Common Variable Attributes): Move "mode" into
13219 alphabetical order.
13220 (Common Type Attributes): Add "mode" attribute.
13222 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
13224 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
13225 stream DECL_ORIGINAL_TYPE.
13226 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
13227 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
13228 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
13229 Do not walk original type.
13230 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
13232 (write_ts_decl_non_common_tree_pointers): Do not stream
13234 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
13235 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
13237 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
13239 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
13240 thread twice from the same starting edge.
13242 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
13244 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
13245 * gimple.c (gimple_call_nonnull_result_p): ...here...
13246 (gimple_call_nonnull_arg): ...and here.
13247 * gimple.h (gimple_call_nonnull_result_p): New.
13248 (gimple_call_nonnull_arg): New.
13250 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
13252 * config/arm/arm-cpus.in: Move information from fpu field of each
13253 cpu definition to the isa field.
13254 * config/arm/parsecpu.awk (fpu): Delete match rule.
13255 (gen_comm_data): Don't add bits from the CPU's FPU entry.
13257 2018-07-11 Richard Biener <rguenther@suse.de>
13260 * dwarf2out.c (init_sections_and_labels): Use
13261 output_asm_line_debug_info consistently.
13262 (dwarf2out_early_finish): Likewise.
13263 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
13266 2018-07-11 Richard Biener <rguenther@suse.de>
13268 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
13269 Rework father_bb setting in a way to avoid propagating constants
13270 multiple times on a loop body.
13272 2018-07-10 Mark Wielaard <mark@klomp.org>
13275 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
13277 2018-07-10 Richard Biener <rguenther@suse.de>
13279 * hash-map.h (hash_map::iterator::operator*): Return
13280 references to key and value.
13282 2018-07-10 Jakub Jelinek <jakub@redhat.com>
13285 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
13286 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
13287 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
13288 (gimplify_omp_for): For composite loops, move outer
13289 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
13290 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
13291 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
13292 TREE_LIST for both the original class iterator and the "last" helper
13293 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
13294 loop, remember has_decl_expr from outer composite loops for the
13295 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
13297 2018-07-09 Martin Sebor <msebor@redhat.com>
13299 PR middle-end/77357
13300 PR middle-end/86428
13301 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
13302 accessing implicitly initialized array elements.
13303 * expr.c (string_constant): Handle string initializers of
13304 character arrays within aggregates.
13305 * gimple-fold.c (fold_array_ctor_reference): Add argument.
13306 Store element offset. As a special case, handle zero size.
13307 (fold_nonarray_ctor_reference): Same.
13308 (fold_ctor_reference): Add argument. Store subobject offset.
13309 * gimple-fold.h (fold_ctor_reference): Add argument.
13311 2018-07-09 Paul Koning <ni1d@arrl.net>
13313 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
13314 (pdp11_insn_cost): New function.
13315 (pdp11_md_asm_adjust): New function.
13316 (TARGET_INVALID_WITHIN_DOLOOP): Define.
13317 (pdp11_rtx_costs): Update to match machine better.
13318 (output_addr_const_pdp11): Correct format mismatch warnings.
13319 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
13320 * config/pdp11/pdp11.md: General change to add base_cost and/or
13321 length attributes for use by new pdp11_insn_cost function.
13322 (MIN_BRANCH): Correct definition.
13324 (doloop_end): Use standard pattern name for looping pattern.
13325 (doloop_end_nocc): New.
13326 (movsf): Add another constraint alternative.
13327 (zero_extendqihi2): Add constraint alternatives for not in place
13329 (zero_extendhisi2): Remove.
13330 (shift patterns): Add CC handling variants.
13334 (define_peephole2): New peephole to recognize mov that sets CC for
13337 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
13339 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
13340 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
13343 2018-07-09 Jakub Jelinek <jakub@redhat.com>
13346 * real.c (real_nextafter): Return true if result is denormal.
13348 2018-07-09 Martin Liska <mliska@suse.cz>
13350 * common.opt: Add back wrongly removed attribute.
13352 2018-07-09 Richard Biener <rguenther@suse.de>
13355 * dwarf2out.c (gen_block_die): For an early generated DIE
13356 always output high/low PC attributes.
13358 2018-07-09 Tom de Vries <tdevries@suse.de>
13360 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
13361 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
13362 onto VAR_DECL with abstract origin.
13364 2018-07-07 Jim Wilson <jimw@sifive.com>
13366 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
13368 2018-07-07 Tom de Vries <tdevries@suse.de>
13370 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
13372 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
13374 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
13376 (combine_bound): Use wide-int overflow calculation instead of
13378 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
13380 * fold-const.c (int_const_binop_2): Same.
13381 (extract_muldiv_1): Same.
13382 (fold_div_compare): Same.
13383 (fold_abs_const): Same.
13385 * poly-int.h (add): Same.
13389 * predict.c (predict_iv_comparison): Same.
13390 * profile-count.c (slow_safe_scale_64bit): Same.
13391 * simplify-rtx.c (simplify_const_binary_operation): Same.
13392 * tree-chrec.c (tree_fold_binomial): Same.
13393 * tree-data-ref.c (split_constant_offset_1): Same.
13394 * tree-if-conv.c (idx_within_array_bound): Same.
13395 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
13396 * tree-ssa-phiopt.c (minmax_replacement): Same.
13397 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
13398 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
13399 * vr-values.c (vr_values::adjust_range_with_scev): Same.
13400 * wide-int.cc (wi::add_large): Same.
13401 (wi::mul_internal): Same.
13402 (wi::sub_large): Same.
13403 (wi::divmod_internal): Same.
13404 * wide-int.h: Change overflow type to overflow_type for neg, add,
13405 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
13406 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
13407 mul_internal, divmod_internal.
13408 (overflow_type): New enum.
13409 (accumulate_overflow): New.
13411 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
13413 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
13414 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
13416 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
13418 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
13419 argument is checked for zero before entering loop, avoid checking again.
13421 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
13423 * gimplify.h (generic_expr_could_trap_p): Set as global function.
13424 * gimplify.h (generic_expr_could_trap_p): Likwise.
13425 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
13427 2018-07-06 Jakub Jelinek <jakub@redhat.com>
13429 PR tree-optimization/86401
13430 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
13431 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
13432 (fold_bit_and_mask): ... here. New helper function for match.pd.
13433 * fold-const.h (fold_bit_and_mask): Declare.
13434 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
13436 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
13439 * target.def (translate_mode_attribute): New hook.
13440 * targhooks.h (default_translate_mode_attribute): Declare.
13441 * targhooks.c (default_translate_mode_attribute): New function.
13442 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
13443 * doc/tm.texi: Regenerate.
13444 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
13445 (rs6000_translate_mode_attribute): New function.
13447 2018-07-06 Paul Koning <ni1d@arrl.net>
13449 * doc/md.texi (define_split): Document DONE and FAIL.
13450 (define_peephole2): Ditto.
13452 2018-07-05 Jeff Law <law@redhat.com>
13454 PR tree-optimization/86010
13455 * tree-ssa-dse.c (compute_trims): More aggressively trim at
13456 both the head and tail of mem* and str* calls.
13458 2018-07-05 Jim Wilson <jimw@sifive.com>
13460 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
13462 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
13464 * config/aarch64/aarch64-simd.md: correct flags text for
13465 MIN_EXPR replacement.
13467 2018-07-05 James Clarke <jrtc27@jrtc27.com>
13469 * configure: Regenerated.
13471 2018-07-05 Carl Love <cel@us.ibm.com>
13473 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
13474 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
13475 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
13476 VSX_BUILTIN_DOUBLEL_V4SF.
13478 2018-07-05 Martin Sebor <msebor@redhat.com>
13481 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
13482 than its domain to compute its the upper bound of a char array.
13484 2018-07-05 Nathan Sidwell <nathan@acm.org>
13486 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
13487 * doc/cpp.texi: Update comment.
13488 * doc/tm.texi: Rebuilt.
13489 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
13490 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
13491 * doc/extend.texi (Backwards Compatibility): Clarify it is system
13492 headers affected by extern "C".
13493 * system.h: Poison NO_IMPLICIT_EXTERN_C.
13494 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
13495 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
13496 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
13497 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
13498 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
13499 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
13500 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
13501 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
13502 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
13503 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
13504 config/sparc/sp64-elf.h, config/spu/spu.h,
13505 config/stormy16/stormy16.h, config/v850/v850.h,
13506 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
13507 define NO_IMPLICIT_EXTERN_C.
13508 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
13510 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13513 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
13514 instead of GET_MODE_SIZE when comparing Units.
13516 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13519 * rtlanal.c (set_noop_p): Constrain on mode change,
13520 include hard-reg-set.h
13522 2018-07-05 Tamar Christina <tamar.christina@arm.com>
13524 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
13526 2018-07-05 Jakub Jelinek <jakub@redhat.com>
13529 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
13532 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
13534 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13536 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
13539 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
13541 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
13543 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
13545 PR middle-end/86380
13546 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
13548 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
13550 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
13551 neg_*_op* variables.
13553 2018-07-04 Martin Liska <mliska@suse.cz>
13555 * tree-switch-conversion.c: Define
13556 max_ratio_for_speed and max_ratio_for_size constants.
13558 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13559 Martin Liska <mliska@suse.cz>
13561 PR middle-end/66240
13564 * common.opt: Rename align options with 'str_' prefix.
13565 * common/config/i386/i386-common.c (set_malign_value): New
13567 (ix86_handle_option): Use it to set -falign-* options/
13568 * config/aarch64/aarch64-protos.h (struct tune_params): Change
13569 type from int to string.
13570 * config/aarch64/aarch64.c: Update default values from int
13572 * config/alpha/alpha.c (alpha_override_options_after_change):
13574 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
13575 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13576 max skip conditionally.
13577 * config/i386/freebsd.h (SUBALIGN_LOG): New.
13578 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13579 max skip conditionally.
13580 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13581 max skip conditionally.
13582 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
13583 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13584 max skip conditionally.
13585 * config/i386/i386.c (struct ptt): Change type from int to
13587 (ix86_default_align): Set default values.
13588 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
13589 max skip conditionally.
13590 * config/i386/iamcu.h (SUBALIGN_LOG): New.
13591 (ASM_OUTPUT_MAX_SKIP_ALIGN):
13592 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
13593 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13594 max skip conditionally.
13595 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
13596 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
13597 * config/i386/x86-64.h (SUBALIGN_LOG): New.
13598 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
13599 max skip conditionally.
13600 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13601 * config/ia64/ia64.c (ia64_option_override): Set default values
13602 for alignment options.
13603 * config/m68k/m68k.c: Handle new str_align_* options.
13604 * config/mips/mips.c (mips_set_compression_mode): Change
13606 (mips_option_override): Set default values for options.
13607 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
13609 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13611 * config/rx/rx.c (rx_option_override): Likewise.
13612 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
13613 (LABEL_ALIGN): Use align_labels_log.
13614 (LOOP_ALIGN): Use align_loops_align.
13615 * config/s390/s390.c (s390_asm_output_function_label): Use new
13617 * config/sh/sh.c (sh_override_options_after_change):
13618 Change type of constants.
13619 * config/spu/spu.c (spu_sched_init): Likewise.
13620 * config/sparc/sparc.c (sparc_option_override): Set default
13621 values for options.
13622 * config/visium/visium.c (visium_option_override): Likewise.
13623 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
13624 emit p2align format with last argument if it's not needed.
13625 * doc/invoke.texi: Document extended format of -falign-*.
13626 * final.c: Use align_labels alignment.
13627 * flags.h (struct target_flag_state): Change type to use
13629 (struct align_flags_tuple): New.
13630 (struct align_flags): Likewise.
13631 (align_loops_log): Redefine macro to use new types.
13632 (align_loops_max_skip): Redefine macro to use new types.
13633 (align_jumps_log): Redefine macro to use new types.
13634 (align_jumps_max_skip): Redefine macro to use new types.
13635 (align_labels_log): Redefine macro to use new types.
13636 (align_labels_max_skip): Redefine macro to use new types.
13637 (align_functions_log): Redefine macro to use new types.
13638 (align_loops): Redefine macro to use new types.
13639 (align_jumps): Redefine macro to use new types.
13640 (align_labels): Redefine macro to use new types.
13641 (align_functions): Redefine macro to use new types.
13642 (align_functions_max_skip): Redefine macro to use new types.
13643 (align_loops_value): New macro.
13644 (align_jumps_value): New macro.
13645 (align_labels_value): New macro.
13646 (align_functions_value): New macro.
13647 * function.c (invoke_set_current_function_hook): Propagate
13648 alignment values from flags to global variables default in
13650 * ipa-icf.c (sem_function::equals_wpa): Use
13651 cl_optimization_option_eq instead of memcmp.
13652 * lto-streamer.h (cl_optimization_stream_out): Support streaming
13654 (cl_optimization_stream_in): Likewise.
13655 * optc-save-gen.awk: Support strings in cl_optimization.
13656 * opth-gen.awk: Likewise.
13657 * opts.c (finish_options): Remove error checking of invalid
13659 (MAX_CODE_ALIGN): Remove.
13660 (MAX_CODE_ALIGN_VALUE): Likewise.
13661 (parse_and_check_align_values): New function.
13662 (check_alignment_argument): Likewise.
13663 (common_handle_option): Use check_alignment_argument.
13664 * opts.h (parse_and_check_align_values): Declare.
13665 * toplev.c (init_alignments): Remove.
13666 (read_log_maxskip): New.
13667 (parse_N_M): Likewise.
13668 (parse_alignment_opts): Likewise.
13669 (backend_init_target): Remove usage of init_alignments.
13670 * toplev.h (parse_alignment_opts): Declare.
13671 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
13673 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
13674 * tree.c (cl_option_hasher::equal): New.
13675 * varasm.c: Use new global macros.
13677 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
13679 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
13680 Use a simpler align directive also if MAXSKIP = ALIGN-1.
13681 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13682 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13683 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13684 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13685 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
13686 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
13687 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
13688 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13689 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13690 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13691 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13693 2018-07-04 Martin Liska <mliska@suse.cz>
13694 Jonathan Wakely <jwakely@redhat.com>
13696 * coverage.c: Use correct type.
13697 * doc/invoke.texi: Language correction.
13699 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
13702 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
13703 ENDBRANCH for non-tail call which may return via indirect branch.
13704 * doc/extend.texi: Document indirect_return attribute.
13706 2018-07-03 Martin Sebor <msebor@redhat.com>
13708 PR tree-optimization/86274
13709 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
13711 (format_floating): Correct handling of infinities and NaNs.
13713 2018-07-03 Martin Sebor <msebor@redhat.com>
13715 * print-tree.c (print_real_cst): New function.
13716 (print_node_brief): Call it.
13717 (print_node): Ditto.
13719 2018-07-03 Jeff Law <law@redhat.com>
13721 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
13722 into a single pattern.
13724 * config/h8300/h8300.md (ors code_iterator): New.
13725 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
13726 a single pattern and single splitter.
13727 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
13728 (iorqi3_1, xorqi3_1): Likewise.
13729 (iorqi3, xorqi3 expanders): Similarly.
13731 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
13732 (movmd_internal) into a single pattern using the P mode iterator.
13733 (movmd splitters): Similarly.
13734 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
13735 (movsd splitters): Similarly.
13737 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
13738 ADDB, ADDW and ADDL into a single ADD attribute which selects the
13739 right table based on the size of the operand.
13740 * config/h8300/h8300.md (length_table): Corresponding changes. All
13741 references to "addb", "addw" and "addl" changed to "add".
13742 (btst patterns): Merge two variants into a single pattern.
13743 (tstqi, tsthi): Likewise.
13744 (addhi3_incdec, addsi3_incdec): Likewise.
13745 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
13746 (mulhi3, mulsi3): Likewise.
13747 (udivhi3, udivsi3): Likewise.
13748 (divhi3, divsi3): Likewise.
13749 (andorqi3, andorhi3, andorsi3): Likewise.
13751 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
13754 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
13755 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
13757 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13759 PR tree-optimization/85694
13760 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
13761 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
13763 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
13764 (<u>avg<mode>3_ceil): New patterns.
13766 2018-07-03 David Malcolm <dmalcolm@redhat.com>
13768 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
13769 scan-tree-dump directive.
13770 * gcc.dg/vect/slp-perm-2.c: Likewise.
13771 * gcc.dg/vect/slp-perm-3.c: Likewise.
13772 * gcc.dg/vect/slp-perm-5.c: Likewise.
13773 * gcc.dg/vect/slp-perm-6.c: Likewise.
13774 * gcc.dg/vect/slp-perm-7.c: Likewise.
13775 * gcc.dg/vect/slp-perm-8.c: Likewise.
13777 2018-07-03 Marek Polacek <polacek@redhat.com>
13779 PR middle-end/86202
13780 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
13782 2018-07-03 Richard Biener <rguenther@suse.de>
13785 * tree-ssa-structalias.c (find_func_clobbers): Properly
13786 handle indirect calls.
13788 2018-07-03 Jeff Law <law@redhat.com>
13790 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
13791 (shifts): New code iterator.
13792 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
13793 expander. Fix HImode handling on H8/SX.
13794 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
13795 (subqi3, subhi3, subsi3 expanders): Likewise.
13796 (andqi3, andhi3, andsi3 expanders): Likewise.
13797 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
13798 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
13799 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
13800 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
13801 (zero_extendqihi2, zero_extendqisi2): Likewise.
13802 (extendqihi2, extendqisi2): Likewise.
13803 (rotlqi3, rotlhi3, rotlsi3): Likewise.
13804 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
13805 (rotlqi3_1, rotlhi3_1): Likewise.
13806 (logicalhi3_sn, logicalsi3_sn): Likewise.
13807 (logicalhi3, logicalsi3): Likewise.
13809 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13811 * tree-vect-patterns.c (vect_recog_rotate_pattern)
13812 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
13813 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
13814 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
13815 type to append_pattern_def_seq instead of creating a stmt_vec_info
13817 (build_mask_conversion): Likewise. Remove vinfo argument.
13818 (vect_add_conversion_to_patterm): Likewise, renaming to...
13819 (vect_add_conversion_to_pattern): ...this.
13820 (vect_recog_mask_conversion_pattern): Update call to
13821 build_mask_conversion. Pass the vector type to
13822 append_pattern_def_seq here too.
13823 (vect_recog_gather_scatter_pattern): Update call to
13824 vect_add_conversion_to_pattern.
13826 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13828 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
13829 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13830 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
13831 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
13832 STMT_VINFO_PATTERN_DEF_SEQ to null here.
13833 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
13834 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
13835 append_pattern_def_seq instead of new_pattern_def_seq.
13836 (vect_recog_divmod_pattern): Do both of the above.
13837 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
13840 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13842 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
13843 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
13844 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
13845 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
13846 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
13847 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
13848 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
13849 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
13850 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
13851 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
13852 parameter with a single stmt_vec_info.
13853 (vect_recog_func_ptr): Likewise.
13854 (vect_recog_gather_scatter_pattern): Likewise, folding in...
13855 (vect_try_gather_scatter_pattern): ...this.
13856 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
13857 the stmt_vec_info of the statement to be matched. Don't clear
13858 STMT_VINFO_RELATED_STMT.
13859 (vect_pattern_recog): Update call accordingly.
13861 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13863 PR tree-optimization/85694
13864 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
13865 (uavgM3_ceil): Document new optabs.
13866 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
13867 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
13869 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
13870 (savg_ceil_optab): New optabs.
13871 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
13872 (vect_vect_recog_func_ptrs): Add it.
13873 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
13874 constant directly from the associated lhs.
13876 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13878 * tree-vect-patterns.c (vect_split_statement): New function.
13879 (vect_convert_input): Use it to try to split an existing cast.
13881 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13883 * poly-int.h (print_hex): New function.
13884 * dumpfile.h (dump_dec, dump_hex): Declare.
13885 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
13886 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
13887 min_input_precision, operation_precision and operation_sign.
13888 * tree-vect-patterns.c (vect_get_range_info): New function.
13889 (vect_same_loop_or_bb_p, vect_single_imm_use)
13890 (vect_operation_fits_smaller_type): Delete.
13891 (vect_look_through_possible_promotion): Add an optional
13892 single_use_p parameter.
13893 (vect_recog_over_widening_pattern): Rewrite to use new
13894 stmt_vec_info infomration. Handle one operation at a time.
13895 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
13896 (vect_truncatable_operation_p, vect_set_operation_type)
13897 (vect_set_min_input_precision): New functions.
13898 (vect_determine_min_output_precision_1): Likewise.
13899 (vect_determine_min_output_precision): Likewise.
13900 (vect_determine_precisions_from_range): Likewise.
13901 (vect_determine_precisions_from_users): Likewise.
13902 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
13903 (vect_vect_recog_func_ptrs): Put over_widening first.
13905 (vect_pattern_recog): Call vect_determine_precisions.
13907 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
13909 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
13910 statements that have been replaced by further pattern statements.
13911 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
13913 2018-07-03 Richard Biener <rguenther@suse.de>
13915 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
13916 always set *dt. Dump vectype in vectype overload.
13917 * dumpfile.h (dump_gimple_expr): New function.
13918 (dump_gimple_expr_loc): Likewise.
13919 * dumpfile.c (dump_gimple_expr): New function.
13920 (dump_gimple_expr_loc): Likewise.
13922 2018-07-02 Jeff Law <law@redhat.com>
13924 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
13925 the H8/300, H8/300H and H8/S variants into a single pattern.
13926 (movhi_h8300, movqi_h8300hs): Similarly.
13927 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
13928 (QHI mode iterator): New.
13930 * config/h8300/h8300.md: Remove trailing whitespace.
13932 2018-07-02 Jim Wilson <jimw@sifive.com>
13934 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
13935 instead of emit_insn for interrupt returns.
13936 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
13937 (riscv_sret, riscv_uret): Likewise.
13939 2018-07-02 David Malcolm <dmalcolm@redhat.com>
13941 * pretty-print.c (selftest::test_pp_format): Move save and restore
13942 of quotes to class auto_fix_quotes, and add an instance.
13943 * selftest.c: Include "intl.h".
13944 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
13945 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
13946 * selftest.h (selftest::auto_fix_quotes): New class.
13948 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
13950 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
13951 (aarch64_sve_prepare_conditional_op): Remove.
13952 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
13953 Allow aarch64_simd_reg_or_zero as select operand; remove
13954 the aarch64_sve_prepare_conditional_op call.
13955 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
13956 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
13957 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
13958 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
13959 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
13960 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
13961 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
13962 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
13963 and a splitters to match all of the *_any patterns.
13964 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
13966 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
13967 (SVE_COND_FP_BINARY_REV): Remove.
13968 (sve_int_op_rev, sve_fp_op_rev): New.
13969 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
13970 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
13971 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
13972 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
13973 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
13974 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
13975 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
13976 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
13977 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
13979 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
13980 Remove match_dup 1 from the inner unspec.
13981 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
13983 * config/aarch64/aarch64.md (movprfx): New attr.
13984 (length): Default movprfx to 8.
13985 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
13986 (*madd<SVE_I>, *msub<SVE_I): Likewise.
13987 (*<su>mul<SVE_I>3_highpart): Likewise.
13988 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
13989 (*v<ASHIFT><SVE_I>3): Likewise.
13990 (*<su><MAXMIN><SVE_I>3): Likewise.
13991 (*<su><MAXMIN><SVE_F>3): Likewise.
13992 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
13993 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
13994 (*div<SVE_F>4): Likewise.
13996 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
13998 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
14001 2018-07-02 Richard Biener <rguenther@suse.de>
14003 PR tree-optimization/86363
14004 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
14005 memset argument refers to a non-variable address.
14007 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
14009 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
14010 {PLUS,MINUS}_EXPR code to...
14011 (adjust_symbolic_bound): ...here,
14012 (combine_bound): ...here,
14013 (set_value_range_with_overflow): ...and here.
14015 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
14017 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
14019 (extract_range_from_abs_expr): ...here.
14021 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
14023 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
14024 -fno-omit-frame-pointer when not optimizing.
14026 2018-07-02 Martin Liska <mliska@suse.cz>
14029 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
14030 (propagate_nothrow): Likewise.
14032 2018-07-02 Martin Liska <mliska@suse.cz>
14035 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
14037 2018-07-02 David Malcolm <dmalcolm@redhat.com>
14039 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
14040 function in r262149, changing "loc" param from source_location to
14041 const dump_location_t &.
14042 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
14043 declaration, as above.
14045 2018-07-01 Paul Koning <ni1d@arrl.net>
14047 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
14048 -munit-asm, -mgnu-asm, -mdec-asm.
14049 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
14050 (pdp11_output_labelref): New.
14051 (pdp11_output_def): New.
14052 (pdp11_output_addr_vec_elt): New.
14053 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
14054 %# and %@ format codes.
14055 (pdp11_option_override): New.
14056 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
14057 (pdp11_output_ident): New.
14058 (pdp11_asm_named_section): New.
14059 (pdp11_asm_init_sections): New.
14060 (pdp11_file_start): New.
14061 (pdp11_file_end): New.
14062 (output_ascii): Use .ascii/.asciz for -mdec-asm.
14063 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
14064 %o, like %c but octal.
14065 (pdp11_option_override): New.
14066 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
14068 (DATA_SECTION_ASM_OP): Ditto.
14069 (READONLY_DATA_SECTION_ASM_OP): New.
14070 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
14071 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
14072 (ASM_OUTPUT_LABELREF): Ditto.
14073 (ASM_OUTPUT_DEF): Ditto.
14074 (ASM_OUTPUT_EXTERNAL): New.
14075 (ASM_OUTPUT_SOURCE_FILENAME): New.
14076 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
14077 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
14078 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
14079 %# and %@ format codes.
14080 * config/pdp11/pdp11.opt (mgnu-asm): New.
14081 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
14082 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
14083 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
14085 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
14087 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
14088 dereferencing path[] beyond its length.
14090 (debug_all_paths): New.
14091 (rewire_first_differing_edge): New.
14092 (adjust_paths_after_duplication): New.
14093 (duplicate_thread_path): Call adjust_paths_after_duplication.
14095 (thread_through_all_blocks): Add new argument to
14096 duplicate_thread_path.
14098 2018-06-30 Jim Wilson <jimw@sifive.com>
14100 * config/riscv/predicates.md (p2m1_shift_operand): New.
14101 (high_mask_shift_operand): New.
14102 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
14103 pattern using p2m1_shift_operand.
14104 (lshsi3_zero_extend_3+2): New combiner pattern using
14105 high_mask_shift_operand.
14107 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14109 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
14111 (vect_recog_rotate_pattern): ...here.
14112 (vect_convert_input): Try to insert casts of invariants in the
14114 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
14115 preheader to be empty.
14117 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14119 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
14120 vector type. If given, install it in the new statement's
14121 STMT_VINFO_VECTYPE.
14122 (vect_element_precision): New function.
14123 (vect_unpromoted_value): New struct.
14124 (vect_unpromoted_value::vect_unpromoted_value): New function.
14125 (vect_unpromoted_value::set_op): Likewise.
14126 (vect_look_through_possible_promotion): Likewise.
14127 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
14128 (vect_widened_op_tree, vect_convert_input): Likewise.
14129 (vect_convert_inputs, vect_convert_output): Likewise.
14130 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
14131 to handle the optional cast of the multiplication result and
14132 vect_widened_op_tree to detect the widened multiplication itself.
14133 Do not require the input and output of promotion casts to have
14134 the same sign, but base the signedness of the operation on the
14135 input rather than the result. If the pattern includes two
14136 promotions, check that those promotions have the same sign.
14137 Do not restrict the MULT_EXPR handling to a double-width result;
14138 handle quadruple-width results and wider. Use vect_convert_inputs
14139 to convert the inputs to the common type.
14140 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
14141 to handle the optional cast of the ABS result. Also allow a sign
14142 change or a sign extension between the ABS and MINUS.
14143 Use vect_widened_op_tree to detect the widened subtraction and use
14144 vect_convert_inputs to convert the inputs to the common type.
14145 (vect_handle_widen_op_by_const): Delete.
14146 (vect_recog_widen_op_pattern): New function.
14147 (vect_recog_widen_mult_pattern): Use it.
14148 (vect_recog_widen_shift_pattern): Likewise.
14149 (vect_recog_widen_sum_pattern): Use
14150 vect_look_through_possible_promotion to handle the promoted
14153 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14155 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
14156 the containing gimple_seq *.
14157 * gimple-iterator.h (gsi_for_stmt): Declare it.
14158 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
14159 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
14160 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
14161 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
14162 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
14164 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
14166 (vect_mark_pattern_stmts): ...here. Handle cases in which the
14167 statement being replaced is part of an existing pattern
14168 definition sequence, inserting the new pattern statements before
14170 (vect_pattern_recog_1): Don't return a bool. If the statement
14171 is already part of a pattern, instead apply pattern matching
14172 to the pattern definition statements. Don't clear the
14173 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
14174 (vect_pattern_recog): Don't break after the first match;
14175 continue processing the pattern definition statements instead.
14176 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
14178 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14180 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
14181 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
14182 (vect_recog_widen_sum_pattern): Use it.
14184 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14186 * tree-vect-loop.c (vectorizable_reduction): Assert that the
14187 phi is not a pattern statement and has not been replaced by
14188 a pattern statement.
14189 * tree-vect-patterns.c (type_conversion_p): Don't check
14190 STMT_VINFO_IN_PATTERN_P.
14191 (vect_recog_vector_vector_shift_pattern): Likewise.
14192 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
14193 the pattern statement rather than the original statement; check
14194 directly for a WIDEN_MULT_EXPR here.
14195 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
14196 vect_is_simple_use to return the pattern statement rather
14197 than the original statement; use is_pattern_stmt_p to check
14198 for such a pattern statement.
14199 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
14200 to return the pattern statement rather than the original statement;
14201 don't do the same transformation here.
14202 (vect_is_simple_use): If the defining statement has been replaced
14203 by a pattern statement, return the pattern statement instead.
14204 Remove the corresponding (local) transformation from the vectype
14207 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
14209 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
14210 end and default to null.
14211 * tree-vect-loop.c (vect_create_epilog_for_reduction)
14212 (vectorizable_reduction): Update calls accordingly, dropping the
14213 gimple ** argument if the passed-back statement isn't needed.
14214 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
14215 (vect_recog_rotate_pattern): Likewise.
14216 (vect_recog_mask_conversion_pattern): Likewise.
14217 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
14218 (vect_mask_constant_operand_p): Likewise.
14219 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
14220 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
14221 (get_group_load_store_type, get_load_store_type): Likewise.
14222 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
14223 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
14224 (vectorizable_conversion, vectorizable_assignment): Likewise.
14225 (vectorizable_shift, vectorizable_operation): Likewise.
14226 (vectorizable_store, vect_is_simple_cond): Likewise.
14227 (vectorizable_condition, vectorizable_comparison): Likewise.
14228 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
14229 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
14230 and move it to the end. Cope with null def_stmt_outs.
14232 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
14234 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
14236 2018-06-29 Jeff Law <law@redhat.com>
14238 * config/v850/v850.c (v850_legitimate_address_p): Handle large
14239 displacements for TARGET_V850E2V3 and newer.
14240 (TARGET_LRA_P): Remove. Defaults to LRA now.
14241 * config/v850/v850.md (sign23byte_load): Remove.
14242 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
14243 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
14245 2018-06-29 Martin Liska <mliska@suse.cz>
14248 * coverage.c (coverage_init): Mangle full path name.
14249 * doc/invoke.texi: Document the change.
14250 * gcov-io.c (mangle_path): New.
14251 * gcov-io.h (mangle_path): Likewise.
14252 * gcov.c (mangle_name): Use mangle_path for path mangling.
14254 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14256 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
14257 if starting source register is not even.
14259 2018-06-29 Martin Liska <mliska@suse.cz>
14261 PR tree-optimization/86263
14262 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
14263 Make edge redirection.
14265 2018-06-29 David Malcolm <dmalcolm@redhat.com>
14267 * dumpfile.c (dump_loc): Add indentation based on scope depth.
14268 (dump_scope_depth): New variable.
14269 (get_dump_scope_depth): New function.
14270 (dump_begin_scope): New function.
14271 (dump_end_scope): New function.
14272 * dumpfile.h (get_dump_scope_depth): New declaration.
14273 (dump_begin_scope): New declaration.
14274 (dump_end_scope): New declaration.
14275 (class auto_dump_scope): New class.
14276 (AUTO_DUMP_SCOPE): New macro.
14277 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
14280 2018-06-29 Richard Biener <rguenther@suse.de>
14282 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
14283 compute_all_dependences succeeds.
14284 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
14285 exceed --param loop-max-datarefs-for-datadeps.
14287 2018-06-29 Jakub Jelinek <jakub@redhat.com>
14289 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
14291 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
14294 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
14295 alternative 0 in preferred_for_speed attribute.
14297 2018-06-28 Paul Koning <ni1d@arrl.net>
14299 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
14300 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
14301 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
14302 * config/pdp11/pdp11.md: Correct "length" attribute calculation
14303 for shift insn patterns.
14305 2018-06-28 David Malcolm <dmalcolm@redhat.com>
14307 * cgraph.c (cgraph_node::get_body): Replace assignments to
14308 "dump_file" with calls to set_dump_file.
14309 * dumpfile.c (alt_dump_file): Make static, and group with...
14310 (alt_flags): ...this definition.
14311 (dumps_are_enabled): New variable.
14312 (refresh_dumps_are_enabled): New function.
14313 (set_dump_file): New function.
14314 (set_alt_dump_file): New function.
14315 (gcc::dump_manager::dump_start): Replace assignments to
14316 "dump_file" and "alt_dump_file" with calls to set_dump_file and
14318 (gcc::dump_manager::dump_finish): Likewise.
14319 * dumpfile.h (alt_dump_file): Delete decl.
14320 (dumps_are_enabled): New variable decl.
14321 (set_dump_file): New function decl.
14322 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
14324 * tree-nested.c (lower_nested_functions): Replace assignments to
14325 "dump_file" with calls to set_dump_file.
14327 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
14329 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
14330 goto_locus of each outgoing edge of each basic block.
14332 2018-06-28 Richard Biener <rguenther@suse.de>
14334 * dwarf2out.c (decl_scope_table): Remove.
14335 (push_decl_scope): Likewise.
14336 (pop_decl_scope): Likewise.
14337 (gen_type_die_for_member): Do not call push/pop_decl_scope.
14338 (gen_struct_or_union_type_die): Likewise.
14339 (gen_tagged_type_die): Likewise.
14340 (dwarf2out_init): Do not initialize decl_scope_table.
14341 (dwarf2out_c_finalize): Do not free it.
14343 2018-06-28 Richard Biener <rguenther@suse.de>
14345 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
14346 deciding whether to not re-use a DIE.
14348 2018-06-28 Richard Biener <rguenther@suse.de>
14350 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
14351 DW_AT_abstract_origin attribute.
14353 2018-06-28 Martin Liska <mliska@suse.cz>
14355 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
14356 Use newly introduced constants.
14357 * tree-switch-conversion.h (struct jump_table_cluster):
14358 Define max_ratio_for_size and max_ratio_for_speed.
14360 2018-06-28 Martin Liska <mliska@suse.cz>
14362 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14363 Add new checking assert to catch invalid state.
14364 (jump_table_cluster::can_be_handled): Handle single case
14366 (jump_table_cluster::is_beneficial): Bail out for such case.
14367 (bit_test_cluster::find_bit_tests):
14368 Add new checking assert to catch invalid state.
14369 (bit_test_cluster::can_be_handled): Handle single case
14371 (bit_test_cluster::is_beneficial): Bail out for such case.
14372 (switch_decision_tree::analyze_switch_statement):
14375 2018-06-28 Martin Liska <mliska@suse.cz>
14377 * common.opt: Introduce -completion option.
14378 * gcc.c (driver_handle_option): Handle it.
14379 (driver::main): Print completions if completion
14381 * opt-suggestions.c (option_proposer::get_completions):
14383 (option_proposer::suggest_completion): Likewise.
14384 (option_proposer::find_param_completions): Likewise.
14385 (verify_autocompletions): Likewise.
14386 (test_completion_valid_options): Likewise.
14387 (test_completion_valid_params): Likewise.
14388 (in_completion_p): Likewise.
14389 (empty_completion_p): Likewise.
14390 (test_completion_partial_match): Likewise.
14391 (test_completion_garbage): Likewise.
14392 (opt_proposer_c_tests): Likewise.
14393 * opt-suggestions.h: Declare new functions.
14394 * opts.c (common_handle_option): Handle OPT__completion_.
14395 * selftest-run-tests.c (selftest::run_tests): Add
14396 opt_proposer_c_tests.
14397 * selftest.c (assert_str_startswith): New.
14398 * selftest.h (assert_str_startswith): Likewise.
14399 (opt_proposer_c_tests): New.
14400 (ASSERT_STR_STARTSWITH): Likewise.
14402 2018-06-28 Martin Liska <mliska@suse.cz>
14404 * Makefile.in: Add opt-suggestions.o.
14405 * gcc-main.c: Include opt-suggestions.h.
14406 * gcc.c (driver::driver): Likewise.
14407 (driver::~driver): Remove m_option_suggestions.
14408 (driver::build_option_suggestions): Moved to option_proposer.
14409 (driver::suggest_option): Likewise.
14410 (driver::handle_unrecognized_options): Use option_proposer.
14411 * gcc.h (class driver): Add new memver m_option_proposer.
14412 * opt-suggestions.c: New file.
14413 * opt-suggestions.h: New file.
14415 2018-06-28 Martin Liska <mliska@suse.cz>
14417 * vec.h (class auto_string_vec): New (moved from auto_argvec).
14418 (auto_string_vec::~auto_string_vec): Likewise.
14420 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
14422 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
14423 prevent_decl_creation_for_types fields up and add reset_location field.
14424 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
14425 statement if id->reset_location is true.
14426 (copy_edges_for_bb): Do not set goto_locus on the new edges if
14427 id->reset_location is true.
14428 (copy_phis_for_bb): Force input_location on the arguments if
14429 id->reset_location is true.
14430 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
14431 is set on the function to be inlined.
14433 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
14435 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
14437 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
14439 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
14440 registers for Pmode.
14441 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
14442 hard registers for the clobbered pseudo.
14444 2018-06-27 Paul Koning <ni1d@arrl.net>
14446 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
14447 mutually exclusive options.
14448 * config/pdp11/constraints.md (h): New constraint.
14449 (O): Update definition to match shift code generation.
14450 (D): New constraint.
14451 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
14453 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
14455 (output_jump): Change arguments.
14456 (pdp11_fixed_cc_regs): New function.
14457 (pdp11_cc_mode): Ditto.
14458 (pdp11_expand_shift): Ditto.
14459 (pdp11_assemble_shift): Ditto.
14460 (pdp11_small_shift): Ditto.
14461 (pdp11_branch_cost): Remove.
14462 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
14464 (pdp11_register_move_cost): Update for CC registers.
14465 (pdp11_rtx_costs): Add case for LSHIFTRT.
14466 (pdp11_output_jump): Add CCNZ mode conditional branches.
14467 (notice_update_cc_on_set): Remove.
14468 (pdp11_cc_mode): New function.
14469 (simple_memory_operand): Correct pre/post decrement case.
14470 (no_side_effect_operand): New function.
14471 (pdp11_regno_reg_class): Add CC_REGS class.
14472 (pdp11_fixed_cc_regs): New function.
14473 (pdp11_small_shift): New function.
14474 (pdp11_expand_shift): New function to expand shift insns.
14475 (pdp11_assemble_shift): New function to output shifts.
14476 (pdp11_branch_cost): Remove.
14477 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
14478 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
14479 (WCHAR_TYPE): Ditto.
14480 (PTRDIFF_TYPE): Ditto.
14481 (ADJUST_INSN_LENGTH): New macro.
14482 (FIXED_REGISTERS): Add CC registers.
14483 (CALL_USED_REGISTERS): Ditto.
14484 (reg_class): Ditto.
14485 (REG_CLASS_NAMES): Ditto.
14486 (REG_CLASS_CONTENTS): Ditto.
14487 (SELECT_CC_MODE): Use new function.
14488 (TARGET_FLAGS_REGNUM): New macro.
14489 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
14490 (cc0_reg_rtx): Remove.
14491 (CC_STATUS_MDEP): Remove.
14492 (CC_STATUS_MDEFP_INIT): Remove.
14493 (CC_IN_FPU): Remove.
14494 (NOTICE_UPDATE_CC): Remove.
14495 (REGISTER_NAMES): Add CC registers.
14496 (BRANCH_COST): Change to constant 1.
14497 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
14499 * config/pdp11/pdp11.opt (mbcopy): Remove.
14500 (mbcopy-builtin): Remove.
14501 (mbranch-cheap): Remove.
14502 (mbranch-expensive): Remove.
14503 * config/pdp11/predicates.md (expand_shift_operand): Update to
14504 match shift code generation.
14505 (ccnz_operator): New predicate.
14506 * doc/invoke.texi (PDP-11 Options): Remove deleted options
14507 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
14508 Remove non-existent option -mabshi, -mno-abshi. Document mutually
14510 * doc/md.texi (PDP-11): Document new D and h constraints. Update
14511 description of O constraint.
14513 2018-06-27 Jeff Law <law@redhat.com>
14514 Austin Law <austinklaw@gmail.com>
14516 * config/v850/v850.md (addsi3_set_flags): New pattern.
14517 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
14518 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
14519 (zero_extendhisi2_v850_set_flags): Likewise.
14520 (zero_extendqisi2_v850_set_flags): Likewise.
14521 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
14522 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
14523 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
14525 * config/v850/v850-protos.h (notice_update_cc): Remove.
14526 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
14527 (v850_print_operand): Handle 'D' and "d".
14528 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
14529 Add handling of arithmetic/logical operations compared against zero.
14530 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
14531 Do not look at v850_compare_op, instead get mode from last argument.
14532 (v850_gen_compare): Remove
14533 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
14534 after reload for prologue insns.
14535 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
14537 (construct_save_jarl): Likewise.
14538 (TARGET_FLAGS_REGNUM): Define.
14539 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
14540 (NOTICE_UPDATE_CC): Remove.
14541 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
14542 than cc0. Conditionalize on reload_completed.
14543 (cmpsi_insn, setfcc_insn): Likewise.
14544 (tst1 splitter): Turn into define_and_split which sets the flags
14546 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
14547 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
14548 (cstoresf4, cstoredf4): Clobber the flags.
14549 (cmpsi, cmpsf, cmpdf): Remove expanders.
14550 (setf_insn): Remove pattern.
14551 (addsi3): Turn into define_and_split which clobbers the flags after
14552 reload and a suitable pattern (addsi3_clobber_flags) for use after
14554 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
14555 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
14556 (ashrsi3, ashrsi3_v850e2): Likewise.
14557 (bins): Clobber the flags.
14558 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
14559 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
14560 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
14561 (call_value_internal_short, call_value_internal_long): Likewise.
14562 (callt_save_interrupt, callt_return_interrupt): Likewise.
14563 (save_interrupt, return_interrupt): Likewise.
14564 (callt_save_all_interrupt, save_all_interrupt): Likewise.
14565 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
14566 (restore_all_interrupt, _restore_all_interrupt): Likewise.
14567 (All FP comparisons): Only allow after reload has completed.
14569 (divh, divhu): Tweak output template.
14570 (branch_z_normal, branch_z_invert): Remove
14571 (branch_nz_normal, branch_nz_invert): Likewise.
14572 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
14574 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
14575 * config/v850/v850.c (notice_update_cc): Remove.
14576 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
14577 (CC_NO_CARRY): Likewise.
14578 (NOTICE_UPDATE_CC): Define to nothing.
14579 * config/v850/v850.md: Remove block comment on cc0 handling
14580 Remove "cc" attribute from all patterns. Remove cc_status handling
14581 from all patterns. Minor formatting fixes.
14583 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14585 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
14586 (cortex-a76.cortex-a55): Likewise.
14587 * config/aarch64/aarch64-tune.md: Regenerate.
14588 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
14589 cortex-a76.cortex-a55.
14591 2018-06-27 Jeff Law <law@redhat.com>
14593 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
14594 (MULTILIB_DIRNAMES): Similarly.
14596 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
14598 * gimple.h (gimple_return_retbnd): Delete.
14599 (gimple_return_set_retbnd): Likewise.
14600 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
14601 gimple_return_set_retbnd.
14602 * gimple-pretty-print.c (dump_gimple_return): Remove call to
14603 gimple_return_retbnd and adjust.
14604 * tree-inline.h (struct copy_body_data): Remove retbnd field.
14605 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
14606 Explicitly return NULL in a couple more cases. Move assertion
14607 on debug statements and remove unreachable code.
14608 (reset_debug_binding): Do not test id->retbnd.
14609 (expand_call_inline): Do not set it.
14611 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
14613 * configure.ac: Add --disable-gcov option.
14614 * configure: Regenerate.
14615 * Makefile.in: Honour @enable_gcov@.
14616 * doc/install.texi: Document --disable-gcov.
14618 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14620 * config/arm/arm-cpus.in (cortex-a76): New entry.
14621 (cortex-a76.cortex-a55): Likewise.
14622 * config/arm/arm-tables.opt: Regenerate.
14623 * config/arm/arm-tune.md: Likewise.
14624 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
14625 * doc/invoke.texi (ARM Options): Document cortex-a76 and
14626 cortex-a76.cortex-a55.
14628 2018-06-27 Tamar Christina <tamar.christina@arm.com>
14631 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
14633 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
14635 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
14637 (EPILOGUE_USES): Likewise.
14639 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
14641 * tree-inline.c (remap_location): New function extracted from...
14642 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
14643 (copy_phis_for_bb): ...here. Call remap_location.
14644 (copy_cfg_body): Adjust call to copy_edges_for_bb.
14646 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
14648 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
14649 unaligned vsx for 16B memset.
14651 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
14654 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
14655 ieee128_float_type_node to long_double_type_node unless
14656 TARGET_LONG_DOUBLE_128 is set.
14658 2018-06-26 David Malcolm <dmalcolm@redhat.com>
14660 * cfgloop.c (get_loop_location): Convert return type from
14661 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
14662 by implicit construction from rtx_insn *, and using
14663 dump_user_location_t::from_function_decl for the fallback case.
14664 * cfgloop.h (get_loop_location): Convert return type from
14665 location_t to dump_user_location_t.
14666 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
14667 dump_printf_loc to pass in a dump_location_t rather than a
14668 location_t, via the gimple stmt.
14669 * coverage.c (get_coverage_counts): Update calls to
14670 dump_printf_loc to pass in dump_location_t rather than a
14672 * doc/optinfo.texi (Dump types): Convert example of
14673 dump_printf_loc from taking "locus" to taking "insn". Update
14674 description of the "_loc" calls to cover dump_location_t.
14675 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
14677 (dump_user_location_t::dump_user_location_t): New constructors,
14678 from gimple *stmt and rtx_insn *.
14679 (dump_user_location_t::from_function_decl): New function.
14680 (dump_loc): Make static.
14681 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
14682 const dump_location_t &.
14683 (dump_generic_expr_loc): Delete.
14684 (dump_printf_loc): Convert param "loc" from location_t to
14685 const dump_location_t &.
14686 (selftest::test_impl_location): New function.
14687 (selftest::dumpfile_c_tests): New function.
14688 * dumpfile.h: Include "profile-count.h".
14689 (class dump_user_location_t): New class.
14690 (struct dump_impl_location_t): New struct.
14691 (class dump_location_t): New class.
14692 (dump_printf_loc): Convert 2nd param from source_location to
14693 const dump_location_t &.
14694 (dump_generic_expr_loc): Delete.
14695 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
14696 const dump_location_t &.
14697 * gimple-fold.c (fold_gimple_assign): Update call to
14698 dump_printf_loc to pass in a dump_location_t rather than a
14699 location_t, via the gimple stmt.
14700 (gimple_fold_call): Likewise.
14701 * gimple-loop-interchange.cc
14702 (loop_cand::analyze_iloop_reduction_var): Update for change to
14703 check_reduction_path.
14704 (tree_loop_interchange::interchange): Update for change to
14705 find_loop_location.
14706 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
14707 change in return-type of find_loop_location.
14708 (graphite_regenerate_ast_isl): Likewise.
14709 * graphite-optimize-isl.c (optimize_isl): Likewise.
14710 * graphite.c (graphite_transform_loops): Likewise.
14711 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
14712 pass in a dump_location_t rather than a location_t, via the
14714 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14715 * ipa.c (walk_polymorphic_call_targets): Likewise.
14716 * loop-unroll.c (report_unroll): Convert "locus" param from
14717 location_t to dump_location_t.
14718 (decide_unrolling): Update for change to get_loop_location's
14720 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
14721 location_t to dump_user_location_t.
14722 (grid_find_single_omp_among_assignments_1): Updates calls to
14723 dump_printf_loc to pass in a dump_location_t rather than a
14724 location_t, via the gimple stmt.
14725 (grid_parallel_clauses_gridifiable): Convert "tloc" from
14726 location_t to dump_location_t. Updates calls to dump_printf_loc
14727 to pass in a dump_location_t rather than a location_t, via the
14729 (grid_inner_loop_gridifiable_p): Likewise.
14730 (grid_dist_follows_simple_pattern): Likewise.
14731 (grid_gfor_follows_tiling_pattern): Likewise.
14732 (grid_target_follows_gridifiable_pattern): Likewise.
14733 (grid_attempt_target_gridification): Convert initialization
14734 of local "grid" from memset to zero-initialization; FIXME: does
14735 this require C++11? Update call to dump_printf_loc to pass in a
14736 optinfo_location rather than a location_t, via the gimple stmt.
14737 * profile.c (read_profile_edge_counts): Updates call to
14738 dump_printf_loc to pass in a dump_location_t rather than a
14740 (compute_branch_probabilities): Likewise.
14741 * selftest-run-tests.c (selftest::run_tests): Call
14743 * selftest.h (dumpfile_c_tests): New decl.
14744 * tree-loop-distribution.c (pass_loop_distribution::execute):
14745 Update for change in return type of find_loop_location.
14746 * tree-parloops.c (parallelize_loops): Likewise.
14747 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
14748 "locus" from location_t to dump_user_location_t.
14749 (canonicalize_loop_induction_variables): Likewise.
14750 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
14751 for change in return type of find_loop_location.
14752 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
14753 to dump_printf_loc to pass in a dump_location_t rather than a
14754 location_t, via the stmt.
14755 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
14757 * tree-vect-loop-manip.c (find_loop_location): Convert return
14758 type from source_location to dump_user_location_t.
14759 (vect_do_peeling): Update for above change.
14760 (vect_loop_versioning): Update for change in type of
14762 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
14763 from location_t to dump_user_location_t.
14764 (vect_estimate_min_profitable_iters): Update for change in type
14766 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
14767 location_t to dump_location_t.
14768 (vect_slp_bb): Update for change in type of vect_location.
14769 * tree-vectorizer.c (vect_location): Convert from source_location
14770 to dump_user_location_t.
14771 (try_vectorize_loop_1): Update for change in vect_location's type.
14772 (vectorize_loops): Likewise.
14773 (increase_alignment): Likewise.
14774 * tree-vectorizer.h (vect_location): Convert from source_location
14775 to dump_user_location_t.
14776 (find_loop_location): Convert return type from source_location to
14777 dump_user_location_t.
14778 (check_reduction_path): Convert 1st param from location_t to
14779 dump_user_location_t.
14780 * value-prof.c (check_counter): Update call to dump_printf_loc to
14781 pass in a dump_user_location_t rather than a location_t; update
14782 call to error_at for change in type of "locus".
14783 (check_ic_target): Update call to dump_printf_loc to
14784 pass in a dump_user_location_t rather than a location_t, via the
14787 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
14789 * config/s390/s390.h (enum processor_flags): Do not use
14790 default tune parameter when -march was specified.
14792 2018-06-26 Jakub Jelinek <jakub@redhat.com>
14795 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
14796 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
14799 2018-06-26 Richard Biener <rguenther@suse.de>
14801 PR tree-optimization/86287
14803 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
14804 (vect_analyze_loop): Initialize n_stmts.
14806 2018-06-26 Richard Biener <rguenther@suse.de>
14808 PR middle-end/86271
14809 * fold-const.c (fold_convertible_p): Pointer extension
14812 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
14815 * dwarf2out.c (loc_list_has_views): Adjust comments.
14816 (dw_loc_list): Split single cross-partition range with
14819 2018-06-25 Jeff Law <law@redhat.com>
14821 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
14822 on -mbig-switch by default.
14824 * config/v850/predicates.md (const_float_1_operand): Fix match_code
14826 (const_float_0_operand): Remove unused predicate.
14827 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
14828 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
14829 (recipsf2): New expander. Original pattern now called
14831 (recipdf2, recipdf2_insn): Similarly.
14832 (rsqrtsf2, rsqrtsf2_insn): Similarly
14833 (rsqrtdf2, rsqrtdf2_insn): Similarly
14835 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
14837 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
14838 Simplify logic for FreeBSD (twice).
14840 2018-06-25 Martin Sebor <msebor@redhat.com>
14842 PR tree-optimization/86204
14843 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
14844 a strnlen result if it's less than the length of the string.
14846 2018-06-25 Martin Sebor <msebor@redhat.com>
14848 PR tree-optimization/85700
14849 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
14850 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
14851 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
14853 2018-06-25 Martin Sebor <msebor@redhat.com>
14855 * doc/extend.texi (Zero-length arrays): Update and clarify.
14857 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
14859 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
14860 added IEEE/IBM long double multilib support on PowerPC little
14861 endian Linux systems.
14862 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
14863 (MULTILIB_DEFAULTS): Likewise.
14864 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14866 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
14867 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
14868 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
14870 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
14872 PR middle-end/86311
14873 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
14874 (REORDER_45): Likewise.
14876 2018-06-25 Jeff Law <law@redhat.com>
14878 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
14879 dividend to 32 bits. Adjust length.
14880 (udivmodhi4): Cleanup output template. Fix length.
14882 2018-06-25 Carl Love <cel@us.ibm.com>
14884 * config/rs6000/vsx.md: Change word selector to prefered location.
14886 2018-06-25 Richard Biener <rguenther@suse.de>
14888 PR tree-optimization/86304
14889 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
14890 epilogue-if-converted loops as well.
14892 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
14894 * lto-section-out.c (lto_begin_section): Do not print section
14895 name for noaddr and unnumbered dumps.
14897 2018-06-25 Richard Biener <rguenther@suse.de>
14899 * tree-vectorizer.h (struct vec_info_shared): New structure
14900 with parts split out from struct vec_info and loop_nest from
14901 struct _loop_vec_info.
14902 (struct vec_info): Adjust accordingly.
14903 (struct _loop_vec_info): Likewise.
14904 (LOOP_VINFO_LOOP_NEST): Adjust.
14905 (LOOP_VINFO_DATAREFS): Likewise.
14906 (LOOP_VINFO_DDRS): Likewise.
14907 (struct _bb_vec_info): Likewise.
14908 (BB_VINFO_DATAREFS): Likewise.
14909 (BB_VINFO_DDRS): Likewise.
14910 (struct _stmt_vec_info): Add dr_aux member.
14911 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
14912 (DR_MISALIGNMENT_UNINITIALIZED): New.
14913 (set_dr_misalignment): Adjust.
14914 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
14915 (vect_analyze_loop): Adjust prototype.
14916 (vect_analyze_loop_form): Likewise.
14917 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14918 Compute dependences lazily.
14919 (vect_record_base_alignments): Use shared datarefs/ddrs.
14920 (vect_verify_datarefs_alignment): Likewise.
14921 (vect_analyze_data_refs_alignment): Likewise.
14922 (vect_analyze_data_ref_accesses): Likewise.
14923 (vect_analyze_data_refs): Likewise.
14924 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
14925 constructor parameter for shared part.
14926 (vect_analyze_loop_form): Pass in shared part and adjust.
14927 (vect_analyze_loop_2): Pass in storage for the number of
14928 stmts. Move loop nest finding to the caller. Compute
14930 (vect_analyze_loop): Pass in shared part.
14931 (vect_transform_loop): Verify shared datarefs are unchanged.
14932 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
14933 constructor parameter for shared part.
14934 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
14935 (vect_slp_bb): Verify shared datarefs are unchanged before
14937 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
14939 (new_stmt_vec_info): Initialize DR_AUX misalignment to
14940 DR_MISALIGNMENT_UNINITIALIZED.
14941 * tree-vectorizer.c (vec_info::vec_info): Add constructor
14942 parameter for shared part.
14943 (vec_info::~vec_info): Adjust.
14944 (vec_info_shared::vec_info_shared): New.
14945 (vec_info_shared::~vec_info_shared): Likewise.
14946 (vec_info_shared::save_datarefs): Likewise.
14947 (vec_info_shared::check_datarefs): Likewise.
14948 (try_vectorize_loop_1): Construct shared part live for analyses
14949 of a single loop for multiple vector sizes.
14950 * tree-parloops.c (gather_scalar_reductions): Adjust.
14952 2018-06-25 Richard Biener <rguenther@suse.de>
14954 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
14955 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
14956 (vect_analyze_data_refs): Remove similar code from here and
14957 simplify accordingly.
14959 2018-06-25 Richard Biener <rguenther@suse.de>
14961 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
14962 for reverse storage order accesses rather than asserting
14963 they cannot happen here.
14965 2018-06-25 Tom de Vries <tdevries@suse.de>
14968 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
14969 Use data16 instead of .byte for insn prefix.
14971 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
14974 * parser.c (make_char_string_pack): Pass this literal chars
14975 through cpp_interpret_string.
14976 (cp_parser_userdef_numeric_literal): Check the result of
14977 make_char_string_pack.
14979 2018-06-24 Maya Rashish <coypu@sdf.org>
14981 * ginclude/stddef.h: Simplify conditions around avoiding
14982 re-definition of __size_t.
14984 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14986 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
14987 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
14989 2018-06-22 Maya Rashish <coypu@sdf.org>
14991 * doc/invoke.texi (mno-fancy-math-387): Update for changes
14992 made to OpenBSD and NetBSD through the years.
14994 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
14996 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
14997 behavior of vec_pack (vector double, vector double) to match
14998 behavior of vec_float2 (vector double, vector double).
15000 2018-06-22 Olivier Hainque <hainque@adacore.com>
15002 * gimplify.c (gimplify_function_tree): Prevent creation
15003 of a trampoline for the address of the current function
15004 passed to entry/exit instrumentation hooks.
15006 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
15009 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
15012 2018-06-22 Martin Liska <mliska@suse.cz>
15014 PR tree-optimization/86263
15015 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15016 Bail out if is_enabled is false.
15017 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
15019 (jump_table_cluster::is_enabled): New function.
15021 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15023 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
15024 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
15025 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
15026 (lto_input_ts_binfo_tree_pointers): Likewise.
15027 * tree-streamer-out.c (streamer_write_tree_bitfields,
15028 write_ts_binfo_tree_pointers): Likewise.
15029 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
15031 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
15033 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
15035 2018-06-22 Martin Liska <mliska@suse.cz>
15037 * symbol-summary.h (get): Make it pure and inline move
15038 functionality from ::get function.
15039 (get): Remove and inline into ::get and ::get_create.
15040 (get_create): Move code from ::get function.
15042 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15045 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
15046 -x assembler-with-cpp.
15048 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15050 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
15051 _FILE_OFFSET_BITS=64 for C++.
15053 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
15055 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
15056 conversion insn that shows up when pr85657-3.c is compiled using
15057 IEEE 128-bit long double.
15058 (neg<mode>2_internal): Use the correct mode to check whether the
15059 mode is IBM extended.
15060 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
15061 multiply and divide external functions from being created more
15064 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15066 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
15068 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
15069 the edge can be forwarded.
15070 (cfg_layout_merge_blocks): Likewise.
15072 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15074 * except.c (finish_eh_generation): Commit edge insertions only after
15075 the EH edges have been redirected from post-landing to landing pads.
15077 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15079 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
15080 create_tmp_var_for to create the FRAME decl.
15081 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
15083 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
15085 * tree-inline.c (copy_edges_for_bb): Minor tweak.
15086 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
15087 debug statement when resetting its value.
15088 (expand_call_inline): Copy the locus of the call onto the assignment
15089 of the return value, if any. Use local variable in more cases.
15091 2018-06-21 Martin Liska <mliska@suse.cz>
15093 * ipa-pure-const.c (propagate_nothrow): Use
15094 funct_state_summaries->get.
15095 (dump_malloc_lattice): Likewise.
15096 (propagate_malloc): Likewise.
15098 2018-06-21 Richard Biener <rguenther@suse.de>
15100 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
15101 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
15102 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
15104 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
15105 BLOCK_ABSTRACT_ORIGIN unconditionally.
15107 2018-06-21 David Malcolm <dmalcolm@redhat.com>
15109 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
15111 * ipa-reference.c (ipa_reference_c_finalize): Delete
15112 ipa_ref_opt_sum_summaries and set it to NULL.
15114 2018-06-21 Tom de Vries <tdevries@suse.de>
15116 PR tree-optimization/85859
15117 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
15118 test with comment from bb_no_side_effects_p.
15120 2018-06-21 Richard Biener <rguenther@suse.de>
15122 PR tree-optimization/86232
15123 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
15124 max for constant niter.
15126 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
15128 * config/aarch64/aarch64-simd.md
15129 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
15131 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
15133 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
15134 Make opernads of the unspec commutative.
15136 2018-06-21 Richard Biener <rguenther@suse.de>
15138 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
15139 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15140 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
15141 (vect_analyze_data_ref_dependence): Re-order checks to deal with
15143 (vect_record_base_alignments): Do not record base alignment
15144 for gathers or scatters.
15145 (vect_compute_data_ref_alignment): Drop return value that is always
15146 true. Bail out early for gathers or scatters.
15147 (vect_enhance_data_refs_alignment): Bail out early for gathers
15149 (vect_find_same_alignment_drs): Likewise.
15150 (vect_analyze_data_refs_alignment): Remove dead code.
15151 (vect_slp_analyze_and_verify_node_alignment): Likewise.
15152 (vect_analyze_data_refs): For possible gathers or scatters do
15153 not create an alternate DR, just check their possible validity
15154 and mark them. Adjust DECL_NONALIASED handling to not rely
15155 on DR_BASE_ADDRESS.
15156 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
15157 update inits of gathers or scatters.
15158 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
15159 Also copy gather/scatter flag to pattern vinfo.
15161 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
15163 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
15164 behavior of vec_packsu (vector unsigned long long, vector unsigned
15165 long long) to match behavior of vec_packs with same signature.
15167 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
15168 Thomas Schwinge <thomas@codesourcery.com>
15169 Cesar Philippidis <cesar@codesourcery.com>
15171 * gimplify.c (gimplify_scan_omp_clauses): Add support for
15172 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
15173 (gimplify_adjust_omp_clauses): Likewise.
15174 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
15175 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
15176 (gimplify_omp_target_update): Update handling of acc update and
15178 * omp-low.c (install_var_field): Remove unused parameter
15179 base_pointers_restrict.
15180 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
15181 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
15183 (omp_target_base_pointers_restrict_p): Delete.
15184 (scan_omp_target): Update call to scan_sharing_clauses.
15185 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
15187 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
15188 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
15189 (convert_local_omp_clauses): Likewise.
15190 * tree-pretty-print.c (dump_omp_clause): Likewise.
15191 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
15193 (omp_clause_code_name): Likewise.
15195 2018-06-20 Jakub Jelinek <jakub@redhat.com>
15198 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
15201 PR tree-optimization/86231
15202 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
15203 anti-range don't overwrite *vr0min before using it to compute *vr0max.
15205 2018-06-20 Tom de Vries <tdevries@suse.de>
15207 PR tree-optimization/86097
15208 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
15209 iv type if signedness of iv type is not the same as that of *nit.
15211 2018-06-20 Jakub Jelinek <jakub@redhat.com>
15213 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
15214 EDGE_EH edges, verify they are all EDGE_EH.
15216 2018-06-20 Maya Rashish <coypu@sdf.org>
15218 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
15220 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15222 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
15223 * config/aarch64/aarch64.c (xgene1_tunings): Add
15224 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
15225 (aarch64_mode_valid_for_sched_fusion_p):
15226 Allow 16-byte modes.
15227 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
15228 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
15230 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
15232 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
15233 * config/aarch64/iterators.md (VQ2): New mode iterator.
15235 2018-06-20 Martin Liska <mliska@suse.cz>
15237 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
15238 Change default ratio from 10 to 8.
15240 2018-06-20 Martin Liska <mliska@suse.cz>
15242 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15244 (bit_test_cluster::find_bit_tests): Likewise.
15245 (switch_decision_tree::analyze_switch_statement): Find clusters.
15246 * tree-switch-conversion.h (struct jump_table_cluster): Document
15249 2018-06-20 Martin Liska <mliska@suse.cz>
15251 * tree-switch-conversion.c (switch_conversion::collect):
15252 Record m_uniq property.
15253 (switch_conversion::expand): Bail out for special conditions.
15254 (group_cluster::~group_cluster): New.
15255 (group_cluster::group_cluster): Likewise.
15256 (group_cluster::dump): Likewise.
15257 (jump_table_cluster::emit): New.
15258 (switch_decision_tree::fix_phi_operands_for_edges): New.
15259 (struct case_node): Remove struct.
15260 (jump_table_cluster::can_be_handled): New.
15261 (case_values_threshold): Moved to header.
15262 (reset_out_edges_aux): Likewise.
15263 (jump_table_cluster::is_beneficial): New.
15264 (bit_test_cluster::can_be_handled): Likewise.
15265 (add_case_node): Remove.
15266 (bit_test_cluster::is_beneficial): New.
15267 (case_bit_test::cmp): New.
15268 (bit_test_cluster::emit): New.
15269 (expand_switch_as_decision_tree_p): Remove.
15270 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
15271 (fix_phi_operands_for_edge): Likewise.
15272 (switch_decision_tree::analyze_switch_statement): New.
15273 (compute_cases_per_edge): Move ...
15274 (switch_decision_tree::compute_cases_per_edge): ... here.
15275 (try_switch_expansion): Likewise.
15276 (switch_decision_tree::try_switch_expansion): Likewise.
15277 (record_phi_operand_mapping): Likewise.
15278 (switch_decision_tree::record_phi_operand_mapping): Likewise.
15279 (emit_case_decision_tree): Likewise.
15280 (switch_decision_tree::emit): Likewise.
15281 (balance_case_nodes): Likewise.
15282 (switch_decision_tree::balance_case_nodes): Likewise.
15283 (dump_case_nodes): Likewise.
15284 (switch_decision_tree::dump_case_nodes): Likewise.
15285 (emit_jump): Likewise.
15286 (switch_decision_tree::emit_jump): Likewise.
15287 (emit_cmp_and_jump_insns): Likewise.
15288 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
15289 (emit_case_nodes): Likewise.
15290 (switch_decision_tree::emit_case_nodes): Likewise.
15291 (conditional_probability): Remove.
15292 * tree-switch-conversion.h (enum cluster_type): New.
15294 (struct cluster): Likewise.
15295 (cluster::cluster): Likewise.
15296 (struct simple_cluster): Likewise.
15297 (simple_cluster::simple_cluster): Likewise.
15298 (struct group_cluster): Likewise.
15299 (struct jump_table_cluster): Likewise.
15300 (struct bit_test_cluster): Likewise.
15301 (struct min_cluster_item): Likewise.
15302 (struct case_tree_node): Likewise.
15303 (case_tree_node::case_tree_node): Likewise.
15304 (jump_table_cluster::case_values_threshold): Likewise.
15305 (struct case_bit_test): Likewise.
15306 (struct switch_decision_tree): Likewise.
15307 (struct switch_conversion): Likewise.
15308 (switch_decision_tree::reset_out_edges_aux): Likewise.
15310 2018-06-20 Martin Liska <mliska@suse.cz>
15312 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
15313 (hoist_edge_and_branch_if_true): Likewise.
15314 (expand_switch_using_bit_tests_p): Likewise.
15315 (struct case_bit_test): Likewise.
15316 (case_bit_test_cmp): Likewise.
15317 (emit_case_bit_tests): Likewise.
15318 (switch_conversion::switch_conversion): New class.
15319 (struct switch_conv_info): Remove old struct.
15320 (collect_switch_conv_info): More to ...
15321 (switch_conversion::collect): ... this.
15322 (check_range): Likewise.
15323 (switch_conversion::check_range): Likewise.
15324 (check_all_empty_except_final): Likewise.
15325 (switch_conversion::check_all_empty_except_final): Likewise.
15326 (check_final_bb): Likewise.
15327 (switch_conversion::check_final_bb): Likewise.
15328 (create_temp_arrays): Likewise.
15329 (switch_conversion::create_temp_arrays): Likewise.
15330 (free_temp_arrays): Likewise.
15331 (gather_default_values): Likewise.
15332 (switch_conversion::gather_default_values): Likewise.
15333 (build_constructors): Likewise.
15334 (switch_conversion::build_constructors): Likewise.
15335 (constructor_contains_same_values_p): Likewise.
15336 (switch_conversion::contains_same_values_p): Likewise.
15337 (array_value_type): Likewise.
15338 (switch_conversion::array_value_type): Likewise.
15339 (build_one_array): Likewise.
15340 (switch_conversion::build_one_array): Likewise.
15341 (build_arrays): Likewise.
15342 (switch_conversion::build_arrays): Likewise.
15343 (gen_def_assigns): Likewise.
15344 (switch_conversion::gen_def_assigns): Likewise.
15345 (prune_bbs): Likewise.
15346 (switch_conversion::prune_bbs): Likewise.
15347 (fix_phi_nodes): Likewise.
15348 (switch_conversion::fix_phi_nodes): Likewise.
15349 (gen_inbound_check): Likewise.
15350 (switch_conversion::gen_inbound_check): Likewise.
15351 (process_switch): Use the newly created class.
15352 (switch_conversion::expand): New.
15353 (switch_conversion::~switch_conversion): New.
15354 * tree-switch-conversion.h: New file.
15356 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15358 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
15359 tree-vect-patterns.c.
15360 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
15361 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
15362 (vect_recog_sad_pattern): Likewise.
15363 (vect_recog_widen_sum_pattern): Likewise.
15364 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
15365 (vect_recog_widen_shift_pattern): Remove the type_in argument.
15366 (vect_recog_rotate_pattern): Likewise.
15367 (vect_recog_mult_pattern): Likewise.
15368 (vect_recog_vector_vector_shift_pattern): Likewise.
15369 (vect_recog_divmod_pattern): Likewise.
15370 (vect_recog_mixed_size_cond_pattern): Likewise.
15371 (vect_recog_bool_pattern): Likewise.
15372 (vect_recog_mask_conversion_pattern): Likewise.
15373 (vect_try_gather_scatter_pattern): Likewise.
15374 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
15375 (vect_recog_over_widening_pattern): Likewise.
15376 (vect_recog_gather_scatter_pattern): Likewise.
15377 (vect_recog_func_ptr): Move from tree-vectorizer.h
15378 (vect_vect_recog_func_ptrs): Move further down the file.
15379 (vect_recog_func): Likewise. Remove the third argument.
15380 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
15381 (vect_pattern_recog_1): Expect the pattern function to do any
15382 necessary target tests. Also expect it to provide a vector type.
15383 Remove the type_in handling.
15385 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15387 * tree-vect-patterns.c (vect_pattern_detected): New function.
15388 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
15389 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
15390 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
15391 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
15392 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
15393 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
15394 (vect_recog_mask_conversion_pattern)
15395 (vect_try_gather_scatter_pattern): Likewise.
15397 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15399 * tree-vect-patterns.c (vect_get_internal_def): New function.
15400 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15401 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
15402 (search_type_for_mask_1): Use it.
15404 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15406 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
15407 redundant WIDEN_SUM_EXPR handling.
15408 (vect_recog_sad_pattern): Likewise.
15410 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15412 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
15413 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
15414 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
15415 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
15416 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
15418 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15420 * tree-vect-stmts.c (vectorizable_call): Make sure that we
15421 use the stmt_vec_info of the original bb statement for the
15422 new zero assignment, even if the call is part of a pattern.
15424 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15426 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
15427 that the sequence is attached to the original statement rather
15428 than the pattern statement.
15429 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
15430 PATTERN_DEF_SEQ from the original statement rather than
15431 the main pattern statement.
15432 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
15433 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
15434 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
15436 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
15438 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
15439 definition statements before the early exit for statements that aren't
15441 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
15443 (vect_transform_loop): ...here. Process pattern definition
15444 statements without first checking whether the main pattern
15445 statement is live or relevant.
15447 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15449 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
15450 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
15452 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
15454 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
15455 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
15456 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
15457 (expand_block_compare): Change select_block_compare_mode call.
15458 (expand_strncmp_align_check): Use new functions, fix comment.
15459 (emit_final_str_compare_gpr): New function.
15460 (expand_strn_compare): Refactor and clean up code.
15461 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
15463 2018-06-19 Tony Reix <tony.reix@atos.com>
15464 Damien Bergamini <damien.bergamini@atos.com>
15465 David Edelsohn <dje.gcc@gmail.com>
15467 * collect2.c (static_obj): New variable.
15468 (static_libs): New variable.
15469 (is_in_list): Uncomment declaration.
15470 (main): Track AIX libraries linked statically.
15471 (is_in_list): Uncomment definition.
15472 (scan_prog_file): Don't add AIX shared libraries initializer
15473 to constructor list if linking statically.
15475 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
15477 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
15479 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
15481 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
15483 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
15486 2018-06-19 Martin Liska <mliska@suse.cz>
15488 * config/i386/i386.c (ix86_can_inline_p): Do not use
15489 ipa_fn_summaries::get_create.
15490 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
15492 (devirtualization_time_bonus): Likewise.
15493 (ipcp_propagate_stage): Likewise.
15494 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
15495 (edge_set_predicate): Likewise.
15496 (evaluate_conditions_for_known_args): Likewise.
15497 (evaluate_properties_for_edge): Likewise.
15498 (ipa_call_summary::reset): Tranform to ...
15499 (ipa_call_summary::~ipa_call_summary): ... this.
15500 (ipa_fn_summary::reset): Transform to ...
15501 (ipa_fn_summary::~ipa_fn_summary): ... this.
15502 (ipa_fn_summary_t::remove): Rename to ...
15503 (ipa_fn_summary_t::remove_callees): ... this.
15504 (ipa_fn_summary_t::duplicate): Use placement new
15505 instead of memory copy.
15506 (ipa_call_summary_t::duplicate): Likewise.
15507 (ipa_call_summary_t::remove): Remove.
15508 (dump_ipa_call_summary): Change get_create to get.
15509 (ipa_dump_fn_summary): Dump only when summary exists.
15510 (analyze_function_body): Use symbol_summary::get instead
15512 (compute_fn_summary): Likewise.
15513 (estimate_edge_devirt_benefit): Likewise.
15514 (estimate_edge_size_and_time): Likewise.
15515 (inline_update_callee_summaries): Likewise.
15516 (remap_edge_change_prob): Likewise.
15517 (remap_edge_summaries): Likewise.
15518 (ipa_merge_fn_summary_after_inlining): Likewise.
15519 (write_ipa_call_summary): Likewise.
15520 (ipa_fn_summary_write): Likewise.
15521 (ipa_free_fn_summary): Likewise.
15522 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
15523 (struct ipa_call_summary): Likewise.
15524 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
15526 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
15527 (estimate_size_after_inlining): Likewise.
15528 (estimate_growth): Likewise.
15529 (growth_likely_positive): Likewise.
15530 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
15531 (inline_call): Likewise.
15532 * ipa-inline.c (caller_growth_limits): Likewise.
15533 (can_inline_edge_p): Likewise.
15534 (can_inline_edge_by_limits_p): Likewise.
15535 (compute_uninlined_call_time): Likewise.
15536 (compute_inlined_call_time): Likewise.
15537 (want_inline_small_function_p): Likewise.
15538 (edge_badness): Likewise.
15539 (update_caller_keys): Likewise.
15540 (update_callee_keys): Likewise.
15541 (inline_small_functions): Likewise.
15542 (inline_to_all_callers_1): Likewise.
15543 (dump_overall_stats): Likewise.
15544 (early_inline_small_functions): Likewise.
15545 (early_inliner): Likewise.
15546 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15547 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15548 * ipa-pure-const.c (malloc_candidate_p): Likewise.
15549 * ipa-split.c (execute_split_functions): Likewise.
15550 * symbol-summary.h: Likewise.
15551 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
15553 2018-06-19 Richard Biener <rguenther@suse.de>
15555 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
15556 (vectorize_loops): ... here. Fix dbgcnt handling.
15557 (try_vectorize_loop): Wrap try_vectorize_loop_1.
15559 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
15562 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
15563 ieee128 argument takes up only one (vector) register, not two (floating
15566 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
15568 * gimplify.c (gimplify_init_constructor): Really never clear for an
15569 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
15571 2018-06-19 Richard Biener <rguenther@suse.de>
15573 PR tree-optimization/86179
15574 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
15575 after failed recognition.
15577 2018-06-18 Martin Sebor <msebor@redhat.com>
15579 PR middle-end/85602
15580 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
15581 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
15582 Handle integer subtraction.
15583 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
15584 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
15586 2018-06-18 David Malcolm <dmalcolm@redhat.com>
15588 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
15589 param from rtx to rtx_insn *.
15590 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
15592 (frv_ifcvt_modify_insn): Likwise.
15593 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
15594 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
15595 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
15596 as_a <rtx_insn *> cast to local "unprotected_region" once
15597 it's been established that it's not NULL or pc_rtx.
15598 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
15599 param "sethi" from rtx to rtx_insn *.
15600 (nds32_group_float_insns): Likewise for param "insn".
15601 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
15603 (vax_output_int_subtract): Likewise.
15604 * config/vax/vax.c (vax_output_int_add): Likewise for param
15606 (vax_output_int_subtract): Likewise.
15607 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
15608 (emit_pattern_after): Likewise for param "after".
15609 (emit_insn_after): Likewise.
15610 (emit_jump_insn_after): Likewise.
15611 (emit_call_insn_after): Likewise.
15612 (emit_debug_insn_after): Likewise.
15613 (emit_pattern_before): Likewise for param "before".
15614 (emit_insn_before): Likewise.
15615 (emit_jump_insn_before): Likewise.
15616 * final.c (get_insn_template): Likewise for param "insn", removing
15618 * output.h (get_insn_template): Likewise for 2nd param.
15619 * rtl.h (emit_insn_before): Likewise.
15620 (emit_jump_insn_before): Likewise.
15621 (emit_debug_insn_before_noloc): Likewise.
15622 (emit_insn_after): Likewise.
15623 (emit_jump_insn_after): Likewise.
15624 (emit_call_insn_after): Likewise.
15625 (emit_debug_insn_after): Likewise.
15626 (set_insn_deleted): Likewise for param.
15628 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
15631 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
15632 floating point modes, so that IFmode is numerically greater than
15633 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
15634 to declare the ordering. This prevents IFmode from being
15635 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
15636 machine. Include rs6000-modes.h to share the fractional values
15637 between genmodes* and the rest of the compiler.
15638 (IFmode): Likewise.
15639 (KFmode): Likewise.
15640 (TFmode): Likewise.
15641 * config/rs6000/rs6000-modes.h: New file.
15642 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
15643 meaning of rs6000_long_double_size so that 126..128 selects an
15644 appropriate 128-bit floating point type.
15645 (rs6000_option_override_internal): Likewise.
15646 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
15647 (TARGET_LONG_DOUBLE_128): Change the meaning of
15648 rs6000_long_double_size so that 126..128 selects an appropriate
15649 128-bit floating point type.
15650 (LONG_DOUBLE_TYPE_SIZE): Update comment.
15651 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
15652 source and destination to match the standard usage.
15653 (truncifkf2): Likewise.
15654 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
15655 ISA 2.07 to use an explicit clobber, instead of passing in a
15657 (copysign<mode>3_soft): Likewise.
15659 2018-06-18 David Malcolm <dmalcolm@redhat.com>
15661 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
15662 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
15663 (vect_slp_analyze_instance_dependence): Likewise.
15664 (vect_enhance_data_refs_alignment): Likewise.
15665 (vect_analyze_data_refs_alignment): Likewise.
15666 (vect_slp_analyze_and_verify_instance_alignment
15667 (vect_analyze_data_ref_accesses): Likewise.
15668 (vect_prune_runtime_alias_test_list): Likewise.
15669 (vect_analyze_data_refs): Likewise.
15670 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
15671 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
15672 (vect_analyze_scalar_cycles_1): Likewise.
15673 (vect_get_loop_niters): Likewise.
15674 (vect_analyze_loop_form_1): Likewise.
15675 (vect_update_vf_for_slp): Likewise.
15676 (vect_analyze_loop_operations): Likewise.
15677 (vect_analyze_loop): Likewise.
15678 (vectorizable_induction): Likewise.
15679 (vect_transform_loop): Likewise.
15680 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
15681 * tree-vect-slp.c (vect_analyze_slp): Likewise.
15682 (vect_make_slp_decision): Likewise.
15683 (vect_detect_hybrid_slp): Likewise.
15684 (vect_slp_analyze_operations): Likewise.
15685 (vect_slp_bb): Likewise.
15686 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
15687 (vectorizable_bswap): Likewise.
15688 (vectorizable_call): Likewise.
15689 (vectorizable_simd_clone_call): Likewise.
15690 (vectorizable_conversion): Likewise.
15691 (vectorizable_assignment): Likewise.
15692 (vectorizable_shift): Likewise.
15693 (vectorizable_operation): Likewise.
15694 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
15696 2018-06-18 Martin Sebor <msebor@redhat.com>
15698 PR tree-optimization/81384
15699 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
15700 * builtins.c (expand_builtin_strnlen): New function.
15701 (expand_builtin): Call it.
15702 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
15703 * builtins.def (BUILT_IN_STRNLEN): New.
15704 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
15705 Warn for bounds in excess of maximum object size.
15706 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
15707 single-value ranges. Handle strnlen.
15708 (handle_builtin_strlen): Handle strnlen.
15709 (strlen_check_and_optimize_stmt): Same.
15710 * doc/extend.texi (Other Builtins): Document strnlen.
15712 2018-06-18 Maya Rashish <coypu@sdf.org>
15714 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
15715 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
15716 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
15718 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
15720 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
15722 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15724 * tree.c (escaped_string::escape): Replace cast to char * by
15725 const_cast<char *> (unescaped).
15727 2018-06-18 Nick Clifton <nickc@redhat.com>
15730 * tree.c (escaped_string): New class. Converts an unescaped
15731 string into its escaped equivalent.
15732 (warn_deprecated_use): Use the new class to convert the
15733 deprecation message, if present.
15734 (test_escaped_strings): New self test.
15735 (test_c_tests): Add test_escaped_strings.
15736 * doc/extend.texi (deprecated): Add a note that the
15737 deprecation message is affected by the -fmessage-length
15738 option, and that control characters will be escaped.
15739 (#pragma GCC error): Document this pragma.
15740 (#pragma GCC warning): Likewise.
15741 * doc/invoke.texi (-fmessage-length): Document this option's
15742 effect on the #warning and #error preprocessor directives and
15743 the deprecated attribute.
15745 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
15747 * tree.c (decl_value_expr_lookup): Revert latest change.
15748 (decl_value_expr_insert): Likewise.
15750 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
15752 * gimplify.c (nonlocal_vlas): Delete.
15753 (nonlocal_vla_vars): Likewise.
15754 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
15756 (gimplify_body): Do not create and destroy nonlocal_vlas.
15757 * tree-nested.c: Include diagnostic.h.
15758 (use_pointer_in_frame): Tweak.
15759 (lookup_field_for_decl): Add assertion and declare the transformation.
15760 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
15761 internal error when the reference is in a wrong context. Do not
15762 create a debug decl by default.
15763 (note_nonlocal_block_vlas): Delete.
15764 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
15765 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
15766 create a debug decl by default.
15767 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
15768 call is in a wrong context.
15769 (fixup_vla_decls): New function.
15770 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
15771 debug variables were created.
15772 * tree.c (decl_value_expr_lookup): Add checking assertion.
15773 (decl_value_expr_insert): Likewise.
15775 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
15777 PR middle-end/82479
15778 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
15779 * tree-scalar-evolution.c (interpret_expr): Likewise.
15780 (expression_expensive_p): Likewise.
15781 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
15782 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
15783 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
15784 (ssa_defined_by_minus_one_stmt_p): New.
15786 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
15788 PR middle-end/64946
15789 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
15790 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
15791 * dojump.c (do_jump): Likewise.
15792 * expr.c (expand_expr_real_2): Check operand type's sign.
15793 * fold-const.c (const_unop): Handle ABSU_EXPR.
15794 (fold_abs_const): Likewise.
15795 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
15796 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
15797 (strip_sign_op_1): Likesise.
15798 * match.pd: Add new pattern to generate ABSU_EXPR.
15799 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
15800 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15801 * tree-eh.c (operation_could_trap_helper_p): Likewise.
15802 * tree-inline.c (estimate_operator_cost): Likewise.
15803 * tree-pretty-print.c (dump_generic_node): Likewise.
15804 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
15805 * tree.def (ABSU_EXPR): New.
15807 2018-06-16 Jakub Jelinek <jakub@redhat.com>
15809 PR middle-end/86095
15810 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
15811 documented as preserved for backward compatibility only.
15812 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
15814 PR rtl-optimization/86108
15815 * bb-reorder.c (create_forwarder_block): Renamed to ...
15816 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
15817 jump from new landing pad to the second part.
15818 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
15821 2018-06-15 Jakub Jelinek <jakub@redhat.com>
15823 PR middle-end/85878
15824 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
15825 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
15826 Only call store_expr for halves if the mode is the same.
15828 PR middle-end/86123
15829 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
15830 Fix up comment formatting.
15832 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15834 * typed-splay-tree.h (typed_splay_tree::remove): New function.
15835 (typed_splay_tree::closure,
15836 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
15837 (typed_splay_tree::typed_splay_tree,
15838 typed_splay_tree::operator =): Declared private.
15839 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
15840 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
15841 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
15842 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
15843 typed_splay_tree::splay_tree_splay,
15844 typed_splay_tree::splay_tree_foreach_helper,
15845 typed_splay_tree::splay_tree_insert,
15846 typed_splay_tree::splay_tree_remove,
15847 typed_splay_tree::splay_tree_lookup,
15848 typed_splay_tree::splay_tree_predecessor,
15849 typed_splay_tree::splay_tree_successor,
15850 typed_splay_tree::splay_tree_min,
15851 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
15852 (typed_splay_tree::root, typed_splay_tree::comp,
15853 typed_splay_tree::delete_key,
15854 typed_splay_tree::delete_value): New data members.
15855 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
15856 typed_splay_tree::remove.
15858 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
15860 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
15861 -mginv and -mno-ginv to the assembler.
15862 * config/mips/mips.opt (-mcrc): New option.
15863 (-mginv): Likewise.
15864 * doc/invoke.text (-mcrc): Document.
15865 (-mginv): Likewise.
15867 2018-06-15 Nick Clifton <nickc@redhat.com>
15870 * tree.c (escaped_string): New class. Converts an unescaped
15871 string into its escaped equivalent.
15872 (warn_deprecated_use): Use the new class to convert the
15873 deprecation message, if present.
15874 (test_escaped_strings): New self test.
15875 (test_c_tests): Add test_escaped_strings.
15876 * doc/extend.texi (deprecated): Add a note that the
15877 deprecation message is affected by the -fmessage-length
15878 option, and that control characters will be escaped.
15879 (#pragma GCC error): Document this pragma.
15880 (#pragma GCC warning): Likewise.
15881 * doc/invoke.texi (-fmessage-length): Document this option's
15882 effect on the #warning and #error preprocessor directives and
15883 the deprecated attribute.
15885 2018-06-15 Richard Biener <rguenther@suse.de>
15887 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
15888 here, also noting vector size used.
15889 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
15890 size used in MSG_OPTIMIZED_LOCATIONS dump.
15891 (pass_slp_vectorize::execute): Adjust.
15893 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
15896 * config/arc/arc.c (arc_return_address_register): Fix
15899 2018-06-15 Richard Biener <rguenther@suse.de>
15901 PR middle-end/86159
15902 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
15903 leave useless conversion stripping to force_gimple_operand_gsi.
15904 (gimplify_build2): Likewise.
15905 (gimplify_build1): Likewise.
15907 2018-06-15 Richard Biener <rguenther@suse.de>
15909 PR middle-end/86076
15910 * tree-cfg.c (move_stmt_op): unshare invariant addresses
15911 before adjusting their block.
15913 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
15915 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
15916 multilibs for *-*-rtems*.
15917 * config/riscv/t-rtems: New file.
15919 2018-06-14 Jakub Jelinek <jakub@redhat.com>
15921 PR middle-end/86122
15922 * match.pd ((A +- CST1) +- CST2): Punt if last resort
15923 unsigned_type_for returns NULL.
15926 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
15927 subregs of multi-word pseudos unless the float mode has word size.
15929 2018-06-14 Richard Biener <rguenther@suse.de>
15931 PR middle-end/86139
15932 * tree-vect-generic.c (build_word_mode_vector_type): Remove
15933 duplicate and harmful type_hash_canon.
15934 * tree.c (type_hash_canon): Assert we didn't find ourselves.
15936 2018-06-14 Richard Biener <rguenther@suse.de>
15939 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
15942 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
15944 * config/rtems.h (STDINT_LONG32): Define.
15946 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
15947 Prachi Godbole <prachi.godbole@imgtec.com>
15949 * config/mips/mips-cpus.def: Define P6600.
15950 * config/mips/mips-tables.opt: Regenerate.
15951 * config/mips/mips.c (mips_ucbranch_type): New enum.
15952 (mips_rtx_cost_data): Add support for P6600.
15953 (mips_issue_rate): Likewise.
15954 (mips_multipass_dfa_lookahead): Likewise.
15955 (mips_avoid_hazard): Likewise.
15956 (mips_reorg_process_insns): Likewise.
15957 (mips_classify_branch_p6600): New function.
15958 * config/mips/mips.h (TUNE_P6600): New define.
15959 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
15960 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
15961 * config/mips/mips.md: Include p6600.md.
15962 (processor): Add p6600.
15963 * config/mips/p6600.md: New file.
15964 * doc/invoke.texi: Add p6600 to supported architectures.
15966 2018-06-13 Martin Sebor <msebor@redhat.com>
15968 PR tree-optimization/86114
15969 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
15971 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
15973 2018-06-13 Richard Biener <rguenther@suse.de>
15975 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
15976 Properly set vector type of the intermediate stmt.
15977 * tree-vect-stmts.c (vectorizable_operation): The destination
15978 var always has vectype_out type.
15980 2018-06-13 Jeff Law <law@redhat.com>
15982 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
15983 integer 0 for argument to print_rtl_with_bb.
15984 (rl78_reorg): Likewise.
15986 2018-06-13 David Malcolm <dmalcolm@redhat.com>
15988 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
15989 from rtx to rtx_insn *.
15990 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
15992 (add_sched_insns_for_speculation): Likewise for local "target",
15993 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
15994 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
15995 from rtx_insn ** to rtx_code_label **.
15996 (reorg_emit_nops): Likewise.
15997 (c6x_reorg): Likewise for local "call_labels".
15998 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
16000 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
16001 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
16002 the loops over LABEL_REFS.
16003 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
16005 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
16006 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
16007 (split_branches): Strengthen local "olabel" from rtx to
16008 rtx_insn *, adding a safe_as_a cast.
16009 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
16011 (add_insn_after): Likewise for first two params.
16012 (add_insn_before): Likewise.
16013 (remove_insn): Likewise for param.
16014 (emit_pattern_before_noloc): Likewise for second and third params.
16015 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
16016 (emit_call_insn_before_noloc): Likewise.
16017 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
16019 (emit_barrier_before): Likewise.
16020 (emit_label_before): Strengthen "label" param from "rtx" to
16021 "rtx_code_label *". Strengthen "before" param from "rtx" to
16023 (emit_insn_after_1): Strengthen "after" param from "rtx" to
16025 (emit_pattern_after_noloc): Likewise.
16026 (emit_insn_after_noloc): Likewise.
16027 (emit_jump_insn_after_noloc): Likewise.
16028 (emit_call_insn_after_noloc): Likewise.
16029 (emit_debug_insn_after_noloc): Likewise.
16030 (emit_barrier_after): Likewise.
16031 (emit_label_after): Likewise for both params.
16032 (emit_pattern_after_setloc): Likewise for "after" param. Convert
16033 "loc" param from "int" to "location_t".
16034 (emit_insn_after_setloc): Likewise.
16035 (emit_jump_insn_after_setloc): Likewise.
16036 (emit_call_insn_after_setloc): Likewise.
16037 (emit_debug_insn_after_setloc): Likewise.
16038 (emit_pattern_before_setloc): Likewise for "before" param. Convert
16039 "loc" param from "int" to "location_t".
16040 (emit_pattern_before): Convert NULL_RTX to NULL.
16041 (emit_insn_before_setloc): Convert "loc" param from "int" to
16043 (emit_jump_insn_before_setloc): Likewise.
16044 (emit_call_insn_before_setloc): Likewise.
16045 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
16046 rtx_insn *. Convert "loc" param from "int" to "location_t".
16047 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
16048 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
16049 Convert 3rd param from "int" to "location_t".
16050 (emit_barrier_before, emit_barrier_after, next_real_insn):
16051 Strengthen param from rtx to rtx_insn *.
16052 (emit_label_before): Strengthen 1st param from "rtx" to
16053 "rtx_code_label *". Strengthen 2nd param from "rtx" to
16055 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
16056 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
16057 Strengthen 2nd param from "rtx" to "rtx_insn *".
16058 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
16059 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
16060 Likewise. Convert 3rd param from "int" to "location_t".
16061 (emit_label_after): Strengthen 1st param from "rtx" to
16062 "rtx_code_label *".
16063 (next_real_insn, remove_insn): Strengthen param from "rtx" to
16065 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
16066 from "rtx" to "rtx_insn *".
16068 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
16070 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
16071 bodies streamed in with -Q.
16072 * dumpfile.c (dump_files): Add lto-stream-out dump file.
16073 * dumpfile.h (tree_dump_index): Add lto_stream_out.
16074 * gimple-streamer-out.c: Include gimple-pretty-print.h
16075 (output_bb): Dump stmts streamed.
16076 * lto-section-out.c: Include print-tree.h
16077 (lto_begin_section): Dump sections created.
16078 (lto_output_decl_index): Dump decl encoded.
16079 * lto-streamer-out.c: Include print-tree.h
16080 (create_output_block): Dump output block created.
16081 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
16082 (output_function): Dump function output.
16083 (output_constructor): Dump constructor streamed.
16084 (write_global_stream): Output indexes encoded.
16085 (produce_asm_for_decls): Dump streams encoded.
16086 * lto-streamer.c (streamer_dump_file): New global var.
16087 * lto-streamer.h (streamer_dump_file): Declare.
16088 * passes.c (ipa_write_summaries): Initialize streamer dump.
16089 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
16092 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
16095 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
16096 offsets for register save directives. Emit a second batch of save
16097 directives, if need be, when the function accesses prior frames.
16099 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16101 * config/arc/fpu.md (fmasf4): Force operand to register.
16102 (fnmasf4): Likewise.
16104 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16106 * config/arc/arc-protos.h (arc_pad_return): Remove.
16107 * config/arc/arc.c (machine_function): Remove force_short_suffix
16109 (arc_print_operand): Adjust printing of '&'.
16110 (arc_verify_short): Remove conditional printing of short suffix.
16111 (arc_final_prescan_insn): Remove reference to size_reason.
16112 (pad_return): New function.
16113 (arc_reorg): Call pad_return.
16114 (arc_pad_return): Remove.
16115 (arc_init_machine_status): Remove reference to force_short_suffix.
16116 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
16117 (attr length): When attribute iscompact is true force to 2
16118 regardless; in the case of maybe check if we want to force the
16119 instruction to have 4 bytes length.
16120 (nopv): Change it to generate 4 byte long nop as well.
16121 (blockage): New pattern.
16122 (simple_return): Remove call to arc_pad_return.
16123 (p_return_i): Likewise.
16125 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16127 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
16129 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16131 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
16134 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16136 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
16137 for ARC700 and ARCv2.
16139 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
16142 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
16143 operands[2] instead of operands[1].
16146 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16148 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
16149 case, check whether the outer register overlaps an unallocatable
16150 register, not just whether it fits the required class.
16152 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16154 * poly-int.h (can_div_trunc_p): Add new overload in which all values
16156 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
16157 (memrefs_conflict_p): Likewise.
16158 (init_alias_analysis): Likewise.
16159 * cfgexpand.c (expand_debug_expr): Likewise.
16160 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
16161 * cse.c (fold_rtx): Likewise.
16162 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
16163 * expr.c (emit_block_move_hints): Likewise.
16164 (clear_storage_hints, push_block, emit_push_insn): Likewise.
16165 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
16166 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
16167 (emit_group_store): Likewise.
16168 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
16169 to read the PRE/POST_MODIFY increment.
16170 * calls.c (store_one_arg): Use strip_offset.
16171 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
16173 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
16175 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
16176 (simplify_binary_operation_1): Extend CONST_INT handling to
16178 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
16179 than a HOST_WIDE_INT.
16180 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
16182 (adjust_mems, add_stores): Update accodingly.
16183 (vt_canonicalize_addr): Track polynomial offsets.
16184 (emit_note_insn_var_location): Likewise.
16185 (vt_add_function_parameter): Likewise.
16186 (vt_initialize): Likewise.
16188 2018-06-12 Jeff Law <law@redhat.com>
16190 * config.gcc (alpha*-*-freebsd*): Remove.
16191 * config/alpha/freebsd.h: Remove.
16193 2018-06-12 David Malcolm <dmalcolm@redhat.com>
16196 * spellcheck-tree.c (levenshtein_distance): Rename to...
16197 (get_edit_distance): ...this, and update for underlying renaming.
16198 * spellcheck-tree.h (levenshtein_distance): Rename to...
16199 (get_edit_distance): ...this.
16200 * spellcheck.c (levenshtein_distance): Rename to...
16201 (get_edit_distance): ...this. Convert from Levenshtein distance
16202 to Damerau-Levenshtein distance by supporting transpositions of
16203 adjacent characters. Rename "v1" to "v_next" and "v0" to
16205 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
16206 (selftest::test_edit_distance_unit_test_oneway): ...this, and
16207 update for underlying renaming.
16208 (selftest::levenshtein_distance_unit_test): Rename to...
16209 (selftest::test_get_edit_distance_unit): ...this, and update for
16210 underlying renaming.
16211 (selftest::test_find_closest_string): Add example from PR 69968
16212 where transposition helps
16213 (selftest::test_metric_conditions): Update for renaming.
16214 (selftest::test_metric_conditions): Likewise.
16215 (selftest::spellcheck_c_tests): Likewise.
16216 * spellcheck.h (levenshtein_distance): Rename both overloads to...
16217 (get_edit_distance): ...this.
16218 (best_match::consider): Update for renaming.
16220 2018-06-12 Martin Sebor <msebor@redhat.com>
16222 PR tree-optimization/85259
16223 * builtins.c (compute_objsize): Handle constant offsets.
16224 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
16225 true iff a warning has been issued.
16226 * gimple.h (gimple_nonartificial_location): New function.
16227 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
16228 gimple_nonartificial_location and handle -Wno-system-headers.
16229 (handle_builtin_stxncpy): Same.
16231 2018-06-12 Martin Sebor <msebor@redhat.com>
16234 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
16236 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
16238 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16239 BUILTIN_VEC_XST entries for pointer to double and long long.
16241 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
16244 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
16246 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
16249 2018-06-12 Martin Liska <mliska@suse.cz>
16251 * doc/options.texi: Document IntegerRange.
16253 2018-06-12 Martin Liska <mliska@suse.cz>
16255 * config/i386/i386.opt: Make MPX-related options as Deprecated.
16256 * opt-functions.awk: Handle Deprecated flag.
16257 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
16259 (read_cmdline_option): Report warning for a deprecated option.
16260 * opts.h (struct cl_option): Add new field cl_deprecated.
16261 (CL_ERR_DEPRECATED): New.
16263 2018-06-12 Martin Liska <mliska@suse.cz>
16265 * doc/options.texi: Document Deprecated option flag.
16267 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16269 * config/arc/arc-arch.h (arc_extras): New enum.
16270 (arc_cpu_t):Add field extra.
16271 (arc_cpu_types): Consider the extras.
16272 * config/arc/arc-cpus.def: Add extras info.
16273 * config/arc/arc-opts.h (processor_type): Consider extra field.
16274 * config/arc/arc.c (arc_override_options): Handle extra field.
16276 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
16278 * config/arc/arc-arch.h: Update ARC_OPTX macro.
16279 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
16281 * config/arc/arc.c (arc_init): Update pic warning.
16282 (irq_range): Update irq range parsing warnings.
16283 (arc_override_options): Update various warning messages.
16284 (arc_handle_aux_attribute): Likewise.
16286 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
16288 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
16290 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16292 * doc/sourcebuild.texi: Document usage of line number 0 in verify
16293 compiler messages directives.
16295 2018-06-12 Matthew Fortune <mfortune@gmail.com>
16297 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
16298 * config/mips/mips-tables.opt: Regenerate.
16299 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
16301 * doc/invoke.texi: Document -march=i6500.
16303 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
16305 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
16306 (i6400_gpmul): Add cpu_unit.
16307 (i6400_gpdiv): Likewise.
16308 (i6400_msa_add_d): Update reservations.
16309 (i6400_msa_int_add) Likewise.
16310 (i6400_msa_short_logic3) Likewise.
16311 (i6400_msa_short_logic2) Likewise.
16312 (i6400_msa_short_logic) Likewise.
16313 (i6400_msa_move) Likewise.
16314 (i6400_msa_cmp) Likewise.
16315 (i6400_msa_short_float2) Likewise.
16316 (i6400_msa_div_d) Likewise.
16317 (i6400_msa_long_logic1) Likewise.
16318 (i6400_msa_long_logic2) Likewise.
16319 (i6400_msa_mult) Likewise.
16320 (i6400_msa_long_float2) Likewise.
16321 (i6400_msa_long_float4) Likewise.
16322 (i6400_msa_long_float5) Likewise.
16323 (i6400_msa_long_float8) Likewise.
16324 (i6400_fpu_fadd): Include frint type.
16325 (i6400_fpu_store): New define_insn_reservation.
16326 (i6400_fpu_load): Likewise.
16327 (i6400_fpu_move): Likewise.
16328 (i6400_fpu_fcmp): Likewise.
16329 (i6400_fpu_fmadd): Likewise.
16330 (i6400_int_mult): Include imul3nc type and update reservation.
16331 (i6400_int_div): Include idiv3 type and update reservation.
16332 (i6400_int_load): Update to check type not move_type.
16333 (i6400_int_store): Likewise.
16334 (i6400_int_prefetch): Set zero latency.
16336 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
16338 * gcc.c: Document new %@{...} sequence.
16339 (LINK_COMMAND_SPEC): Use it for the -L switches.
16340 (cpp_unique_options): Use it for the -I switches.
16341 (at_file_argbuf): New global variable.
16342 (in_at_file): Likewise.
16343 (alloc_args): Create at_file_argbuf.
16344 (clear_args): Truncate at_file_argbuf.
16345 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
16346 (open_at_file): New function.
16347 (close_at_file): Likewise.
16348 (create_at_file): Delete.
16349 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
16352 (validate_switches_from_spec): Deal with %@{...} sequence.
16353 (validate_switches): Likewise.
16354 (driver::finalize): Call clear_args.
16356 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
16358 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
16360 2018-06-11 Martin Sebor <msebor@redhat.com>
16362 * doc/invoke.texi (-Wall): List -Wc++17-compat.
16363 (Wno-class-memaccess): Add @opindex.
16364 (Wno-templates, Wno-multiple-inheritance): Same.
16365 (Wno-virtual-inheritance, Wno-namespaces): Same.
16366 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
16367 (Wno-format-overflow, Wno-format-truncation): Same.
16368 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
16369 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
16370 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
16371 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
16372 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
16373 (Wno-misspelled-isr): Same.
16375 2018-06-11 Martin Sebor <msebor@redhat.com>
16377 * PR tree-optimization/86083
16378 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
16380 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
16382 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
16384 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
16387 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
16388 on the correct operand.
16389 (*movdi_internal64): Ditto.
16391 2018-06-11 Martin Liska <mliska@suse.cz>
16393 PR tree-optimization/86089
16394 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
16396 2018-06-11 Julia Koval <julia.koval@intel.com>
16398 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
16399 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
16400 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
16402 2018-06-11 Olivier Hainque <hainque@adacore.com>
16404 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
16405 for Ada with strict dwarf2.
16407 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
16410 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
16413 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
16415 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
16417 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
16419 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
16422 2018-06-08 Martin Liska <mliska@suse.cz>
16424 * tree-cfg.h (debug_function): Fix argument type to match
16427 2018-06-08 Martin Liska <mliska@suse.cz>
16429 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
16430 Remove usage of MPX-related (and removed) fields.
16431 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
16433 2018-06-08 David Malcolm <dmalcolm@redhat.com>
16435 * cfg.c (debug): Use TDF_NONE rather than 0.
16436 * cfghooks.c (debug): Likewise.
16437 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
16438 (struct dump_option_value_info): Convert to...
16439 (struct kv_pair): ...this template type.
16440 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
16442 (optinfo_verbosity_options): Likewise.
16443 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
16445 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
16446 than int for "optgroup_flags" param.
16447 (dump_generic_expr_loc): Use dump_flags_t rather than int for
16449 (dump_dec): Likewise.
16450 (dump_finish): Use TDF_NONE rather than 0.
16451 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
16452 rather than int for "optgroup_flags" param. Use TDF_NONE rather
16453 than 0. Update for change to option_ptr.
16454 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
16455 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
16456 0. Update for changes to optinfo_verbosity_options and
16458 (opt_info_switch_p): Convert optgroup_flags from int to
16460 (dump_basic_block): Use dump_flags_t rather than int
16461 for "dump_kind" param.
16462 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
16463 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
16464 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
16465 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
16466 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
16467 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
16468 TDF_NONE): Convert from macros to...
16469 (enum dump_flag): ...this new enum.
16470 (dump_flags_t): Update to use enum.
16471 (operator|, operator&, operator~, operator|=, operator&=):
16472 Implement for dump_flags_t.
16473 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
16474 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
16475 Convert from macros to...
16476 (enum optgroup_flag): ...this new enum.
16477 (optgroup_flags_t): New typedef.
16478 (operator|, operator|=): Implement for optgroup_flags_t.
16479 (struct dump_file_info): Convert field "alt_flags" to
16480 dump_flags_t. Convert field "optgroup_flags" to
16482 (dump_basic_block): Use dump_flags_t rather than int for param.
16483 (dump_generic_expr_loc): Likewise.
16484 (dump_dec): Likewise.
16485 (dump_register): Convert param "optgroup_flags" to
16487 (opt_info_enable_passes): Likewise.
16488 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
16490 * gimple-pretty-print.c (debug): Likewise.
16491 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
16492 (merged_store_group::apply_stores): Likewise.
16493 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
16494 * gimple.c (verify_gimple_pp): Likewise.
16495 * graphite-poly.c (print_pbb_body): Likewise.
16496 * passes.c (pass_manager::register_one_dump_file): Convert
16497 local "optgroup_flags" to optgroup_flags_t.
16498 * print-tree.c (print_node): Use TDF_NONE rather than 0.
16500 (debug_body): Likewise.
16501 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
16502 to optgroup_flags_t.
16503 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
16505 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
16506 (convert_mult_to_fma): Likewise.
16507 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
16508 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
16509 * tree-vect-data-refs.c (dump_lower_bound): Convert param
16510 "dump_kind" to dump_flags_t.
16512 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
16514 * config/rs6000/rs6000.c (min, max): Delete.
16516 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
16518 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
16519 -mabi=spe and -mabi=no-spe.
16521 2018-06-08 Martin Liska <mliska@suse.cz>
16523 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
16524 where we expect an existing summary.
16526 2018-06-08 Martin Liska <mliska@suse.cz>
16528 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
16529 * ipa-inline.h (estimate_edge_growth): Likewise.
16531 2018-06-08 Martin Liska <mliska@suse.cz>
16533 * cgraph.c (function_version_hasher::hash): Use
16534 cgraph_node::get_uid ().
16535 (function_version_hasher::equal):
16536 * cgraph.h (cgraph_node::get_uid): New method.
16537 * ipa-inline.c (update_caller_keys): Use
16538 cgraph_node::get_uid ().
16539 (update_callee_keys): Likewise.
16540 * ipa-utils.c (searchc): Likewise.
16541 (ipa_reduced_postorder): Likewise.
16542 * lto-cgraph.c (input_node): Likewise.
16543 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
16544 * symbol-summary.h (symtab_insertion): Likewise.
16545 (symtab_removal): Likewise.
16546 (symtab_duplication): Likewise.
16547 * tree-pretty-print.c (dump_function_header): Likewise.
16548 * tree-sra.c (convert_callers_for_node): Likewise.
16550 2018-06-08 Martin Liska <mliska@suse.cz>
16552 * cgraph.c (symbol_table::create_edge): Always assign a new
16554 (symbol_table::free_edge): Do not recycle numbers.
16555 * cgraph.h (cgraph_edge::get): New method.
16556 * symbol-summary.h (symtab_removal): Use it.
16557 (symtab_duplication): Likewise.
16558 (call_summary::hashable_uid): Remove.
16560 2018-06-08 Martin Liska <mliska@suse.cz>
16562 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
16563 (initialize_growth_caches): Remove.
16564 (free_growth_caches): Likewise.
16565 (do_estimate_edge_time): Use edge_growth_cache.
16566 (do_estimate_edge_size): Likewise.
16567 (do_estimate_edge_hints): Likewise.
16568 * ipa-inline.c (reset_edge_caches): Likewise.
16569 (recursive_inlining): Likewise.
16570 (inline_small_functions): Likewise.
16571 * ipa-inline.h (initialize_growth_caches): Remove.
16572 (estimate_edge_size): Likewise.
16573 (estimate_edge_time): Likewise.
16574 (estimate_edge_hints): Likewise.
16575 (reset_edge_growth_cache): Likewise.
16576 * symbol-summary.h (call_summary::remove): New method.
16578 2018-06-08 Martin Liska <mliska@suse.cz>
16580 * ipa-cp.c (class edge_clone_summary): New summary.
16581 (grow_edge_clone_vectors): Remove.
16582 (ipcp_edge_duplication_hook): Remove.
16583 (class edge_clone_summary_t): New call_summary class.
16584 (ipcp_edge_removal_hook): Remove.
16585 (edge_clone_summary_t::duplicate): New function.
16586 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
16587 (create_specialized_node): Likewise.
16588 (ipcp_driver): Initialize edge_clone_summaries and do not
16591 2018-06-08 Martin Liska <mliska@suse.cz>
16593 * symbol-summary.h (get): New function.
16594 (call_summary::m_initialize_when_cloning): New class member.
16596 2018-06-08 Martin Liska <mliska@suse.cz>
16598 * cgraph.c (cgraph_node::remove): Do not recycle uid.
16599 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
16600 (symbol_table::allocate_cgraph_symbol): Do not set uid.
16601 * passes.c (uid_hash_t): Record removed_nodes by their uids.
16602 (remove_cgraph_node_from_order): Use the removed_nodes set.
16603 (do_per_function_toporder): Likwise.
16604 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
16605 instead of summary_uid.
16606 (symtab_removal): Likewise.
16607 (symtab_duplication): Likewise.
16609 2018-06-08 Martin Liska <mliska@suse.cz>
16611 * ipa-cp.c (ipcp_store_bits_results): Use
16612 ipcp_transformation_sum.
16613 (ipcp_store_vr_results): Likewise.
16614 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
16616 (ipcp_transformation_initialize): ... this.
16617 (ipa_set_node_agg_value_chain):
16618 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
16619 (write_ipcp_transformation_info): Likewise.
16620 (read_ipcp_transformation_info): Likewise.
16621 (ipcp_update_bits): Likewise.
16622 (ipcp_update_vr): Likewise.
16623 (ipcp_transform_function): Likewise.
16624 * ipa-prop.h: Rename ipcp_transformation_summary to
16625 ipcp_transformation.
16626 (class ipcp_transformation_t): New function summary.
16627 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
16628 (ipa_get_agg_replacements_for_node): Likewise.
16630 2018-06-08 Martin Liska <mliska@suse.cz>
16632 * ipa-pure-const.c (struct funct_state_d): Do it class instead
16634 (class funct_state_summary_t): New function_summary class.
16635 (has_function_state): Remove.
16636 (get_function_state): Likewise.
16637 (set_function_state): Likewise.
16638 (add_new_function): Likewise.
16639 (funct_state_summary_t::insert): New function.
16640 (duplicate_node_data): Remove.
16641 (remove_node_data): Remove.
16642 (funct_state_summary_t::duplicate): New function.
16643 (register_hooks): Create new funct_state_summaries.
16644 (pure_const_generate_summary): Use it.
16645 (pure_const_write_summary): Likewise.
16646 (pure_const_read_summary): Likewise.
16647 (propagate_pure_const): Likewise.
16648 (propagate_nothrow): Likewise.
16649 (dump_malloc_lattice): Likewise.
16650 (propagate_malloc): Likewise.
16651 (execute): Do not register hooks, just remove summary
16653 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
16656 2018-06-08 Martin Liska <mliska@suse.cz>
16658 * ipa-reference.c (remove_node_data): Remove.
16659 (duplicate_node_data): Likewise.
16660 (class ipa_ref_var_info_summary_t): New class.
16661 (class ipa_ref_opt_summary_t): Likewise.
16662 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
16663 (get_reference_optimization_summary): Use
16664 ipa_ref_opt_sum_summaries.
16665 (set_reference_vars_info): Remove.
16666 (set_reference_optimization_summary): Likewise.
16667 (ipa_init): Create summaries.
16668 (init_function_info): Use function summary.
16669 (ipa_ref_opt_summary_t::duplicate): New function.
16670 (ipa_ref_opt_summary_t::remove): New function.
16671 (get_read_write_all_from_node): Fix GNU coding style.
16672 (propagate): Use function summary.
16673 (write_node_summary_p): Fix GNU coding style.
16674 (stream_out_bitmap): Likewise.
16675 (ipa_reference_read_optimization_summary): Use function summary.
16676 (ipa_reference_c_finalize): Do not release hooks.
16678 2018-06-08 Martin Liska <mliska@suse.cz>
16680 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
16681 (analyze_function_body): Extract multiple calls of get_create.
16682 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
16683 * ipa-inline.c (recursive_inlining): Use ::get method.
16684 * ipa-inline.h (estimate_edge_growth): Likewise.
16686 2018-06-08 Martin Liska <mliska@suse.cz>
16688 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
16690 (hsa_function_summary::hsa_function_summary): Use the new enum
16692 (hsa_gpu_implementation_p): Use hsa_summaries::get.
16693 * hsa-gen.c (hsa_get_host_function): Likewise.
16694 (get_brig_function_name): Likewise.
16695 * ipa-hsa.c (process_hsa_functions): Likewise.
16696 (ipa_hsa_write_summary): Likewise.
16697 * symbol-summary.h (symtab_duplication): Use ::get function/
16698 (get): New function.
16700 2018-06-08 Martin Liska <mliska@suse.cz>
16702 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
16704 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
16705 (hsa_register_kernel): Likewise.
16706 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
16707 * hsa-gen.c (hsa_get_host_function): Likewise.
16708 (get_brig_function_name): Likewise.
16709 (generate_hsa): Likewise.
16710 (pass_gen_hsail::execute): Likewise.
16711 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
16712 (devirtualization_time_bonus): Likewise.
16713 (ipcp_propagate_stage): Likewise.
16714 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
16715 (edge_set_predicate): Likewise.
16716 (evaluate_conditions_for_known_args): Likewise.
16717 (evaluate_properties_for_edge): Likewise.
16718 (ipa_fn_summary::reset): Likewise.
16719 (ipa_fn_summary_t::duplicate): Likewise.
16720 (dump_ipa_call_summary): Likewise.
16721 (ipa_dump_fn_summary): Likewise.
16722 (analyze_function_body): Likewise.
16723 (compute_fn_summary): Likewise.
16724 (estimate_edge_devirt_benefit): Likewise.
16725 (estimate_edge_size_and_time): Likewise.
16726 (estimate_calls_size_and_time): Likewise.
16727 (estimate_node_size_and_time): Likewise.
16728 (inline_update_callee_summaries): Likewise.
16729 (remap_edge_change_prob): Likewise.
16730 (remap_edge_summaries): Likewise.
16731 (ipa_merge_fn_summary_after_inlining): Likewise.
16732 (ipa_update_overall_fn_summary): Likewise.
16733 (read_ipa_call_summary): Likewise.
16734 (inline_read_section): Likewise.
16735 (write_ipa_call_summary): Likewise.
16736 (ipa_fn_summary_write): Likewise.
16737 (ipa_free_fn_summary): Likewise.
16738 * ipa-hsa.c (process_hsa_functions): Likewise.
16739 (ipa_hsa_write_summary): Likewise.
16740 (ipa_hsa_read_section): Likewise.
16741 * ipa-icf.c (sem_function::merge): Likewise.
16742 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
16743 (do_estimate_edge_time): Likewise.
16744 (estimate_size_after_inlining): Likewise.
16745 (estimate_growth): Likewise.
16746 (growth_likely_positive): Likewise.
16747 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
16748 (inline_call): Likewise.
16749 * ipa-inline.c (caller_growth_limits): Likewise.
16750 (can_inline_edge_p): Likewise.
16751 (can_inline_edge_by_limits_p): Likewise.
16752 (compute_uninlined_call_time): Likewise.
16753 (compute_inlined_call_time): Likewise.
16754 (want_inline_small_function_p): Likewise.
16755 (edge_badness): Likewise.
16756 (update_caller_keys): Likewise.
16757 (update_callee_keys): Likewise.
16758 (recursive_inlining): Likewise.
16759 (inline_small_functions): Likewise.
16760 (inline_to_all_callers_1): Likewise.
16761 (dump_overall_stats): Likewise.
16762 (early_inline_small_functions): Likewise.
16763 (early_inliner): Likewise.
16764 * ipa-inline.h (estimate_edge_growth): Likewise.
16765 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16766 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16767 * ipa-prop.h (IPA_NODE_REF): Likewise.
16768 (IPA_EDGE_REF): Likewise.
16769 * ipa-pure-const.c (malloc_candidate_p): Likewise.
16770 (propagate_malloc): Likewise.
16771 * ipa-split.c (execute_split_functions): Likewise.
16772 * symbol-summary.h: Rename get to get_create.
16774 (get_create): Likewise.
16775 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
16777 2018-06-08 Martin Liska <mliska@suse.cz>
16779 * symbol-summary.h (release): Move definition out of class
16781 (symtab_removal): Likewise.
16782 (symtab_duplication): Likewise.
16784 2018-06-08 Martin Liska <mliska@suse.cz>
16786 * symbol-summary.h (function_summary): Move constructor
16787 implementation out of class declaration.
16788 (release): Likewise.
16789 (symtab_insertion): Likewise.
16790 (symtab_removal): Likewise.
16791 (symtab_duplication): Likewise.
16794 2018-06-08 Martin Liska <mliska@suse.cz>
16796 * Makefile.in: Remove support for MPX (macros, related functions,
16797 fields in cgraph_node, ...).
16798 * builtin-types.def (BT_BND): Likewise.
16799 (BT_FN_BND_CONST_PTR): Likewise.
16800 (BT_FN_CONST_PTR_BND): Likewise.
16801 (BT_FN_VOID_PTR_BND): Likewise.
16802 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
16803 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
16804 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
16805 (expand_builtin_mempcpy_with_bounds): Likewise.
16806 (expand_builtin_memset_with_bounds): Likewise.
16807 (expand_builtin_memset_args): Likewise.
16808 (std_expand_builtin_va_start): Likewise.
16809 (expand_builtin): Likewise.
16810 (expand_builtin_with_bounds): Likewise.
16811 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
16812 (DEF_LIB_BUILTIN_CHKP): Likewise.
16813 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
16814 (DEF_CHKP_BUILTIN): Likewise.
16815 (BUILT_IN_MEMCPY): Likewise.
16816 (BUILT_IN_MEMMOVE): Likewise.
16817 (BUILT_IN_MEMPCPY): Likewise.
16818 (BUILT_IN_MEMSET): Likewise.
16819 (BUILT_IN_STPCPY): Likewise.
16820 (BUILT_IN_STRCAT): Likewise.
16821 (BUILT_IN_STRCHR): Likewise.
16822 (BUILT_IN_STRCPY): Likewise.
16823 (BUILT_IN_STRLEN): Likewise.
16824 (BUILT_IN_MEMCPY_CHK): Likewise.
16825 (BUILT_IN_MEMMOVE_CHK): Likewise.
16826 (BUILT_IN_MEMPCPY_CHK): Likewise.
16827 (BUILT_IN_MEMSET_CHK): Likewise.
16828 (BUILT_IN_STPCPY_CHK): Likewise.
16829 (BUILT_IN_STRCAT_CHK): Likewise.
16830 (BUILT_IN_STRCPY_CHK): Likewise.
16831 * calls.c (store_bounds): Likewise.
16832 (emit_call_1): Likewise.
16833 (special_function_p): Likewise.
16834 (maybe_warn_nonstring_arg): Likewise.
16835 (initialize_argument_information): Likewise.
16836 (finalize_must_preallocate): Likewise.
16837 (compute_argument_addresses): Likewise.
16838 (expand_call): Likewise.
16839 * cfgexpand.c (expand_call_stmt): Likewise.
16840 (expand_return): Likewise.
16841 (expand_gimple_stmt_1): Likewise.
16842 (pass_expand::execute): Likewise.
16843 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
16844 (cgraph_node::remove): Likewise.
16845 (cgraph_node::dump): Likewise.
16846 (cgraph_node::verify_node): Likewise.
16847 * cgraph.h (chkp_function_instrumented_p): Likewise.
16848 (symtab_node::get_alias_target): Likewise.
16849 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
16850 (cgraph_local_p): Likewise.
16851 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
16852 (cgraph_edge::rebuild_references): Likewise.
16853 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
16854 (walk_polymorphic_call_targets): Likewise.
16855 (cgraph_node::expand_thunk): Likewise.
16856 (symbol_table::output_weakrefs): Likewise.
16857 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
16858 (ix86_handle_option): Likewise.
16859 * config/i386/constraints.md: Likewise.
16860 * config/i386/i386-builtin-types.def (BND): Likewise.
16864 * config/i386/i386-builtin.def (BDESC_END): Likewise.
16865 (BDESC_FIRST): Likewise.
16867 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
16868 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
16869 * config/i386/i386.c (enum reg_class): Likewise.
16870 (ix86_target_string): Likewise.
16871 (ix86_option_override_internal): Likewise.
16872 (ix86_conditional_register_usage): Likewise.
16873 (ix86_valid_target_attribute_inner_p): Likewise.
16874 (ix86_set_indirect_branch_type): Likewise.
16875 (ix86_set_current_function): Likewise.
16876 (ix86_function_arg_regno_p): Likewise.
16877 (init_cumulative_args): Likewise.
16878 (ix86_function_arg_advance): Likewise.
16879 (ix86_function_arg): Likewise.
16880 (ix86_pass_by_reference): Likewise.
16881 (ix86_function_value_regno_p): Likewise.
16882 (ix86_function_value_1): Likewise.
16883 (ix86_function_value_bounds): Likewise.
16884 (ix86_return_in_memory): Likewise.
16885 (ix86_setup_incoming_vararg_bounds): Likewise.
16886 (ix86_va_start): Likewise.
16887 (indirect_thunk_need_prefix): Likewise.
16888 (print_reg): Likewise.
16889 (ix86_print_operand): Likewise.
16890 (ix86_expand_call): Likewise.
16891 (ix86_output_function_return): Likewise.
16892 (reg_encoded_number): Likewise.
16893 (BDESC_VERIFYS): Likewise.
16894 (ix86_init_mpx_builtins): Likewise.
16895 (ix86_init_builtins): Likewise.
16896 (ix86_emit_cmove): Likewise.
16897 (ix86_emit_move_max): Likewise.
16898 (ix86_expand_builtin): Likewise.
16899 (ix86_builtin_mpx_function): Likewise.
16900 (ix86_get_arg_address_for_bt): Likewise.
16901 (ix86_load_bounds): Likewise.
16902 (ix86_store_bounds): Likewise.
16903 (ix86_load_returned_bounds): Likewise.
16904 (ix86_store_returned_bounds): Likewise.
16905 (ix86_class_likely_spilled_p): Likewise.
16906 (ix86_hard_regno_mode_ok): Likewise.
16907 (x86_order_regs_for_local_alloc): Likewise.
16908 (ix86_mitigate_rop): Likewise.
16909 (ix86_bnd_prefixed_insn_p): Likewise.
16910 (ix86_mpx_bound_mode): Likewise.
16911 (ix86_make_bounds_constant): Likewise.
16912 (ix86_initialize_bounds): Likewise.
16913 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
16914 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
16915 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
16916 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
16917 (TARGET_CHKP_BOUND_MODE): Likewise.
16918 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
16919 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
16920 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
16921 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
16922 * config/i386/i386.h (TARGET_MPX): Likewise.
16923 (TARGET_MPX_P): Likewise.
16924 (VALID_BND_REG_MODE): Likewise.
16925 (FIRST_BND_REG): Likewise.
16926 (LAST_BND_REG): Likewise.
16927 (enum reg_class): Likewise.
16928 (BND_REG_P): Likewise.
16929 (BND_REGNO_P): Likewise.
16930 (BNDmode): Likewise.
16931 (ADJUST_INSN_LENGTH): Likewise.
16932 * config/i386/i386.md: Likewise.
16933 * config/i386/i386.opt: Likewise.
16934 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
16935 (defined): Likewise.
16936 (LINK_MPX): Likewise.
16937 (MPX_SPEC): Likewise.
16938 (LIBMPX_SPEC): Likewise.
16939 (LIBMPXWRAPPERS_SPEC): Likewise.
16940 (CHKP_SPEC): Likewise.
16941 * config/i386/predicates.md: Likewise.
16942 * dbxout.c (dbxout_type): Likewise.
16943 * doc/extend.texi: Likewise.
16944 * doc/invoke.texi: Likewise.
16945 * doc/md.texi: Likewise.
16946 * doc/tm.texi: Likewise.
16947 * doc/tm.texi.in: Likewise.
16948 * dwarf2out.c (is_base_type): Likewise.
16949 (gen_formal_types_die): Likewise.
16950 (gen_subprogram_die): Likewise.
16951 (gen_type_die_with_usage): Likewise.
16952 (gen_decl_die): Likewise.
16953 (dwarf2out_late_global_decl): Likewise.
16954 * expr.c (expand_assignment): Likewise.
16955 (emit_storent_insn): Likewise.
16956 (store_expr_with_bounds): Likewise.
16957 (store_expr): Likewise.
16958 (expand_expr_real_1): Likewise.
16959 * expr.h (store_expr_with_bounds): Likewise.
16960 * function.c (use_register_for_decl): Likewise.
16961 (struct bounds_parm_data): Likewise.
16962 (assign_parms_augmented_arg_list): Likewise.
16963 (assign_parm_find_entry_rtl): Likewise.
16964 (assign_parm_is_stack_parm): Likewise.
16965 (assign_parm_load_bounds): Likewise.
16966 (assign_bounds): Likewise.
16967 (assign_parms): Likewise.
16968 (expand_function_start): Likewise.
16969 * gcc.c (CHKP_SPEC): Likewise.
16970 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16971 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
16972 (wrestrict_dom_walker::check_call): Likewise.
16973 * gimple.c (gimple_build_call_from_tree): Likewise.
16974 * gimple.h (enum gf_mask): Likewise.
16975 (gimple_call_with_bounds_p): Likewise.
16976 (gimple_call_set_with_bounds): Likewise.
16977 * gimplify.c (gimplify_init_constructor): Likewise.
16978 * ipa-cp.c (initialize_node_lattices): Likewise.
16979 (propagate_constants_across_call): Likewise.
16980 (find_more_scalar_values_for_callers_subset): Likewise.
16981 * ipa-hsa.c (process_hsa_functions): Likewise.
16982 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
16983 * ipa-icf.c (sem_function::merge): Likewise.
16984 * ipa-inline.c (early_inliner): Likewise.
16985 * ipa-pure-const.c (warn_function_noreturn): Likewise.
16986 (warn_function_cold): Likewise.
16987 (propagate_pure_const): Likewise.
16988 * ipa-ref.h (enum GTY): Likewise.
16989 * ipa-split.c (find_retbnd): Likewise.
16990 (consider_split): Likewise.
16991 (split_function): Likewise.
16992 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
16993 * ipa.c (walk_polymorphic_call_targets): Likewise.
16994 (symbol_table::remove_unreachable_nodes): Likewise.
16995 (process_references): Likewise.
16996 (cgraph_build_static_cdtor_1): Likewise.
16997 * lto-cgraph.c (lto_output_node): Likewise.
16998 (output_refs): Likewise.
16999 (compute_ltrans_boundary): Likewise.
17000 (input_overwrite_node): Likewise.
17001 (input_node): Likewise.
17002 (input_cgraph_1): Likewise.
17003 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
17004 * passes.c (pass_manager::execute_early_local_passes): Likewise.
17005 (class pass_chkp_instrumentation_passes): Likewise.
17006 (make_pass_chkp_instrumentation_passes): Likewise.
17007 * passes.def: Likewise.
17008 * rtl.h (struct GTY): Likewise.
17009 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
17010 * stor-layout.c (layout_type): Likewise.
17011 * symtab.c: Likewise.
17012 * target.def: Likewise.
17013 * targhooks.c (default_chkp_bound_type): Likewise.
17014 (default_chkp_bound_mode): Likewise.
17015 (default_builtin_chkp_function): Likewise.
17016 (default_chkp_function_value_bounds): Likewise.
17017 (default_chkp_make_bounds_constant): Likewise.
17018 (default_chkp_initialize_bounds): Likewise.
17019 * targhooks.h (default_chkp_bound_type): Likewise.
17020 (default_chkp_bound_mode): Likewise.
17021 (default_builtin_chkp_function): Likewise.
17022 (default_chkp_function_value_bounds): Likewise.
17023 (default_chkp_make_bounds_constant): Likewise.
17024 (default_chkp_initialize_bounds): Likewise.
17025 * toplev.c (compile_file): Likewise.
17026 (process_options): Likewise.
17027 * tree-core.h (DEF_BUILTIN): Likewise.
17028 (DEF_BUILTIN_CHKP): Likewise.
17029 * tree-inline.c (declare_return_variable): Likewise.
17030 (remap_gimple_stmt): Likewise.
17031 (copy_bb): Likewise.
17032 (initialize_inlined_parameters): Likewise.
17033 (expand_call_inline): Likewise.
17034 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
17035 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
17036 (make_pass_ipa_chkp_produce_thunks): Likewise.
17037 (make_pass_chkp): Likewise.
17038 (make_pass_chkp_opt): Likewise.
17039 (make_pass_chkp_instrumentation_passes): Likewise.
17040 * tree-pretty-print.c (dump_generic_node): Likewise.
17041 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
17042 * tree-ssa-dce.c (propagate_necessity): Likewise.
17043 (eliminate_unnecessary_stmts): Likewise.
17044 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
17045 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
17046 * tree-ssa-sccvn.h: Likewise.
17047 * tree-ssa-strlen.c (get_string_length): Likewise.
17048 (valid_builtin_call): Likewise.
17049 (adjust_last_stmt): Likewise.
17050 (handle_builtin_strchr): Likewise.
17051 (handle_builtin_strcpy): Likewise.
17052 (handle_builtin_stxncpy): Likewise.
17053 (handle_builtin_memcpy): Likewise.
17054 (handle_builtin_strcat): Likewise.
17055 (strlen_check_and_optimize_stmt): Likewise.
17056 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
17057 * tree-streamer-in.c: Likewise.
17058 * tree-streamer.c (record_common_node): Likewise.
17059 * tree.c (tree_code_size): Likewise.
17060 (wide_int_to_tree_1): Likewise.
17061 (type_contains_placeholder_1): Likewise.
17062 (build_common_tree_nodes): Likewise.
17063 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
17064 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
17065 (POINTER_BOUNDS_P): Likewise.
17066 (BOUNDED_TYPE_P): Likewise.
17067 (BOUNDED_P): Likewise.
17068 (CALL_WITH_BOUNDS_P): Likewise.
17069 (pointer_bounds_type_node): Likewise.
17070 * value-prof.c (gimple_ic): Likewise.
17071 * var-tracking.c (vt_add_function_parameters): Likewise.
17072 * varasm.c (make_decl_rtl): Likewise.
17073 (assemble_start_function): Likewise.
17074 (output_constant): Likewise.
17075 (maybe_assemble_visibility): Likewise.
17076 * varpool.c (ctor_for_folding): Likewise.
17077 * chkp-builtins.def: Remove.
17078 * ipa-chkp.c: Remove.
17079 * ipa-chkp.h: Remove.
17080 * rtl-chkp.c: Remove.
17081 * rtl-chkp.h: Remove.
17082 * tree-chkp-opt.c: Remove.
17083 * tree-chkp.c: Remove.
17084 * tree-chkp.h: Remove.
17086 2018-06-07 Carl Love <cel@us.ibm.com>
17088 * config/rs6000/vsx.md (vextract_fp_from_shorth,
17089 vextract_fp_from_shortl): Add BE support.
17091 2018-06-07 Paul Koning <ni1d@arrl.net>
17093 * compare-elim.c (try_merge_compare): Don't merge compare if
17094 address contains a side effect.
17095 (try_eliminate_compare): Likewise.
17097 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
17099 * config.gcc: Support "tremont".
17100 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
17101 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17103 * config/i386/i386.c (m_TREMONT): Define.
17104 (processor_target_table): Add "tremont".
17105 (PTA_TREMONT): Define.
17106 (ix86_lea_outperforms): Add TARGET_TREMONT.
17107 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
17108 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
17109 and M_INTEL_GOLDMONT_PLUS.
17110 (fold_builtin_cpu): Add "tremont".
17111 (ix86_add_stmt_cost): Add TARGET_TREMONT.
17112 (ix86_option_override_internal): Add "tremont".
17113 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
17114 (processor_type): Add PROCESSOR_TREMONT.
17115 * config/i386/x86-tune.def: Add m_TREMONT.
17116 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
17118 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17120 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
17121 symbol defined for msp430i* devices to be lower case.
17123 2018-06-07 Richard Biener <rguenther@suse.de>
17125 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
17126 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
17127 Properly wrap signed arithmetic if overflow wraps.
17129 2018-06-07 Jakub Jelinek <jakub@redhat.com>
17131 PR tree-optimization/69615
17132 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
17133 of a cast from a same precision integral SSA_NAME in a bb dominated
17134 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
17135 cast to utype if rhs2 has already a compatible type.
17137 2018-06-07 Richard Biener <rguenther@suse.de>
17139 PR tree-optimization/85935
17140 * graphite-scop-detection.c (find_params_in_bb): Analyze
17141 condition operands with respect to the correct loop. Assert
17142 the analysis doesn't fail.
17144 2018-06-04 Carl Love <cel@us.ibm.com>
17146 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
17147 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
17148 as it is slightly cheaper.
17149 (first_match_or_eos_index_<mode>):
17150 Calculate index using natural element order.
17151 (first_match_index_<mode>):
17152 Calculate index using natural element order.
17153 (first_match_or_eos_index_<mode>):
17154 Calculate index using natural order.
17155 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
17156 for BE and LE modes.
17157 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
17158 P9V_BUILTIN_VCLZLSBB_V16QI.
17159 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
17162 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
17164 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
17165 indentation and line wrap for many prototypes. Add missing
17166 @smallexample directives around block of prototypes for vec_xl and
17169 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
17171 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
17172 track if we pass or return IEEE 128-bit floating point.
17173 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
17174 C++ mangling that is compatible with GCC 8.1.
17175 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
17176 (init_cumulative_args): Note if we pass or return IEEE 128-bit
17177 floating point types.
17178 (rs6000_function_arg_advance_1): Likewise.
17179 (rs6000_mangle_type): Optionally generate mangled names that match
17180 what GCC 8.1 generated for IEEE 128-bit floating point types.
17181 (rs6000_globalize_decl_name): If we have an external function that
17182 passes or returns IEEE 128-bit types, generate a weak reference
17183 from the mangled name used in GCC 8.1 to the current mangled
17185 (rs6000_init_builtins): Make __ibm128 use the long double type if
17186 long double is IBM extended double. Make __float128 use the long
17187 double type if long double is IEEE 128-bit.
17190 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
17191 macro for __ibm128 built-in functions.
17192 (PACK_IF): Add __ibm128 pack/unpack functions.
17193 (UNPACK_IF): Likewise.
17194 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
17195 enable long double built-in functions if long double is IEEE
17196 128-bit floating point.
17197 (rs6000_invalid_builtin): Update long double built-in function
17199 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
17200 functions, adjust the built-in function to use the long double
17201 built-in function if __ibm128 and long double are the same type.
17202 * doc/extend.texi (PowerPC builtins): Update documention for
17203 __builtin_{,un}pack_longdouble. Add documentation for
17204 __builtin_{,un}pack_ibm128.
17206 2018-06-06 Jim Wilson <jimw@sifive.com>
17208 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
17209 (struct machine_function): New field interrupt_mode.
17210 (riscv_handle_type_attribute): New function. Add forward declaration.
17211 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
17212 (riscv_expand_epilogue): Check interrupt_mode field.
17213 (riscv_set_current_function): Check interrupt attribute args and
17214 set interrupt_mode field.
17215 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
17216 (riscv_sret, riscv_uret): New.
17217 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
17218 new arguments to interrupt attribute.
17220 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
17223 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
17224 Don't handle -mcpu=power8 if -mpower9-vector is also used.
17226 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
17228 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
17229 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
17230 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
17231 several redundant entries.
17233 2018-06-06 David Malcolm <dmalcolm@redhat.com>
17235 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
17236 type from "rtx" to "rtx_insn *".
17237 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
17238 for local "call_insn", removing cast.
17239 (ix86_expand_call): Likewise, introducing a "call_insn" local.
17241 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
17243 PR tree-optimization/86066
17244 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
17245 for BIT_INSERT_EXPR stores.
17247 2018-06-06 Richard Biener <rguenther@suse.de>
17249 PR tree-optimization/86062
17250 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
17251 component refs ontop
17252 of to be offsetted base.
17254 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17256 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
17257 to be static and remove check on interrupt attribute name.
17259 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
17261 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
17262 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
17264 2018-06-05 Steve Ellcey <sellcey@cavium.com>
17267 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
17269 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
17270 Remove second argument, change how error is called.
17271 (aarch64_layout_arg): Remove second argument from
17272 aarch64_err_no_fpadvsimd call.
17273 (aarch64_init_cumulative_args): Ditto.
17274 (aarch64_gimplify_va_arg_expr): Ditto.
17275 * config/aarch64/aarch64.md (mov<mode>): Ditto.
17277 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
17279 * config/i386/i386.md (simple_return_indirect_internal): New expander.
17280 (*simple_return_indirect_internal<mode>): Rename from
17281 simple_return_indirect_internal. Use W mode iterator.
17282 (rstorssp): New expander.
17283 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
17284 (clrssbsy): New expander.
17285 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
17287 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
17289 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
17291 (cmse_check_pointed_object): Likewise.
17293 2018-06-05 Martin Liska <mliska@suse.cz>
17295 PR gcov-profile/47618
17296 * doc/invoke.texi: Document how -fprofile-dir format
17299 2018-06-05 Richard Biener <rguenther@suse.de>
17301 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
17302 removal pretend DOM info isn't available so we do not update
17303 it and only remove edges, not dominated blocks. Actually free
17304 DOM info in case we removed something. Remove unreachable blocks.
17305 (mfb_keep_latches): Work with either DOM info or marked backedges.
17306 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
17307 first. Mark backedges if DOM info isn't available.
17308 (Re-)compute DOM info after cleanup_control_flow_pre.
17310 2018-06-05 Richard Biener <rguenther@suse.de>
17312 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
17313 (locus_discrim_hasher::hash): Adjust.
17314 (locus_discrim_hasher::equal): Likewise.
17315 (next_discriminator_for_locus): Work on line directly.
17316 (same_line_p): Pass in expanded locus1 as well.
17317 (assign_discriminators): Avoid redundant location expansions.
17319 2018-06-05 Richard Biener <rguenther@suse.de>
17321 PR tree-optimization/86046
17322 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
17323 if required after clearing TREE_ADDRESSABLE.
17325 2018-06-05 Richard Biener <rguenther@suse.de>
17327 PR tree-optimization/86047
17328 * tree-ssa-loop.c (for_each_index): Glob handling of all
17329 decls and constants and really handle all of them.
17331 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17334 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
17335 qualifier_void_pointer and qualifier_const_void_pointer.
17336 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
17337 (arm_init_builtins): Handle the above.
17338 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
17339 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
17340 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
17343 2018-06-05 Martin Liska <mliska@suse.cz>
17345 * auto-profile.c (read_autofdo_file): Do not use
17346 gcov_ctr_summary struct.
17347 (afdo_callsite_hot_enough_for_early_inline): Likewise.
17348 * coverage.c (struct counts_entry): Likewise.
17349 (read_counts_file): Read just single summary entry.
17350 (get_coverage_counts): Use gcov_summary struct.
17351 * coverage.h (get_coverage_counts): Likewise.
17352 * gcov-dump.c (dump_working_sets): Likewise.
17353 (tag_summary): Dump just single summary.
17354 * gcov-io.c (gcov_write_summary): Write just histogram
17356 (gcov_read_summary): Read just single summary.
17357 (compute_working_sets): Use gcov_summary struct.
17358 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
17359 of GCOV_COUNTERS_SUMMABLE.
17360 (GCOV_COUNTERS_SUMMABLE): Remove.
17361 (GCOV_FIRST_VALUE_COUNTER): Replace with
17362 GCOV_COUNTER_V_INTERVAL.
17363 (struct gcov_ctr_summary): Remove.
17364 (struct gcov_summary): Directly use fields of former
17366 (compute_working_sets): Use gcov_summary struct.
17367 * gcov.c (read_count_file): Do not use ctrs fields.
17368 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
17370 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
17372 * profile.c: Likewise.
17373 * profile.h: Likewise.
17375 2018-06-05 Martin Liska <mliska@suse.cz>
17377 PR gcov-profile/84846
17378 * gcov.c (output_lines): Print working directory only
17379 in intermediate format.
17381 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
17383 * config/s390/s390-builtin-types.def: Add void function type.
17384 * config/s390/s390-builtins.def: Use the function type for the
17387 2018-06-04 Jim Wilson <jimw@sifive.com>
17389 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
17391 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
17392 and maybe_eh_return. Change regno to unsigned int. Use new args to
17393 handle EH_RETURN_DATA_REGNO registers properly.
17394 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
17395 (riscv_expand_epilogue): Update comment. Change argument name and
17396 type. Update code to use new name and type. Pass new args to
17397 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
17399 * config/riscv/riscv.md (NORMAL_RETURN): New.
17400 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
17401 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
17402 (eh_return): Call gen_eh_return_internal and emit barrier.
17403 (eh_return_internal): Call riscv_expand_epilogue.
17405 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
17407 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
17408 bit_insertion field and declare can_be_merged_into method.
17409 (merged_store_group::can_be_merged_into): New method.
17410 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
17411 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
17412 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
17414 2018-06-04 Richard Biener <rguenther@suse.de>
17416 PR tree-optimization/85955
17417 * builtins.c (fold_builtin_sincos): Convert pointers to
17418 destination to appropriate type before dereferencing.
17420 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
17422 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
17424 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
17426 * expr.c (expand_expr_real_1): Force the operand into memory if
17427 its TYPE_MODE is BLKmode and if there is no integer mode for
17428 the number of bits being extracted.
17430 2018-06-04 Jakub Jelinek <jakub@redhat.com>
17434 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
17435 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
17437 2018-06-04 Richard Biener <rguenther@suse.de>
17439 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
17440 (cleanup_tree_cfg_noloop): ... single caller. Do
17441 start_recording_case_labels later.
17443 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
17445 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
17446 to _IMMINTRIN_H_INCLUDED.
17447 * config/i386/pconfigintrin.h: Ditto.
17448 * config/i386/waitpkgintrin.h: Ditto.
17449 * config/i386/immintrin.h: Add includes for sgxintrin.h,
17450 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
17451 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
17452 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
17453 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
17454 waitpkgintrin.h and cldemoteintrin.h.
17456 2018-06-04 Richard Biener <rguenther@suse.de>
17458 PR tree-optimization/86038
17459 * tracer.c (find_best_successor): Check probability for
17460 being initialized, bail out if not.
17462 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
17465 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
17466 of bits to ignore when comparing architectures.
17468 2018-06-04 Jakub Jelinek <jakub@redhat.com>
17470 PR tree-optimization/69615
17471 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
17472 maximum or minimum of the type, try to merge it also as if
17473 range1 is + [-, x - 1] or + [x + 1, -].
17476 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
17478 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
17480 PR tree-optimization/86034
17481 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
17482 the unsigned bitfield type in a bit insertion sequence if it does not
17483 have a larger precision than the bitfield size.
17484 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
17486 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
17488 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
17490 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
17492 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
17493 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
17494 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
17495 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
17497 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
17499 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
17500 Disable -fdelete-null-pointer-checks for ELF toolchain.
17502 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
17503 Kito Cheng <kito.cheng@gmail.com>
17505 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
17506 (nds32le-*-*, nds32be-*-*): Integrate checking process.
17507 (nds32*-*-*): Add glibc and uclibc conditions.
17508 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
17509 (TARGET_EXCEPT_UNWIND_INFO): Define.
17510 * config/nds32/elf.h: New file.
17511 * config/nds32/linux.h: New file.
17512 * config/nds32/nds32-elf.opt: New file.
17513 * config/nds32/nds32-linux.opt: New file.
17514 * config/nds32/nds32-fp-as-gp.c
17515 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
17516 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
17518 (nds32_asm_file_end): Ditto.
17519 (nds32_print_operand): Ditto.
17520 (nds32_insert_attributes): Ditto.
17521 (nds32_init_libfuncs): New function.
17522 (TARGET_HAVE_TLS): Define.
17523 (TARGET_INIT_LIBFUNCS): Define.
17524 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
17526 (TARGET_ELF): Apply different mcmodel setting.
17527 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
17528 been migrated into elf.h and linux.h files.
17529 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
17530 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
17531 (mcmodel): The content has been migrated into nds32-elf.opt and
17532 nds32-linux.opt files.
17533 * config/nds32/t-elf: New file.
17534 * config/nds32/t-linux: New file.
17536 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
17537 Shiva Chen <shiva0217@gmail.com>
17539 * config/nds32/constants.md (unspec_volatile_element): Add
17540 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
17541 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
17543 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
17544 (make_pass_nds32_fp_as_gp): Declare.
17545 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
17547 (nds32_asm_function_end_prologue): Remove unused asm output.
17548 (nds32_asm_function_begin_epilogue): Remove unused asm output.
17549 (nds32_asm_file_start): Output necessary fp_as_gp information.
17550 (nds32_option_override): Adjust register usage.
17551 (nds32_expand_prologue): Consider fp_as_gp situation.
17552 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
17553 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
17556 (simple_return): Ditto.
17557 (omit_fp_begin): Output special directive for fp_as_gp.
17558 (omit_fp_end): Output special directive for fp_as_gp.
17559 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
17560 mforbid-fp-as-gp): New options.
17562 2018-06-01 Mark Wielaard <mark@klomp.org>
17564 * dwarf2out.c (dwarf2out_finish): Remove generation of
17565 DW_AT_loclists_base.
17567 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
17569 * gimple-ssa-store-merging.c: Include gimple-fold.h.
17570 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
17571 (struct merged_store_group): Add bit_insertion field.
17572 (dump_char_array): Use standard hexadecimal format.
17573 (merged_store_group::merged_store_group): Set bit_insertion to false.
17574 (merged_store_group::apply_stores): Use optimal buffer size. Deal
17575 with BIT_INSERT_EXPR stores. Move up code updating the mask and
17576 also print the mask in the dump file.
17577 (pass_store_merging::gate): Minor tweak.
17578 (imm_store_chain_info::coalesce_immediate): Fix wrong association
17579 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
17580 stores with INTEGER_CST stores.
17581 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
17582 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
17583 and use it throughout. Generate bit insertion sequences if need be.
17584 (pass_store_merging::process_store): Remove redundant condition.
17585 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
17587 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
17589 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
17590 the 128-bit floating point types. Fix function comment.
17592 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17594 * config/aarch64/aarch64-simd.md
17595 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
17597 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
17600 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
17602 PR tree-optimization/85989
17603 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
17605 (backprop::intersect_uses): Check it when deciding whether this
17606 is a backedge reference.
17607 (backprop::process_block): Add each phi to m_visited_phis
17608 after visiting it, then clear it at the end.
17610 2018-06-01 Richard Biener <rguenther@suse.de>
17612 * tree-vectorizer.h (vect_dr_stmt): New function.
17613 (vect_get_load_cost): Adjust.
17614 (vect_get_store_cost): Likewise.
17615 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
17616 Use vect_dr_stmt instead of DR_SMTT.
17617 (vect_record_base_alignments): Likewise.
17618 (vect_calculate_target_alignment): Likewise.
17619 (vect_compute_data_ref_alignment): Likewise and make static.
17620 (vect_update_misalignment_for_peel): Likewise.
17621 (vect_verify_datarefs_alignment): Likewise.
17622 (vector_alignment_reachable_p): Likewise.
17623 (vect_get_data_access_cost): Likewise. Pass down
17624 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
17625 (vect_get_peeling_costs_all_drs): Likewise.
17626 (vect_peeling_hash_get_lowest_cost): Likewise.
17627 (vect_enhance_data_refs_alignment): Likewise.
17628 (vect_find_same_alignment_drs): Likewise.
17629 (vect_analyze_data_refs_alignment): Likewise.
17630 (vect_analyze_group_access_1): Likewise.
17631 (vect_analyze_group_access): Likewise.
17632 (vect_analyze_data_ref_access): Likewise.
17633 (vect_analyze_data_ref_accesses): Likewise.
17634 (vect_vfa_segment_size): Likewise.
17635 (vect_small_gap_p): Likewise.
17636 (vectorizable_with_step_bound_p): Likewise.
17637 (vect_prune_runtime_alias_test_list): Likewise.
17638 (vect_analyze_data_refs): Likewise.
17639 (vect_supportable_dr_alignment): Likewise.
17640 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
17641 (vect_gen_prolog_loop_niters): Likewise.
17642 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
17643 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
17645 (vect_recog_mask_conversion_pattern): Likewise.
17646 (vect_try_gather_scatter_pattern): Likewise.
17647 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
17648 to vect_get_store_cost.
17649 (vect_get_store_cost): Get stmt_info instead of DR.
17650 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
17651 (vect_get_load_cost): Get stmt_info instead of DR.
17653 2018-06-01 Richard Biener <rguenther@suse.de>
17655 PR middle-end/86017
17656 * gimple-fold.c (var_decl_component_p): Also allow offsetted
17657 vars wrapped in MEM_REFs.
17659 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
17661 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17662 Fix subreg tests so that we only return a choice between
17663 GENERAL_REGS and FP_REGS if the original classes included both.
17665 2018-06-01 Richard Biener <rguenther@suse.de>
17668 * tree-ssa-structalias.c (get_function_part_constraint):
17669 Handle NULL fi->decl.
17670 (find_func_aliases_for_call): Properly handle indirect
17671 fi from direct call.
17672 (find_func_clobbers): Likewise.
17673 (ipa_pta_execute): Likewise.
17674 (create_variable_info_for): For functions that are ifunc_resolver
17675 resolve to a varinfo that contains the result of the resolver call.
17676 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
17679 2018-05-31 Michael Collison <michael.collison@arm.com>
17681 * config/aarch64/aarch64.md:
17682 (*fix_to_zero_extenddfdi2): New pattern.
17683 * gcc.target/aarch64/fix_extend1.c: New testcase.
17685 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
17687 PR middle-end/78809
17688 PR middle-end/83026
17689 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
17690 and BUILT_IN_STRNCMP_EQ.
17691 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
17692 BUILT_IN_STRNCMP_EQ.
17693 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
17694 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
17695 (gimple_fold_builtin): Likewise.
17696 * tree-ssa-strlen.c (compute_string_length): New function.
17697 (determine_min_obsize): New function.
17698 (handle_builtin_string_cmp): New function to handle calls to
17699 string compare functions.
17700 (strlen_optimize_stmt): Add handling to builtin string compare
17702 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17703 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
17704 * tree.c (build_common_builtin_nodes): Add new defines of
17705 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
17707 2018-05-31 Jakub Jelinek <jakub@redhat.com>
17710 * bb-reorder.c (pass_partition_blocks::gate): Return false for
17711 functions with naked attribute.
17713 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
17715 * config/i386/sse.md (avx_vec_concat<mode>):
17716 Substitute concat_tg_mode mode attribute with xtg_mode.
17717 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
17718 (concat_tg_mode): Remove mode attribute.
17720 2018-05-31 Martin Sebor <msebor@redhat.com>
17723 * calls.c (alloc_max_size): Correct a logic error/typo.
17724 Treat excessive arguments as infinite. Warn for invalid arguments.
17725 * doc/invoke.texi (-Walloc-size-larger-than): Update.
17727 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
17730 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
17731 and movx for Haswell.
17733 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
17734 Cesar Philippidis <cesar@codesourcery.com>
17736 PR middle-end/85879
17737 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
17738 when emitting error on private/firstprivate reductions.
17739 * omp-low.c (lower_omp_target): Avoid reference-type processing
17740 on pointers for firstprivate clause.
17742 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
17744 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
17747 * config/aarch64/aarch64-simd.md
17748 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
17749 (aarch64_ld1_x3_<mode>): Likewise
17750 (aarch64_st1x2<VALLDIF:mode>): Likewise
17751 (aarch64_st1_x2_<mode>): Likewise
17752 (aarch64_st1x3<VALLDIF:mode>): Likewise
17753 (aarch64_st1_x3_<mode>): Likewise
17754 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
17755 (vld1_s8_x3): Likewise.
17756 (vld1_u16_x3): Likewise.
17757 (vld1_s16_x3): Likewise.
17758 (vld1_u32_x3): Likewise.
17759 (vld1_s32_x3): Likewise.
17760 (vld1_u64_x3): Likewise.
17761 (vld1_s64_x3): Likewise.
17762 (vld1_f16_x3): Likewise.
17763 (vld1_f32_x3): Likewise.
17764 (vld1_f64_x3): Likewise.
17765 (vld1_p8_x3): Likewise.
17766 (vld1_p16_x3): Likewise.
17767 (vld1_p64_x3): Likewise.
17768 (vld1q_u8_x3): Likewise.
17769 (vld1q_s8_x3): Likewise.
17770 (vld1q_u16_x3): Likewise.
17771 (vld1q_s16_x3): Likewise.
17772 (vld1q_u32_x3): Likewise.
17773 (vld1q_s32_x3): Likewise.
17774 (vld1q_u64_x3): Likewise.
17775 (vld1q_s64_x3): Likewise.
17776 (vld1q_f16_x3): Likewise.
17777 (vld1q_f32_x3): Likewise.
17778 (vld1q_f64_x3): Likewise.
17779 (vld1q_p8_x3): Likewise.
17780 (vld1q_p16_x3): Likewise.
17781 (vld1q_p64_x3): Likewise.
17782 (vst1_s64_x2): Likewise.
17783 (vst1_u64_x2): Likewise.
17784 (vst1_f64_x2): Likewise.
17785 (vst1_s8_x2): Likewise.
17786 (vst1_p8_x2): Likewise.
17787 (vst1_s16_x2): Likewise.
17788 (vst1_p16_x2): Likewise.
17789 (vst1_s32_x2): Likewise.
17790 (vst1_u8_x2): Likewise.
17791 (vst1_u16_x2): Likewise.
17792 (vst1_u32_x2): Likewise.
17793 (vst1_f16_x2): Likewise.
17794 (vst1_f32_x2): Likewise.
17795 (vst1_p64_x2): Likewise.
17796 (vst1q_s8_x2): Likewise.
17797 (vst1q_p8_x2): Likewise.
17798 (vst1q_s16_x2): Likewise.
17799 (vst1q_p16_x2): Likewise.
17800 (vst1q_s32_x2): Likewise.
17801 (vst1q_s64_x2): Likewise.
17802 (vst1q_u8_x2): Likewise.
17803 (vst1q_u16_x2): Likewise.
17804 (vst1q_u32_x2): Likewise.
17805 (vst1q_u64_x2): Likewise.
17806 (vst1q_f16_x2): Likewise.
17807 (vst1q_f32_x2): Likewise.
17808 (vst1q_f64_x2): Likewise.
17809 (vst1q_p64_x2): Likewise.
17810 (vst1_s64_x3): Likewise.
17811 (vst1_u64_x3): Likewise.
17812 (vst1_f64_x3): Likewise.
17813 (vst1_s8_x3): Likewise.
17814 (vst1_p8_x3): Likewise.
17815 (vst1_s16_x3): Likewise.
17816 (vst1_p16_x3): Likewise.
17817 (vst1_s32_x3): Likewise.
17818 (vst1_u8_x3): Likewise.
17819 (vst1_u16_x3): Likewise.
17820 (vst1_u32_x3): Likewise.
17821 (vst1_f16_x3): Likewise.
17822 (vst1_f32_x3): Likewise.
17823 (vst1_p64_x3): Likewise.
17824 (vst1q_s8_x3): Likewise.
17825 (vst1q_p8_x3): Likewise.
17826 (vst1q_s16_x3): Likewise.
17827 (vst1q_p16_x3): Likewise.
17828 (vst1q_s32_x3): Likewise.
17829 (vst1q_s64_x3): Likewise.
17830 (vst1q_u8_x3): Likewise.
17831 (vst1q_u16_x3): Likewise.
17832 (vst1q_u32_x3): Likewise.
17833 (vst1q_u64_x3): Likewise.
17834 (vst1q_f16_x3): Likewise.
17835 (vst1q_f32_x3): Likewise.
17836 (vst1q_f64_x3): Likewise.
17837 (vst1q_p64_x3): Likewise.
17839 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17841 * config/msp430/msp430.c (msp430_output_labelref): Prepend
17842 user_label_prefix to name.
17844 * tree-core.h: Update comment about the format of NAME string
17845 passed to handler in attribute_spec.
17847 * config/msp430/msp430.md: Remove erroneous subreg expression from
17848 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
17849 zero_extend{q,h}isi2.
17851 2018-05-30 Borislav Petkov <bp@suse.de>
17853 * doc/extend.texi: Document some architecture specific
17854 constraints and sort entries.
17856 2018-05-30 Martin Sebor <msebor@redhat.com>
17858 PR middle-end/85369
17859 * builtins.c (expand_builtin_stpcpy_1): New function.
17860 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
17861 only if the former succeeds.
17863 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
17865 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
17868 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17870 * doc/invoke.texi (-flinker-output): Document
17872 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17874 * passes.c (ipa_write_summaries): Only modify statements if body
17876 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
17877 incrementally linking.
17878 (ipa_passes): Likewise.
17879 * lto-cgraph.c (lto_output_node): When incrementally linking do not
17880 pass down resolution info.
17881 * common.opt (flag_incremental_link): Update info.
17882 * gcc.c (plugin specs): Turn flinker-output=* to
17883 -plugin-opt=-linker-output-known
17884 * toplev.c (compile_file): Also cut compilation when doing incremental
17886 * flag-types. (enum lto_partition_model): Add
17887 LTO_LINKER_OUTPUT_NOLTOREL.
17888 (invoke.texi): Add -flinker-output docs.
17889 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
17890 link same way as WPA; do not stream in dead initializers.
17892 * dwarf2out.c (dwarf2out_die_ref_for_decl,
17893 darf2out_register_external_decl): Support incremental link.
17895 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17897 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
17899 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
17901 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
17902 it down to simple_object_copy_lto_debug_sections.
17903 (run_gcc): Determine incremental LTO link time and configure
17904 lto1 into non-wpa mode, disable renaming of debug sections.
17906 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
17908 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
17909 descriptions of various incorrectly documented functions.
17911 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
17914 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
17915 address check not strict.
17917 2018-05-30 Richard Biener <rguenther@suse.de>
17919 PR tree-optimization/85964
17920 * tracer.c (better_p): Drop initialized count check, we only
17921 call the function with initialized counts now.
17922 (find_best_successor): Do find a best edge if one
17923 has uninitialized count.
17924 (find_best_predecessor): Likewise. Do BB frequency check only
17925 if count is initialized.
17927 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
17929 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
17930 (aarch64_ldrstr_offset_compare): New.
17931 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
17932 load/store orderings.
17933 (aarch64_gen_adjusted_ldpstp): Likewise.
17935 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17937 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17938 Check for subset of GENERAL_REGS and FP_REGS.
17939 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
17942 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
17944 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
17945 and wi::to_poly_offset. Add the current offset and then check
17946 whether the sum fits, rather than using an unchecked addition of
17947 a checked term. Check for a shwi rather than a uhwi.
17948 * expr.c (get_bit_range): Use tree_to_poly_uint64.
17949 (store_constructor): Use poly_int_tree_p.
17950 (expand_expr_real_1): Likewise.
17951 * function.c (assign_temp): Likewise.
17952 * fold-const.c (const_binop): Use poly_int_tree_p and
17953 wi::to_poly_offset.
17954 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
17956 * ipa-icf-gimple.c (func_checker::compare_operand): Use
17957 to_poly_offset for MEM offsets.
17958 * ipa-icf.c (sem_variable::equals): Likewise.
17959 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
17960 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
17961 wi::to_poly_offset for BIT_FIELD_REF offsets.
17962 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
17963 wi::to_poly_offset.
17964 * var-tracking.c (emit_note_insn_var_location): Use
17965 tree_to_poly_uint64.
17967 2018-05-29 Jim Wilson <jimw@sifive.com>
17969 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
17971 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
17974 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
17975 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
17977 (sse4_1_round<mode>2): Use nonimmediate_operand
17978 for operand 1 predicate.
17980 2018-05-29 Martin Sebor <msebor@redhat.com>
17981 Richard Biener <rguenther@suse.de>
17984 * calls.c (get_size_range): Call determine_value_range instead
17985 of get_value_range..
17986 * tree-vrp.h (determine_value_range): Declared new function.
17987 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
17989 2018-05-29 Richard Biener <rguenther@suse.de>
17991 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
17992 sure to use non-pattern stmts for get_earlier_stmt arguments.
17993 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
17994 called on pattern stmts.
17995 (get_later_stmt): Likewise.
17997 2018-05-29 Martin Liska <mliska@suse.cz>
17999 PR gcov-profile/85759
18000 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
18003 2018-05-29 Jakub Jelinek <jakub@redhat.com>
18005 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
18007 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
18011 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
18012 VEC_PACK_FLOAT_EXPR): New tree codes.
18013 * tree-pretty-print.c (op_code_prio): Handle
18014 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
18015 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
18016 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
18017 * tree-inline.c (estimate_operator_cost): Likewise.
18018 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
18019 * fold-const.c (const_binop): Likewise.
18020 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
18021 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
18022 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
18023 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
18024 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
18025 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
18026 * expr.c (expand_expr_real_2): Likewise.
18027 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
18028 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
18029 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
18031 * optabs.c (expand_widen_pattern_expr): For
18032 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
18033 sign from result type rather than operand's type.
18034 (expand_binop_directly): For vec_packu_float_optab and
18035 vec_packs_float_optab allow result type to be different from operand's
18037 * optabs-tree.c (optab_for_tree_code): Handle
18038 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
18039 VEC_PACK_FLOAT_EXPR. Formatting fixes.
18040 * tree-vect-generic.c (expand_vector_operations_1): Handle
18041 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
18042 VEC_PACK_FLOAT_EXPR.
18043 * tree-vect-stmts.c (supportable_widening_operation): Handle
18045 (supportable_narrowing_operation): Handle FLOAT_EXPR.
18046 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
18047 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
18048 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
18049 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
18051 (vec_pack<floatprefix>_float_<mode>): New expander.
18052 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
18054 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
18055 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
18056 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
18057 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
18058 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
18060 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
18061 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
18062 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
18063 VEC_PACK_FLOAT_EXPR): Document.
18065 2018-05-29 Richard Biener <rguenther@suse.de>
18067 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
18069 (stmt_vec_info_vec): Make pointer.
18070 (init_stmt_vec_info_vec): Remove.
18071 (free_stmt_vec_info_vec): Likewise.
18072 (set_stmt_vec_info_vec): New function.
18073 (free_stmt_vec_infos): Likewise.
18074 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
18075 (set_vinfo_for_stmt): Likewise.
18076 (get_earlier_stmt): Likewise.
18077 (get_later_stmt): Likewise.
18078 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
18079 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
18080 (vec_info::~vec_info): Free stmt_vec_infos.
18081 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
18082 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
18083 (pass_slp_vectorize::execute): Likewise.
18084 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
18085 (free_stmt_vec_info_vec): Likewise.
18086 (set_stmt_vec_info_vec): New function.
18087 (free_stmt_vec_infos): Likewise.
18088 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
18089 the global stmt_vec_info_vec.
18090 * tree-parloops.c (gather_scalar_reductions): Use
18091 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
18094 2018-05-29 Richard Biener <rguenther@suse.de>
18096 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
18098 2018-05-29 Martin Liska <mliska@suse.cz>
18099 David Malcolm <dmalcolm@redhat.com>
18101 * vec.c (test_reverse): New.
18102 (vec_c_tests): Add new test.
18103 * vec.h (vl_ptr>::reverse): New function.
18105 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
18107 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
18109 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
18112 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
18114 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
18116 2018-05-28 Richard Biener <rguenther@suse.de>
18118 PR tree-optimization/85933
18119 * tree-vect-data-refs.c (vect_record_base_alignments): Only
18120 look at stmts marked as vectorizable.
18122 2018-05-28 Richard Biener <rguenther@suse.de>
18124 PR tree-optimization/85934
18125 * tree-vect-generic.c (expand_vector_operations_1): Hoist
18126 vector boolean check before scalar optimization.
18128 2018-05-28 Jakub Jelinek <jakub@redhat.com>
18130 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
18133 2018-05-28 Mark Wielaard <mark@klomp.org>
18135 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
18136 if it is an expression containing a minus sign.
18138 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
18140 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
18142 2018-05-27 Paul Koning <ni1d@arrl.net>
18144 * config/pdp11/pdp11.md (truncsihi2): Remove.
18146 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
18147 Chung-Ju Wu <jasonwucj@gmail.com>
18149 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
18151 (unaligned_store_dw): Ditto.
18152 * config/nds32/nds32-memory-manipulation.c
18153 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
18154 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
18155 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
18156 (emit_setmem_word_loop): Rename to ...
18157 (emit_setmem_doubleword_loop): ... this.
18158 (nds32_gen_dup_4_byte_to_word_value): New function.
18159 (nds32_gen_dup_8_byte_to_double_word_value): New function.
18160 (nds32_expand_setmem_loop): Refine implementation.
18161 (nds32_expand_setmem_loop_v3m): Ditto.
18162 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
18165 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
18167 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
18169 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
18171 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
18172 (nds32_init_machine_status): Initialize machine->attr_naked_p and
18173 machine->attr_no_prologue_p.
18174 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
18175 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
18176 (nds32_expand_epilogue): Consider attr_naked_p.
18177 (nds32_expand_epilogue_v3pop): Likewise.
18178 (nds32_can_use_return_insn): Likewise.
18179 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
18180 attr_no_prologue_p fields.
18181 * config/nds32/nds32.opt (mret-in-naked-func): New option.
18183 2018-05-27 Jakub Jelinek <jakub@redhat.com>
18186 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
18188 * config/i386/sse.md
18189 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
18191 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
18193 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
18195 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
18197 (*<floatsuffix>floatv2div2sf2): Rename to ...
18198 (*float<floatunssuffix>v2div2sf2): ... this.
18199 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
18200 (float<floatunssuffix>v2div2sf2_mask): ... this.
18201 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
18202 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
18203 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
18205 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
18207 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
18209 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
18211 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
18213 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
18215 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
18216 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
18217 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
18218 gen_ufix_truncv8dfv8si2.
18219 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
18220 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
18221 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
18222 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
18223 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
18224 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
18225 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
18226 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
18228 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
18232 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
18234 2018-05-25 Jim Wilson <jimw@sifive.com>
18236 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
18237 * config/riscv/riscv.c (struct machine_function): Add
18238 interrupt_handler_p and attribute_checked_p fields.
18239 (riscv_attribute_table): Add interrupt.
18240 (riscv_interrupt_type_p): New.
18241 (riscv_save_reg_p): Save extra regs for interrupt handler.
18242 (riscv_use_save_libcall): Return false for interrupt handler.
18243 (riscv_first_stack_step): Add forward declaration.
18244 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
18245 for interrupt handler with large frame. Use it for saved reg list.
18246 (riscv_expand_prologue): Move flag_stack_usage_info support to
18247 eliminate duplication.
18248 (riscv_expand_epilogue): Generate mret for interrupt handler.
18249 (riscv_epilogue_uses): New.
18250 (riscv_can_use_return_insn): Return false for interrupt handler.
18251 (riscv_function_ok_for_sibcall): Likewise.
18252 (riscv_set_current_function): Add interrupt handler support.
18253 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
18254 * config/riscv/riscv.md (UNSPECV_MRET): New.
18256 (riscv_frflags, riscv_fsflags): Use tab after opcode.
18258 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
18260 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
18262 PR tree-optimization/85712
18263 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
18264 this candidate has already been replaced in-situ by a copy.
18266 2018-05-25 Jason Merrill <jason@redhat.com>
18268 PR c++/80485 - inline function non-zero address.
18269 * symtab.c (nonzero_address): Check DECL_COMDAT.
18271 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
18274 * config/alpha/alpha.md (ashlsi3): New insn pattern.
18275 (*ashlsi_se): Rename from *ashldi_se. Define as sign
18276 extension of SImode operation. Use const123_operand predicate.
18277 (*saddsi_1): Remove.
18278 (*saddl_se_1): Ditto.
18279 (*ssubsi_1): Ditto.
18280 (*ssubl_se_1): Ditto.
18281 * config/alpha/predicates.md (const123_operand): New predicate.
18282 * config/alpha/constraints.md (P): Use IN_RANGE.
18284 2018-05-25 Richard Biener <rguenther@suse.de>
18286 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
18288 (ref_maybe_used_by_stmt_p): Likewise.
18289 (stmt_may_clobber_ref_p): Likewise.
18290 (stmt_may_clobber_ref_p_1): Likewise.
18291 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
18293 (ref_maybe_used_by_stmt_p): Likewise.
18294 (stmt_may_clobber_ref_p): Likewise.
18295 (stmt_may_clobber_ref_p_1): Likewise.
18296 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
18297 the alias oracle to disambiguate DRs with stmts DR analysis
18299 (vect_analyze_data_refs): Do not give up on not analyzable
18300 DRs for BB vectorization. Remove code truncating the dataref
18303 2018-05-25 Jakub Jelinek <jakub@redhat.com>
18306 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
18307 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
18308 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
18310 2018-05-25 Richard Biener <rguenther@suse.de>
18312 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
18313 function, combining stmt data ref gathering and fatal analysis
18315 (vect_analyze_data_refs): Remove now redudnant code and simplify.
18316 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
18317 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
18318 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
18319 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
18321 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18323 PR tree-optimization/85720
18324 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
18325 SCC if all partitions are builtins.
18326 (version_loop_by_alias_check): New parameter. Generate cancelable
18327 runtime alias check if all partitions are builtins.
18328 (distribute_loop): Update call to above function.
18330 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18332 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
18333 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
18334 (parm_default_def_partition_arg): Ditto.
18335 (set_parm_default_def_partition): Ditto.
18336 (get_parm_default_def_partitions): Ditto and make it static.
18337 (get_undefined_value_partitions): Ditto and make it static.
18338 (remove_ssa_form): Refactor call to init_var_map here.
18339 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
18340 computation for loop region.
18341 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
18342 (register_default_def): Delete.
18343 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
18344 (parm_default_def_partition_arg): Ditto.
18345 (set_parm_default_def_partition): Ditto.
18346 (get_parm_default_def_partitions): Ditto and make it static.
18347 (get_undefined_value_partitions): Ditto and make it static.
18348 (coalesce_with_default, coalesce_with_default): Update comment.
18349 (create_coalesce_list_for_region): New func factored out from
18350 create_outofssa_var_map.
18351 (populate_coalesce_list_for_outofssa): New func factored out from
18352 create_outofssa_var_map and coalesce_ssa_name.
18353 (create_outofssa_var_map): Delete.
18354 (coalesce_ssa_name): Refactor to support live range computation.
18355 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
18356 (get_parm_default_def_partitions): Delete.
18357 (get_undefined_value_partitions): Ditto.
18358 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
18359 computation for loop region.
18360 (new_tree_live_info, loe_visit_block): Ditto.
18361 (live_worklist, set_var_live_on_entry): Ditto.
18362 (calculate_live_on_exit, verify_live_on_entry): Ditto.
18363 * tree-ssa-live.h (struct _var_map): New fields.
18364 (init_var_map): Change decl.
18365 (region_contains_p): New.
18367 2018-05-25 Bin Cheng <bin.cheng@arm.com>
18369 * tree-ssa-live.h (live_merge_and_clear): Delete.
18371 2018-05-25 Richard Biener <rguenther@suse.de>
18374 * tree-dump.c (dequeue_and_dump): Remove access to removed
18375 operand 2 of a SWITCH_EXPR.
18377 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18379 * doc/sourcebuild.texi (vect_double_cond_arith): Include
18380 multiplication and division.
18381 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
18382 (cond_udiv@var{m}, cond_umod@var{m}): Document.
18383 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
18384 (cond_udiv_optab, cond_umod_optab): New optabs.
18385 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
18386 (IFN_COND_RDIV): New internal functions.
18387 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
18388 TRUNC_MOD_EXPR and RDIV_EXPR.
18389 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
18390 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
18392 (SVE_INT_BINARY): Include mult.
18393 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
18394 (optab, sve_int_op): Handle mult.
18395 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
18397 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
18398 for SVE_INT_BINARY_SD.
18400 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18402 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
18403 (optab, sve_int_op): Handle div and udiv.
18404 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
18405 for SVE_INT_BINARY_SD.
18406 (*<optab><mode>3): New insn for the same.
18408 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18410 * tree-vect-patterns.c: Include predict.h.
18411 (vect_recog_divmod_pattern): Restrict check for division support
18412 to when optimizing for size.
18414 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18416 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
18417 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
18418 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
18419 (gimple_match_op::set_op): Likewise.
18420 (gimple_resimplify4): Declare.
18421 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
18422 (expr::gen_transform): Likewise.
18423 (decision_tree::gen): Generate a simplification routine for 4 operands.
18424 * gimple-match-head.c (gimple_simplify): Add an overload for
18425 4 operands. In the top-level function, handle up to 4 call
18426 arguments and call gimple_resimplify4.
18427 (gimple_resimplify4): New function.
18428 (build_call_internal): Pass a fourth operand.
18429 (maybe_push_to_seq): Likewise.
18430 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
18431 Fold VEC_COND_EXPRs of an operation and a default value into
18432 an IFN_COND_* function if possible.
18433 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
18435 (SVE_COND_FP_BINARY): Include them.
18436 (optab, sve_fp_op): Handle them.
18437 (SVE_INT_BINARY_REV): New code iterator.
18438 (SVE_COND_FP_BINARY_REV): New int iterator.
18439 (commutative): New int attribute.
18440 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
18442 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
18444 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
18445 (*cond_<optab><mode>): New patterns for reversed operands.
18447 2018-05-25 Richard Biener <rguenther@suse.de>
18449 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
18450 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
18451 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
18452 (STMT_VINFO_GROUPED_ACCESS): Adjust.
18453 * tree-vect-data-refs.c (everywhere): Adjust users.
18454 * tree-vect-loop.c (everywhere): Likewise.
18455 * tree-vect-slp.c (everywhere): Likewise.
18456 * tree-vect-stmts.c (everywhere): Likewise.
18457 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
18459 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
18461 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
18463 (gcc_cv_as_section_exclude): ... this.
18464 Try Solaris as #exclude syntax.
18465 * configure: Regenerate.
18466 * config.in: Regenerate.
18467 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
18469 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
18470 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
18472 * varasm.c (default_elf_asm_named_section): Don't check if
18473 HAVE_GAS_SECTION_EXCLUDE is defined.
18475 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18477 * doc/md.texi: Update the documentation of the cond_* optabs
18478 to mention the new final operand. Fix GET_MODE_NUNITS call.
18479 Describe the scalar case too.
18480 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
18481 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
18483 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
18484 (get_conditional_internal_fn): Update comment.
18485 * tree-vect-loop.c (vectorizable_reduction): Pass the original
18486 accumulator value as a final argument to conditional functions.
18487 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
18488 a define_expand and add an "else" operand. Assert for now that
18489 the else operand is equal to operand 2. Use SVE_INT_BINARY and
18490 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
18491 (*cond_<optab><mode>): New patterns.
18492 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
18493 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
18494 (UNSPEC_COND_EOR): Delete.
18495 (optab): Remove associated mappings.
18496 (SVE_INT_BINARY): New code iterator.
18497 (sve_int_op): Remove int attribute and add "minus" to the code
18499 (SVE_COND_INT_OP): Delete.
18500 (SVE_COND_FP_OP): Rename to...
18501 (SVE_COND_FP_BINARY): ...this.
18503 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
18505 * optabs.c (can_reuse_operands_p): New function.
18506 (maybe_legitimize_operands): Try to reuse the results for
18509 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
18511 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
18512 Add {q} suffix to insn mnemonic.
18514 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18516 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
18517 (msp430_warn_func_return): New.
18519 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
18521 * fold-const.c (tree_nonzero_bits): New function.
18522 * fold-const.h (tree_nonzero_bits): Likewise.
18523 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
18524 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
18526 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
18530 * varasm.c (assemble_alias): Check ifunc_resolver only on
18533 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
18536 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
18537 when memory input operand is handled.
18539 2018-05-24 Luis Machado <luis.machado@linaro.org>
18541 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
18543 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
18545 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18547 * match.pd: Delay FMA folds until after vectorization.
18549 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
18552 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
18553 address check not strict.
18555 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
18557 * gimple-match.h (gimple_match_op): New class.
18558 (mprts_hook): Replace parameters with a gimple_match_op *.
18559 (maybe_build_generic_op): Likewise.
18560 (gimple_simplified_result_is_gimple_val): Replace parameters with
18561 a const gimple_match_op *.
18562 (gimple_simplify): Replace code_helper * and tree * parameters with
18563 a gimple_match_op * parameter.
18564 (gimple_resimplify1): Replace code_helper *, tree and tree *
18565 parameters with a gimple_match_op * parameter.
18566 (gimple_resimplify2): Likewise.
18567 (gimple_resimplify3): Likewise.
18568 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
18569 parameters with a gimple_match_op * parameter.
18570 * gimple-match-head.c (gimple_simplify): Change prototypes of
18571 auto-generated functions to take a gimple_match_op * instead of
18572 separate code_helper * and tree * parameters. Make the same
18573 change in the top-level overload and update calls to the
18574 gimple_resimplify routines. Update calls to the auto-generated
18575 functions and to maybe_push_res_to_seq in the publicly-facing
18576 operation-specific gimple_simplify overloads.
18577 (gimple_match_op::MAX_NUM_OPS): Define.
18578 (gimple_resimplify1): Replace rcode and ops with a single res_op
18579 parameter. Update call to gimple_simplify.
18580 (gimple_resimplify2): Likewise.
18581 (gimple_resimplify3): Likewise.
18582 (mprts_hook): Replace parameters with a gimple_match_op *.
18583 (maybe_build_generic_op): Likewise.
18584 (build_call_internal): Replace type, nargs and ops with
18585 a gimple_match_op *.
18586 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
18587 with a single gimple_match_op *. Update calls to mprts_hook,
18588 build_call_internal and gimple_simplified_result_is_gimple_val.
18589 Factor out code that is common to the tree_code and combined_fn cases.
18590 * genmatch.c (expr::gen_transform): Replace tem_code and
18591 tem_ops with a gimple_match_op called tem_op. Update calls
18592 to the gimple_resimplify functions and maybe_push_res_to_seq.
18593 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
18594 res_ops. Update call to the gimple_resimplify functions.
18595 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
18596 (decision_tree::gen): Make the functions take a gimple_match_op *
18597 called res_op instead of separate res_code and res_ops parameters.
18598 Update call accordingly.
18599 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
18600 and ops with a single res_op parameter. Update calls to
18601 maybe_build_generic_op and maybe_push_res_to_seq.
18602 (fold_stmt_1): Update calls to gimple_simplify and
18603 replace_stmt_with_simplification.
18604 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
18605 and gimple_simplified_result_is_gimple_val.
18606 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
18608 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
18609 with a gimple_match_op *.
18610 (vn_nary_build_or_lookup): Likewise. Update call to
18611 vn_nary_build_or_lookup_1.
18612 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
18613 gimple_match_op *. Update calls to the gimple_resimplify routines
18614 and to gimple_simplified_result_is_gimple_val.
18615 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
18616 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
18617 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
18618 (visit_nary_op): Likewise.
18619 (visit_reference_op_load): Likewise.
18621 2018-05-23 Luis Machado <luis.machado@linaro.org>
18623 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
18624 modifier for printing the step amount.
18626 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18629 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
18632 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
18634 * doc/sourcebuild.texi (Endianness): New subsubsection.
18636 2018-05-23 Luis Machado <luis.machado@linaro.org>
18638 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18639 <prefetch_dynamic_strides>: New const bool field.
18640 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18641 prefetch_dynamic_strides.
18642 (exynosm1_prefetch_tune): Likewise.
18643 (thunderxt88_prefetch_tune): Likewise.
18644 (thunderx_prefetch_tune): Likewise.
18645 (thunderx2t99_prefetch_tune): Likewise.
18646 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
18648 (aarch64_override_options_internal): Update to set
18649 PARAM_PREFETCH_DYNAMIC_STRIDES.
18650 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
18651 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
18652 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
18653 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
18654 prefetch-dynamic-strides setting.
18656 2018-05-23 Luis Machado <luis.machado@linaro.org>
18658 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
18659 <minimum_stride>: New const int field.
18660 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
18661 minimum_stride field defaulting to -1.
18662 (exynosm1_prefetch_tune): Likewise.
18663 (thunderxt88_prefetch_tune): Likewise.
18664 (thunderx_prefetch_tune): Likewise.
18665 (thunderx2t99_prefetch_tune): Likewise.
18666 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
18667 <default_opt_level>: Set to 3.
18668 (aarch64_override_options_internal): Update to set
18669 PARAM_PREFETCH_MINIMUM_STRIDE.
18670 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
18671 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
18672 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
18673 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
18674 stride is constant and is below the minimum stride threshold.
18676 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18678 * config/arm/arm-cpus.in (mode26): Delete.
18679 (armv4): Delete mode26 reference.
18680 * config/arm/arm.c (arm_configure_build_target): Delete use of
18683 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
18685 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
18687 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
18688 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
18690 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
18691 (floatunsdidf2): Ditto.
18693 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
18695 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
18696 (fixuns_trunc<mode>si2_avx512f): Ditto.
18697 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
18698 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
18699 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
18701 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
18703 PR rtl-optimization/79985
18704 * df-scan.c (df_insn_refs_collect): Remove special case for
18705 global registers and asm statements.
18707 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
18709 * extend.texi (Global Register Variables): Rewrite the bullet list.
18710 Note that the register is available for allocation. Note that access
18711 via inline asm must use constraints. Add note about async-signal
18712 handlers. Remove paragraph about automagic register selection.
18714 2018-05-23 Richard Biener <rguenther@suse.de>
18716 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
18717 of fixed offset from memset VN.
18719 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
18721 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
18722 first_interp field.
18723 (alloc_cand_and_find_basis): Initialize first_interp field.
18724 (slsr_process_mul): Modify first_interp field.
18725 (slsr_process_add): Likewise.
18726 (slsr_process_cast): Modify first_interp field for each new
18728 (slsr_process_copy): Likewise.
18729 (dump_candidate): Dump first_interp field.
18730 (replace_mult_candidate): Process all interpretations, not just
18732 (replace_rhs_if_not_dup): Likewise.
18733 (replace_one_candidate): Likewise.
18735 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
18737 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
18739 (aarch64_needs_frame_chain): New function.
18740 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
18742 2018-05-23 Sudakshina Das <sudi.das@arm.com>
18745 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
18746 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
18747 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
18748 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
18749 as true for strict-align.
18750 (aarch64_can_inline_p): Perform checks even when callee has no
18751 attributes to check for strict alignment.
18752 * doc/extend.texi (AArch64 Function Attributes): Document
18754 * doc/invoke.texi: (AArch64 Options): Likewise.
18756 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
18758 PR tree-optimization/85853
18759 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
18760 the handling of the root of the node to...
18761 (vect_slp_analyze_node_operations_1): ...this new function,
18762 and run the whole thing with the child nodes' def types
18763 set according to their SLP node's def type.
18765 2018-05-23 Richard Biener <rguenther@suse.de>
18767 PR middle-end/85874
18768 * tree-data-ref.c (create_runtime_alias_checks): Defer
18769 and ignore overflow warnings.
18771 2018-05-23 Yury Gribov <tetra2005@gmail.com>
18773 PR tree-optimization/85822
18774 * tree-vrp.c (is_masked_range_test): Fix handling of negative
18777 2018-05-23 Richard Biener <rguenther@suse.de>
18779 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
18780 memset constants via native_interpret_expr.
18782 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
18785 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
18787 (cgraph_node::create_alias): Likewise.
18788 (cgraph_node::get_availability): Check ifunc_resolver instead
18789 of looking up ifunc attribute.
18790 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
18791 * varasm.c (do_assemble_alias): Likewise.
18792 (assemble_alias): Likewise.
18793 (default_binds_local_p_3): Likewise.
18794 * cgraph.h (cgraph_node): Add ifunc_resolver.
18795 (cgraph_node::only_called_directly_or_aliased_p): Return false
18796 for IFUNC resolver.
18797 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
18799 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
18800 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
18801 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
18802 instead of looking up ifunc attribute.
18804 2018-05-22 Luis Machado <luis.machado@linaro.org>
18806 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
18808 2018-05-22 Martin Sebor <msebor@redhat.com>
18810 PR middle-end/85359
18811 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
18812 only when expasion succeeds.
18813 (expand_builtin_strcmp): Same.
18814 (expand_builtin_strncmp): Same.
18816 2018-05-22 Martin Sebor <msebor@redhat.com>
18818 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
18820 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
18821 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18823 * config/aarch64/aarch64-ldpstp.md: Replace uses of
18824 aarch64_mem_pair_operand with memory_operand and delete operand swapping
18826 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
18827 Add check for legitimate_address.
18828 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
18829 (aarch64_swap_ldrstr_operands): New.
18830 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
18833 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
18834 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18836 * config/aarch64/aarch64.md: New patterns to generate stp
18838 (store_pair_sw, store_pair_dw): New patterns to generate stp for
18839 single words and double words.
18840 (load_pair_sw, load_pair_dw): Likewise.
18841 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
18843 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
18845 * config/aarch64/aarch64-ldpstp.md: Modify peephole
18846 for different mode ldpstp and add peephole for merged zero stores.
18847 Likewise for loads.
18848 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
18850 (aarch64_gen_store_pair): Rename calls to match new patterns.
18851 (aarch64_gen_load_pair): Rename calls to match new patterns.
18852 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
18853 (load_pair<DREG:mode><DREG2:mode>): ... This.
18854 (store_pair<mode>): Rename to...
18855 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
18856 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
18857 New mode iterators.
18858 (V_INT_EQUIV): Handle SImode.
18859 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
18862 2018-05-22 Martin Sebor <msebor@redhat.com>
18865 * calls.c (maybe_warn_nonstring_arg): Use string length to set
18866 or ajust the presumed bound on an operation to avoid unnecessary
18869 2018-05-22 Martin Sebor <msebor@redhat.com>
18871 PR tree-optimization/85826
18872 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
18873 assuming that a DECL necesarily has a constant size.
18875 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
18877 PR middle-end/85862
18878 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
18880 2018-05-22 Richard Biener <rguenther@suse.de>
18882 PR tree-optimization/85834
18883 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
18884 non-constant and non-zero memset arguments.
18886 2018-05-22 Martin Liska <mliska@suse.cz>
18889 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
18891 2018-05-22 Richard Biener <rguenther@suse.de>
18893 PR tree-optimization/85863
18894 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
18895 comparisons when vectype is specified.
18896 (vectorizable_condition): Do not specify vectype for
18897 vect_is_simple_cond when SLP vectorizing.
18899 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
18902 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
18903 define __ibm128 as long double.
18904 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
18905 as a distinct type when IEEE 128-bit support is enabled.
18906 (init_float128_ieee): Fix up conversions between IFmode and IEEE
18907 128-bit types to use the correct functions.
18908 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
18909 convert between 128-bit floating point types that have different
18910 modes but the same representation, instead of using gen_lowpart to
18912 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
18914 (IFKF_reg): New attributes to give the register constraints for
18916 (extend<mode>tf2_internal): New insns to mark an explicit
18917 conversion between 128-bit floating point types that have a
18918 different mode but share the same representation.
18920 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
18922 PR tree-optimization/85814
18923 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
18924 a null return from get_strinfo when unsharing the next
18925 strinfo in the chain.
18927 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
18930 * varasm.c (weak_finish): Clean up weak_decls.
18932 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18934 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
18935 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
18936 UNSPEC_UADALP values.
18937 * config/aarch64/iterators.md (ABAL): New int iterator.
18940 (sur): Add mappings for the above.
18941 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
18943 (aarch64_<sur>abal<mode>_4): Likewise.
18944 (aarch64_<sur>adalp<mode>_3): Likewise.
18945 (<sur>sadv16qi): New define_expand.
18947 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
18949 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
18950 (*movdf_internal): Ditto.
18951 (*rcpsf2_sse): Ditto.
18952 (*rsqrtsf2_sse): Ditto.
18953 (*sqrt<mode>2_sse): Ditto.
18955 2018-05-21 Tamar Christina <tamar.christina@arm.com>
18957 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
18959 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
18960 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
18961 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
18962 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
18964 * config/aarch64/arm_neon.h: Likewise.
18965 * config/aarch64/iterators.md (VQ_I): New.
18967 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
18969 * config.gcc: Add arc/t-multilib-linux to tmake_file for
18971 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
18974 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
18976 * config/nds32/constraints.md (S): New constraint.
18977 * config/nds32/nds32.md (call_internal): Use constraint S.
18978 (call_value_internal): Likewise.
18979 (sibcall_internal): Likewise.
18980 (sibcall_value_internal): Likewise.
18982 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
18983 Chung-Ju Wu <jasonwucj@gmail.com>
18985 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
18986 into consideration.
18988 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
18989 Chung-Ju Wu <jasonwucj@gmail.com>
18991 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
18992 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
18993 (nds32_rtx_costs_impl): Simplify.
18994 (nds32_address_cost_impl): Simplify.
18995 (nds32_init_rtx_costs): New function.
18996 (nds32_rtx_costs_speed_prefer): Likewise.
18997 (nds32_rtx_costs_size_prefer): Likewise.
18998 (nds32_address_cost_speed_prefer): Likewise.
18999 (nds32_address_cost_speed_fwprop): Likewise.
19000 (nds32_address_cost_size_prefer): Likewise.
19001 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
19002 * config/nds32/nds32.c (nds32_option_override): Use
19003 nds32_init_rtx_costs function.
19005 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
19007 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
19008 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
19009 (TARGET_PIPELINE_N8): Likewise.
19010 (TARGET_PIPELINE_N10): Likewise.
19011 (TARGET_PIPELINE_N13): Likewise.
19012 (TARGET_PIPELINE_GRAYWOLF): Likewise.
19014 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
19016 * config/nds32/nds32-fpu.md: Update copyright year.
19018 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19020 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
19022 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19024 * config/nds32/nds32.c
19025 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
19026 * config/nds32/nds32.opt (minline-asm-r15): New option.
19028 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19030 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
19032 * config/nds32/nds32.md (abssi2): New pattern.
19034 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
19036 * config/i386/i386.md (rex64namesuffix): New mode attribute.
19037 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
19038 Merge insn pattern from sse_cvtsi2ss<round_name> and
19039 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
19040 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
19041 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
19042 using SWI48 mode iterator.
19043 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
19044 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
19045 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
19046 pattern from sse_cvttss2si<round_saeonly_name>
19047 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
19048 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
19049 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
19050 using SWI48 mode iterator.
19051 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
19052 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
19053 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
19054 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
19055 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
19056 using SWI48 mode iterator.
19057 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
19058 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
19059 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
19060 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
19061 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
19062 SWI48 mode iterator.
19063 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
19064 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
19065 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
19066 pattern from sse_cvttsd2si<round_saeonly_name>
19067 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
19069 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19071 * config/nds32/nds32-md-auxiliary.c
19072 (nds32_valid_smw_lwm_base_p): Refine.
19073 (nds32_output_smw_single_word): Refine.
19074 (nds32_output_smw_double_word): New.
19075 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
19077 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19079 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
19080 (nds32_output_stack_pop): Refine.
19081 (nds32_expand_unaligned_load): Refine.
19082 (nds32_expand_unaligned_store): Refine.
19084 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19085 Chung-Ju Wu <jasonwucj@gmail.com>
19087 * config/nds32/constants.md: Add TP_REGNUM constant.
19088 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
19089 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
19091 * config/nds32/nds32-doubleword.md: Consider flag_pic.
19092 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
19093 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
19094 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
19095 and PIC code generation.
19096 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
19098 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
19100 * config/nds32/nds32.md: Support TLS and PIC.
19101 * config/nds32/nds32.c: Support TLS and PIC.
19102 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
19103 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
19106 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
19108 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
19109 mode with E_ prefix.
19111 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
19112 Chung-Ju Wu <jasonwucj@gmail.com>
19114 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
19115 * config/nds32/nds32-md-auxiliary.c
19116 (symbolic_reference_mentioned_p): New.
19117 (nds32_legitimize_ict_address): New.
19118 (nds32_expand_ict_move): New.
19119 (nds32_indirect_call_referenced_p): New.
19120 (nds32_symbol_binds_local_p): Delete.
19121 (nds32_long_call_p): Modify.
19122 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
19123 * config/nds32/nds32-protos.h
19124 (symbolic_reference_mentioned_p): Declare.
19125 (nds32_legitimize_ict_address): Declare.
19126 (nds32_expand_ict_move): Declare.
19127 (nds32_indirect_call_referenced_p): Declare.
19128 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
19129 (nds32_relax_group): Use nds32_ict_const_p as condition.
19130 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
19131 (nds32_asm_file_start): Output ict_model directive in asm code.
19132 (nds32_legitimate_address_p): Consider indirect call.
19133 (nds32_print_operand): Consider indirect call.
19134 (nds32_print_operand_address): Consider indirect call.
19135 (nds32_insert_attributes): Handle "indirect_call" attribute.
19136 (TARGET_LEGITIMATE_ADDRESS_P): Define.
19137 (TARGET_LEGITIMATE_CONSTANT_P): Define.
19138 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
19139 (TARGET_DELEGITIMIZE_ADDRESS): Define.
19140 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
19141 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
19142 (TARGET_ICT_MODEL_SMALL): Define.
19143 (TARGET_ICT_MODEL_LARGE): Define.
19144 * config/nds32/nds32.md (movsi): Consider ict model.
19145 (call, call_value): Consider ict model.
19146 (sibcall, sibcall_value): Consider ict model.
19147 * config/nds32/nds32.opt (mict-model): New option.
19148 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
19151 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
19152 Monk Chiang <sh.chiang04@gmail.com>
19153 Jim Wilson <jimw@sifive.com>
19155 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
19156 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
19157 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
19158 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
19159 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
19160 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
19161 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
19162 compute save_libcall_adjustment properly.
19163 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
19164 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
19165 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
19166 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
19167 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
19168 (ABI_SPEC): Handle mabi=ilp32e.
19169 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
19170 (RVE): Add RVE mask.
19171 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
19172 <-march>: Add rv32e as an example.
19174 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
19177 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
19178 (intra_create_variable_infos): Handle C++ constructors.
19180 2018-05-18 Martin Liska <mliska@suse.cz>
19182 * passes.def: Remove a redundant pass.
19184 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
19187 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
19189 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19191 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
19193 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
19195 (armv2, armv2a, armv3, armv3m): Delete architectures.
19196 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
19197 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
19198 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
19200 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
19201 (*mulsidi3adddi): Likewise.
19202 (mulsidi3): Likewise.
19203 (*mulsidi3_nov6): Likewise.
19204 (umulsidi3): Likewise.
19205 (umulsidi3_nov6): Likewise.
19206 (umaddsidi4): Likewise.
19207 (*umulsidi3adddi): Likewise.
19208 (smulsi3_highpart): Likewise.
19209 (*smulsi3_highpart_nov6): Likewise.
19210 (umulsi3_highpart): Likewise.
19211 (*umulsi3_highpart_nov6): Likewise.
19212 * config/arm/arm.h (arm_arch3m): Delete.
19213 * config/arm/arm.c (arm_arch3m): Delete.
19214 (arm_option_override_internal): Update armv3-related comment.
19215 (arm_configure_build_target): Delete use of isa_bit_mode32.
19216 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
19217 (arm_rtx_costs_internal): Delete check of arm_arch3m.
19218 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
19219 (mulsa3): Likewise.
19220 (mulusa3): Likewise.
19221 * config/arm/arm-protos.h (arm_arch3m): Delete.
19222 * config/arm/arm-tables.opt: Regenerate.
19223 * config/arm/arm-tune.md: Likewise.
19224 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
19225 deleted architectures.
19227 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19229 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
19230 (armv5t, armv5te): New features.
19231 (ARMv5, ARMv5e): Delete fgroups.
19232 (ARMv5t, ARMv5te): Adjust for above changes.
19233 (ARMv6m): Likewise.
19234 (armv5, armv5e): Delete arches.
19235 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
19237 (*call_reg_arm): Likewise.
19238 (*call_value_reg_armv5): Likewise.
19239 (*call_value_reg_arm): Likewise.
19240 (*call_symbol): Likewise.
19241 (*call_value_symbol): Likewise.
19242 (*sibcall_insn): Likewise.
19243 (*sibcall_value_insn): Likewise.
19244 (clzsi2): Likewise.
19245 (prefetch): Likewise.
19246 (define_split and define_peephole2 dependent on arm_arch5):
19248 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
19250 (TARGET_ARM_QBIT): Likewise.
19251 (TARGET_DSP_MULTIPLY): Likewise.
19252 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
19253 (arm_arch5, arm_arch5e): Delete.
19254 (arm_arch5t, arm_arch5te): Declare.
19255 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
19256 (arm_arch5t): Declare.
19257 (arm_option_reconfigure_globals): Update for the above.
19258 (arm_options_perform_arch_sanity_checks): Update comment, replace
19259 use of arm_arch5 with arm_arch5t.
19260 (use_return_insn): Likewise.
19261 (arm_emit_call_insn): Likewise.
19262 (output_return_instruction): Likewise.
19263 (arm_final_prescan_insn): Likewise.
19264 (arm_coproc_builtin_available): Likewise.
19265 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
19266 arm_arch5e with arm_arch5t and arm_arch5te.
19267 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
19268 (arm_arch5t, arm_arch5te): Declare.
19269 * config/arm/arm-tables.opt: Regenerate.
19270 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
19271 * config/arm/t-multilib: Likewise.
19272 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
19273 instead of arm_arch5.
19274 (*call_reg_thumb1): Likewise.
19275 (*call_value_reg_thumb1_v5): Likewise.
19276 (*call_value_reg_thumb1): Likewise.
19277 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
19279 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
19281 2018-05-18 Martin Liska <mliska@suse.cz>
19283 PR gcov-profile/84846
19284 * doc/gcov.texi: Document -t option of gcov tool.
19286 2018-05-18 Martin Liska <mliska@suse.cz>
19288 PR gcov-profile/84846
19289 * gcov.c (print_usage): Add new -t option.
19290 (process_args): Handle the option.
19291 (generate_results): Use stdout as output when requested by
19294 2018-05-18 Martin Liska <mliska@suse.cz>
19296 PR gcov-profile/84846
19297 * coverage.c (coverage_init): Write PWD to .gcno file.
19298 * doc/gcov.texi: Document how working directory is printed.
19299 * gcov-dump.c (dump_gcov_file): Print PWD.
19300 * gcov.c (output_intermediate_file): Likewise.
19301 (read_graph_file): Read PWD string.
19302 (output_lines): Print PWD.
19304 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19306 PR middle-end/85817
19307 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
19308 for retval and return false if all args to phi are zero.
19310 2018-05-18 Richard Biener <rguenther@suse.de>
19312 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
19314 (evrp_dom_walker::before_dom_children): Call it.
19316 2018-05-18 Richard Biener <rguenther@suse.de>
19318 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
19319 results when processing array refs with variable index.
19321 2018-05-18 Toon Moene <toon@moene.org>
19323 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
19324 directly after that of -floop-interchange. Indicate that both
19325 options are enabled by default when specifying -O3.
19327 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19329 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
19330 iterator. Delete separate integer-mode vec_set<mode> expander.
19331 (aarch64_simd_vec_setv2di): Delete.
19332 (vec_setv2di): Delete.
19333 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
19334 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
19335 the "w, r" alternative.
19337 2018-05-18 Martin Liska <mliska@suse.cz>
19339 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
19340 * tree-pass.h (make_pass_lower_switch_O0): New function.
19341 * tree-switch-conversion.c (node_has_low_bound): Remove.
19342 (node_has_high_bound): Likewise.
19343 (node_is_bounded): Likewise.
19344 (class pass_lower_switch): Make it a template type and create
19346 (pass_lower_switch::execute): Add template argument.
19347 (make_pass_lower_switch): New function.
19348 (make_pass_lower_switch_O0): New function.
19349 (do_jump_if_equal): Remove.
19350 (emit_case_nodes): Simplify to just handle all 3 cases and leave
19351 all the hard work to tree optimization passes.
19353 2018-05-18 Martin Liska <mliska@suse.cz>
19355 * dbgcnt.c (limit_low): Renamed from limit.
19356 (limit_high): New variable.
19357 (dbg_cnt_is_enabled): Check for upper limit.
19358 (dbg_cnt): Adjust dumping.
19359 (dbg_cnt_set_limit_by_index): Add new argument for high
19361 (dbg_cnt_set_limit_by_name): Likewise.
19362 (dbg_cnt_process_single_pair): Parse new format.
19363 (dbg_cnt_process_opt): Use strtok.
19364 (dbg_cnt_list_all_counters): Remove 'value' and add
19366 * doc/invoke.texi: Document changes.
19368 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
19370 * doc/sourcebuild.texi (scalar_all_fma): Document.
19371 * tree.def (FMA_EXPR): Delete.
19372 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
19373 * internal-fn.c (ternary_direct): New macro.
19374 (expand_ternary_optab_fn): Likewise.
19375 (direct_ternary_optab_supported_p): Likewise.
19376 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
19377 * builtins.c (fold_builtin_fma): Delete.
19378 (fold_builtin_3): Don't call it.
19379 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
19380 * expr.c (expand_expr_real_2): Likewise.
19381 * fold-const.c (operand_equal_p): Likewise.
19382 (fold_ternary_loc): Likewise.
19383 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
19384 * gimple.c (DEFTREECODE): Likewise.
19385 * gimplify.c (gimplify_expr): Likewise.
19386 * optabs-tree.c (optab_for_tree_code): Likewise.
19387 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19388 * tree-eh.c (operation_could_trap_p): Likewise.
19389 (stmt_could_throw_1_p): Likewise.
19390 * tree-inline.c (estimate_operator_cost): Likewise.
19391 * tree-pretty-print.c (dump_generic_node): Likewise.
19392 (op_code_prio): Likewise.
19393 * tree-ssa-loop-im.c (stmt_cost): Likewise.
19394 * tree-ssa-operands.c (get_expr_operands): Likewise.
19395 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
19396 * fold-const-call.h (fold_fma): Delete.
19397 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
19398 CFN_FNMA and CFN_FNMS.
19399 (fold_fma): Delete.
19400 * genmatch.c (combined_fn): New enum.
19401 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
19402 (commutative_op): New function.
19403 (commutate): Use it. Handle more than 2 operands.
19404 (dt_operand::gen_gimple_expr): Use commutative_op.
19405 (parser::parse_expr): Allow :c to be used with non-binary
19406 operators if the commutative operand is known.
19407 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
19408 CFN_FMS, CFN_FNMA and CFN_FNMS.
19409 (backprop::process_assign_use): Remove FMA_EXPR handling.
19410 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
19411 (gen_hsa_fma): New function.
19412 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
19413 IFN_FNMA and IFN_FNMS.
19414 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
19415 * gimple-fold.h (follow_all_ssa_edges): Declare.
19416 * gimple-fold.c (follow_all_ssa_edges): New function.
19417 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
19418 gimple_build interface and use follow_all_ssa_edges to fold the result.
19419 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
19420 instead of checking for optabs directly.
19421 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
19422 rather than FMA_EXPRs.
19423 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
19424 call to IFN_FMA instead of an FMA_EXPR.
19426 2018-05-17 Jim Wilson <jimw@sifive.com>
19428 * expr.c (do_tablejump): When converting index to Pmode, if we have a
19429 sign extended promoted subreg, and the range does not have the sign bit
19430 set, then do a sign extend.
19432 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
19433 test, check for sign extended subreg and/or constant operands, and
19434 do a sign extend in that case.
19436 2018-05-17 Steve Ellcey <sellcey@cavium.com>
19438 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
19439 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
19441 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
19442 Change logics_shift_reg to logics_shift_imm.
19443 (thunderx2t99_fp_loadpair_basic): Delete.
19444 (thunderx2t99_fp_storepair_basic): Delete.
19445 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
19446 (thunderx2t99_asimd_polynomial): Delete.
19447 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
19448 and neon_fp_mul_d_scalar_q.
19449 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
19450 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
19451 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
19452 (thunderx2t99_asimd_lut): Add missing tbl types.
19453 (thunderx2t99_asimd_ext): Delete.
19454 (thunderx2t99_asimd_load1_1_mult): Delete.
19455 (thunderx2t99_asimd_load1_2_mult): Delete.
19456 (thunderx2t99_asimd_load1_ldp): New.
19457 (thunderx2t99_asimd_load1): New.
19458 (thunderx2t99_asimd_load2): Add missing *load2* types.
19459 (thunderx2t99_asimd_load3): New.
19460 (thunderx2t99_asimd_load4): New.
19461 (thunderx2t99_asimd_store1_1_mult): Delete.
19462 (thunderx2t99_asimd_store1_2_mult): Delete.
19463 (thunderx2t99_asimd_store2_mult): Delete.
19464 (thunderx2t99_asimd_store2_onelane): Delete.
19465 (thunderx2t99_asimd_store_stp): New.
19466 (thunderx2t99_asimd_store1): New.
19467 (thunderx2t99_asimd_store2): New.
19468 (thunderx2t99_asimd_store3): New.
19469 (thunderx2t99_asimd_store4): New.
19471 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
19473 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
19474 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
19476 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
19477 Segher Boessenkool <segher@kernel.crashing.org>
19480 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
19483 2018-05-17 Richard Biener <rguenther@suse.de>
19485 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
19486 for pruning loop and prune defs feeding only already visited PHIs.
19488 2018-05-17 Richard Biener <rguenther@suse.de>
19490 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
19492 2018-05-17 Bin Cheng <bin.cheng@arm.com>
19493 Richard Biener <rguenther@suse.de>
19495 PR tree-optimization/85793
19496 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
19497 for VMAT_ELEMENTWISE.
19499 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
19501 * internal-fn.h (lookup_internal_fn): Declare
19502 * internal-fn.c (lookup_internal_fn): New function.
19503 * gimple.c (gimple_build_call_from_tree): Handle calls to
19504 internal functions.
19505 * gimple-pretty-print.c (dump_gimple_call): Print "." before
19506 internal function names.
19507 * tree-pretty-print.c (dump_generic_node): Likewise.
19508 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
19510 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
19512 * gimple-fold.h (gimple_build): Make the function forms take
19513 combined_fn rather than built_in_function.
19514 (gimple_simplify): Likewise.
19515 * gimple-match-head.c (gimple_simplify): Likewise.
19516 * gimple-fold.c (gimple_build): Likewise.
19517 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
19518 rather than gimple_build_call_internal.
19519 (get_initial_defs_for_reduction): Likewise.
19520 (vect_create_epilog_for_reduction): Likewise.
19521 (vectorizable_live_operation): Likewise.
19523 2018-05-17 Martin Liska <mliska@suse.cz>
19525 * gimple-ssa-sprintf.c (format_directive): Do not use
19526 space in between 'G_' and '('.
19528 2018-05-17 Jakub Jelinek <jakub@redhat.com>
19531 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
19532 even if the mask is not all ones.
19535 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
19537 (ix86_gimple_fold_builtin): Likewise.
19540 * config/i386/i386.c: Include tree-vector-builder.h.
19541 (ix86_vector_shift_count): New function.
19542 (ix86_fold_builtin): Fold shift builtins by scalar count.
19543 (ix86_gimple_fold_builtin): Likewise.
19545 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
19546 _mm512_setzero): New intrinsics.
19548 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
19549 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19551 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
19552 code generation for cases where splatting a value is not useful.
19553 * simplify-rtx.c (simplify_ternary_operation): Simplify
19554 vec_merge across a vec_duplicate and a paradoxical subreg forming
19555 a vector mode to a vec_concat.
19557 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
19559 * config.gcc: Support "goldmont-plus".
19560 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19562 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19563 PROCESSOR_GOLDMONT_PLUS.
19564 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
19565 (processor_target_table): Add "goldmont-plus".
19566 (PTA_GOLDMONT_PLUS): Define.
19567 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
19568 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
19569 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
19570 (fold_builtin_cpu): Add "goldmont-plus".
19571 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
19572 (ix86_option_override_internal): Add "goldmont-plus".
19573 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
19574 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
19575 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
19576 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
19578 2018-05-17 Richard Biener <rguenther@suse.de>
19580 PR tree-optimization/85757
19581 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
19582 remove defs that only feed that PHI from further processing.
19584 2018-05-16 Jim Wilson <jimw@sifive.com>
19586 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
19588 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
19590 2018-05-16 Mark Wielaard <mark@klomp.org>
19592 * dwarf2out.c (count_index_strings): New function.
19593 (output_indirect_strings): Call count_index_strings and generate
19594 header for dwarf_version >= 5.
19596 2018-05-16 Mark Wielaard <mark@klomp.org>
19598 * dwarf2out.c (dwarf_FORM): New function.
19599 (set_indirect_string): Use dwarf_FORM.
19600 (reset_indirect_string): Likewise.
19601 (size_of_die): Likewise.
19602 (value_format): Likewise.
19603 (output_die): Likewise.
19604 (add_skeleton_AT_string): Likewise.
19605 (output_macinfo_op): Likewise.
19606 (index_string): Likewise.
19607 (output_index_string_offset): Likewise.
19608 (output_index_string): Likewise.
19609 (count_index_strings): Likewise.
19611 2018-05-16 Carl Love <cel@us.ibm.com>
19613 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
19614 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
19616 2018-05-16 Martin Jambor <mjambor@suse.cz>
19618 * ipa-prop.c (ipa_free_all_edge_args): Remove.
19619 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
19621 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
19623 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
19624 (fnma<mode>4): Likewise.
19625 (fms<mode>4): Likewise.
19626 (fnms<mode>4): Likewise.
19627 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
19628 (aarch64_fnma<mode>4): Likewise.
19629 (aarch64_fms<mode>4): Likewise.
19630 (aarch64_fnms<mode>4): Likewise.
19631 (aarch64_fnmadd<mode>4): Likewise.
19633 2018-05-16 Jason Merrill <jason@redhat.com>
19635 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
19637 2018-05-16 Richard Biener <rguenther@suse.de>
19639 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
19640 (dump_stmt_cost): Declare.
19641 (add_stmt_cost): Dump cost we add.
19642 (add_stmt_costs): New function.
19643 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
19644 No longer exported.
19645 (vect_analyze_stmt): Adjust prototype.
19646 (vectorizable_condition): Likewise.
19647 (vectorizable_live_operation): Likewise.
19648 (vectorizable_reduction): Likewise.
19649 (vectorizable_induction): Likewise.
19650 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
19651 cost vector to pass to vectorizable_ and record afterwards.
19652 (vect_model_reduction_cost): Take cost vector argument and adjust.
19653 (vect_model_induction_cost): Likewise.
19654 (vectorizable_reduction): Likewise.
19655 (vectorizable_induction): Likewise.
19656 (vectorizable_live_operation): Likewise.
19657 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
19658 SLP_TREE_NUMBER_OF_VEC_STMTS.
19659 (vect_analyze_slp_cost_1): Remove.
19660 (vect_analyze_slp_cost): Likewise.
19661 (vect_slp_analyze_node_operations): Take visited args and
19662 a target cost vector. Avoid processing already visited stmt sets.
19663 (vect_slp_analyze_operations): Use a local cost vector to gather
19664 costs and register those of non-discarded instances.
19665 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
19666 (vect_schedule_slp_instance): Remove copying of
19667 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
19669 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
19670 adding cost. Record cost entry location.
19671 (vect_prologue_cost_for_slp_op): Function to compute cost of
19672 a constant or invariant generated for SLP vect in the prologue,
19673 split out from vect_analyze_slp_cost_1.
19674 (vect_model_simple_cost): Make static. Adjust for SLP costing.
19675 (vect_model_promotion_demotion_cost): Likewise.
19676 (vect_model_store_cost): Likewise, make static.
19677 (vect_model_load_cost): Likewise.
19678 (vectorizable_bswap): Add cost vector arg and adjust.
19679 (vectorizable_call): Likewise.
19680 (vectorizable_simd_clone_call): Likewise.
19681 (vectorizable_conversion): Likewise.
19682 (vectorizable_assignment): Likewise.
19683 (vectorizable_shift): Likewise.
19684 (vectorizable_operation): Likewise.
19685 (vectorizable_store): Likewise.
19686 (vectorizable_load): Likewise.
19687 (vectorizable_condition): Likewise.
19688 (vectorizable_comparison): Likewise.
19689 (can_vectorize_live_stmts): Likewise.
19690 (vect_analyze_stmt): Likewise.
19691 (vect_transform_stmt): Adjust calls to vectorizable_*.
19692 * tree-vectorizer.c: Include gimple-pretty-print.h.
19693 (dump_stmt_cost): New function.
19695 2018-05-16 Richard Biener <rguenther@suse.de>
19697 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
19698 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
19699 * tree-ssa-dse.c: Include tree-ssa-loop.h.
19700 (check_name): New callback.
19701 (dse_classify_store): Track cycles via a visited bitmap of PHI
19702 defs and simplify handling of in-loop and across loop dead stores
19703 and properly fail for loop-variant refs. Handle byte-tracking with
19704 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
19707 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
19709 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
19710 (vect_get_mask_type_for_stmt): Likewise.
19711 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
19713 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
19714 to determine the statement's vector type and the vector type that
19715 should be used for calculating nunits. Deal with cases in which
19716 the type has to be deferred.
19717 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
19718 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
19719 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
19720 (vect_determine_vf_for_stmt): New functions, split out from...
19721 (vect_determine_vectorization_factor): ...here.
19722 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
19723 (vect_get_mask_type_for_stmt): New functions, split out from
19724 vect_determine_vectorization_factor.
19726 2018-05-16 Richard Biener <rguenther@suse.de>
19728 * tree-cfg.c (verify_gimple_assign_ternary): Properly
19729 verify the [VEC_]COND_EXPR embedded comparison.
19731 2018-05-15 Martin Sebor <msebor@redhat.com>
19733 PR tree-optimization/85753
19734 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
19735 RECORD_TYPE in addition to ARRAY_TYPE.
19737 2018-05-15 Martin Sebor <msebor@redhat.com>
19739 PR middle-end/85643
19740 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
19742 2018-05-15 Richard Biener <rguenther@suse.de>
19744 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
19745 add by_clobber_p one. Change algorithm to collect all defs
19746 representing uses we need to walk and try reducing them to
19747 a single one before failing.
19748 (dse_dom_walker::dse_optimize_stmt): Adjust.
19750 2018-05-13 Mark Wielaard <mark@klomp.org>
19752 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
19753 (size_of_loc_descr): Likewise.
19754 (output_loc_operands): Likewise.
19755 (output_loc_operands_raw): Likewise.
19756 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
19757 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
19758 (hash_loc_operands): Likewise.
19759 (compare_loc_operands): Likewise.
19761 2018-05-14 Mark Wielaard <mark@klomp.org>
19763 * dwarf2out.c (count_index_addrs): New function.
19764 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
19766 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19768 PR tree-optimization/83648
19769 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
19770 return value as malloc candidate.
19772 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19775 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
19776 param as true in call to suggest_attribute.
19778 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
19780 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
19781 -mreadonly-in-sdata.
19783 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19785 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
19787 (aarch64_crypto_aesd_fused): Likewise.
19789 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
19791 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
19792 (movsi_aarch64): Likewise.
19793 (load_pairsi): Likewise.
19794 (load_pairdi): Likewise.
19795 (store_pairsi): Likewise.
19796 (store_pairdi): Likewise.
19797 (load_pairsf): Likewise.
19798 (load_pairdf): Likewise.
19799 (store_pairsf): Likewise.
19800 (store_pairdf): Likewise.
19801 (zero_extend): Likewise.
19802 (trunc): Swap alternatives.
19803 (fcvt_target): Add '?' to prefer w over r.
19805 2018-05-14 Jakub Jelinek <jakub@redhat.com>
19808 * config/i386/i386.md: Disallow non-commutative arithmetics in
19809 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
19810 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
19811 in the peephole2 before it.
19813 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
19815 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
19816 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
19817 (ix86_handle_option): Handle -mcldemote.
19818 * config.gcc: New header.
19819 * config/i386/cldemoteintrin.h: New file.
19820 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
19821 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
19823 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19824 OPTION_MASK_ISA_CLDEMOTE.
19825 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
19826 (ix86_valid_target_attribute_inner_p): Ditto.
19827 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
19828 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
19829 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
19830 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
19831 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
19833 * config/i386/i386.opt: Add -mcldemote.
19834 * config/i386/x86intrin.h: New header.
19835 * doc/invoke.texi: Add -mcldemote.
19837 2018-05-14 Richard Biener <rguenther@suse.de>
19839 * doc/match-and-simplify.texi: Adjust :s documentation.
19841 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
19843 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
19844 intended memcpy size.
19845 (REORDER_45): Likewise.
19847 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
19849 * sort.cc: New file.
19850 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
19851 * vec.c (qsort_chk): Use gcc_qsort.
19852 * Makefile.in (OBJS-libcommon): Add sort.o.
19853 (build/sort.o): New target. Use it...
19854 (BUILD_RTL): ... here, and...
19855 (build/gencfn-macros): ... here, and...
19856 (build/genmatch): ... here.
19858 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19859 Chung-Ju Wu <jasonwucj@gmail.com>
19861 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
19862 * config/nds32/nds32-graywolf.md: New file.
19863 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
19864 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
19866 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
19867 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
19868 * config/nds32/nds32.md (pipeline_model): Add graywolf.
19869 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
19870 * config/nds32/pipelines.md: Include n15 settings.
19872 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19873 Chung-Ju Wu <jasonwucj@gmail.com>
19875 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
19876 * config/nds32/nds32-n13.md: New file.
19877 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
19878 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
19880 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
19881 * config/nds32/nds32.md (pipeline_model): Add n13.
19882 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
19883 * config/nds32/pipelines.md: Include n13 settings.
19885 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
19886 Chung-Ju Wu <jasonwucj@gmail.com>
19888 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
19889 * config/nds32/nds32-n10.md: New file.
19890 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
19891 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
19893 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
19894 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
19895 * config/nds32/nds32.md (pipeline_model): Add n10.
19896 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
19897 * config/nds32/pipelines.md: Include n10 settings.
19899 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
19900 Kito Cheng <kito.cheng@gmail.com>
19901 Chung-Ju Wu <jasonwucj@gmail.com>
19903 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
19904 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19905 Add enum values for DSP extension instructions.
19906 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
19908 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
19909 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
19910 New code iterators.
19911 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
19912 * config/nds32/nds32-dspext.md: New file for DSP implementation.
19913 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
19914 * config/nds32/nds32-intrinsic.md: Likewise.
19915 * config/nds32/nds32_intrinsic.h: Likewise.
19916 * config/nds32/nds32-md-auxiliary.c: Likewise.
19917 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
19918 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
19919 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
19920 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
19921 * config/nds32/nds32-protos.h: New declarations for DSP extension.
19922 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
19923 TYPE_DMAC in switch statement.
19924 * config/nds32/nds32.c: New checking and implementation for DSP
19925 extension instructions.
19926 * config/nds32/nds32.h: Likewise.
19927 * config/nds32/nds32.md: Likewise.
19928 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
19929 * config/nds32/predicates.md: Implement new predicates for DSP
19932 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19934 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
19935 Reformat alternatives and attributes so it is easier to identify
19936 which constraints/attributes go with which instruction.
19937 (mov<mode>_hardfloat32, FMOVE64): Likewise.
19938 (mov<mode>_softfloat32, FMOVE64): Likewise.
19939 (mov<mode>_hardfloat64, FMOVE64): Likewise.
19940 (mov<mode>_softfloat64, FMOVE64): Likewise.
19942 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19944 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19946 (Basic PowerPC Built-in Functions): The new name of the
19947 subsection previously known as "PowerPC Built-in Functions".
19948 (Basic PowerPC Built-in Functions Available on all Configurations):
19950 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
19951 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
19952 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
19953 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
19955 2018-05-11 Martin Jambor <mjambor@suse.cz>
19958 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
19961 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
19964 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
19966 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
19968 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
19969 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
19970 (ix86_handle_option): Handle -mwaitpkg.
19971 * config.gcc: New header.
19972 * config/i386/cpuid.h (bit_WAITPKG): New bit.
19973 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
19974 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
19976 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19977 OPTION_MASK_ISA_WAITPKG.
19978 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
19979 (ix86_option_override_internal): Add PTA_WAITPKG.
19980 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
19981 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
19982 IX86_BUILTIN_TPAUSE.
19983 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
19984 __builtin_ia32_umwait and __builtin_ia32_tpause.
19985 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
19986 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
19987 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
19988 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
19989 UNSPECV_TPAUSE): New.
19990 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
19991 * config/i386/i386.opt: Add -mwaitpkg.
19992 * config/i386/waitpkgintrin.h: New file.
19993 * config/i386/x86intrin.h: New header.
19994 * doc/invoke.texi: Add -mwaitpkg.
19996 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
19999 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
20001 (cortex-m0): Use armv6s-m isa.
20002 (cortex-m0plus): Likewise.
20003 (cortex-m1): Likewise.
20004 (cortex-m0.small-multiply): Likewise.
20005 (cortex-m0plus.small-multiply): Likewise.
20006 (cortex-m1.small-multiply): Likewise.
20008 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
20009 Jakub Jelinek <jakub@redhat.com>
20011 PR tree-optimization/85692
20012 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
20013 source permute as well.
20015 2018-05-11 Martin Liska <mliska@suse.cz>
20018 * doc/extend.texi: Document LLVM style format for no_sanitize
20021 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
20023 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
20024 mode_supports_vsx_dform_quad to mode_supports_dq_form.
20025 (mode_supports_vsx_dform_quad): Likewise.
20026 (mode_supports_vmx_dform): Move these functions to be next to the
20027 other mode_supports functions.
20028 (mode_supports_dq_form): Likewise.
20029 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
20030 mode_supports_dq_form.
20031 (reg_offset_addressing_ok_p): Likewise.
20032 (offsettable_ok_by_alignment): Likewise.
20033 (rs6000_legitimate_offset_address_p): Likewise.
20034 (legitimate_lo_sum_address_p): Likewise.
20035 (rs6000_legitimize_address): Likewise.
20036 (rs6000_legitimize_reload_address): Likewise.
20037 (rs6000_secondary_reload_inner): Likewise.
20038 (rs6000_preferred_reload_class): Likewise.
20039 (rs6000_output_move_128bit): Likewise.
20041 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
20043 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
20044 Generate SImode target register for null target.
20045 <case IX86_BUILTIN_XGETBV>: Ditto.
20046 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
20047 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
20049 2018-05-10 Carl Love <cel@us.ibm.com>
20051 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
20052 dcbtt and dcbtstt if operands[2] is 0.
20054 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
20057 * config/i386/sse.md (usadv64qi): New expander.
20059 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
20061 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
20062 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
20063 -maltivec=be support.
20064 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
20065 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
20066 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20067 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
20068 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
20069 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
20070 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
20071 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
20072 altivec_vsumsws): Adjust.
20073 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
20074 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
20075 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
20077 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
20078 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
20079 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
20080 (altivec_lve<VI_char>x): Delete expand.
20081 (*altivec_lve<VI_char>x_internal): Rename to...
20082 (altivec_lve<VI_char>x): ... this.
20083 (altivec_lvxl_<mode>): Delete expand.
20084 (*altivec_lvxl_<mode>_internal): Rename to ...
20085 (altivec_lvxl_<mode>): ... this.
20086 (altivec_stvxl_<mode>): Delete expand.
20087 (*altivec_stvxl_<mode>_internal): Rename to ...
20088 (altivec_stvxl_<mode>): ... this.
20089 (altivec_stve<VI_char>x): Delete expand.
20090 (*altivec_stve<VI_char>x_internal): Rename to ...
20091 (altivec_stve<VI_char>x): ... this.
20092 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
20093 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
20094 reduc_plus_scal_<mode>): Adjust.
20095 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
20097 (rs6000_cpu_cpp_builtins): Adjust.
20098 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
20099 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
20100 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
20101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
20102 -maltivec=be support.
20103 (rs6000_split_vec_extract_var): Adjust.
20104 (rs6000_split_v4si_init): Adjust.
20105 (swap_selector_for_mode): Delete.
20106 (altivec_expand_lvx_be, altivec_expand_stvx_be,
20107 altivec_expand_stvex_be): Delete.
20108 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
20109 -maltivec=be support.
20110 (rs6000_gimple_fold_builtin): Ditto.
20111 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
20113 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
20114 (TARGET_DIRECT_MOVE_64BIT): Adjust.
20115 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
20116 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
20117 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
20118 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
20119 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
20120 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
20121 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
20122 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
20123 anonymous split): Adjust.
20124 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
20125 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
20127 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
20129 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
20130 when --with-gxx-include-dir is also specified.
20131 * configure: Regenerate.
20133 2018-05-09 Jim Wilson <jimw@sifive.com>
20136 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
20137 * config/riscv/t-withmultilib: New.
20138 * config/riscv/withmultilib.h: New.
20139 * doc/install.texi: Document RISC-V --with-multilib-list support.
20141 2018-05-09 Richard Biener <rguenther@suse.de>
20143 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
20145 (vect_bb_vectorization_profitable_p): Adjust. Compute
20146 actual scalar cost using the cost vector and the add_stmt_cost
20149 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20151 PR rtl-optimization/85645
20152 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
20153 in the REG_CFA_REGISTER note for LR, don't leave it empty.
20155 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20157 PR rtl-optimization/85645
20158 * shrink-wrap.c (spread_components): Return a boolean saying if
20159 anything was changed.
20160 (try_shrink_wrapping_separate): Iterate spread_components until
20161 nothing changes anymore.
20163 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20165 PR rtl-optimization/85645
20166 * regrename.c (build_def_use): Also kill the chains that include the
20167 destination of a REG_CFA_REGISTER note.
20169 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
20171 PR rtl-optimization/85645
20172 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
20173 insn that has a REG_CFA_REGISTER note.
20175 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
20177 * cfgexpand.c (expand_clobber): New function.
20178 (expand_gimple_stmt_1): Use it.
20179 * tree-vect-stmts.c (vect_clobber_variable): New function,
20181 (vectorizable_simd_clone_call): ...here.
20182 (vectorizable_store): Emit a clobber either side of an
20183 IFN_STORE_LANES sequence.
20184 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
20186 2018-05-09 Tom de Vries <tom@codesourcery.com>
20189 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
20190 (define_insn "trap_if_false"): Add exit after trap.
20192 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
20194 PR rtl-optimization/85638
20195 * bb-reorder.c: Include common/common-target.h.
20196 (create_forwarder_block): New function extracted from...
20197 (fix_up_crossing_landing_pad): ...here. Rename into...
20198 (dw2_fix_up_crossing_landing_pad): ...this.
20199 (sjlj_fix_up_crossing_landing_pad): New function.
20200 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
20201 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
20202 from both partitions and exit the loop after one iteration.
20204 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20207 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20209 (Basic PowerPC Built-in Functions): The new name of the
20210 subsection previously known as "PowerPC Built-in Functions".
20211 (Basic PowerPC Built-in Functions Available on all Configurations):
20213 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
20215 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
20217 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
20219 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
20222 2018-05-08 Jim Wilson <jimw@sifive.com>
20224 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
20225 (LD_EMUL_SUFFIX): New.
20226 (LINK_SPEC): Use it.
20228 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20230 * doc/extend.texi (PowerPC Built-in Functions): Rename this
20232 (Basic PowerPC Built-in Functions): The new name of the
20233 subsection previously known as "PowerPC Built-in Functions".
20234 (Basic PowerPC Built-in Functions Available on all Configurations):
20236 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
20238 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
20240 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
20242 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
20245 2018-05-08 Jakub Jelinek <jakub@redhat.com>
20248 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
20249 after cmpelim optimization.
20251 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
20253 * config.gcc: Support "goldmont".
20254 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
20255 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20256 PROCESSOR_GOLDMONT.
20257 * config/i386/i386.c (m_GOLDMONT): Define.
20258 (processor_target_table): Add "goldmont".
20259 (PTA_GOLDMONT): Define.
20260 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
20261 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
20262 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
20263 (fold_builtin_cpu): Add "goldmont".
20264 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
20265 (ix86_option_override_internal): Add "goldmont".
20266 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
20267 (processor_type): Add PROCESSOR_GOLDMONT.
20268 * config/i386/i386.md: Add CPU "glm".
20269 * config/i386/glm.md: New file.
20270 * config/i386/x86-tune.def: Add m_GOLDMONT.
20271 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
20273 2018-05-08 Jakub Jelinek <jakub@redhat.com>
20276 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
20278 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
20279 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
20280 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
20283 * config/i386/i386.c (ix86_fold_builtin): Handle
20284 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
20287 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
20288 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
20290 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
20293 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
20294 (check_arch): Likewise.
20295 (check_fpu): Return the result rather than printing it.
20296 (end arch): Fix operator precedence.
20297 (end cpu): Likewise.
20298 (END): Print the result from check_fpu.
20300 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
20301 Alan Hayward <alan.hayward@arm.com>
20302 David Sherwood <david.sherwood@arm.com>
20304 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
20305 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
20306 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
20307 (*fcmuo<mode>_and): New patterns.
20309 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
20311 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
20312 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
20313 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
20314 (cmp_op, sve_imm_con): New code attributes.
20315 (SVE_COND_INT_CMP, imm_con): Delete.
20316 (cmp_op): Remove above unspecs from int attribute.
20317 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
20319 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
20320 comparison-specific unspecs.
20321 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
20322 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
20323 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
20324 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
20325 (*vec_fcm<cmp_op><mode>): Rename to...
20326 (*fcm<cmp_op><mode>): ...this and adjust likewise.
20327 (*vec_fcmuo<mode>): Rename to...
20328 (*fcmuo<mode>): ...this and adjust likewise.
20329 (*pred_fcm<cmp_op><mode>): New pattern.
20330 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
20331 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
20333 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
20335 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
20336 (aarch64_emit_sve_predicated_cond): New function.
20337 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
20338 (aarch64_emit_unspec_cond_or): Replace with...
20339 (aarch64_emit_sve_or_conds): ...this new function. Use
20340 aarch64_emit_sve_ptrue_op for the individual comparisons and
20341 aarch64_emit_binop to OR them together.
20342 (aarch64_emit_inverted_unspec_cond): Replace with...
20343 (aarch64_emit_sve_inverted_cond): ...this new function. Use
20344 aarch64_emit_sve_ptrue_op for the comparison and
20345 aarch64_emit_unop to invert the result.
20346 (aarch64_expand_sve_vec_cmp_float): Update after the above
20347 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
20349 2018-05-07 Nathan Sidwell <nathan@acm.org>
20351 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
20352 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
20353 (Backwards Compatibility): Likewise.
20355 2018-05-07 Luis Machado <luis.machado@linaro.org>
20359 2018-05-07 Luis Machado <luis.machado@linaro.org>
20361 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20362 <prefetch_dynamic_strides>: New const bool field.
20363 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20364 prefetch_dynamic_strides.
20365 (exynosm1_prefetch_tune): Likewise.
20366 (thunderxt88_prefetch_tune): Likewise.
20367 (thunderx_prefetch_tune): Likewise.
20368 (thunderx2t99_prefetch_tune): Likewise.
20369 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
20371 (aarch64_override_options_internal): Update to set
20372 PARAM_PREFETCH_DYNAMIC_STRIDES.
20373 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20374 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20375 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20376 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20377 prefetch-dynamic-strides setting.
20379 2018-05-07 Luis Machado <luis.machado@linaro.org>
20381 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20382 <minimum_stride>: New const int field.
20383 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20384 minimum_stride field.
20385 (exynosm1_prefetch_tune): Likewise.
20386 (thunderxt88_prefetch_tune): Likewise.
20387 (thunderx_prefetch_tune): Likewise.
20388 (thunderx2t99_prefetch_tune): Likewise.
20389 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
20390 (aarch64_override_options_internal): Update to set
20391 PARAM_PREFETCH_MINIMUM_STRIDE.
20392 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20393 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20394 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20395 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20396 stride is constant and is below the minimum stride threshold.
20398 2018-05-07 Luis Machado <luis.machado@linaro.org>
20400 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
20403 2018-05-07 Luis Machado <luis.machado@linaro.org>
20405 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20406 <prefetch_dynamic_strides>: New const bool field.
20407 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20408 prefetch_dynamic_strides.
20409 (exynosm1_prefetch_tune): Likewise.
20410 (thunderxt88_prefetch_tune): Likewise.
20411 (thunderx_prefetch_tune): Likewise.
20412 (thunderx2t99_prefetch_tune): Likewise.
20413 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
20415 (aarch64_override_options_internal): Update to set
20416 PARAM_PREFETCH_DYNAMIC_STRIDES.
20417 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20418 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20419 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20420 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20421 prefetch-dynamic-strides setting.
20423 2018-05-07 Luis Machado <luis.machado@linaro.org>
20425 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20426 <minimum_stride>: New const int field.
20427 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20428 minimum_stride field.
20429 (exynosm1_prefetch_tune): Likewise.
20430 (thunderxt88_prefetch_tune): Likewise.
20431 (thunderx_prefetch_tune): Likewise.
20432 (thunderx2t99_prefetch_tune): Likewise.
20433 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
20434 (aarch64_override_options_internal): Update to set
20435 PARAM_PREFETCH_MINIMUM_STRIDE.
20436 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20437 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20438 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20439 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20440 stride is constant and is below the minimum stride threshold.
20442 2018-05-06 Jakub Jelinek <jakub@redhat.com>
20445 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
20446 the type is addressable. Don't force op into register if it has
20449 2018-05-05 Roland McGrath <mcgrathr@google.com>
20452 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
20453 any section for which we don't know a specific type it should have,
20454 regardless of name. Previously this was done only for the exact
20455 names ".init_array", ".fini_array", and ".preinit_array".
20456 (default_elf_asm_named_section): Add comment about
20457 relationship with default_section_type_flags and SECTION_NOTYPE.
20458 (get_section): Don't consider it a type conflict if one side has
20459 SECTION_NOTYPE and the other doesn't, as long as neither has the
20460 SECTION_BSS et al used in the default_section_type_flags logic.
20462 2018-05-05 Tom de Vries <tom@codesourcery.com>
20465 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
20466 (workaround_barsyncs): New function.
20467 (nvptx_reorg): Use workaround_barsyncs.
20468 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
20469 (define_expand "nvptx_membar_cta"): New define_expand.
20470 (define_insn "*nvptx_membar_cta"): New insn.
20472 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20474 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
20475 To improve optimization opportunities.
20476 * builtin-types.def: The new needed builtin types for the above.
20478 2018-05-04 Richard Biener <rguenther@suse.de>
20480 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
20481 * gimple-ssa-store-merging.c
20482 (imm_store_chain_info::output_merged_store): Remove redundant create,
20483 release split_store vector contents on failure.
20484 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
20485 scalar stmt vector on cache hit.
20487 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
20489 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20491 * config.gcc (powerpc-xilinx-eabi*): Remove.
20492 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
20494 (fusion_addis_mem_combo_load): Ditto.
20495 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
20497 (rs6000_cpu_cpp_builtins): Ditto.
20498 * config/rs6000/rs6000-linux.c
20499 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
20500 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
20501 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
20503 (rs6000_setup_reg_addr_masks): Ditto.
20504 (rs6000_init_hard_regno_mode_ok): Ditto.
20505 (rs6000_option_override_internal): Ditto.
20506 (legitimate_lo_sum_address_p): Ditto.
20507 (rs6000_legitimize_address): Ditto.
20508 (rs6000_legitimize_reload_address): Ditto.
20509 (rs6000_legitimate_address_p): Ditto.
20510 (abi_v4_pass_in_fpr): Ditto.
20511 (setup_incoming_varargs): Ditto.
20512 (rs6000_gimplify_va_arg): Ditto.
20513 (rs6000_split_multireg_move): Ditto.
20514 (rs6000_savres_strategy): Ditto.
20515 (rs6000_emit_prologue_components): Ditto.
20516 (rs6000_emit_epilogue_components): Ditto.
20517 (rs6000_emit_prologue): Ditto.
20518 (rs6000_emit_epilogue): Ditto.
20519 (rs6000_elf_file_end): Ditto.
20520 (rs6000_function_value): Ditto.
20521 (rs6000_libcall_value): Ditto.
20522 * config/rs6000/rs6000.h: Ditto.
20523 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
20524 (TARGET_MINMAX): ... this. New.
20525 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
20526 * config/rs6000/rs6000.md: Remove Xilinx FP support.
20527 (*movsi_internal1_single): Delete.
20528 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
20529 mfpu=, mxilinx-fpu): Delete.
20530 * config/rs6000/singlefp.h: Delete.
20531 * config/rs6000/sysv4.h: Remove Xilinx FP support.
20532 * config/rs6000/t-rs6000: Ditto.
20533 * config/rs6000/t-xilinx: Delete.
20534 * config/rs6000/titan.md: Adjust for fp_type removal.
20535 * config/rs6000/vsx.md: Remove Xilinx FP support.
20536 (VStype_simple): Delete.
20537 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
20538 * config/rs6000/xfpu.h: Delete.
20539 * config/rs6000/xfpu.md: Delete.
20540 * config/rs6000/xilinx.h: Delete.
20541 * config/rs6000/xilinx.opt: Delete.
20542 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
20543 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
20545 2018-05-04 Tom de Vries <tom@codesourcery.com>
20548 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
20551 2018-05-04 Richard Biener <rguenther@suse.de>
20553 PR middle-end/85627
20554 * tree-complex.c (update_complex_assignment): We are always in SSA form.
20555 (expand_complex_div_wide): Likewise.
20556 (expand_complex_operations_1): Likewise.
20557 (expand_complex_libcall): Preserve EH info of the original stmt.
20558 (tree_lower_complex): Handle removed blocks.
20559 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
20560 on complex multiplication and division libcall builtins.
20562 2018-05-04 Richard Biener <rguenther@suse.de>
20564 PR middle-end/85574
20565 * fold-const.c (negate_expr_p): Restrict negation of operand
20566 zero of a division to when we know that can happen without
20568 (fold_negate_expr_1): Likewise.
20570 2018-05-04 Jakub Jelinek <jakub@redhat.com>
20573 * real.h (real_nextafter): Declare.
20574 * real.c (real_nextafter): New function.
20575 * fold-const-call.c (fold_const_nextafter): New function.
20576 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
20577 CASE_CFN_NEXTTOWARD.
20578 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
20579 even when arg1_mode is different from arg0_mode.
20581 2018-05-03 Nathan Sidwell <nathan@acm.org>
20583 * doc/extend.texi (Deprecated Features): Remove
20584 -ffriend-injection.
20585 (Backwards Compatibility): Likewise.
20586 * doc/invoke.texi (C++ Language Options): Likewise.
20587 (C++ Dialect Options): Likewise.
20589 2018-05-03 Jakub Jelinek <jakub@redhat.com>
20592 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
20593 _mm512_mask_mullox_epi64): New intrinsics.
20595 2018-05-03 Tom de Vries <tom@codesourcery.com>
20598 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20599 dump files): Add offload-tree.
20601 2018-05-03 Richard Biener <rguenther@suse.de>
20603 PR tree-optimization/85615
20604 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
20605 to loops not nested in BBs loop father to avoid creating multi-entry
20608 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20610 PR tree-optimization/70291
20611 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
20612 arguments. Change return type to tree. Emit libcall as a new
20613 statement rather than replacing existing one when inplace_p is true.
20614 (expand_complex_multiplication_components): New function.
20615 (expand_complex_multiplication): Expand floating-point complex
20616 multiplication using the above.
20617 (expand_complex_division): Rename inner_type parameter to type.
20618 Update expand_complex_libcall call-site.
20619 (expand_complex_operations_1): Update expand_complex_multiplication
20620 and expand_complex_division call-sites.
20622 2018-05-02 Jakub Jelinek <jakub@redhat.com>
20625 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
20626 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
20627 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
20628 the highest significant bit of the shift count mask is clear. In
20629 check whether and[sq]i3 is needed verify that all significant bits
20630 of the shift count other than the highest are set.
20632 2018-05-02 Tom de Vries <tom@codesourcery.com>
20635 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
20636 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
20637 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
20638 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
20639 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
20640 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
20641 __builtin_goacc_parlevel_size.
20643 2018-05-02 Richard Biener <rguenther@suse.de>
20645 PR tree-optimization/85597
20646 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
20647 do not use split vect_get_vec_defs call but call vect_get_slp_defs
20650 2018-05-02 Tom de Vries <tom@codesourcery.com>
20653 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20654 dump files): Add ltrans-tree.
20656 2018-05-02 Tom de Vries <tom@codesourcery.com>
20659 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
20660 dump files): Add wpa-ipa.
20662 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
20664 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
20665 powerpc*-*-linux*paired* target.
20666 * config/rs6000/750cl.h: Delete.
20667 * config/rs6000/paired.h: Delete.
20668 * config/rs6000/paired.md: Delete.
20669 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
20671 * config/rs6000/rs6000-builtin.def: Remove paired float support.
20672 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20673 comment. Remove paired float support.
20674 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
20675 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
20677 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
20678 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
20680 * config/rs6000/rs6000.c: Remove paired float support.
20681 (paired_expand_vector_init, paired_expand_vector_move,
20682 paired_emit_vector_compare, paired_emit_vector_cond_expr,
20683 (paired_expand_lv_builtin, paired_expand_stv_builtin,
20684 paired_expand_builtin, paired_expand_predicate_builtin,
20685 paired_init_builtins): Delete.
20686 * config/rs6000/rs6000.h: Remove paired float support.
20687 * config/rs6000/rs6000.md: Remove paired float support.
20688 (move_from_CR_ov_bit): Delete.
20689 * config/rs6000/rs6000.opt (mpaired): Delete.
20690 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
20691 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
20693 2018-05-02 Richard Biener <rguenther@suse.de>
20695 PR middle-end/85567
20696 * gimplify.c (gimplify_save_expr): When in SSA form allow
20697 SAVE_EXPRs to compute to SSA vars.
20699 2018-05-02 Jakub Jelinek <jakub@redhat.com>
20702 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
20703 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
20704 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
20705 clobber operands[2], instead use a new pseudo. Formatting fixes.
20707 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
20709 PR tree-optimization/85586
20710 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
20711 exit early for statements in the same group if the accesses are
20714 2018-05-02 Tom de Vries <tom@codesourcery.com>
20717 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
20720 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
20722 PR tree-optimization/85143
20723 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
20725 2018-05-01 Tom de Vries <tom@codesourcery.com>
20728 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
20729 not found" error message.
20731 2018-05-01 Tom de Vries <tom@codesourcery.com>
20734 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
20735 * vec.c (test_ordered_remove_if): New function.
20736 (vec_c_tests): Call test_ordered_remove_if.
20737 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
20738 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
20739 * tree-vect-patterns.c (vect_pattern_recog_1): Use
20740 VEC_ORDERED_REMOVE_IF.
20742 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20744 PR tree-optimization/82665
20745 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
20746 pointer subtraction where arguments come from a memchr call.
20748 2018-05-01 Jakub Jelinek <jakub@redhat.com>
20750 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
20751 --push-state --as-needed and --pop-state instead of --as-needed and
20752 --no-as-needed if ld supports it.
20753 * configure: Regenerated.
20756 * doc/install.texi2html: Replace _002d with - and _002a with * in
20757 generated html files using sed.
20759 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20762 * gcc-rich-location.c (blank_line_before_p): New function.
20763 (use_new_line): New function.
20764 (gcc_rich_location::add_fixit_insert_formatted): New function.
20765 * gcc-rich-location.h
20766 (gcc_rich_location::add_fixit_insert_formatted): New function.
20768 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20770 * selftest.c (assert_streq): Rename "expected" and "actual" to
20771 "val1" and "val2". Extend NULL-handling to cover both inputs
20772 symmetrically, while still requiring both to be non-NULL for a pass.
20773 * selftest.h (assert_streq): Rename "expected" and "actual" to
20775 (ASSERT_EQ): Likewise.
20776 (ASSERT_EQ_AT): Likewise.
20777 (ASSERT_KNOWN_EQ): Likewise.
20778 (ASSERT_KNOWN_EQ_AT): Likewise.
20779 (ASSERT_NE): Likewise.
20780 (ASSERT_MAYBE_NE): Likewise.
20781 (ASSERT_MAYBE_NE_AT): Likewise.
20782 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
20783 the assertion to pass.
20784 (ASSERT_STREQ_AT): Likewise.
20786 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
20788 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
20789 interaction with -pie.
20791 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20793 * selftest.h: Fix alphabetization of per-source-file selftest
20796 2018-04-30 Jason Merrill <jason@redhat.com>
20798 PR c++/61982 - dead stores to destroyed objects.
20799 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
20802 2018-04-30 Jason Merrill <jason@redhat.com>
20804 * tree.c (build_clobber): New.
20805 * tree.h: Declare it.
20806 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
20808 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20810 * diagnostic-show-locus.c (layout::layout): Update for
20811 location_get_source_line returning a char_span.
20812 (struct char_span): Move to input.h.
20813 (struct correction): Update for fields in char_span becoming
20815 (struct source_line): Update for location_get_source_line
20816 returning a char_span.
20817 (layout::print_line): Likewise.
20818 * edit-context.c (edited_file::print_content): Likewise.
20819 (edited_file::print_diff_hunk): Likewise.
20820 (edited_file::print_run_of_changed_lines): Likewise.
20821 (edited_file::get_num_lines): Likewise.
20822 (edited_line::edited_line): Likewise.
20823 * final.c (asm_show_source): Likewise.
20824 * input.c (location_get_source_line): Convert return type
20825 from const char * to char_span, losing the final "line_len"
20827 (dump_location_info): Update for the above.
20828 (get_substring_ranges_for_loc): Likewise. Use a char_span
20829 when handling the literal within the line.
20830 (test_reading_source_line): Update for location_get_source_line
20831 returning a char_span.
20832 * input.h (class char_span): Move here from
20833 diagnostic-show-locus.c, converting from a struct to a class.
20834 Make data members private.
20835 (char_span::operator bool): New.
20836 (char_span::length): New.
20837 (char_span::get_buffer): New.
20838 (char_span::operator[]): New.
20839 (char_span::subspan): Make const.
20840 (char_span::xstrdup): New.
20841 (location_get_source_line): Convert return type from const char *
20842 to char_span, losing the final "line_size" param.
20844 2018-04-30 Jan Hubicka <jh@suse.cz>
20846 * lto-wrapper.c (ltrans_priorities): New static var.
20847 (cmp_priority): New.
20848 (run_gcc): Read priorities and if doing parallel build order
20849 the Makefile by them.
20851 2018-04-30 David Malcolm <dmalcolm@redhat.com>
20853 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
20855 2018-04-30 Richard Biener <rguenther@suse.de>
20857 * tree-cfg.c (verify_address): Remove base argument, add
20858 flag whether to check TREE_ADDRESSABLE and do that.
20859 (verify_expr): Remove.
20860 (verify_types_in_gimple_reference): Add pieces from verify_expr.
20861 (verify_gimple_assign_single): Likewise.
20862 (verify_gimple_switch): Likewise.
20863 (verify_expr_location_1): Dereference tp once. Add (disabled)
20864 piece from verify_expr.
20865 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
20867 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20869 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
20871 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20873 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
20874 (small_data_pattern): Likewise.
20875 (arc_rewrite_small_data): Likewise.
20876 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
20877 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
20878 (get_symbol_alignment): New function.
20879 (legitimate_small_data_address_p): Likewise.
20880 (legitimate_scaled_address): Update, call
20881 legitimate_small_data_address_p.
20882 (output_sdata): New static variable.
20883 (arc_print_operand): Update how we handle small data operands.
20884 (arc_print_operand_address): Likewise.
20885 (arc_legitimate_address_p): Update, use
20886 legitimate_small_data_address_p.
20887 (arc_rewrite_small_data_p): Remove.
20888 (arc_rewrite_small_data_1): Likewise.
20889 (arc_rewrite_small_data): Likewise.
20890 (small_data_pattern): Likewise.
20891 (compact_sda_memory_operand): Update to use
20892 legitimate_small_data_address_p and get_symbol_alignment.
20893 (prepare_move_operands): Don't rewite sdata pattern.
20894 (prepare_extend_operands): Remove.
20895 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
20897 (zero_extendqisi2): Likewise.
20898 (zero_extendhisi2): Likewise.
20899 (extendqihi2): Likewise.
20900 (extendqisi2): Likewise.
20901 (extendhisi2): Likewise.
20902 (addsi3): Likewise.
20903 (subsi3): Likewise.
20904 (andsi3): Likewise.
20905 * config/arc/constraints.md (Usd): Change it to memory constraint.
20907 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
20909 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
20910 as source of std instructions.
20911 * config/arc/arc.md (movsi_insn): Update pattern predicate to
20912 allow 6-bit constants as source for store instructions.
20913 (movdi_insn): Update instruction pattern to allow 6-bit constants
20914 as source for store instructions.
20916 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
20918 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
20920 2018-04-30 Nathan Sidwell <nathan@acm.org>
20921 Sandra Loosemore <sandra@codesourcery.com>
20923 * dumpfile.c (dump_open): Allow '-' for stdout.
20924 * doc/invoke.texi (Developer Options): Document dump filename
20925 determination early. Document stdin/stdout selection.
20927 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
20929 Microblaze Target: PIC data text relative
20931 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
20932 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
20934 * config/microblaze/microblaze.h (microblaze_constant_address_p):
20935 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
20936 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
20937 New addressing mode for data-text relative position indepenedent code.
20938 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
20939 'ADDRESS_SYMBOLIC_TXT_REL'.
20940 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
20941 (microblaze_legitimate_pic_operand): Exclude function calls from
20942 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
20943 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
20945 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
20946 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
20947 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
20948 for 'address + offset'.
20949 (microblaze_expand_prologue): Add new function prologue call for
20951 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
20952 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
20953 table in case of TARGET_PIC_DATA_TEXT_REL.
20954 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
20955 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
20956 Add new macros 'UNSPEC_TEXT',
20957 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
20958 + exclude function calls from 'UNSPEC_PLT' in case of data text
20960 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
20961 new target hook for generating address diff vector tables in case of
20963 * doc/tm.texi : Regenerate.
20964 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
20965 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
20966 of addr diff vector generation.
20967 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
20968 target hook definition.
20969 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
20970 Add default function for generate_pic_addr_diff_vec -> flag_pic.
20971 * doc/invoke.texi (Add new pic option): Add new microblaze pic
20972 option for data text relative.
20974 2018-04-30 Richard Biener <rguenther@suse.de>
20976 * tree-chrec.h (evolution_function_is_constant_p): Remove
20978 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
20980 2018-04-30 Richard Biener <rguenther@suse.de>
20983 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
20985 2018-04-30 Richard Biener <rguenther@suse.de>
20987 PR tree-optimization/28364
20988 PR tree-optimization/85275
20989 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
20990 copying first exit test.
20992 2018-04-28 Mark Wielaard <mark@klomp.org>
20994 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
20995 dwarf_version >= 5.
20996 (dwarf_AT): Handle DW_AT_addr_base.
20997 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
20999 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
21002 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
21003 (*ashl<dwi>3_doubleword_mask_1): Ditto.
21004 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
21005 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
21007 2018-04-28 Richard Biener <rguenther@suse.de>
21009 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
21010 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
21011 to reflect use. Only add interesting stmts.
21013 2018-04-27 Martin Jambor <mjambor@suse.cz>
21016 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
21017 the jump function allows for passing through aggregate values.
21019 2018-04-27 David Malcolm <dmalcolm@redhat.com>
21021 * input.h (in_system_header_at): Convert from macro to inline
21023 (from_macro_expansion_at): Likewise.
21024 (from_macro_definition_at): Likewise.
21026 2018-04-27 Jeff Law <law@redhat.com>
21028 * config.gcc: Mark tile* targets as deprecated/obsolete.
21030 2018-04-27 Richard Biener <rguenther@suse.de>
21032 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
21035 2018-04-27 Richard Biener <rguenther@suse.de>
21037 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
21039 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
21041 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
21042 with Yd constraint. Set "preferred_for_speed" attribute from
21043 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
21044 with Yd constraint.
21045 (*movdi_internal): Ditto.
21046 (movti_interunit splitters): Remove
21047 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
21048 (movdi_interunit splitters): Ditto.
21049 * config/i386/constraints.md (Ye): Remove.
21050 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
21052 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21055 * config/aarch64/constraints.md (Usg): Limit to 31.
21056 (Usj): Limit to 63.
21058 2018-04-27 Jakub Jelinek <jakub@redhat.com>
21060 PR tree-optimization/85529
21061 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
21062 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
21063 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
21064 zero extension or masking of the MSB bit.
21065 (optimize_range_tests): Add FIRST_BB argument, pass it through
21066 to optimize_range_tests_var_bound.
21067 (maybe_optimize_range_tests, reassociate_bb): Adjust
21068 optimize_range_tests callers.
21070 2018-04-26 Richard Biener <rguenther@suse.de>
21071 Jakub Jelinek <jakub@redhat.com>
21073 * cgraph.h (symbol_table): Just declare debug method here.
21074 * symtab.c (symbol_table::debug): Define.
21076 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
21078 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
21080 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
21082 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
21083 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
21084 (*movdi_internal): Substitute Yi and Yj constraint with x
21085 and Ym and Yn constraint with y constraint. Update "isa"
21086 attribute and set "preferred_for_speed" attribute from
21087 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
21088 (*movsi_internal): Ditto.
21089 (*movdf_internal): Ditto.
21090 (*movsf_internal): Ditto.
21091 (*zero_extendsidi2): Ditto.
21092 * config/i386/sse.md (vec_set<mode>_0): Ditto.
21093 (sse2_loadld): Ditto.
21094 (*vec_extract<ssevecmodelower>_0): Ditto.
21095 (*vec_extractv4si_0_zext_sse4): Ditto.
21096 (vec_concatv2di): Ditto.
21097 (*vec_dup<mode>): Ditto.
21098 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
21099 * config/i386/constraints.md (Yi): Remove.
21104 2018-04-26 Nathan Sidwell <nathan@acm.org>
21106 * dumpfile.c (dump_open): New.
21107 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
21108 (dump_finish): Detect stdio/stderr by value not name.
21110 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
21112 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
21114 2018-04-26 Tom de Vries <tom@codesourcery.com>
21117 * config/nvptx/nvptx.c (verify_neutering_jumps)
21118 (verify_neutering_labels): New function
21119 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
21121 2018-04-26 Tom de Vries <tom@codesourcery.com>
21124 * config/nvptx/nvptx.c (needs_neutering_p): New function.
21125 (nvptx_single): Use needs_neutering_p to skip over insns that do not
21128 2018-04-26 Richard Biener <rguenther@suse.de>
21129 Tom de Vries <tom@codesourcery.com>
21132 * lto-streamer-out.c (output_function): Fixup loops if required to match
21133 discovery done in the reader.
21135 2018-04-26 Richard Biener <rguenther@suse.de>
21137 PR tree-optimization/85116
21138 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
21139 have a loop exit from the single latch predecessor. Remove
21140 case of header with just condition.
21141 (ch_base::copy_headers): Exclude infinite loops from any
21143 (pass_ch::execute): Record exits.
21145 2018-04-26 Richard Biener <rguenther@suse.de>
21147 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
21148 prologue cost vector and pass it to vect_get_load_cost.
21149 (vect_get_peeling_costs_all_drs): Likewise.
21150 (vect_peeling_hash_get_lowest_cost): Likewise.
21151 (vect_enhance_data_refs_alignment): Likewise.
21153 2018-04-26 Richard Biener <rguenther@suse.de>
21155 PR middle-end/85450
21156 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
21157 checking of integer<->pointer conversions.
21158 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
21159 sign-/zero-extending pointer types.
21160 (expand_omp_for_static_chunk): Likewise.
21162 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
21163 Jean Lee <xiaoyur347@gmail.com>
21165 * config/mips/mips.c (mips_asan_shadow_offset): New function.
21166 (TARGET_ASAN_SHADOW_OFFSET): Define.
21167 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
21168 true for -fsanitize=address.
21170 2018-04-25 Mark Wielaard <mark@klomp.org>
21172 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
21175 2018-04-25 Jakub Jelinek <jakub@redhat.com>
21177 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
21178 than "alu", remove explicit "memory" and "imm_disp" attributes.
21179 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
21181 PR middle-end/85414
21182 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
21183 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
21184 gen_lowpart_no_emit.
21186 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
21189 * config/i386/i386.c (ix86_expand_builtin): Change memory
21190 operand to XI, extend p0 to Pmode.
21191 * config/i386/i386.md: Change unspec volatile and operand
21192 1 mode to XI, change operand 0 mode to P.
21194 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21196 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
21197 GET_MODE_MASK before any checking.
21198 (nds32_can_use_bset_p): Likewise.
21199 (nds32_can_use_btgl_p): Likewise.
21201 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21203 * config/nds32/nds32-doubleword.md: New define_split pattern for
21204 illegal register number.
21206 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21208 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
21210 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
21212 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
21214 2018-04-25 Richard Biener <rguenther@suse.de>
21216 * lto-streamer.h (LTO_major_version): Bump to 8.
21218 2018-04-25 Jakub Jelinek <jakub@redhat.com>
21220 * BASE-VER: Set to 9.0.0.
21222 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
21224 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
21225 in __abskf2 and __powikf2.
21227 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21230 * config/aarch64/constraints.md (Usg, Usj): New constraints.
21231 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
21232 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
21233 Use the above on operand 2. Reindent.
21234 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
21236 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
21239 * common/config/i386/i386-common.c (ix86_handle_option): Don't
21241 * config/i386/i386.opt (mcet): Removed.
21242 * doc/install.texi: Remove -mcet documentation.
21243 * doc/invoke.texi: Likewise.
21245 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
21248 * doc/install.texi: Remove -mcet from bootstrap-cet.
21250 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21253 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
21254 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
21258 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
21259 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
21260 containing a CONST_VECTOR.
21262 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
21264 * doc/install.texi: Update newlib dependency for nvptx.
21266 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21269 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
21270 instead of INTVAL when shifting x left.
21272 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
21274 PR tree-optimization/85478
21275 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
21276 vect_grouped_store_supported for single element vectors.
21278 2018-04-24 Richard Biener <rguenther@suse.de>
21281 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
21282 load cost increase to the case of non-constant step.
21284 2018-04-24 Jakub Jelinek <jakub@redhat.com>
21287 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
21288 destination if any_malformed_asm.
21290 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
21292 PR middle-end/85496
21293 * expr.c (store_field): In the bitfield case, if the value comes from
21294 a function call and is returned in registers by means of a PARALLEL,
21295 do not change the mode of the temporary unless BLKmode and VOIDmode.
21297 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
21299 PR rtl-optimization/85423
21300 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
21301 dependencies to debug insns when the previous insn is non-debug.
21303 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
21305 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
21306 enums into a single definition.
21307 (fls): Fix predicates and printing.
21310 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
21312 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
21313 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
21314 and short u6 immediate.
21315 (check_if_valid_sleep_operand): Remove.
21316 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
21318 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21320 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
21321 flag_always_save_lp condition.
21322 * config/nds32/nds32.opt (malways-save-lp): New option.
21324 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21326 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
21327 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
21328 * config/nds32/nds32.h
21329 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
21330 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
21332 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21334 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
21335 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
21337 2018-04-22 Shiva Chen <shiva0217@gmail.com>
21338 Chung-Ju Wu <jasonwucj@gmail.com>
21340 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
21342 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
21343 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
21345 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21347 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
21349 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21351 * config/nds32/nds32-protos.h (nds32_data_alignment,
21352 nds32_local_alignment): Declare.
21353 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
21354 nds32_local_alignment): New functions.
21355 (TARGET_CONSTANT_ALIGNMENT): Define.
21356 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
21358 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21360 * config/nds32/nds32.c
21361 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
21362 (TARGET_MODES_TIEABLE_P): Likewise.
21364 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
21366 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
21367 level Ofast and Og.
21369 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
21370 Chung-Ju Wu <jasonwucj@gmail.com>
21372 * config/nds32/constants.md (unspec_volatile_element): Add enum values
21373 for unaligned access.
21374 * config/nds32/nds32-intrinsic.c: Implementation of expanding
21376 * config/nds32/nds32-intrinsic.md: Likewise.
21377 * config/nds32/nds32_intrinsic.h: Likewise.
21378 * config/nds32/nds32.h (nds32_builtins): Likewise.
21379 * config/nds32/nds32.opt (munaligned-access): New option.
21380 * config/nds32/nds32.c (nds32_asm_file_start): Display
21381 flag_unaligned_access status.
21383 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
21385 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
21386 -mno-relax is present.
21387 * config/riscv/linux.h (LINK_SPEC): Ditto.
21389 2018-04-20 Martin Sebor <msebor@redhat.com>
21392 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
21394 (gimple_fold_builtin_stxcpy_chk): Same.
21395 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
21397 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
21400 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
21401 __powikf2 when long double is IEEE 128-bit.
21403 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
21405 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
21406 step to make sure stack always aligned.
21408 2018-04-20 Carl Love <cel@us.ibm.com>
21411 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
21412 size check for arg0.
21414 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
21415 Tom de Vries <tom@codesourcery.com>
21418 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
21419 Emit insns for calls too.
21420 (nvptx_find_par): Always look for worker-level predecessor insn.
21421 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
21423 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
21424 (nvptx_process_pars): Propagate frames for calls.
21426 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
21429 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
21431 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
21432 (ix86_handle_option): Don't handle OPT_mibt.
21433 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
21435 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
21437 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
21438 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
21439 (ix86_target_macros): Define __CET__ with flag_cf_protection
21440 for -fcf-protection.
21441 * config/i386/i386.c (isa2_opts): Remove -mibt.
21442 * config/i386/i386.h (TARGET_IBT): Removed.
21443 (TARGET_IBT_P): Likewise.
21444 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
21445 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
21446 * config/i386/i386.opt (mcet): Update help message.
21447 (mshstk): Likewise.
21449 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
21450 -mcet as an alias for -mshstk.
21452 2018-04-20 Richard Biener <rguenther@suse.de>
21454 PR middle-end/85475
21455 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
21456 complexity by forcing a single use of the multiply operand.
21458 2018-04-20 Martin Jambor <mjambor@suse.cz>
21461 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
21462 recursion dependency to only apply to non-clones.
21464 2018-04-20 Martin Jambor <mjambor@suse.cz>
21467 * ipa-cp.c (create_specialized_node): Check that clones of
21468 self-recursive edges exist during IPA-CP.
21470 2018-04-19 Toon Moene <toon@moene.org>
21472 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
21475 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21477 PR tree-optimization/85467
21478 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
21479 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
21480 VECTOR_CST element to type.
21482 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21485 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
21486 * config/i386/i386.md (builtin_setjmp_setup): Removed.
21487 (builtin_longjmp): Likewise.
21488 (save_stack_nonlocal): New pattern.
21489 (restore_stack_nonlocal): Likewise.
21491 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21494 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
21495 Replace ASM_OUTPUT_LABEL with fprintf.
21497 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
21500 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
21501 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
21502 * config/i386/i386-c.c (ix86_target_macros_internal): Also
21503 define __IBT__ and __SHSTK__ for -fcf-protection.
21504 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
21506 (ix86_trampoline_init): Likewise.
21507 (x86_output_mi_thunk): Likewise.
21508 (ix86_notrack_prefixed_insn_p): Likewise.
21509 (ix86_option_override_internal): Don't disallow -fcf-protection.
21510 * config/i386/i386.md (rdssp<mode>): Also enable for
21512 (incssp<mode>): Likewise.
21513 (nop_endbr): Likewise.
21514 * config/i386/i386.opt (mcet): Change help message to built-in
21517 (mshstk): Likewise.
21518 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
21519 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
21520 enable CET built-in functions.
21522 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
21524 * common/config/i386/i386-common.c
21525 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
21526 OPTION_MASK_ISA_MOVDIRI_UNSET,
21527 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
21528 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
21529 * config.gcc (movdirintrin.h): New header.
21530 * config/i386/cpuid.h (bit_MOVDIRI,
21531 bit_MOVDIR64B): New bits.
21532 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
21534 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
21535 (VOID, PVOID, PCVOID)): New function types.
21536 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
21537 __builtin_ia32_directstoreu_u64,
21538 __builtin_ia32_movdir64b): New builtins.
21539 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
21540 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
21542 (ix86_valid_target_attribute_inner_p): Ditto.
21543 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
21544 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
21545 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
21546 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
21547 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
21548 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
21549 (movdiri<mode>, movdir64b_<mode>): New.
21550 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
21551 * config/i386/immintrin.h: Include movdirintrin.h.
21552 * config/i386/movdirintrin.h: New file.
21553 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
21555 2018-04-19 Richard Biener <rguenther@suse.de>
21557 PR middle-end/85455
21558 * cfg.c (clear_bb_flags): When loop state says we have
21559 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
21561 2018-04-19 Richard Biener <rguenther@suse.de>
21563 PR tree-optimization/84737
21564 * tree-vect-data-refs.c (vect_copy_ref_info): New function
21565 copying restrict info.
21566 (vect_setup_realignment): Use it.
21567 * tree-vectorizer.h (vect_copy_ref_info): Declare.
21568 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
21569 the first DR to all generated stores.
21570 (vectorizable_load): Likewise for loads.
21572 2018-04-19 Jakub Jelinek <jakub@redhat.com>
21574 PR tree-optimization/85446
21575 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
21576 the integral and pointer types to have the same precision.
21578 * doc/install.texi: Document --disable-cet being the default and
21581 2018-04-18 Martin Liska <mliska@suse.cz>
21583 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
21586 2018-04-18 Martin Liska <mliska@suse.cz>
21589 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
21592 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
21593 arguments if they are comparable.
21595 2018-04-18 Martin Liska <mliska@suse.cz>
21598 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21601 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21604 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
21607 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
21608 ENDBR after calling __morestack.
21610 2018-04-18 David Malcolm <dmalcolm@redhat.com>
21613 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
21614 by using gcc_base_ver to generate a gcc_driver_version, and use
21615 it when generating GCC_DRIVER_NAME.
21616 * configure: Regenerate.
21618 2018-04-18 Jakub Jelinek <jakub@redhat.com>
21621 * config.gcc: Obsolete powerpc*-*-*spe*.
21623 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21626 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
21627 (stabstr_D): Change type of unum from unsigned int to
21628 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
21631 2018-04-17 Jim Wilson <jimw@sifive.com>
21634 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
21635 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
21636 Set arg_pointer_offset after using pretend_args_size.
21638 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21640 PR rtl-optimization/85431
21641 * dse.c (record_store): Ignore zero width stores.
21644 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
21645 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
21646 __builtin_stack_restore rather than after it.
21647 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
21648 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
21649 argument instead of virtual_dynamic_stack_rtx.
21651 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
21653 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
21655 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21656 Add note to error message to explain internal mapping of overloaded
21657 built-in function name to non-overloaded built-in function name.
21658 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
21661 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
21664 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
21665 where the inputs overlap with the output.
21667 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21669 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
21670 (=v, v) alternative and explicit "memory" attribute.
21671 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
21672 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21674 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
21675 "sselog1" type instead of "sselog".
21676 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
21677 "sselog". Remove explicit "memory" attribute.
21678 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
21679 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21681 (vec_extract_hi_v32hi): Merge all alternatives into one, use
21682 "sselog1" type instead of "sselog". Remove explicit "memory"
21684 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
21685 use "sselog1" type instead of "sselog". Remove explicit "memory"
21687 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
21688 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
21690 (vec_extract_hi_v64qi): Merge all alternatives into one, use
21691 "sselog1" type instead of "sselog". Remove explicit "memory"
21693 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
21694 use "sselog1" type instead of "sselog". Remove explicit "memory"
21698 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
21700 PR middle-end/85414
21701 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
21704 2018-04-17 Martin Jambor <mjambor@suse.cz>
21707 * ipa-cp.c (create_specialized_node): Call
21708 expand_all_artificial_thunks if necessary.
21710 2018-04-17 Martin Liska <mliska@suse.cz>
21713 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
21714 in message, remote space in between '_G' and '('.
21716 2018-04-17 Jakub Jelinek <jakub@redhat.com>
21719 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
21720 avx512f_vmcmp<mode>3<round_saeonly_name>,
21721 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
21722 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
21723 avx512f_rndscale<mode><round_saeonly_name>,
21724 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
21725 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
21726 Use %<iptr>2 instead of %2 for -masm=intel.
21727 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
21728 avx512f_vcvttss2usi<round_saeonly_name>,
21729 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
21731 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
21732 avx512f_vcvttsd2usi<round_saeonly_name>,
21733 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
21734 Use %q1 instead of %1 for -masm=intel.
21735 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
21736 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
21737 of %3 for -masm=intel.
21738 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
21740 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
21742 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
21744 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
21745 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
21747 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
21749 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
21750 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
21751 %g1 and one with %0 and %1.
21752 (avx512er_vmrcp28<mode><round_saeonly_name>,
21753 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
21754 %1 for -masm=intel.
21755 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
21756 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
21757 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
21758 of %0 and %{%4%} for -masm=intel.
21759 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
21760 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
21761 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
21762 order of %0 and %{%5%}%{z%} for -masm=intel.
21764 2018-04-17 Jan Hubicka <jh@suse.cz>
21767 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
21769 2018-04-17 Martin Liska <mliska@suse.cz>
21772 * multiple_target.c (create_dispatcher_calls): Set apostrophes
21773 for target_clone error message. Make default implementation
21774 clone to be a local declaration.
21775 (separate_attrs): Add new argument and check for an empty
21777 (expand_target_clones): Handle it.
21778 (ipa_target_clone): Make redirection just for target_clones
21781 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
21782 Tom de Vries <tom@codesourcery.com>
21784 PR middle-end/84955
21785 * omp-expand.c (expand_oacc_for): Add dummy false branch for
21786 tiled basic blocks without omp continue statements.
21788 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
21791 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
21792 vec_extract expression as having side effects to make sure it gets
21795 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
21798 * config/i386/i386.c (get_builtin_code_for_version): Check
21801 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
21804 * config.gcc: Support "skylake".
21805 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21807 * config/i386/i386.c (m_SKYLAKE): Define.
21808 (processor_target_table): Add "skylake".
21809 (ix86_option_override_internal): Add "skylake".
21810 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
21811 PROCESSOR_CANNONLAKE.
21812 (get_builtin_code_for_version): Fix priority for
21813 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
21814 PROCESSOR_SKYLAKE-AVX512.
21815 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
21816 (processor_type): Add PROCESSOR_SKYLAKE.
21818 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
21819 Jason Merrill <jason@redhat.com>
21822 * convert.c (convert_to_integer_1): Use direct recursion for
21823 enumeral types and types with a precision less than the number
21824 of bits in their mode.
21826 2018-04-16 Julia Koval <julia.koval@intel.com>
21829 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
21830 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
21832 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
21835 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
21836 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
21837 and -mno-direct-move.
21839 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
21842 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
21843 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
21844 Ensure negative shifts result in {0}.
21846 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
21848 PR rtl-optimization/79916
21849 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
21850 regs (if any) to define how to gnerate SD moves when LRA is in
21853 2018-04-13 Jakub Jelinek <jakub@redhat.com>
21855 PR rtl-optimization/85393
21856 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
21857 * except.c (expand_dw2_landing_pad_for_region): Make static.
21858 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
21859 a label and unconditional jump to old_bb, rather than
21860 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
21863 PR rtl-optimization/85376
21864 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
21865 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
21866 instead of a specific value.
21868 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
21869 Bin Cheng <bin.cheng@arm.com>
21871 PR tree-optimization/82965
21872 PR tree-optimization/83991
21873 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
21874 by_profile_only parameter.
21875 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
21876 information if the loop was predicted to iterate too many times.
21877 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
21879 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
21882 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
21885 2018-04-13 Martin Liska <mliska@suse.cz>
21886 Jakub Jelinek <jakub@redhat.com>
21888 PR middle-end/81657
21889 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
21890 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
21891 * builtins.c (expand_builtin_memory_copy_args): Use
21892 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
21893 handle dest_addr == pc_rtx.
21895 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21898 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
21899 asked to not generate direct moves.
21900 (fix_trunc<mode>si2_stfiwx): Similar.
21901 (fix_trunc<mode>si2_internal): Similar.
21903 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21906 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
21907 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
21908 lookup if dest in some wider mode is known to be const0_rtx and
21909 if so, record permanent equivalence for it to be ZERO_EXTEND of
21910 the narrower mode destination.
21912 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
21914 * lto-streamer-out.c (output_function): Revert 259346.
21915 * omp-expand.c (expand_oacc_for): Likewise.
21917 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
21919 PR rtl-optimization/85354
21920 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
21921 * sel-sched.c (sel_global_init): ... here.
21923 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
21926 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
21927 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
21928 mode for PE-COFF targets.
21929 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
21930 (i386_pe_asm_lto_end): Likewise.
21931 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
21932 (TARGET_ASM_LTO_END): Likewise.
21933 * config/i386/winnt.c (saved_debug_info_level): New static variable.
21934 (i386_pe_asm_lto_start): New function.
21935 (i386_pe_asm_lto_end): Likewise.
21937 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
21938 Richard Biener <rguenther@suse.de>
21940 PR middle-end/84955
21941 * lto-streamer-out.c (output_function): Fix CFG loop state before
21943 * omp-expand.c (expand_oacc_for): Handle calls to internal
21944 functions like regular functions.
21946 2018-04-12 Richard Biener <rguenther@suse.de>
21949 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
21950 for the early LTO debug to properly generate references to it
21951 during DIE emission. Do not re-use that for the skeleton for
21953 (dwarf2out_early_finish): Likewise.
21955 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21958 * config/i386/sse.md
21959 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
21960 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
21961 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
21962 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
21963 and output is a reg, avoid creating invalid lowpart subreg, but
21964 instead split into a 512-bit move. Don't split if not AVX512VL,
21965 input is xmm16+ reg and output is a mem.
21966 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
21967 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
21968 xmm16+ reg and output is a mem.
21970 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21972 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
21973 also for flag_dwarf2_cfi_asm.
21975 2018-04-12 Jakub Jelinek <jakub@redhat.com>
21977 PR rtl-optimization/85342
21978 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
21979 a bool scalar var inside of the loop instead. Don't try to update
21980 recog_data.operand after failed apply_change_group.
21982 2018-04-12 Tom de Vries <tom@codesourcery.com>
21985 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
21986 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
21987 array with flexible array member as array without given dimension.
21988 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
21989 argument for undefined param to true.
21991 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
21994 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
21995 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
21996 from PowerPC section.
21997 * config/rs6000/sysv4.opt (mcall-): Improve help text.
21998 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
21999 help text that is too long.
22000 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
22001 help text that is too long.
22002 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
22003 help text that is too long.
22005 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
22007 * config/alpha/alpha.md (stack_probe_internal): Rename
22008 from "probe_stack". Update all callers.
22010 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22012 PR rtl-optimization/84566
22013 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
22014 sched_macro_fuse_insns.
22016 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22019 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
22020 (compute_block_dependences): ... from here.
22022 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22024 PR tree-optimization/85331
22025 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
22026 from int to HOST_WIDE_INT.
22028 2018-04-11 Martin Jambor <mjambor@suse.cz>
22031 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
22032 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
22033 not the same as the source val.
22034 (cgraph_edge_brings_value_p): New parameter.
22035 (gather_edges_for_value): Pass destination value to
22036 cgraph_edge_brings_value_p.
22037 (perhaps_add_new_callers): Likewise.
22038 (get_info_about_necessary_edges): Likewise and exclude values brought
22039 only by self-recursive edges.
22040 (create_specialized_node): Redirect only clones of self-calling edges.
22041 (+self_recursive_pass_through_p): New function.
22042 (find_more_scalar_values_for_callers_subset): Use it.
22043 (find_aggregate_values_for_callers_subset): Likewise.
22044 (known_aggs_to_agg_replacement_list): Removed.
22045 (decide_whether_version_node): Re-calculate known constants for all
22046 remaining context clones.
22048 2018-04-11 Richard Biener <rguenther@suse.de>
22051 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
22052 from early DWARF output.
22053 (dwarf2out_early_finish): Output line info unconditionally into
22054 early DWARF and add reference to it.
22056 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22059 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
22060 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
22061 other than V2DFmode using iptr mode attribute.
22062 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
22064 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
22066 PR rtl-optimization/84659
22067 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
22069 2018-04-11 Jakub Jelinek <jakub@redhat.com>
22072 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
22074 (output_loc_list): Pass address of a dummy size variable even in the
22075 locview handling loop.
22076 (index_location_lists): Add comment on why skip_loc_list_entry can't
22079 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
22082 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
22085 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
22088 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
22089 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
22090 and -mstring-compare-inline-limit.
22092 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22095 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
22096 for stack clash protection in a register whenever we need it to be in
22099 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22101 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
22102 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
22104 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
22107 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
22109 (mlong-double-): Ditto.
22110 * config/rs6000/sysv4.opt (msdata=): Ditto.
22111 (mtls-size=): Ditto.
22113 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22115 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
22116 erroneous entries for
22117 "vector int vec_ldl (int, long int *)", and
22118 "vector unsigned int vec_ldl (int, unsigned long int *)".
22119 Add comments and entries for
22120 "vector bool char vec_ldl (int, bool char *)",
22121 "vector bool short vec_ldl (int, bool short *)",
22122 "vector bool int vec_ldl (int, bool int *)",
22123 "vector bool long long vec_ldl (int, bool long long *)",
22124 "vector pixel vec_ldl (int, pixel *)",
22125 "vector long long vec_ldl (int, long long *)",
22126 "vector unsigned long long vec_ldl (int, unsigned long long *)".
22127 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
22128 type tree bool_long_long_type_node and correct definition of
22129 bool_V2DI_type_node to make reference to this new type tree.
22130 (rs6000_mangle_type): Replace erroneous reference to
22131 bool_long_type_node with bool_long_long_type_node.
22132 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
22133 comments to emphasize sign distinctions for char and int types and
22134 replace RS6000_BTI_bool_long constant with
22135 RS6000_BTI_bool_long_long constant. Also add comment to restrict
22136 use of RS6000_BTI_pixel.
22137 (bool_long_type_node): Remove this macro definition.
22138 (bool_long_long_type_node): New macro definition
22140 2018-04-10 Jakub Jelinek <jakub@redhat.com>
22142 PR rtl-optimization/85300
22143 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
22144 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
22145 simplify_unary_operation fails.
22147 2018-04-10 Martin Liska <mliska@suse.cz>
22149 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
22150 cgraph_edge and ipa_ref.
22152 2018-04-10 Jakub Jelinek <jakub@redhat.com>
22156 * config/i386/sse.md
22157 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
22158 computation of the VEC_MERGE selector from mask.
22159 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
22160 Fix decoding of the VEC_MERGE selector into mask.
22162 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
22164 PR tree-optimization/85286
22165 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
22167 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
22169 * final.c (final_1): Set insn_last_address as well as
22170 insn_current_address.
22172 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22175 * explow.c (emit_stack_probe): Call validize_mem on memory location
22176 before passing it to gen_probe_stack. Create address operand and
22177 legitimize it for the probe_stack_address case.
22179 2018-04-09 Jan Hubicka <jh@suse.cz>
22182 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
22183 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
22184 * tree.c (free_lang_data_in_type): Fix handling of binfos;
22186 (free_lang_data): Rebuild type inheritance graph.
22188 2018-04-09 Martin Sebor <msebor@redhat.com>
22190 * invoke.texi (-finline-small-functions): Mention other optimization
22192 (-findirect-inlining, -fpartial-inlining): Same.
22193 (-finline-functions-called-once): Same.
22194 (-freorder-blocks-and-partition): Same.
22196 2018-04-09 Jan Hubicka <jh@suse.cz>
22199 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
22200 jumps; choose last target that matches the criteria (i.e.
22201 no partition changes for non-crossing jumps).
22202 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
22203 support for redirecting crossing jumps to non-crossing.
22205 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
22207 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
22208 also for naked functions.
22210 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
22212 * config/arc/arc.md (add_shift): New pattern.
22213 (add_shift2): Likewise.
22214 (sub_shift): Likewise.
22215 (sub_shift_cmp0_noout): Likewise.
22216 (compare_si_ashiftsi): Likewise.
22217 (xbfu_cmp0_noout): New combine pattern.
22218 (xbfu_cmp0"): Likewise.
22219 (movsi_set_cc_insn): Place the predicable variant first.
22220 (commutative_binary_cmp0_noout): Remove clobber.
22221 (commutative_binary_cmp0): New pattern.
22222 (noncommutative_binary_cmp0): Likewise.
22223 (noncommutative_binary_cmp0_noout): Likewise.
22224 (noncommutative_binary_comparison_result_used): Removed.
22225 (rsub_cmp0): New pattern.
22226 (rsub_cmp0_noout): Likewise.
22227 (extzvsi): Changed, keep only meaningful variants.
22228 (SQH, SEZ): New iterators.
22229 (SQH_postfix): New mode attribute.
22230 (SEZ_prefix): New code attribute.
22231 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
22232 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
22233 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
22234 of numerical value.
22235 (noncommutative_operator): Check the availability of barrel
22238 2018-04-09 Richard Biener <rguenther@suse.de>
22240 PR tree-optimization/85284
22241 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
22242 Only use the niter constraining form of simple_iv when the exit
22243 is always executed.
22245 2018-04-09 Tom de Vries <tom@codesourcery.com>
22248 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22249 (define_expand "*memory_barrier"): New define_expand.
22250 (define_insn "memory_barrier"): New insn.
22252 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22254 PR rtl-optimization/80463
22255 PR rtl-optimization/83972
22256 PR rtl-optimization/83480
22258 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
22259 correct producer for the insn.
22260 (tidy_control_flow): Fixup seqnos in case of debug insns.
22262 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22264 PR rtl-optimization/83913
22266 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
22267 different sched-times when merging exprs.
22269 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22271 PR rtl-optimization/83962
22273 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
22274 tidy_fallthru_edge and tidy_control_flow.
22276 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
22278 PR rtl-optimization/83530
22280 * sel-sched.c (force_next_insn): New global variable.
22281 (remove_insn_for_debug): When force_next_insn is true, also leave only
22282 next insn in the ready list.
22283 (sel_sched_region): When the region wasn't scheduled, make another pass
22284 over it with force_next_insn set to 1.
22286 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
22288 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
22290 * config/nds32/constants.md (unspec_volatile_element): Add enum values
22291 for interrupt control.
22292 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
22293 functions for interrupt control.
22294 * config/nds32/nds32-intrinsic.md: Likewise.
22295 * config/nds32/nds32_intrinsic.h: Likewise.
22296 * config/nds32/nds32.h (nds32_builtins): Likewise.
22298 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
22300 * config/nds32/nds32.c (nds32_init_machine_status,
22301 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
22302 strict_aligned_p field.
22303 (nds32_expand_to_rtl_hook): New function.
22304 (TARGET_EXPAND_TO_RTL_HOOK): Define.
22305 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
22307 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22308 Chung-Ju Wu <jasonwucj@gmail.com>
22310 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
22311 * config/nds32/nds32-n7.md: New file.
22312 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
22313 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
22315 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
22316 * config/nds32/nds32.md (pipeline_model): Add n7.
22317 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
22318 * config/nds32/pipelines.md: Include n7 settings.
22320 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22321 Chung-Ju Wu <jasonwucj@gmail.com>
22323 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
22324 * config/nds32/nds32-e8.md: New file.
22325 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
22326 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
22328 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
22329 * config/nds32/nds32.md (pipeline_model): Add e8.
22330 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
22331 * config/nds32/pipelines.md: Include e8 settings.
22333 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22334 Chung-Ju Wu <jasonwucj@gmail.com>
22336 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
22337 * config/nds32/nds32-n8.md: New file.
22338 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
22339 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
22341 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
22342 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
22343 * config/nds32/nds32.md (pipeline_model): Add n8.
22344 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
22345 * config/nds32/pipelines.md: Include n8 settings.
22347 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
22348 Chung-Ju Wu <jasonwucj@gmail.com>
22350 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
22351 * config/nds32/nds32-n9-2r1w.md: New file.
22352 * config/nds32/nds32-n9-3r2w.md: New file.
22353 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
22354 nds32_register_ports): New or modify for cpu n9.
22355 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
22357 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
22358 * config/nds32/nds32-utils.c: New file.
22359 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
22360 TARGET_MUL_SLOW): Define.
22361 * config/nds32/nds32.md (pipeline_model): New attribute.
22362 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
22363 New options that support cpu n9.
22364 * config/nds32/pipelines.md: Include n9 settings.
22365 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
22367 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
22369 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
22370 information if necessary.
22371 (output_cond_branch_compare_zero): Likewise.
22372 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
22373 (nds32_target_alignment): Refine for alignment.
22374 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
22375 (FUNCTION_BOUNDARY): Modify.
22376 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
22378 * config/nds32/nds32.opt (malways-align, malign-functions): New.
22380 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
22382 * config/nds32/constants.md (unspec_volatile_element): Add values for
22383 TLB operation and data prefetch.
22384 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
22385 functions for TLB operation and data prefetch.
22386 * config/nds32/nds32-intrinsic.md: Likewise.
22387 * config/nds32/nds32_intrinsic.h: Likewise.
22388 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
22389 (nds32_print_operand): Likewise.
22390 * config/nds32/nds32.h (nds32_builtins): Likewise.
22392 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22393 Andrew Pinski <pinsika@gcc.gnu.org>
22395 PR middle-end/82976
22396 * match.pd: Use constant_boolean_node of correct type instead of
22397 boolean_true_node or boolean_false_node for simplifying
22398 pointer comparisons to zero.
22400 2018-04-07 Jakub Jelinek <jakub@redhat.com>
22402 PR tree-optimization/80021
22403 * tree.c (verify_type_variant): Make error call in verify_variant_match
22404 translatable and remove final full stop.
22406 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22408 * config/nds32/constants.md (unspec_volatile_element): Add
22409 UNSPEC_VOLATILE_EH_RETURN.
22410 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
22411 nds32_output_stack_pop): Support dwarf exception handling process.
22412 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
22413 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
22414 exception handling process.
22415 (nds32_compute_stack_frame): Likewise.
22416 (nds32_return_addr_rtx): Likewise.
22417 (nds32_initial_elimination_offset): Likewise.
22418 (nds32_expand_prologue): Likewise.
22419 (nds32_expand_epilogue): Likewise.
22420 (nds32_dynamic_chain_address): New function.
22421 * config/nds32/nds32.h (machine_function): Add fields for dwarf
22422 exception handling.
22423 (DYNAMIC_CHAIN_ADDRESS): Define.
22424 (EH_RETURN_DATA_REGNO): Define.
22425 (EH_RETURN_STACKADJ_RTX): Define.
22426 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
22427 patterns for dwarf exception handling.
22429 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22431 * config/nds32/nds32.h: Clean up obsolete macros.
22433 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22435 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22436 Add enum values for particular instructions.
22437 * config/nds32/nds32-intrinsic.c: Implementation of expanding
22438 particular intrinsic functions.
22439 * config/nds32/nds32-intrinsic.md: Likewise.
22440 * config/nds32/nds32_intrinsic.h: Likewise.
22441 * config/nds32/nds32.h (nds32_builtins): Likewise.
22442 * config/nds32/nds32.md (type): Add pbsad and pbsada.
22443 (btst, ave): New patterns for particular instructions.
22445 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22447 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22448 Add enum values for atomic load/store and memory sync.
22449 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
22451 * config/nds32/nds32-intrinsic.md: Likewise.
22452 * config/nds32/nds32_intrinsic.h: Likewise.
22453 * config/nds32/nds32.h (nds32_builtins): Likewise.
22455 2018-04-07 Jakub Jelinek <jakub@redhat.com>
22457 PR tree-optimization/85257
22458 * fold-const.c (native_encode_vector): If not all elts could fit
22459 and off is -1, return 0 rather than offset.
22460 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
22461 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
22462 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
22463 adjust buffer in native_interpret_expr call.
22465 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22467 * config/nds32/constants.md (unspec_volatile_element): Add cache
22468 control enum values.
22469 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
22470 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
22471 * config/nds32/nds32.c (nds32_cctl_names): New.
22472 (nds32_print_operand): Handle cache control register names.
22473 * config/nds32/nds32.h (nds32_builtins): New enum values.
22474 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
22476 * config/nds32/nds32.md (type): Add mmu.
22477 * config/nds32/pipelines.md (simple_insn): Add mmu.
22479 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22481 * config/nds32/nds32.md (type): Remove call.
22482 * config/nds32/pipelines.md (simple_insn): Likewise.
22484 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22486 * config/nds32/constants.md (unspec_volatile_element): Add
22487 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
22488 UNSPEC_VOLATILE_FMFCFG.
22489 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
22490 description for fmfcfg and fmfcsr.
22491 (bdesc_1arg): Add fmtcsr.
22492 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
22493 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
22494 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
22495 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
22496 unspec_fmfcfg): New patterns.
22497 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
22498 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
22499 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
22500 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
22501 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
22502 __nds32__fmfcfg): Define.
22504 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
22506 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
22507 intrinsic register names.
22508 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
22509 intrinsic register enum values and macros.
22511 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
22513 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
22514 for load/store addressing form.
22515 (nds32_print_operand_address): Likewise.
22517 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
22520 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
22521 based on LABEL_REF. Remove useless assertion.
22522 (pic_address_needs_scratch): Fix formatting.
22523 (sparc_legitimize_pic_address): Minor tweaks.
22524 (sparc_delegitimize_address): Adjust assertion accordingly.
22525 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
22526 into symbolic_operand.
22527 (movsi_high_pic_label_ref): Likewise.
22528 (movsi_lo_sum_pic_label_ref): Likewise.
22529 (movdi_pic_label_ref): Likewise.
22530 (movdi_high_pic_label_ref): Likewise.
22531 (movdi_lo_sum_pic_label_ref): Likewise.
22533 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
22535 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
22536 custom LIB_SPEC setup.
22538 2018-04-06 Ruslan Bukin <br@bsdpad.com>
22539 Kito Cheng <kito.cheng@gmail.com>
22541 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
22542 * config/riscv/freebsd.h: New.
22544 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22546 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
22547 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
22550 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22551 Kito Cheng <kito.cheng@gmail.com>
22553 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
22554 nds32_output_call, nds32_symbol_binds_local_p): New functions.
22555 * config/nds32/nds32-protos.h (nds32_output_call,
22556 nds32_output_return): Declare.
22557 * config/nds32/nds32.md: Refine all the call and return patterns.
22559 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22562 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
22563 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
22565 PR rtl-optimization/84872
22566 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
22567 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
22568 EDGE_CROSSING edge.
22570 2018-04-06 Tamar Christina <tamar.christina@arm.com>
22572 * expr.c (copy_blkmode_to_reg): Revert 254862.
22573 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
22575 2018-04-06 Richard Biener <rguenther@suse.de>
22577 PR middle-end/85244
22578 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
22579 after seeing a component reference with an adjacent field. Treat
22580 refs to arrays at struct end of external decls similar to
22581 refs to unconstrained commons.
22583 2018-04-06 Jakub Jelinek <jakub@redhat.com>
22586 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
22587 look through SAVE_EXPRs with non-side-effects argument. Adjust
22589 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
22592 2018-04-06 Richard Biener <rguenther@suse.de>
22594 PR middle-end/85180
22595 * alias.c (find_base_term): New wrapper around find_base_term
22596 unwinding CSELIB_VAL_PTR changes.
22597 (find_base_term): Do not restore CSELIB_VAL_PTR during the
22600 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22602 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
22604 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
22605 constant definitions.
22606 ("nop"): lr 0,0 -> nopr r0
22607 ("nop_lr0", "nop_lr1"): New insn definitions.
22609 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
22611 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
22612 NDS32_V3PUSH_AVAILABLE_P macro.
22614 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
22615 Chung-Ju Wu <jasonwucj@gmail.com>
22617 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
22618 (nds32*-*-*): Add float and fpu_config into supported_defaults.
22619 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
22620 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
22621 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
22622 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
22623 * config/nds32/constraints.md: New constraints and checking for hard
22624 float configuration.
22625 * config/nds32/iterators.md: New mode iterator and attribute for hard
22626 float configuration.
22627 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
22629 * config/nds32/nds32-fpu.md: New file.
22630 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
22631 deal with hard float code generation.
22632 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
22634 (abi_type, float_reg_number): New enum type.
22635 * config/nds32/nds32-predicates.c: New predicates for hard float.
22636 * config/nds32/nds32-protos.h: Declare functions for hard float.
22637 * config/nds32/nds32.c: Implementation for hard float configuration.
22638 * config/nds32/nds32.h: Definitions for hard float configuration.
22639 * config/nds32/nds32.md: Include hard float machine description and
22640 modify patterns for hard float configuration.
22641 * config/nds32/nds32.opt: New options for hard float configuration.
22642 * config/nds32/predicates.md: New predicates for hard float
22645 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22647 * common/config/nds32/nds32-common.c
22648 (nds32_option_optimization_table): Enable -mreleax-hint by default.
22650 2018-04-05 Jakub Jelinek <jakub@redhat.com>
22652 PR middle-end/85195
22653 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
22654 CONSTRUCTOR_ELT (ctor, ...)->value.
22656 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
22659 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
22661 2018-04-05 Tom de Vries <tom@codesourcery.com>
22664 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
22667 2018-04-05 Shiva Chen <shiva0217@gmail.com>
22668 Kito Cheng <kito.cheng@gmail.com>
22670 * config/nds32/constraints.md (U33): Fine-tune checking condition.
22671 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
22672 * config/nds32/nds32.h (nds32_16bit_address_type): Add
22673 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
22675 2018-04-05 Shiva Chen <shiva0217@gmail.com>
22676 Kito Cheng <kito.cheng@gmail.com>
22678 * config/nds32/constraints.md (Ufe): New memory constraint.
22679 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
22680 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
22681 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
22683 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
22684 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
22686 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22688 * config/nds32/nds32.md: Use optimize_size in the condition for
22689 alu-shift instructions.
22691 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22693 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
22695 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22697 * config/nds32/nds32.md (negsi2): Refine pattern.
22699 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
22700 Chung-Ju Wu <jasonwucj@gmail.com>
22702 * config/nds32/iterators.md (shift_rotate): New code iterator.
22703 (shift): New code attribute.
22704 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
22705 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
22706 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
22707 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
22708 bit-wise operations.
22709 (andsi3, *andsi3): Ditto.
22710 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
22711 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
22712 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
22713 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
22714 nds32_ior_operand, nds32_xor_operand): New predicates.
22716 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22718 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
22719 (addsi3, subsi3): ... this.
22721 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22723 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
22725 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22727 * config/nds32/nds32.md: Adjust indention.
22729 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
22731 * config/nds32/nds32.md (feature): New attribute.
22733 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
22735 * config/nds32/nds32.md (subtype): New attribute.
22737 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22740 * config/arm/arm-builtins.c (arm_expand_builtin): Change
22741 expansion to perform a bitwise AND of the argument followed by a
22742 boolean negation of the result.
22744 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
22746 PR rtl-optimization/84878
22747 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
22748 the basic block. Assert the use reference is not artificial and that
22749 it has an associated insn.
22751 2018-04-04 Michael Matz <matz@suse.de>
22753 * builtins.c (compute_objsize): Pass correct operand
22754 to array_at_struct_end_p.
22756 2018-04-04 Richard Biener <rguenther@suse.de>
22759 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
22760 from contexts for DINFO_LEVEL_TERSE and below.
22762 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22764 * config/nds32/nds32-doubleword.md (move_<mode>): Require
22765 resiter_operand condition.
22766 * config/nds32/nds32.md (*move<mode>): Ditto.
22768 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22769 Monk Chiang <sh.chiang04@gmail.com>
22771 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
22773 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22775 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
22777 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22778 Kito Cheng <kito.cheng@gmail.com>
22780 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
22781 nds32_cond_code_str, output_cond_branch,
22782 output_cond_branch_compare_zero, nds32_expand_cbranch,
22783 nds32_expand_cstore, nds32_expand_movcc,
22784 nds32_output_cbranchsi4_equality_zero,
22785 nds32_output_cbranchsi4_equality_reg,
22786 nds32_output_cbranchsi4_equality_reg_or_const_int,
22787 nds32_output_cbranchsi4_greater_less_zero: New functions.
22788 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
22789 nds32_expand_cstore, nds32_expand_movcc,
22790 nds32_output_cbranchsi4_equality_zero,
22791 nds32_output_cbranchsi4_equality_reg,
22792 nds32_output_cbranchsi4_equality_reg_or_const_int,
22793 nds32_output_cbranchsi4_greater_less_zero): Declare.
22794 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
22795 nds32_rimm11s_operand): New predicates.
22796 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
22797 * config/nds32/nds32.md: Rewrite all the branch and conditional move
22800 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
22802 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
22803 * config/nds32/nds32.md: Ditto.
22804 * config/nds32/pipelines.md: Ditto.
22806 2018-04-04 Richard Biener <rguenther@suse.de>
22808 PR tree-optimization/85168
22809 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
22810 propagating abnormals.
22812 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22814 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
22816 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
22817 Kito Cheng <kito.cheng@gmail.com>
22819 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
22820 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
22821 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
22822 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
22823 * config/nds32/nds32.md (sibcall_internal): New.
22824 (sibcall_register): Remove.
22825 (sibcall_immediate): Remove.
22826 (sibcall_value_internal): New.
22827 (sibcall_value_register): Remove.
22828 (sibcall_value_immediate): Remove.
22829 * config/nds32/predicates.md (nds32_general_register_operand): New.
22830 (nds32_call_address_operand): New.
22832 2018-04-03 Jakub Jelinek <jakub@redhat.com>
22834 PR rtl-optimization/85167
22835 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
22836 bb_defs if *split_p, instead preinitialize it to NULL.
22838 PR tree-optimization/85156
22839 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
22840 evaluating the argument multiple times.
22842 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
22844 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
22846 (_mm_cvtpd_ps): Likewise.
22847 (_mm_cvttpd_epi32): Likewise.
22848 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
22849 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
22850 vector, pixel, and bool following altivec.h include.
22852 2018-04-03 Martin Sebor <msebor@redhat.com>
22854 * doc/extend.texi (Common Function Attributes): Clarify.
22855 (const attribute): Likewise.
22856 (pure attribute): Likewise.
22858 2018-04-03 Jakub Jelinek <jakub@redhat.com>
22861 * config/i386/i386.c (ix86_expand_vector_set): Use
22862 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
22864 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
22866 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
22867 instructions when changing rounding bits to preserve precision bits
22868 in the x87 control word.
22870 2018-04-03 Martin Liska <mliska@suse.cz>
22872 PR tree-optimization/82491
22873 * rtl.h (strip_offset_and_add): Replace += suboffset with
22874 poly_uint64 () + suboffset.
22876 2018-03-29 Martin Liska <mliska@suse.cz>
22877 Martin Jambor <mjambor@suse.cz>
22880 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
22881 param_type is not an integral or pointer type.
22883 2018-04-03 Richard Biener <rguenther@suse.de>
22885 * sese.h (recompute_all_dominators): Remove.
22887 2018-04-02 Martin Sebor <msebor@redhat.com>
22889 * doc/invoke.texi (-Wrestrict): Fix typos.
22891 2018-04-02 Jim Wilson <jimw@sifive.com>
22893 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
22894 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
22895 (<optab>di3, <optab>si3_extend): Likewise.
22896 (<optab>si3_mask, <optab>si3_mask_1): New.
22897 (<optab>di3_mask, <optab>di3_mask_1): New.
22898 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
22899 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
22900 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
22902 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
22904 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
22907 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
22909 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
22910 (nds32_canonicalize_comparison): New function.
22912 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22913 Kito Cheng <kito.cheng@gmail.com>
22914 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22916 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
22917 * config/nds32/constants.md (unspec_volatile_element): Add
22918 UNSPEC_VOLATILE_RELAX_GROUP.
22919 * config/nds32/nds32-relax-opt.c: New file.
22920 * config/nds32/nds32-predicates.c
22921 (nds32_symbol_load_store_p): New function.
22922 * config/nds32/nds32-protos.h
22923 (nds32_symbol_load_store_p): Declare function.
22924 (make_pass_nds32_relax_opt): Declare new rtl pass function.
22925 * config/nds32/nds32.c
22926 (nds32_register_pass): New function to register pass.
22927 (nds32_register_passes): New function to register passes.
22928 * config/nds32/nds32.md (relax_group): New pattern.
22929 * config/nds32/nds32.opt (mrelax-hint): New option.
22930 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
22932 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
22934 * config/nds32/t-nds32: Modify files dependency.
22936 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22938 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
22939 (PROFILE_HOOK): Define its implementation.
22941 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
22943 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
22944 type and 32-bit size.
22946 2018-04-01 Jakub Jelinek <jakub@redhat.com>
22948 PR middle-end/85090
22949 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
22950 (V_128_256): New mode iterator.
22951 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
22952 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
22953 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
22955 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
22956 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
22958 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22961 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
22962 NaN inputs correctly.
22964 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
22967 * config/rs6000/vsx.md (??r): New mode attribute.
22968 (*vsx_mov<mode>_64bit): Use it.
22969 (*vsx_mov<mode>_32bit): Likewise.
22971 2018-03-30 Martin Sebor <msebor@redhat.com>
22973 PR tree-optimization/84818
22974 * builtins.c (check_access): Use warning_n.
22976 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22979 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
22981 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
22984 2018-03-30 Julia Koval <julia.koval@intel.com>
22987 * x86-tune.def (movx, partial_reg_dependency): Enable for
22990 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
22992 PR inline-asm/84985
22993 * lra-constraints.c (process_alt_operands): Move setting
22994 this_alternative_matches below.
22996 2018-03-29 Martin Liska <mliska@suse.cz>
22999 * doc/invoke.texi: Document how LTO works with debug info.
23000 Describe auto-load support of binutils. Mention 'x86-64'
23001 as valid option value of -march option.
23003 2018-03-29 Jakub Jelinek <jakub@redhat.com>
23005 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
23008 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
23009 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
23010 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
23013 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
23016 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
23017 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
23018 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
23019 for RS6000_BTM_POWERPC64.
23020 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
23021 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
23022 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
23025 (DIVDEU): Likewise.
23027 2018-03-28 Carl Love <cel@us.ibm.com>
23030 2017-09-27 Carl Love <cel@us.ibm.com>
23032 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
23033 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
23034 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
23037 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23039 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
23040 instead of __vector bool.
23041 (_mm_max_pu8): Likewise.
23042 (_mm_min_pi16): Likewise.
23044 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
23047 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
23048 (DIVWEUO): Likewise.
23049 (DIVDEO): Likewise.
23050 (DIVDEUO): Likewise.
23051 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
23052 DIVWEUO and DIVDEUO.
23053 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
23054 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
23055 (div_extend): Likewise.
23056 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
23058 (__builtin_divweuo): Likewise.
23059 (__builtin_divdeo): Likewise.
23060 (__builtin_divdeuo): Likewise.
23062 2018-03-28 Jakub Jelinek <jakub@redhat.com>
23065 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
23066 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
23068 PR tree-optimization/82004
23069 * gimple-match-head.c (optimize_pow_to_exp): New function.
23070 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
23071 Don't fold to exp if optimize_pow_to_exp is false.
23073 2018-03-28 Martin Liska <mliska@suse.cz>
23076 * calls.c (initialize_argument_information): Fix trailing space.
23077 * common.opt: Fix typo and provide better explanation for
23078 -fsanitize-coverage option.
23079 * config/i386/i386.opt: Fix typo.
23081 2018-03-28 Jakub Jelinek <jakub@redhat.com>
23082 Martin Liska <mliska@suse.cz>
23085 * gimplify.c (asan_poison_variable): Don't do the check for
23086 gimplify_omp_ctxp here.
23087 (gimplify_decl_expr): Do it here.
23088 (gimplify_target_expr): Likewise.
23090 2018-03-28 Martin Liska <mliska@suse.cz>
23093 * config/i386/i386.c (ix86_function_arg_advance): Do not call
23094 chkp_type_bounds_count if MPX is not enabled.
23096 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
23098 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
23100 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23103 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
23104 function to create the function decl for complex long double
23105 multiply and divide for -mabi=ieeelongdouble.
23106 (init_float128_ieee): Call it.
23108 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
23111 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
23112 -fcf-protection=branch -mibt.
23113 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
23115 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23118 * config/arm/arm.c (arm_valid_symbolic_address): Handle
23119 arm_word_relocations.
23121 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
23124 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
23125 extern array declarations.
23127 2018-03-27 Richard Biener <rguenther@suse.de>
23129 PR middle-end/84067
23130 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
23131 explicit single_use checks.
23133 2018-03-27 Richard Biener <rguenther@suse.de>
23135 PR tree-optimization/85082
23136 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
23139 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23141 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
23142 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
23143 Turn on fasynchronous-unwind-tables and funwind-tables.
23145 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
23148 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
23149 (*bmi_blsr_<mode>_ccz): Ditto.
23151 2018-03-26 Tom de Vries <tom@codesourcery.com>
23153 PR tree-optimization/85063
23154 * omp-general.c (offloading_function_p): New function. Factor out
23156 * omp-offload.c (pass_omp_target_link::gate): ... here.
23157 * omp-general.h (offloading_function_p): Declare.
23158 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
23159 with attribute omp declare target for offloading functions.
23161 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
23163 PR tree-optimization/84005
23164 * tree-data-ref.h (get_base_for_alignment): Declare.
23165 * tree-data-ref.c (get_base_for_alignment_1): New function.
23166 (get_base_for_alignment): Likewise.
23167 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23168 get_base_for_alignment to find a suitable base object, instead
23169 of always using drb->base_address.
23171 2018-03-23 Jakub Jelinek <jakub@redhat.com>
23173 PR inline-asm/85022
23174 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
23175 known size by default.
23177 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
23179 PR inline-asm/85030
23180 * lra-constraints.c (process_alt_operands): Don't match BLKmode
23181 and non BLKmode operands.
23183 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23186 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
23187 Clean up attributes.
23189 2018-03-23 Richard Biener <rguenther@suse.de>
23192 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
23193 we are going to emit early debug for LTO.
23195 2018-03-23 Jakub Jelinek <jakub@redhat.com>
23197 PR inline-asm/85034
23198 * function.c (match_asm_constraints_1): Don't optimize if input
23199 doesn't satisfy general_operand predicate for output's mode.
23201 PR inline-asm/85022
23202 * alias.c (write_dependence_p): Don't require for x_canonicalized
23203 non-VOIDmode if x has VOIDmode.
23206 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
23207 just don't try to optimize it rather than assert it never happens.
23209 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
23211 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
23212 macro expansions for definition of ST_INTERNAL_<mode> and
23213 LD_INTERNAL_<mode> builtins.
23214 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
23216 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
23218 (altivec_expand_st_builtin): Likewise.
23219 (altivec_expand_builtin): Remove calls to deleted functions.
23220 (rs6000_address_for_altivec): Delete this function.
23221 * config/rs6000/vector.md: Remove expands for
23222 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
23224 2018-03-22 Sudakshina Das <sudi.das@arm.com>
23227 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
23228 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
23229 re-computing once computed.
23230 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
23231 (arm_init_machine_status): Initialize
23232 machine->static_chain_stack_bytes.
23234 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
23237 * doc/extend.texi: Add four new prototypes for vec_ld.
23238 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
23239 definitions for more logical presentation.
23240 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
23241 entries for V1TI variants of __builtin_altivec_ld builtin.
23242 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
23243 handling of V1TI variant of LVX icode pattern.
23244 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
23245 (rs6000_gimple_fold_builtin): Likewise.
23246 (altivec_init_builtins): Add code to define
23247 __builtin_altivec_lvx_v1ti function.
23249 2018-03-22 Jakub Jelinek <jakub@redhat.com>
23251 PR inline-asm/84941
23252 * function.c (match_asm_constraints_1): Don't do the optimization
23253 if input isn't a REG, SUBREG, MEM or constant.
23255 2018-03-22 Tom de Vries <tom@codesourcery.com>
23257 PR tree-optimization/84956
23258 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
23259 bb_has_abnormal_pred.
23261 2018-03-22 Jakub Jelinek <jakub@redhat.com>
23264 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
23265 DECL_INITIAL (decl) to decl at the end.
23266 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
23267 adjust the comment.
23269 2018-03-21 Joseph Myers <joseph@codesourcery.com>
23271 * doc/extend.texi (__builtin_tgmath): Document when complex
23272 integer types are treated as _Complex _Float64.
23274 2018-03-21 Tom de Vries <tom@codesourcery.com>
23276 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
23278 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23280 PR tree-optimization/84960
23281 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
23282 if it is ENTRY block, move them into single succ of ENTRY in that case.
23284 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
23286 PR tree-optimization/84811
23287 * poly-int.h (poly_span_traits): Remove the T3 parameter and
23288 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
23289 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
23290 (known_subrange_p): Update accordingly. Cast each value involved
23291 in the size comparison, rather than casting the result of the
23294 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23296 PR tree-optimization/84982
23297 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
23298 by flipping the least significant bit rather than all bits from
23299 bitpos to bitpos + bitsize - 1.
23301 2018-03-21 Nathan Sidwell <nathan@acm.org>
23303 * doc/extend.texi (Deprecated Features): Remove mention of
23304 long-deleted deprecations.
23306 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23309 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
23310 * configure: Regenerate.
23312 2018-03-21 Tom de Vries <tom@codesourcery.com>
23314 PR tree-optimization/83126
23315 * tree-parloops.c (num_phis): New function.
23316 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
23318 2018-03-21 Nathan Sidwell <nathan@acm.org>
23320 * doc/extend.texi (Deprecated Features): Update deprecated flags,
23321 mention anon-struct/union members and trailing attributes.
23323 2018-03-21 Bin Cheng <bin.cheng@arm.com>
23325 PR tree-optimization/84969
23326 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
23327 builtin memset partitions if they set different rhs values.
23329 2018-03-21 Jakub Jelinek <jakub@redhat.com>
23331 PR rtl-optimization/84989
23332 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
23333 VEC_DUPLICATE with scalar result mode.
23335 2018-03-21 Martin Liska <mliska@suse.cz>
23338 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
23339 not intended return statement.
23341 2018-03-21 Martin Liska <mliska@suse.cz>
23344 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
23345 (chkp_find_bound_slots_1): Limit number of iterations.
23347 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
23350 * Minor grammar fixes for x86 options.
23352 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23355 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
23356 holding REG_CFA_RESTORE notes, instead turn them into a USE.
23358 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
23361 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
23362 (altivec_lvx_<mode>_1op): Likewise.
23363 (altivec_stvx_<mode>_2op): Likewise.
23364 (altivec_stvx_<mode>_1op): Likewise.
23365 (altivec_lvx_<VM2:mode>): New define_expand.
23366 (altivec_stvx_<VM2:mode>): Likewise.
23367 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
23368 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
23369 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
23370 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
23371 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
23372 (rs6000_gen_lvx): Likewise.
23373 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
23374 (altivec_expand_stv_builtin): Likewise.
23375 (altivec_expand_builtin): Likewise.
23376 * config/rs6000/vector.md: Likewise.
23378 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23381 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
23384 2018-03-20 Richard Biener <rguenther@suse.de>
23387 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
23388 sign-conversions as zero, fall back to standard scalar_stmt
23391 2018-03-20 Martin Liska <mliska@suse.cz>
23394 * predict.c (rebuild_frequencies): Handle case when we have
23395 PROFILE_ABSENT, but flag_guess_branch_prob is false.
23397 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23400 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
23401 flag_section_anchors.
23402 * varasm.c (use_blocks_for_decl_p): Remove hack for
23403 dw2_force_const_mem.
23406 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
23408 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
23409 be created, use lowpart_subreg of operands[0] rather than operands[0]
23411 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
23412 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
23413 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
23414 and n constraint instead of aarch64_shift_imm_di and Usd.
23415 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
23416 (*aarch64_<optab>_reg_minus<mode>3): ... this.
23418 2018-03-20 Sudakshina Das <sudi.das@arm.com>
23421 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
23422 to favor GPR over NEON registers.
23423 (<shift>di3_neon): Likewise.
23425 2018-03-20 Tom de Vries <tom@codesourcery.com>
23428 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
23429 (nvptx_process_pars): Emit bar.sync asap and alap.
23431 2018-03-20 Tom de Vries <tom@codesourcery.com>
23434 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
23435 seen_label if seen_label is already set.
23437 2018-03-20 Jakub Jelinek <jakub@redhat.com>
23440 * config/i386/i386.c (fold_builtin_cpu): For features above 31
23441 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
23442 Use 1U instead of 1. Formatting fixes.
23445 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
23446 instead of TREE_TYPE (s1) for the return value.
23448 2018-03-19 Jakub Jelinek <jakub@redhat.com>
23450 PR tree-optimization/84946
23451 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
23452 bitsize + bitsize in poly_uint64 rather than poly_int64.
23455 * dwarf2asm.c: Include fold-const.c.
23456 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
23457 of decl rather than decl itself.
23459 PR rtl-optimization/84643
23460 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
23462 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
23465 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
23466 calling assemble_variable.
23468 2018-03-19 Sudakshina Das <sudi.das@arm.com>
23471 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
23472 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
23474 2018-03-19 Jim Wilson <jimw@sifive.com>
23477 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
23478 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
23479 (riscv_first_stack_step): Likewise.
23480 (riscv_option_override): Use STACK_BOUNDARY instead of
23481 MIN_STACK_BOUNDARY.
23482 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
23483 MIN_STACK_BOUNDARY.
23484 (BIGGEST_ALIGNMENT): Set to 128.
23485 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
23486 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
23489 2018-03-19 Richard Biener <rguenther@suse.de>
23491 PR tree-optimization/84933
23492 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
23493 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
23495 2018-03-19 Richard Biener <rguenther@suse.de>
23497 PR tree-optimization/84859
23498 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
23499 (cond_if_else_store_replacement): Perform sinking operation on
23500 single-store BBs regardless of MAX_STORES_TO_SINK setting.
23501 Generalize what a BB with a single eligible store is.
23503 2018-03-19 Richard Biener <rguenther@suse.de>
23505 PR tree-optimization/84929
23506 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
23507 chrec_is_positive against non-chrec arg.
23509 2018-03-19 Tamar Christina <tamar.christina@arm.com>
23512 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
23514 2018-03-18 Martin Liska <mliska@suse.cz>
23516 PR rtl-optimization/84635
23517 * regrename.c (build_def_use): Use matches_mode only when
23520 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
23522 PR tree-optimization/84913
23523 * tree-vect-loop.c (vectorizable_reduction): Don't try to
23524 vectorize chains of COND_EXPRs.
23526 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23528 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
23530 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23532 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
23534 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
23536 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
23538 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23539 Kito Cheng <kito.cheng@gmail.com>
23541 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
23542 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
23543 (nds32_adjust_reg_alloc_order): New function.
23544 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
23546 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
23548 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
23549 nds32_print_operand, nds32_print_operand_address): Use
23550 HOST_WIDE_INT_PRINT_DEC instead.
23552 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
23554 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
23556 2018-03-17 Jakub Jelinek <jakub@redhat.com>
23559 * config/i386/i386.c (initial_ix86_tune_features,
23560 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
23561 unsigned long long.
23562 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
23563 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
23564 rather than 1u << ix86_tune. Formatting fix.
23565 (ix86_option_override_internal): Change ix86_arch_mask from
23566 unsigned int to unsigned HOST_WIDE_INT, initialize to
23567 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
23568 (ix86_function_specific_restore): Likewise.
23570 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23573 * postreload.c (reload_combine_recognize_pattern): Perform
23574 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
23575 truncate_int_for_mode the result for the destination's mode.
23578 * hsa-gen.c (mem_type_for_type): Fix comment typo.
23579 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
23581 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23584 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
23587 * lra-assigns.c (lra_split_hard_reg_for): Don't use
23588 regno_allocno_class_array and sorted_pseudos.
23589 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
23590 insns where regno is used.
23592 2018-03-16 Martin Liska <mliska@suse.cz>
23595 * multiple_target.c (create_dispatcher_calls): Redirect
23596 reference in the symbol table.
23598 2018-03-16 Martin Liska <mliska@suse.cz>
23601 * multiple_target.c (create_dispatcher_calls): Redirect also
23604 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23608 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
23609 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
23610 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
23612 2018-03-16 Julia Koval <julia.koval@intel.com>
23614 * doc/invoke.texi (Skylake Server): Add CLWB.
23615 Cannonlake): Remove CLWB.
23617 2018-03-16 Jakub Jelinek <jakub@redhat.com>
23619 PR tree-optimization/84841
23620 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
23622 (FLOAT_ONE_CONST_TYPE): Define.
23623 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
23624 (sort_by_operand_rank): Put entries with higher constant_type last
23625 rather than first to match comments.
23627 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
23629 * config/nios2/nios2.md (movsi_internal): Fix thinko in
23632 2018-03-15 Jakub Jelinek <jakub@redhat.com>
23635 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
23636 check and use address of target always.
23638 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
23641 * config/i386/i386.c (indirect_thunk_needed): Update comments.
23642 (indirect_thunk_bnd_needed): Likewise.
23643 (indirect_thunks_used): Likewise.
23644 (indirect_thunks_bnd_used): Likewise.
23645 (indirect_return_needed): New.
23646 (indirect_return_bnd_needed): Likewise.
23647 (output_indirect_thunk_function): Add a bool argument for
23649 (output_indirect_thunk_function): Don't generate alias for
23650 function return thunk.
23651 (ix86_code_end): Call output_indirect_thunk_function to generate
23652 function return thunks.
23653 (ix86_output_function_return): Set indirect_return_bnd_needed
23654 and indirect_return_needed instead of indirect_thunk_bnd_needed
23655 and indirect_thunk_needed.
23657 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
23659 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
23660 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
23661 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
23663 2018-03-15 David Malcolm <dmalcolm@redhat.com>
23664 Paul Hua <paul.hua.gm@gmail.com>
23667 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
23669 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
23671 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
23672 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
23673 resp. SFmode cases.
23675 2018-03-15 Tamar Christina <tamar.christina@arm.com>
23678 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
23679 instead of GET_MODE_SIZE when comparing Units.
23681 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
23684 * varasm.c (hash_section): Return an unchangeble hash value
23685 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
23686 Return !aarch64_can_use_per_function_literal_pools_p ().
23688 2018-03-15 Jakub Jelinek <jakub@redhat.com>
23691 * optabs.c (emit_conditional_move): Pass address of cmode's copy
23692 rather than address of cmode as last argument to prepare_cmp_insn.
23694 2018-03-15 Julia Koval <julia.koval@intel.com>
23696 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
23697 F_AVX512VNNI, F_AVX512BITALG): New.
23699 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
23702 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
23703 insn for floating-point loads and stores.
23705 2018-03-14 Carl Love <cel@us.ibm.com>
23707 * config/rs6000/rs6000-c.c: Add macro definitions for
23708 ALTIVEC_BUILTIN_VEC_PERMXOR.
23709 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
23710 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
23711 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
23712 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
23714 * config/doc/extend.texi: Add prototypes for vec_permxor.
23716 2018-03-14 David Malcolm <dmalcolm@redhat.com>
23719 * diagnostic-show-locus.c (class layout_point): Convert m_line
23720 from int to linenum_type.
23721 (line_span::comparator): Use linenum "compare" function when
23722 comparing line numbers.
23723 (test_line_span): New function.
23724 (layout_range::contains_point): Convert param "row" from int to
23726 (layout_range::intersects_line_p): Likewise.
23727 (layout::will_show_line_p): Likewise.
23728 (layout::print_source_line): Likewise.
23729 (layout::should_print_annotation_line_p): Likewise.
23730 (layout::print_annotation_line): Likewise.
23731 (layout::print_leading_fixits): Likewise.
23732 (layout::annotation_line_showed_range_p): Likewise.
23733 (struct line_corrections): Likewise for field m_row.
23734 (line_corrections::line_corrections): Likewise for param "row".
23735 (layout::print_trailing_fixits): Likewise.
23736 (layout::get_state_at_point): Likewise.
23737 (layout::get_x_bound_for_row): Likewise.
23738 (layout::print_line): Likewise.
23739 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
23740 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
23741 * input.c (selftest::test_linenum_comparisons): New function.
23742 (selftest::input_c_tests): Call it.
23743 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
23744 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
23745 * selftest.h (ASSERT_GT): New macro.
23746 (ASSERT_GT_AT): New macro.
23747 (ASSERT_LT): New macro.
23748 (ASSERT_LT_AT): New macro.
23750 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
23752 PR rtl-optimization/84780
23753 * combine.c (distribute_links): Don't make a link based on pc_rtx.
23755 2018-03-14 Martin Liska <mliska@suse.cz>
23757 * tree.c (record_node_allocation_statistics): Use
23758 get_stats_node_kind.
23759 (get_stats_node_kind): New function extracted from
23760 record_node_allocation_statistics.
23761 (free_node): Use get_stats_node_kind.
23763 2018-03-14 Richard Biener <rguenther@suse.de>
23765 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
23766 that the value-set of ANTIC_IN doesn't grow.
23769 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
23771 (BB_VISITED_WITH_VISITED_SUCCS): New define.
23772 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
23774 2018-03-14 Julia Koval <julia.koval@intel.com>
23776 * config.gcc (icelake-client, icelake-server): New.
23778 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
23779 (initial_ix86_arch_features): Ditto.
23780 (PTA_SKYLAKE): Add SGX.
23781 (PTA_ICELAKE): Remove.
23782 (PTA_ICELAKE_CLIENT): New.
23783 (PTA_ICELAKE_SERVER): New.
23784 (ix86_option_override_internal): Split up icelake on icelake client and
23786 (get_builtin_code_for_version): Ditto.
23787 (fold_builtin_cpu): Ditto.
23788 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
23789 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
23790 * config/i386/i386.h (processor_type): Ditto.
23791 * doc/invoke.texi: Ditto.
23793 2018-03-14 Jakub Jelinek <jakub@redhat.com>
23796 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
23797 INTEGER_CST offset, add it together with bitpos / 8 and
23798 sign extend based on POINTER_SIZE.
23802 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
23805 * config/i386/constraints.md (Yc): New register constraint.
23806 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
23807 Use Yc constraint for alternative 2 of operand 0. Remove
23808 preferred_for_speed attribute.
23810 2018-03-14 Richard Biener <rguenther@suse.de>
23812 PR tree-optimization/84830
23813 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
23814 with the old one to avoid oscillations.
23816 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
23819 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
23821 (assign_by_spills): Return a flag of reload assignment failure.
23822 Do not process the reload assignment failures. Do not spill other
23823 reload pseudos if they has the same reg class. Update n if
23825 (lra_assign): Add a return arg. Set up from the result of
23826 assign_by_spills call.
23827 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23828 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23829 usage_insns if it is not NULL.
23830 (spill_hard_reg_in_range): New function.
23831 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23832 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23833 function prototypes.
23834 (lra_assign): Change prototype.
23835 * lra.c (lra): Add code to deal with fails by splitting hard reg
23838 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
23840 * config/riscv/riscv.opt (mrelax): New option.
23841 * config/riscv/riscv.c (riscv_file_start): Emit ".option
23842 "norelax" when riscv_mrelax is disabled.
23843 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
23845 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23848 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
23849 reassociation for int modes.
23851 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23853 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
23854 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
23856 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
23857 * config/aarch64/aarch64-sve.md
23858 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
23859 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
23860 (*extend<mode><Vwide>2): Rename to...
23861 (aarch64_sve_extend<mode><Vwide>2): ...this.
23862 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
23863 renaming the old pattern to...
23864 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
23866 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
23867 define_expand, renaming the old pattern to...
23868 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
23869 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
23870 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
23871 account when deciding which SVE instruction the optab should use.
23872 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
23874 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23876 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
23877 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
23878 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
23879 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
23880 (tlsdesc_small_<mode>): Turn a define_expand and use
23881 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
23882 (tlsdesc_small_advsimd_<mode>): ...this.
23883 (tlsdesc_small_sve_<mode>): New pattern.
23885 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
23887 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
23888 (UNSPEC_UMUL_HIGHPART): New constants.
23889 (MUL_HIGHPART): New int iteraor.
23890 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
23891 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
23893 (*<su>mul<mode>3_highpart): New define_insn.
23895 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23898 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23901 2018-03-13 Martin Liska <mliska@suse.cz>
23904 * (sem_item_optimizer::sem_item_optimizer): Initialize new
23906 (sem_item_optimizer::~sem_item_optimizer): Release it.
23907 (sem_item_optimizer::merge_classes): Register variable aliases.
23908 (sem_item_optimizer::fixup_pt_set): New function.
23909 (sem_item_optimizer::fixup_points_to_sets): Likewise.
23910 * ipa-icf.h: Declare new variables and functions.
23912 2018-03-13 Jakub Jelinek <jakub@redhat.com>
23914 PR middle-end/84834
23915 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
23916 integer_pow2p@2 and test integer_pow2p in condition.
23917 (A < 0 ? C : 0): Similarly for @1.
23919 PR middle-end/84831
23920 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
23921 characters starting at p contain '\0' character, don't look beyond
23925 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
23926 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
23929 * reg-stack.c (change_stack): Change update_end var from int to
23930 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
23931 also call set_block_for_insn on the newly added insns and rescan.
23934 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
23935 on the last operand.
23938 * tree.c (stabilize_reference_1): Return save_expr (e) for
23939 STATEMENT_LIST even if it doesn't have side-effects.
23941 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
23943 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
23945 2018-03-12 Renlin Li <renlin.li@arm.com>
23947 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
23948 aarch64_output_scalar_simd_mov_immediate.
23950 2018-03-12 Martin Sebor <msebor@redhat.com>
23952 PR tree-optimization/83456
23953 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
23954 for perfectly overlapping calls to memcpy.
23955 (gimple_fold_builtin_memory_chk): Same.
23956 (gimple_fold_builtin_strcpy): Handle no-warning.
23957 (gimple_fold_builtin_stxcpy_chk): Same.
23958 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
23960 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
23962 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
23963 parameter. Use it for SFmode.
23964 (rs6000_function_arg_advance_1): Adjust.
23965 (rs6000_function_arg): Adjust.
23966 (rs6000_gimplify_va_arg): Pass false for that new parameter.
23968 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
23970 PR rtl-optimization/84169
23971 PR rtl-optimization/84780
23972 * combine.c (can_combine_p): Check for a 2-insn combination whether
23973 the destination register is used between the two insns, too.
23975 2018-03-12 Richard Biener <rguenther@suse.de>
23977 PR tree-optimization/84803
23978 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
23979 for refs DR analysis didn't process.
23981 2018-03-12 Richard Biener <rguenther@suse.de>
23983 PR tree-optimization/84777
23984 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
23985 force-vectorize loops ignore whether we are optimizing for size.
23987 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
23989 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
23990 (TARGET_MD_ASM_ADJUST): Define.
23992 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
23993 Kito Cheng <kito.cheng@gmail.com>
23994 Chung-Ju Wu <jasonwucj@gmail.com>
23996 * config/nds32/nds32.c (nds32_compute_stack_frame,
23997 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
23998 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
23999 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
24000 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
24001 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
24002 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
24003 * config/nds32/nds32.md (prologue, epilogue): Use macro
24004 NDS32_V3PUSH_AVAILABLE_P to do checking.
24006 2018-03-11 Jakub Jelinek <jakub@redhat.com>
24009 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
24010 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
24011 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
24012 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
24013 addition of most attributes on !orig_type_die or the attribute not
24014 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
24016 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24017 Chung-Ju Wu <jasonwucj@gmail.com>
24019 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
24020 __NDS32_VH__ macro.
24021 * config/nds32/nds32.opt (mvh): New option.
24023 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24024 Chung-Ju Wu <jasonwucj@gmail.com>
24026 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
24028 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
24029 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
24032 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
24033 Chung-Ju Wu <jasonwucj@gmail.com>
24035 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
24037 * config/nds32/nds32-multiple.md (strlensi): New pattern.
24038 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
24040 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
24041 Kito Cheng <kito.cheng@gmail.com>
24042 Chung-Ju Wu <jasonwucj@gmail.com>
24044 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
24045 UNSPEC_FFMISM and UNSPEC_FLMISM.
24046 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
24047 for ffb, ffmism and flmism.
24048 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
24049 (unspec_ffmism): Ditto.
24050 (unspec_flmism): Ditto.
24051 (nds32_expand_builtin_impl): Check if string extension is available.
24052 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
24053 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
24055 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
24058 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
24061 * lra-assigns.c (assign_by_spills): Return a flag of reload
24062 assignment failure. Do not process the reload assignment
24063 failures. Do not spill other reload pseudos if they has the same
24065 (lra_assign): Add a return arg. Set up from the result of
24066 assign_by_spills call.
24067 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24068 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24069 usage_insns if it is not NULL.
24070 (spill_hard_reg_in_range): New function.
24071 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24072 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24073 function prototypes.
24074 (lra_assign): Change prototype.
24075 * lra.c (lra): Add code to deal with fails by splitting hard reg
24078 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
24081 * config/i386/i386.opt: Replace Enforcment with Enforcement.
24083 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
24086 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
24087 (dw_val_node): Add val_symbolic_view.
24088 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
24089 (symview_upper_bound): New.
24090 (new_line_info_table): Initialize symviews_since_reset.
24091 (dwarf2out_source_line): Count symviews_since_reset and set
24092 symview_upper_bound.
24093 (dw_val_equal_p): Handle symview.
24094 (add_AT_symview): New.
24095 (print_dw_val): Handle symview.
24096 (attr_checksum, attr_checksum_ordered): Likewise.
24097 (same_dw_val_p, size_of_die): Likewise.
24098 (value_format, output_die): Likewise.
24099 (add_high_low_attributes): Use add_AT_symview for entry_view.
24100 (dwarf2out_finish): Reset symview_upper_bound, clear
24103 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
24106 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
24107 Add strict argument and use it.
24108 (rs6000_split_multireg_move): Update for new strict argument.
24109 (mem_operand_gpr): Disallow all non-offsettable addresses.
24110 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
24112 2018-03-09 Jakub Jelinek <jakub@redhat.com>
24115 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
24116 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
24117 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
24120 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
24121 decl, use remap_type if we want to use the type.
24123 2018-03-09 Martin Sebor <msebor@redhat.com>
24125 PR tree-optimization/84526
24126 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
24128 (builtin_access::generic_overlap): Be prepared to handle non-array
24131 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
24133 PR rtl-optimization/84682
24134 * lra-constraints.c (process_address_1): Check is_address flag
24135 for address constraints.
24136 (process_alt_operands): Likewise.
24137 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
24138 preprocess_constraints.
24139 * recog.h (preprocess_constraints): Add oploc parameter.
24141 * recog.c (preprocess_constraints): Test address_operand for
24142 CT_ADDRESS constraints.
24144 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
24147 * lra-assigns.c (assign_by_spills): Return a flag of reload
24148 assignment failure. Do not process the reload assignment
24149 failures. Do not spill other reload pseudos if they has the same
24151 (lra_assign): Add a return arg. Set up from the result of
24152 assign_by_spills call.
24153 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
24154 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
24155 usage_insns if it is not NULL.
24156 (spill_hard_reg_in_range): New function.
24157 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
24158 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
24159 function prototypes.
24160 (lra_assign): Change prototype.
24161 * lra.c (lra): Add code to deal with fails by splitting hard reg
24164 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24167 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
24168 Accept complain bool parameter. Only emit errors if it is true.
24169 (arm_parse_cpu_option_name): Likewise.
24170 (arm_target_thumb_only): Adjust callers of the above.
24171 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
24172 prototype to take a default true bool parameter.
24173 (arm_parse_arch_option_name): Likewise.
24175 2018-03-09 David Malcolm <dmalcolm@redhat.com>
24176 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
24180 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
24181 * configure: Regenerate.
24182 * configure.ac ("linker --version-script option"): New.
24183 ("linker soname option"): New.
24185 2018-03-09 Richard Biener <rguenther@suse.de>
24187 PR tree-optimization/84775
24188 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
24189 immediate uses of predicate stmts and mark them modified.
24192 PR tree-optimization/84178
24193 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
24195 (version_loop_for_if_conversion): Delay update_ssa call.
24196 (tree_if_conversion): Delay update_ssa until after predicate
24199 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
24202 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
24203 when the function accesses prior frames.
24205 2018-03-08 Jakub Jelinek <jakub@redhat.com>
24208 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
24209 gen_llsym, otherwise call maybe_gen_llsym.
24211 PR inline-asm/84742
24212 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
24213 has ',' character inside of it.
24215 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24218 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
24219 as clobbering CC_REGNUM.
24221 2018-03-08 Richard Biener <rguenther@suse.de>
24223 PR middle-end/84552
24224 * tree-scalar-evolution.c: Include tree-into-ssa.h.
24225 (follow_copies_to_constant): Do not follow SSA names registered
24228 2018-03-08 Richard Biener <rguenther@suse.de>
24230 PR tree-optimization/84178
24231 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
24233 (version_loop_for_if_conversion): Delay update_ssa call.
24234 (tree_if_conversion): Delay update_ssa until after predicate
24237 2018-03-08 David Malcolm <dmalcolm@redhat.com>
24239 PR tree-optimization/84178
24240 * tree-if-conv.c (release_bb_predicate): Remove the
24241 the assertion that the stmts have NULL use_ops.
24242 Discard the statements, asserting that they haven't
24243 yet been added to a BB.
24245 2018-03-08 Richard Biener <rguenther@suse.de>
24247 PR tree-optimization/84746
24248 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
24249 (phi_translate): Pass in destination ANTIC_OUT set.
24250 (phi_translate_1): Likewise. For a simplified result lookup
24251 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
24252 (phi_translate_set): Adjust.
24253 (do_pre_regular_insertion): Likewise.
24254 (do_pre_partial_partial_insertion): Likewise.
24256 2018-03-08 Martin Liska <mliska@suse.cz>
24258 PR gcov-profile/84735
24259 * doc/gcov.texi: Document usage of profile files.
24260 * gcov-io.h: Document changes in the format.
24262 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
24266 * dwarf2out.c (struct dw_line_info_table): Update comments for
24268 (FORCE_RESET_NEXT_VIEW): New.
24269 (FORCE_RESETTING_VIEW_P): New.
24270 (RESETTING_VIEW_P): Check for -1 too.
24271 (ZERO_VIEW_P): Likewise.
24272 (new_line_info_table): Force-reset next view.
24273 (dwarf2out_begin_function): Likewise.
24274 (dwarf2out_source_line): Simplify zero_view_p initialization.
24275 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
24276 view directly. Omit view when omitting .loc at line 0.
24278 2018-03-08 Jakub Jelinek <jakub@redhat.com>
24280 PR tree-optimization/84740
24281 * tree-switch-conversion.c (process_switch): Call build_constructors
24282 only if info.phi_count is non-zero.
24284 PR tree-optimization/84739
24285 * tree-tailcall.c (find_tail_calls): Check call arguments against
24286 DECL_ARGUMENTS (current_function_decl) rather than
24287 DECL_ARGUMENTS (func) when checking for tail recursion.
24289 2018-03-07 Jakub Jelinek <jakub@redhat.com>
24291 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
24292 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
24293 Volker Reichelt's entry and add entries for people that perform
24294 GCC fuzzy testing and report numerous bugs.
24296 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24299 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
24300 readonly data in sdata, if that is disabled.
24301 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
24302 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
24303 -mreadonly-in-sdata option.
24305 2018-03-07 Martin Sebor <msebor@redhat.com>
24307 PR tree-optimization/84468
24308 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
24309 basic block when looking for nul assignment.
24311 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
24314 * except.h (output_function_exception_table): Adjust prototype.
24315 * except.c (output_function_exception_table): Remove FNNAME parameter
24316 and add SECTION parameter. Ouput one part of the table at a time.
24317 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
24318 the first part of the exception table and emit unwind directives.
24319 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
24320 (i386_pe_seh_cold_init): Likewise.
24321 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
24322 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
24323 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
24324 (ix86_output_call_insn): Emit a nop in one more case for SEH.
24325 * config/i386/winnt.c: Include except.h.
24326 (struct seh_frame_state): Add reg_offset, after_prologue and
24327 in_cold_section fields.
24328 (i386_pe_seh_end_prologue): Set seh->after_prologue.
24329 (i386_pe_seh_cold_init): New function.
24330 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
24331 to seh->in_cold_section.
24332 (seh_emit_push): Record the offset of the push.
24333 (seh_emit_save): Record the offet of the save.
24334 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
24335 Test seh->after_prologue to disregard the epilogue.
24336 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
24337 (i386_pe_end_cold_function): New function.
24339 2018-03-07 Jakub Jelinek <jakub@redhat.com>
24342 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
24343 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
24346 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
24348 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
24349 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
24351 PR middle-end/84723
24352 * multiple_target.c: Include tree-inline.h and intl.h.
24353 (expand_target_clones): Diagnose and fail if node->definition and
24354 !tree_versionable_function_p (node->decl).
24356 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
24358 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
24360 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
24361 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
24362 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
24364 2018-03-06 Jakub Jelinek <jakub@redhat.com>
24367 * combine.c (try_combine): Use reg_or_subregno instead of handling
24368 just paradoxical SUBREGs and REGs.
24370 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
24372 * config/arc/arc.c (arc_finalize_pic): Remove function.
24373 (arc_must_save_register): We use single base PIC register, remove
24374 checks to save/restore the PIC register.
24375 (arc_expand_prologue): Likewise.
24376 * config/arc/arc-protos.h (arc_set_default_type_attributes):
24378 (arc_verify_short): Likewise.
24379 (arc_attr_type): Likewise.
24380 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
24381 (walk_stores): Likewise.
24382 (arc_address_cost): Make it static.
24383 (arc_verify_short): Likewise.
24384 (branch_dest): Likewise.
24385 (arc_attr_type): Likewise.
24386 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
24387 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
24388 (arc_final_prescan_insn): Remove inserting the nops due to
24389 hardware hazards. It is done in reorg step.
24390 (insn_length_variant_t): Remove.
24391 (insn_length_parameters_t): Likewise.
24392 (arc_insn_length_parameters): Likewise.
24393 (arc_get_insn_variants): Likewise.
24394 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
24396 2018-03-06 Jakub Jelinek <jakub@redhat.com>
24398 PR inline-asm/84683
24399 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
24402 PR tree-optimization/84687
24403 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
24405 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
24407 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24409 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
24410 Rename to ppc_speculation_barrier.
24411 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
24412 __builtin_ppc_speculation_barrier.
24414 2018-03-05 Jakub Jelinek <jakub@redhat.com>
24417 * combine.c (combine_simplify_rtx): Don't try to simplify if
24418 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
24421 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
24423 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
24424 to 32 bytes when compiling for POWER9.
24426 2018-03-05 Jakub Jelinek <jakub@redhat.com>
24429 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
24430 regparm >= 3 with no arg reg available also for calls with
24431 flag_force_indirect_call. Pass decl to ix86_function_regparm.
24434 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
24436 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
24438 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
24441 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
24443 2018-03-05 Richard Biener <rguenther@suse.de>
24445 PR tree-optimization/84486
24446 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
24447 When inserting a __builtin_assume_aligned call set the LHS
24448 SSA name alignment info accordingly.
24450 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
24452 PR tree-optimization/84114
24453 * config/aarch64/aarch64.c (aarch64_reassociation_width)
24454 Avoid reassociation of FLOAT_MODE addition.
24456 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
24458 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
24459 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
24460 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
24461 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
24462 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
24463 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
24464 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
24466 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
24467 __builtin_ia32_wbinvd): New builtins.
24468 (SPECIAL_ARGS2): New.
24469 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
24470 (SPECIAL_ARGS2): New.
24471 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
24472 (ix86_valid_target_attribute_inner_p): Ditto.
24473 (ix86_init_mmx_sse_builtins): Add special_args2.
24474 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
24475 TARGET_WBNOINVD_P): New.
24476 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
24477 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
24478 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
24479 * config/i386/immintrin.h (_wbinvd): New intrinsic.
24480 * config/i386/pconfigintrin.h: New file.
24481 * config/i386/wbnoinvdintrin.h: Ditto.
24482 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
24484 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
24486 2018-03-05 Richard Biener <rguenther@suse.de>
24488 PR tree-optimization/84670
24489 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
24491 (BB_VISITED_WITH_VISITED_SUCCS): New define.
24492 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
24493 (compute_antic_aux): Only assert the number of values in ANTIC_IN
24494 doesn't grow if all successors (recursively) were visited at least
24497 2018-03-05 Richard Biener <rguenther@suse.de>
24499 PR tree-optimization/84650
24500 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
24501 if executed in the loop pipeline.
24503 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
24505 * doc/configfiles.texi (Configuration Files): Move info about
24506 conditionalizing $target-protos.h to...
24507 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
24508 differs from $target-protos.h.
24510 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
24511 Chung-Ju Wu <jasonwucj@gmail.com>
24513 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
24514 * config/nds32/nds32-multiple.md (setmemsi): Define.
24515 * config/nds32/nds32-memory-manipulation.c
24516 (nds32_gen_dup_4_byte_to_word_value): New.
24517 (emit_setmem_word_loop): New.
24518 (emit_setmem_byte_loop): New.
24519 (nds32_expand_setmem_loop): New.
24520 (nds32_expand_setmem_loop_v3m): New.
24521 (nds32_expand_setmem_unroll): New.
24522 (nds32_expand_setmem): New.
24524 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24525 Chung-Ju Wu <jasonwucj@gmail.com>
24527 * config/nds32/nds32-memory-manipulation.c
24528 (nds32_emit_load_store): New.
24529 (nds32_emit_post_inc_load_store): New.
24530 (nds32_emit_mem_move): New.
24531 (nds32_emit_mem_move_block): New.
24532 (nds32_expand_movmemsi_loop_unknown_size): New.
24533 (nds32_expand_movmemsi_loop_known_size): New.
24534 (nds32_expand_movmemsi_loop): New.
24535 (nds32_expand_movmemsi_unroll): New.
24536 (nds32_expand_movmemqi): Rename ...
24537 (nds32_expand_movmemsi): ... to this.
24538 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
24539 (movmemsi): ... to this.
24540 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
24541 (nds32_expand_movmemsi): ... to this.
24543 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24544 Monk Chiang <sh.chiang04@gmail.com>
24545 Chung-Ju Wu <jasonwucj@gmail.com>
24547 * config/nds32/nds32-protos.h
24548 (nds32_expand_load_multiple): New arguments.
24549 (nds32_expand_store_multiple): Ditto.
24550 (nds32_valid_multiple_load_store): Rename ...
24551 (nds32_valid_multiple_load_store_p): ... to this.
24552 * config/nds32/nds32-memory-manipulation.c
24553 (nds32_expand_load_multiple): Refine implementation.
24554 (nds32_expand_store_multiple): Ditto.
24555 * config/nds32/nds32-multiple.md
24556 (load_multiple): Update nds32_expand_load_multiple interface.
24557 (store_multiple): Update nds32_expand_store_multiple interface.
24558 * config/nds32/nds32-predicates.c
24559 (nds32_valid_multiple_load_store): Rename ...
24560 (nds32_valid_multiple_load_store_p): ... to this and refine
24562 * config/nds32/predicates.md
24563 (nds32_load_multiple_and_update_address_operation): New predicate.
24564 (nds32_store_multiple_and_update_address_operation): New predicate.
24566 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
24567 Chung-Ju Wu <jasonwucj@gmail.com>
24569 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
24570 (combo): New attribute.
24571 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
24573 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
24575 * config/nds32/nds32.opt: Change -mcmodel= default value.
24577 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
24578 Monk Chiang <sh.chiang04@gmail.com>
24579 Chung-Ju Wu <jasonwucj@gmail.com>
24581 * config/nds32/constants.md (unspec_element): New enum.
24582 * config/nds32/constraints.md (Umw): New constraint.
24583 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
24584 * config/nds32/nds32-intrinsic.md: Likewise.
24585 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
24586 (nds32_valid_smw_lwm_base_p): New.
24587 (nds32_output_smw_single_word): New.
24588 (nds32_output_lmw_single_word): New.
24589 (nds32_expand_unaligned_load): New.
24590 (nds32_expand_unaligned_store): New.
24591 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
24592 (nds32_output_smw_single_word): Declare.
24593 (nds32_output_lmw_single_word): Declare.
24594 (nds32_expand_unaligned_load): Declare.
24595 (nds32_expand_unaligned_store): Declare.
24596 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
24597 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
24598 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
24599 NDS32_BUILTIN_UASTORE_DW.
24600 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
24603 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
24604 Kito Cheng <kito.cheng@gmail.com>
24605 Chung-Ju Wu <jasonwucj@gmail.com>
24607 * config/nds32/nds32-intrinsic.c
24608 (nds32_expand_builtin_null_ftype_reg): Delete.
24609 (nds32_expand_builtin_reg_ftype_imm): Ditto.
24610 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
24611 (nds32_read_argument): New.
24612 (nds32_legitimize_target): Ditto.
24613 (nds32_legitimize_argument): Ditto.
24614 (nds32_check_constant_argument): Ditto.
24615 (nds32_expand_unop_builtin): Ditto.
24616 (nds32_expand_unopimm_builtin): Ditto.
24617 (nds32_expand_binop_builtin): Ditto.
24618 (nds32_builtin_decl_impl): Ditto.
24619 (builtin_description): Ditto.
24620 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
24621 (nds32_init_builtins_impl): Ditto.
24622 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
24623 (nds32_builtin_decl): New.
24624 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
24625 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
24627 2018-03-02 Jeff Law <law@redhat.com>
24629 * reorg.c (stop_search_p): Handle DEBUG_INSN.
24630 (redundant_insn, fill_simple_delay_slots): Likewise.
24631 (fill_slots_from_thread): Likewise.
24632 * resource.c (mark_referenced_resources): Likewise.
24633 (mark_set_resources, find_dead_or_set_registers): Likewise.
24635 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24637 * substring-locations.h (format_warning_va): Formatting fix for
24638 ATTRIBUTE_GCC_DIAG.
24639 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
24641 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
24642 * substring-locations.c: Include intl.h.
24643 (format_warning_va): Turned into small wrapper around
24644 format_warning_n_va, renamed to ...
24645 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
24646 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
24648 (format_warning_at_substring_n): New function.
24649 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
24650 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
24651 format_warning_at_substring with just a shorter name instead of
24652 const function pointer.
24653 (fmtwarn_n): New function.
24654 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
24655 appropriate, get rid of all the fmtstr temporaries, move conditionals
24656 with G_() wrapped string literals directly into fmtwarn arguments,
24657 cast dir.len to (int), formatting fixes.
24659 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
24661 * doc/invoke.texi: Remove "Cilk Plus" references.
24663 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24664 Richard Biener <rguenther@suse.de>
24667 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
24668 for error or warning attributes if CALL_FROM_THUNK_P is set.
24671 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24674 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
24675 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
24678 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
24679 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
24681 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
24682 instead of -1U in last predictors element's probability member.
24684 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
24687 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
24688 arguments if they are comparable.
24690 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
24692 PR tree-optimization/84634
24693 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
24694 masks and masked_loop_p with a single loop_masks, making sure it's
24695 null for bb vectorization.
24697 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
24699 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
24700 (vect_analyze_data_ref_access): Use loop->safe_len rather than
24701 loop->force_vectorize to check whether there is no alias.
24703 2018-03-02 Jakub Jelinek <jakub@redhat.com>
24706 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
24708 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
24710 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
24711 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
24712 instead of a loop around prev_real_insn.
24713 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
24716 PR inline-asm/84625
24717 * config/i386/i386.c (ix86_print_operand): Use conditional
24718 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
24721 2018-03-02 Richard Biener <rguenther@suse.de>
24723 PR tree-optimization/84427
24724 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
24725 (bitmap_set_subtract_values): Rewrite to handle multiple
24728 (prune_clobbered_mems): Likewise.
24729 (phi_translate): Take edge instead of pred/phiblock.
24730 (phi_translate_1): Likewise.
24731 (phi_translate_set): Likewise. Insert all translated
24732 exprs for a value into the set, keeping possibly multiple
24733 expressions per value.
24734 (compute_antic_aux): Adjust for phi_translate changes.
24735 When intersecting union the expressions and prune those
24736 not in the final value set, keeping possibly multiple
24737 expressions per value. Do not use value-insertion
24738 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
24739 all expressions. Add verification that the value-sets
24740 only shrink during iteration.
24741 (compute_partial_antic_aux): Adjust for the phi_translate changes.
24742 (do_pre_regular_insertion): Likewise.
24743 (do_pre_partial_partial_insertion): Likewise.
24745 2018-03-02 Richard Biener <rguenther@suse.de>
24748 * config/darwin.c (saved_debug_info_level): New static global.
24749 (darwin_asm_lto_start): Disable debug info generation for LTO out.
24750 (darwin_asm_lto_end): Restore debug info generation settings.
24752 2018-03-01 Martin Liska <mliska@suse.cz>
24755 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
24756 volatile arguments.
24758 2018-03-01 Richard Biener <rguenther@suse.de>
24761 * dwarf2out.c (gen_variable_die): Properly handle late VLA
24762 type annotation with LTO when debug was disabled at compile-time.
24764 2018-03-01 Matthew Fortune <mfortune@gmail.com>
24766 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
24768 (mips_final_postscan_insn): Likewise.
24770 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
24772 PR rtl-optimization/84528
24773 * alias.c (init_alias_target): Add commentary.
24774 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
24775 a unique base value if the frame pointer is not eliminated
24776 to the stack pointer.
24778 2018-03-01 Tom de Vries <tom@codesourcery.com>
24780 PR rtl-optimization/83327
24781 * lra-int.h (hard_regs_spilled_into): Declare.
24782 * lra.c (hard_regs_spilled_into): Define.
24783 (init_reg_info): Init hard_regs_spilled_into.
24784 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
24785 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
24786 (process_bb_lives): Handle hard_regs_spilled_into.
24787 (lra_create_live_ranges_1): Before doing liveness propagation, clear
24788 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
24790 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
24792 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
24793 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
24794 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
24795 * config/rs6000/aix72.h: New file.
24797 2018-02-28 Jakub Jelinek <jakub@redhat.com>
24799 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
24800 instead of warning_at with conditional singular and plural messages
24804 * stor-layout.c (update_alignment_for_field): For
24805 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
24806 && !DECL_PACKED (field), do the alignment update, just use
24807 only desired_align instead of MAX (type_align, desired_align)
24809 (place_field): Don't do known_align < desired_align handling
24810 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
24811 is non-NULL, instead do it after rli->prev_field handling and
24812 only if not within a bitfield word. For DECL_PACKED (field)
24813 use type_align of BITS_PER_UNIT.
24815 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
24817 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
24818 superfluous parentheses and trailing spaces.
24820 2018-02-28 Richard Biener <rguenther@suse.de>
24822 PR tree-optimization/84584
24823 * graphite-scop-detection.c (scop_detection::add_scop): Discard
24824 SCoPs with fake exit edge.
24826 2018-02-28 Martin Liska <mliska@suse.cz>
24829 * timevar.c (timer::print): Fix format to properly print 100%
24832 2018-02-28 Richard Biener <rguenther@suse.de>
24834 PR middle-end/84607
24835 * genmatch.c (capture_info::walk_match): Do not mark
24836 captured expressions without operands as expr_p given
24837 they act more like predicates and should be subject to
24838 "lost tail" side-effect preserving.
24840 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
24842 PR rtl-optimization/81611
24843 * auto-inc-dec.c (attempt_change): Move dead note from
24844 mem_insn if it's the next use of regno
24845 (find_address): Take address use of reg holding
24846 non-incremented value. Add parm to limit search to the named
24848 (merge_in_block): Attempt to use a mem insn that is the next
24849 use of the original regno.
24851 2018-02-27 Martin Sebor <msebor@redhat.com>
24854 * doc/invoke.texi (-Wmissing-attributes): New option.
24855 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
24857 2018-02-27 Martin Sebor <msebor@redhat.com>
24859 PR translation/84207
24860 * diagnostic-core.h (warning_n, error_n, inform_n): Change
24861 n argument to unsigned HOST_WIDE_INT.
24862 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
24863 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
24864 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
24865 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
24867 2018-02-27 Richard Biener <rguenther@suse.de>
24869 PR tree-optimization/84512
24870 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
24871 Do not use the estimate returned from record_stmt_cost for
24872 the scalar iteration cost but sum properly using add_stmt_cost.
24874 2018-02-27 Richard Biener <rguenther@suse.de>
24876 PR tree-optimization/84466
24877 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
24878 Adjust last change to less strictly validate use operands.
24880 2018-02-27 Martin Liska <mliska@suse.cz>
24882 PR gcov-profile/84548
24883 * gcov.c (process_file): Allow partial overlap and consider it
24884 also as group functions.
24885 (output_lines): Properly calculate range of lines for a group.
24887 2018-02-27 Martin Liska <mliska@suse.cz>
24889 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
24890 'ggc' suffixes. Change first column width.
24891 (timer::print): Fix formatting of the column.
24893 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
24895 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
24896 preserve inline entry blocks for the sake of debug inline
24897 entry point markers alone.
24898 (remove_unused_locals): Suggest in comments a better place to
24899 force the preservation of inline entry blocks that are
24900 otherwise unused, but do not preserve them.
24902 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24904 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
24906 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24909 * config/i386/constraints.md (Bs): Replace
24910 ix86_indirect_branch_register with
24911 TARGET_INDIRECT_BRANCH_REGISTER.
24913 * config/i386/i386.md (indirect_jump): Likewise.
24914 (tablejump): Likewise.
24915 (*sibcall_memory): Likewise.
24916 (*sibcall_value_memory): Likewise.
24917 Peepholes of indirect call and jump via memory: Likewise.
24918 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
24919 (*sibcall_value_GOT_32): Likewise.
24920 * config/i386/predicates.md (indirect_branch_operand): Likewise.
24921 (GOT_memory_operand): Likewise.
24922 (call_insn_operand): Likewise.
24923 (sibcall_insn_operand): Likewise.
24924 (GOT32_symbol_operand): Likewise.
24925 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
24927 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
24929 PR rtl-optimization/83496
24930 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
24931 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
24932 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
24933 redundant insn, if any.
24934 (relax_delay_slots): Likewise.
24935 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
24937 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
24939 PR tree-optimization/83965
24940 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
24941 that grouped statements are part of a reduction chain. Return
24942 true if the statement is not marked as a reduction itself but
24943 is part of a group.
24944 (vect_recog_dot_prod_pattern): Don't check whether the statement
24945 is part of a group here.
24946 (vect_recog_sad_pattern): Likewise.
24947 (vect_recog_widen_sum_pattern): Likewise.
24949 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
24952 * final.c (rest_of_clean_state): Also look for calls inside sequences.
24954 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
24957 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
24959 (ix86_output_indirect_function_return): New prototype.
24960 (ix86_split_simple_return_pop_internal): Likewise.
24961 * config/i386/i386.c (indirect_return_via_cx): New.
24962 (indirect_return_via_cx_bnd): Likewise.
24963 (indirect_thunk_name): Handle return va CX_REG.
24964 (output_indirect_thunk_function): Create alias for
24965 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
24966 (ix86_output_indirect_jmp): Remove the bool argument.
24967 (ix86_output_indirect_function_return): New function.
24968 (ix86_split_simple_return_pop_internal): Likewise.
24969 * config/i386/i386.md (*indirect_jump): Don't pass false
24970 to ix86_output_indirect_jmp.
24971 (*tablejump_1): Likewise.
24972 (simple_return_pop_internal): Change it to define_insn_and_split.
24973 Call ix86_split_simple_return_pop_internal to split it for
24974 -mfunction-return=.
24975 (simple_return_indirect_internal): Call
24976 ix86_output_indirect_function_return instead of
24977 ix86_output_indirect_jmp.
24979 2018-02-26 Jakub Jelinek <jakub@redhat.com>
24982 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
24983 memset and value initialization afterwards.
24985 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
24987 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
24989 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24992 * common/config/aarch64/aarch64-common.c
24993 (aarch_option_optimization_table[]): Switch
24994 off fomit-frame-pointer
24996 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
24997 Chung-Ju Wu <jasonwucj@gmail.com>
24999 * config/nds32/nds32-multiple.md (load_multiple): Disallow
25001 (store_multiple): Ditto.
25003 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
25005 * config.gcc: Add --with-cpu support for nds32 target.
25006 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
25007 * config/nds32/nds32.opt: Add -mcpu= option.
25009 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
25011 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
25012 isel=yes): Warn for these deprecated options.
25014 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
25016 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
25017 ISA_2_5_MASKS_EMBEDDED.
25019 2018-02-23 Jakub Jelinek <jakub@redhat.com>
25021 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
25022 p->max as pointers rather than using iterative_hash_expr.
25024 2018-02-23 Carl Love <cel@us.ibm.com>
25026 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
25027 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
25029 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
25030 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
25031 P8V_BUILTIN_VEC_VUNSIGNED2.
25033 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
25036 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
25037 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
25039 * lra-constraints.c (curr_insn_transform): Set up
25040 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
25042 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
25043 * lra-eliminations.c (spill_pseudos): Ditto.
25044 (process_insn_for_elimination): Ditto.
25045 * lra-lives.c (reg_early_clobber_p): Use the new macros.
25046 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
25047 LRA_NON_CLOBBERED_ALT.
25049 2018-02-22 Martin Sebor <msebor@redhat.com>
25051 PR tree-optimization/84480
25052 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
25053 to maybe_diag_stxncpy_trunc. Call it.
25054 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
25055 from gimple_fold_builtin_strcpy. Print inlining stack.
25056 (handle_builtin_stxncpy): Print inlining stack.
25057 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
25059 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
25062 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
25063 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
25064 and -fcheck-pointer-bounds are used together.
25065 (indirect_thunk_prefix): New enum.
25066 (indirect_thunk_need_prefix): New function.
25067 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
25068 "_nt" instead of "_bnd" for NOTRACK prefix.
25069 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
25070 (output_indirect_thunk_function): Likewise.
25072 (ix86_code_end): Update output_indirect_thunk_function calls.
25073 (ix86_output_indirect_branch_via_reg): Replace
25074 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
25075 (ix86_output_indirect_branch_via_push): Likewise.
25076 (ix86_output_function_return): Likewise.
25077 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
25078 incompatible with -fcf-protection=branch and
25079 -fcheck-pointer-bounds.
25081 2018-02-22 Steve Ellcey <sellcey@cavium.com>
25084 * config/aarch64/aarch64.c (aarch64_print_address_internal):
25085 Change gcc_assert call to output_operand_lossage.
25087 2018-02-22 Steve Ellcey <sellcey@cavium.com>
25089 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
25091 2018-02-22 DJ Delorie <dj@redhat.com>
25092 Sebastian Perta <sebastian.perta@renesas.com>
25093 Oleg Endo <olegendo@gcc.gnu.org>
25095 * config/rx/rx.c (rx_rtx_costs): New function.
25096 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
25098 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
25100 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
25102 2018-02-22 Martin Liska <mliska@suse.cz>
25105 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
25106 Add "native" as a possible value.
25108 2018-02-22 Martin Liska <mliska@suse.cz>
25111 * config/i386/i386.c (ix86_option_override_internal):
25112 Add "native" as a possible value for -march and -mtune.
25114 2018-02-22 Jakub Jelinek <jakub@redhat.com>
25117 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
25118 to all type variants.
25120 PR tree-optimization/84503
25121 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
25122 width as info->bitpos + info->bitsize - start.
25123 (merged_store_group::merge_overlapping): Simplify width computation.
25124 (check_no_overlap): New function.
25125 (imm_store_chain_info::try_coalesce_bswap): Compute expected
25126 start + width and last_order of the group, fail if check_no_overlap
25128 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
25129 to group if check_no_overlap fails.
25131 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
25133 * config/rs6000/altivec.md: Delete contraint arguments to
25134 define_expand, define_split, and define_peephole2, and in
25135 define_insn_and_split if always unused.
25136 * config/rs6000/darwin.md: Ditto.
25137 * config/rs6000/dfp.md: Ditto.
25138 * config/rs6000/rs6000.md: Ditto.
25139 * config/rs6000/sync.md: Ditto.
25140 * config/rs6000/vector.md: Ditto.
25141 * config/rs6000/vsx.md: Ditto.
25143 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
25145 * config/rs6000/altivec.md: Write output control strings as braced
25146 blocks instead of double-quoted strings.
25147 * config/rs6000/darwin.md: Ditto.
25148 * config/rs6000/rs6000.md: Ditto.
25149 * config/rs6000/vector.md: Ditto.
25150 * config/rs6000/vsx.md: Ditto.
25152 2018-02-21 Jason Merrill <jason@redhat.com>
25154 PR c++/84314 - ICE with templates and fastcall attribute.
25155 * attribs.c (build_type_attribute_qual_variant): Remove assert.
25157 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
25159 * ipa-cp.c (determine_versionability): Fix comment typos.
25161 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
25164 * ipa-cp.c (determine_versionability): Do not version functions caling
25167 2018-02-21 Martin Liska <mliska@suse.cz>
25170 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
25171 Add "native" as a possible value.
25172 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
25173 the macro when native cpu detection is available.
25175 2018-02-21 Martin Liska <mliska@suse.cz>
25178 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
25179 Add "native" as a possible value.
25180 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
25181 when native cpu detection is available.
25183 2018-02-21 Jakub Jelinek <jakub@redhat.com>
25184 Martin Sebor <msebor@redhat.com>
25186 PR tree-optimization/84478
25187 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
25189 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
25190 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
25191 support which is conservatively correct, for 2 only stay conservative
25192 for maxlen. Formatting and comment capitalization fixes. Add STRICT
25193 argument to the 2 argument get_range_strlen, adjust 6 arg
25194 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
25196 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
25197 (gimple_fold_builtin_strlen): Pass true as last argument to
25200 2018-02-20 Martin Sebor <msebor@redhat.com>
25202 PR middle-end/84095
25203 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
25204 (builtin_memref::set_base_and_offset): Same. Handle inner references.
25205 (builtin_memref::builtin_memref): Factor out parts into
25206 set_base_and_offset and call it.
25208 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
25210 PR middle-end/84406
25211 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
25212 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
25213 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
25214 search at the associated MODE_INT.
25216 2018-02-20 Jeff Law <law@redhat.com>
25218 PR middle-end/82123
25219 PR tree-optimization/81592
25220 PR middle-end/79257
25221 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
25222 for range data rather than using global data.
25223 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
25224 range data rather than using global data.
25225 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
25226 pass it to children as needed.
25227 (struct directive::fmtresult): Similarly.
25228 (struct directive::set_width): Similarly.
25229 (struct directive::set_precision): Similarly.
25230 (format_integer, format_directive, parse_directive): Similarly.
25231 (format_none): Accept unnamed vr_values parameter.
25232 (format_percent, format_floating, format_character): Similarly.
25233 (format_string, format_plain): Similarly.
25234 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
25235 the EVRP range analyzer for range data rather than using global data.
25236 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
25237 gimple-ssa-evrp-analyze.h
25238 (class sprintf_dom_walker): Add after_dom_children member function.
25239 Add evrp_range_analyzer member.
25240 (sprintf_dom_walker::before_dom_children): Call into the EVRP
25241 range analyzer as needed.
25242 (sprintf_dom_walker::after_dom_children): New member function.
25243 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
25245 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
25246 (evrp_range_analyzer::pop_to_marker): Likewise.
25248 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
25250 PR tree-optimization/84419
25251 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
25252 with the required type if its current type is compatible but
25255 2018-02-20 Jakub Jelinek <jakub@redhat.com>
25257 PR middle-end/82004
25258 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
25259 after vectorization.
25261 2018-02-20 Martin Liska <mliska@suse.cz>
25264 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
25265 possible values if we don't have a hint.
25267 2018-02-20 Martin Liska <mliska@suse.cz>
25271 * final.c (shorten_branches): Build align_tab array with one
25273 * opts.c (finish_options): Add alignment option limit check.
25274 (MAX_CODE_ALIGN): Likewise.
25275 (MAX_CODE_ALIGN_VALUE): Likewise.
25276 * doc/invoke.texi: Document maximum allowed option value for
25277 all -falign-* options.
25279 2018-02-19 Jakub Jelinek <jakub@redhat.com>
25282 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
25283 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
25284 * var-tracking.c (emit_note_insn_var_location): Remove all references
25285 to NOTE_INSN_CALL_ARG_LOCATION.
25286 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
25287 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
25288 Use copy_rtx_if_shared.
25289 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
25290 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
25291 (dwarf2out_var_location): Remove handling of
25292 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
25294 * final.c (final_scan_insn): Remove all references to
25295 NOTE_INSN_CALL_ARG_LOCATION.
25296 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
25297 before dumping final insns.
25298 * except.c (emit_note_eh_region_end): Remove all references to
25299 NOTE_INSN_CALL_ARG_LOCATION.
25300 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
25301 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
25302 * config/arc/arc.c (hwloop_optimize): Likewise.
25303 * config/arm/arm.c (create_fix_barrier): Likewise.
25304 * config/s390/s390.c (s390_chunkify_start): Likewise.
25305 * config/sh/sh.c (find_barrier): Likewise.
25306 * config/i386/i386.c (rest_of_insert_endbranch,
25307 ix86_seh_fixup_eh_fallthru): Likewise.
25308 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
25309 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
25310 * config/frv/frv.c (frv_function_prologue): Likewise.
25311 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
25313 (note_outside_basic_block_p): Remove all references to
25314 NOTE_INSN_CALL_ARG_LOCATION.
25315 * gengtype.c (adjust_field_rtx_def): Likewise.
25316 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
25318 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
25319 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
25322 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
25325 PR tree-optimization/84452
25326 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
25327 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
25330 2018-02-19 Martin Liska <mliska@suse.cz>
25333 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
25335 2018-02-19 Martin Liska <mliska@suse.cz>
25336 Richard Sandiford <richard.sandiford@linaro.org>
25338 PR tree-optimization/82491
25339 * gimple-fold.c (get_base_constructor): Make earlier bail out
25342 2018-02-19 Carl Love <cel@us.ibm.com>
25344 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
25345 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
25347 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
25348 P8V_BUILTIN_VEC_NEG.
25350 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
25352 * config/rl78/rl78.md (movdf): New define expand.
25354 2018-02-19 Martin Liska <mliska@suse.cz>
25357 * doc/invoke.texi: Fix typo.
25358 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
25360 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
25362 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
25363 handle rs6000_single_float and rs6000_double_float specially for
25366 2018-02-16 Jeff Law <law@redhat.com>
25368 * config/rx/rx.c (add_pop_cfi_notes): New function.;
25369 (pop_regs): Use it.
25371 2018-02-16 Jakub Jelinek <jakub@redhat.com>
25374 * ipa-inline.c (inline_small_functions): Fix a typo.
25376 2018-02-16 Nathan Sidwell <nathan@acm.org>
25378 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
25380 2018-02-16 Carl Love <cel@us.ibm.com>
25382 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
25383 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
25384 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
25385 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
25386 expansion to P8V_BUILTIN_VEC_FLOAT2.
25388 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
25390 PR rtl-optimization/70023
25391 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
25392 src_regno into account.
25394 2018-02-16 Carl Love <cel@us.ibm.com>
25396 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
25397 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
25398 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
25399 * config/rs6000/rs6000.c: Remove case statements for
25400 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
25401 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
25402 and P9V_BUILTIN_VEC_VINSERT4B.
25403 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
25404 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
25405 * config/rs6000/vsx.md:
25406 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
25409 2018-02-16 Carl Love <cel@us.ibm.com>
25411 * config/rs6000/altivec.h: Add builtin names vec_extract4b
25413 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
25415 * config/rs6000/rs6000-c.c: Add the definitions for
25416 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
25417 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
25418 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
25419 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
25420 definition for insert4b and define insn *insert3b_internal.
25421 * doc/extend.texi: Add documentation for vec_extract4b.
25423 2018-02-16 Nathan Sidwell <nathan@acm.org>
25425 * doc/extend.texi (Backwards Compatibility): Mention friend
25426 injection. Note for-scope is deprecated.
25427 * doc/invoke.texi (-ffriend-injection): Deprecate.
25429 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
25431 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
25432 that moved to I2, also allow destinations that are a paradoxical
25433 subreg (instead of a normal reg).
25435 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
25438 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
25441 2018-02-16 Richard Biener <rguenther@suse.de>
25443 PR tree-optimization/84037
25444 PR tree-optimization/84016
25446 * config/i386/i386.c (ix86_builtin_vectorization_cost):
25447 Adjust vec_construct for the fact we need additional higher latency
25448 128bit inserts for AVX256 and AVX512 vector builds.
25449 (ix86_add_stmt_cost): Scale vector construction cost for
25452 2018-02-16 Richard Biener <rguenther@suse.de>
25454 PR tree-optimization/84417
25455 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
25456 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
25457 (non_rewritable_lvalue_p): Likewise, use poly-ints.
25459 2018-02-16 Martin Liska <mliska@suse.cz>
25462 * internal-fn.def (ASAN_CHECK): Set proper flags.
25463 (ASAN_MARK): Likewise.
25465 2018-02-16 Julia Koval <julia.koval@intel.com>
25467 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
25468 from PTA_CANNONLAKE.
25470 2018-02-16 Jakub Jelinek <jakub@redhat.com>
25473 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
25474 Use ++iter rather than iter++ for std::list iterators.
25475 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
25476 defer deleting them until all nodes in the forest are processed. Do
25477 free even leaf nodes. Change to_process into auto_vec.
25480 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
25481 * vec.h (vec_default_construct): Use memset instead of placement new
25482 if BROKEN_VALUE_INITIALIZATION is defined.
25483 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
25484 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
25487 PR rtl-optimization/83723
25488 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
25489 * lra.c (lra_substitute_pseudo): Likewise. If true, use
25490 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
25492 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
25494 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
25496 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
25498 PR rtl-optimization/81443
25499 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
25500 from inner REGs to paradoxical SUBREGs.
25502 2018-02-16 Richard Biener <rguenther@suse.de>
25504 PR tree-optimization/84399
25505 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
25506 For operands we can analyze at their definition make sure we can
25507 analyze them at each use as well.
25509 2018-02-16 Richard Biener <rguenther@suse.de>
25511 PR tree-optimization/84190
25512 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
25513 volatile accesses if the decl isn't volatile.
25515 2018-02-15 Jason Merrill <jason@redhat.com>
25517 PR c++/84314 - ICE with templates and fastcall attribute.
25518 * attribs.c (build_type_attribute_qual_variant): Don't clobber
25519 TYPE_CANONICAL on an existing type.
25521 2018-02-15 Jakub Jelinek <jakub@redhat.com>
25523 PR tree-optimization/84383
25524 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
25525 dstoff nor call operand_equal_p if dstbase is NULL.
25527 PR tree-optimization/84334
25528 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
25529 also a CONSTANT_CLASS_P, punt.
25531 2018-02-14 Jim Wilson <jimw@sifive.com>
25533 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
25534 first SMALL_OPERAND check. New local min_second_step. Move assert
25535 to where locals are set. Add TARGET_RVC support.
25536 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
25538 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
25540 * doc/invoke.texi: Correct -Wformat-overflow code sample.
25542 2018-02-14 Martin Sebor <msebor@redhat.com>
25544 PR tree-optimization/83698
25545 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
25546 arrays constrain the offset range to their bounds.
25547 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
25548 (builtin_access::overlap): Avoid setting the size of overlap if it's
25550 (maybe_diag_overlap): Also consider arrays when deciding what values
25551 of offsets to include in diagnostics.
25553 2018-02-14 Martin Sebor <msebor@redhat.com>
25556 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
25557 that correspond to the kind of a declaration.
25559 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
25562 * config/pa/pa.md: Load address of PIC label using the linkage table
25563 if the label is nonlocal.
25565 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
25567 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
25568 warning message if user requests -maltivec=be.
25569 * doc/invoke.texi: Document deprecation of -maltivec=be.
25571 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
25574 * config/rs6000/rs6000-c.c: Update definitions for
25575 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
25576 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
25578 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25581 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
25582 add _get_ssp intrinsics. Remove argument from
25583 __builtin_ia32_rdssp[d|q].
25584 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
25585 * config/i386/i386-builtin.def: Remove argument from
25586 __builtin_ia32_rdssp[d|q].
25587 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
25588 ix86_expand_special_args_builtin for _rdssp[d|q].
25589 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
25590 Clear register before usage.
25591 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
25592 Add documentation for new _get_ssp and _inc_ssp intrinsics.
25594 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
25596 PR tree-optimization/84357
25597 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
25598 operand 1 of an ARRAY_REF too.
25600 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
25603 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
25604 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
25606 (set_of_reg): New struct.
25607 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
25608 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
25609 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
25611 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
25612 Split into bitclr, bitset, bitinvert patterns if appropriate.
25613 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
25614 use rx_fuse_in_memory_bitop.
25615 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
25616 to named insn, correct maximum insn length.
25618 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
25621 * machmode.def: Define a complex mode for PARTIAL_INT.
25622 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
25624 * doc/rtl.texi: Document CSPImode.
25625 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
25627 (msp430_hard_regno_nregs_with_padding): Likewise.
25629 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
25632 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
25634 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25636 PR rtl-optimization/84169
25637 * combine.c (try_combine): New variable split_i2i3. Set it to true if
25638 we generated a parallel as new i3 and we split that to new i2 and i3
25639 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
25640 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
25641 those to i2, not i1. Partially rewrite this scan code.
25643 2018-02-13 Jakub Jelinek <jakub@redhat.com>
25646 * stor-layout.c (place_field): For variable length fields, adjust
25647 offset_align afterwards not just based on the field's alignment,
25648 but also on the size.
25650 PR middle-end/84309
25651 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
25652 of exps and logs in the use_exp2 case.
25654 2018-02-13 Jeff Law <law@redhat.com>
25656 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
25657 entry for "vector".
25659 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
25662 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
25666 * common.opt (gas-loc-support, gas-locview-support): New.
25667 (ginline-points, ginternal-reset-location-views): New.
25668 * doc/invoke.texi: Document them. Use @itemx where intended.
25669 (gvariable-location-views): Adjust.
25670 * target.def (reset_location_view): New.
25671 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
25672 (TARGET_RESET_LOCATION_VIEW): New.
25673 * doc/tm.texi: Rebuilt.
25674 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
25675 (dwarf2out_default_as_locview_support): New.
25676 (output_asm_line_debug_info): Use option variables.
25677 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
25678 (output_loc_list): Likewise.
25679 (add_high_low_attributes): Check option variables.
25680 Don't output entry view attribute in strict mode.
25681 (gen_inlined_subroutine_die): Check option variables.
25682 (dwarf2out_inline_entry): Likewise.
25683 (init_sections_and_labels): Likewise.
25684 (dwarf2out_early_finish): Likewise.
25685 (maybe_reset_location_view): New, from...
25686 (dwarf2out_var_location): ... here. Call it.
25687 * debug.h (dwarf2out_default_as_loc_support): Declare.
25688 (dwarf2out_default_as_locview_support): Declare.
25689 * hooks.c (hook_int_rtx_insn_0): New.
25690 * hooks.h (hook_int_rtx_insn_0): Declare.
25691 * toplev.c (process_options): Take -gas-loc-support and
25692 -gas-locview-support from dwarf2out. Enable
25693 -gvariable-location-views by default only with locview
25694 assembler support. Enable -ginternal-reset-location-views by
25695 default only if the target defines the corresponding hook.
25696 Enable -ginline-points by default if location views are
25697 enabled; force it disabled if statement frontiers are
25699 * tree-inline.c (expand_call_inline): Check option variables.
25700 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
25702 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
25704 PR tree-optimization/84321
25705 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
25706 handling. Also check whether the anti-range contains any values
25707 that satisfy the mask; switch to a VR_RANGE if not.
25709 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
25712 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
25714 2018-02-13 Martin Jambor <mjambor@suse.cz>
25717 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
25718 of call statements, also set location of a load to a temporary.
25720 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
25722 * config/rl78/rl78.c (add_vector_labels): New function.
25723 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
25724 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
25725 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
25726 which checks that no arguments are passed.
25727 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
25728 * doc/extend.texi: Documentation for the new attribute.
25730 2018-02-13 Andreas Schwab <schwab@suse.de>
25732 * config/riscv/linux.h (CPP_SPEC): Define.
25734 2018-02-13 Jakub Jelinek <jakub@redhat.com>
25737 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
25738 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
25739 OPTION_MASK_ISA_AES as first argument to def_builtin_const
25740 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
25741 instead of OPTION_MASK_ISA_PCLMUL as first argument to
25742 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
25743 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
25744 temporarily for AES and PCLMUL builtins.
25746 PR tree-optimization/84339
25747 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
25748 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
25751 PR middle-end/84309
25752 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
25753 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
25754 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
25756 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
25758 * omp-simd-clone.h: New file.
25759 * omp-simd-clone.c: Include omp-simd-clone.h.
25760 (expand_simd_clones): No longer static.
25761 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
25762 cgraph.h and omp-simd-clone.h.
25763 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
25764 (vect_recog_widen_shift_pattern): Formatting fix.
25765 (vect_pattern_recog_1): Don't check optab for calls.
25768 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
25769 operands[2] into a REG before using gen_lowpart on it.
25771 2018-02-12 Jeff Law <law@redhat.com>
25774 * config/sh/sh.c (find_barrier): Consider a sibling call
25777 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
25778 successfully back substituting a reg.
25780 2018-02-12 Richard Biener <rguenther@suse.de>
25782 PR tree-optimization/84037
25783 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
25784 parameter, move visited init to caller.
25785 (vect_slp_analyze_operations): Separate cost from validity
25786 check, initialize visited once for all instances.
25787 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
25789 * tree-vect-stmts.c (vect_model_simple_cost): Make early
25791 (vect_model_promotion_demotion_cost): Likewise.
25792 (vectorizable_bswap): Guard cost modeling with !slp_node
25793 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
25795 (vectorizable_call): Likewise.
25796 (vectorizable_conversion): Likewise.
25797 (vectorizable_assignment): Likewise.
25798 (vectorizable_shift): Likewise.
25799 (vectorizable_operation): Likewise.
25800 (vectorizable_store): Likewise.
25801 (vectorizable_load): Likewise.
25802 (vectorizable_condition): Likewise.
25803 (vectorizable_comparison): Likewise.
25805 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
25808 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
25809 (ASAN_MARK): Fix fnspec to account for return value, change pointer
25810 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
25812 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
25814 PR middle-end/83665
25815 * params.def (inline-min-speedup): Increase from 8 to 15.
25816 (max-inline-insns-auto): Decrease from 40 to 30.
25817 * ipa-split.c (consider_split): Add some buffer for function to
25818 be considered inlining candidate.
25819 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
25822 2018-02-12 Richard Biener <rguenther@suse.de>
25824 PR tree-optimization/84037
25825 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
25826 matched stmts if we cannot swap the non-matched ones.
25828 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
25830 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
25831 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
25832 _mm_maskz_scalef_round_ss): New intrinsics.
25833 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
25834 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
25835 __builtin_ia32_scalefss_round): Remove.
25836 (__builtin_ia32_scalefsd_mask_round,
25837 __builtin_ia32_scalefss_mask_round): New intrinsics.
25838 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
25839 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
25840 ((match_operand:VF_128 2 "<round_nimm_predicate>"
25841 "<round_constraint>")): Changed to ...
25842 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
25843 "<round_scalar_constraint>")): ... this.
25844 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
25845 %0, %1, %2<round_op3>}"): Changed to ...
25846 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
25847 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
25848 %2<round_scalar_mask_op3>}"): ... this.
25849 * config/i386/subst.md (round_scalar_nimm_predicate): New.
25851 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
25853 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
25854 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
25855 (_mm_maskz_sqrt_round_ss): New intrinsics.
25856 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
25857 (__builtin_ia32_sqrtsd_mask_round)
25858 (__builtin_ia32_sqrtss_mask_round): New builtins.
25859 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
25860 (__builtin_ia32_sqrtss_round): Remove.
25861 (__builtin_ia32_sqrtsd_mask_round)
25862 (__builtin_ia32_sqrtss_mask_round): New builtins.
25863 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
25864 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
25865 ((match_operand:VF_128 1 "vector_operand"
25866 "xBm,<round_constraint>")): Changed to ...
25867 ((match_operand:VF_128 1 "vector_operand"
25868 "xBm,<round_scalar_constraint>")): ... this.
25869 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
25870 %0, %2, %<iptr>1<round_op3>}): Changed to ...
25871 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
25872 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
25873 %<iptr>1<round_scalar_mask_op3>}): ... this.
25874 ((set_attr "prefix" "<round_prefix>")): Changed to ...
25875 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
25877 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
25880 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
25881 Cast vec_cmpeq result to correct type.
25882 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
25883 Cast vec_cmpgt result to correct type.
25885 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
25887 * final.c (final_scan_insn_1): Renamed from...
25888 (final_scan_insn): ... this. New wrapper, to recover
25889 seen from the outermost call in recursive ones.
25890 * config/sparc/sparc.c (output_return): Drop seen from call.
25891 (output_sibcall): Likewise.
25892 * config/visium/visium.c (output_branch): Likewise.
25894 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
25896 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
25899 2018-02-10 Alan Modra <amodra@gmail.com>
25902 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
25903 Specify LR as an input.
25905 2018-02-10 Jakub Jelinek <jakub@redhat.com>
25908 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
25909 remove_member_access_dummy_vars): New functions.
25910 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
25911 lower_omp_1, execute_lower_omp): Use them.
25913 PR rtl-optimization/84308
25914 * shrink-wrap.c (spread_components): Release todo vector.
25916 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
25918 PR rtl-optimization/57193
25919 * ira-color.c (struct allocno_color_data): Add member
25920 conflict_allocno_hard_prefs.
25921 (update_conflict_allocno_hard_prefs): New.
25922 (bucket_allocno_compare_func): Add a preference based on
25923 conflict_allocno_hard_prefs.
25924 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
25925 (color_allocnos): Remove a dead code. Initiate
25926 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
25928 2018-02-09 Jakub Jelinek <jakub@redhat.com>
25931 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
25932 constraint from =wa to wa. Avoid a subreg on the output operand,
25933 instead use a pseudo and subreg it in a move.
25934 (p9_xxbrd_<mode>): Changed to ...
25935 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
25936 (p9_xxbrd_v2df): New expander.
25937 (p9_xxbrw_<mode>): Changed to ...
25938 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
25939 (p9_xxbrw_v4sf): New expander.
25941 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
25943 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
25945 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
25948 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
25949 multiply in 32-bit mode.
25950 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
25951 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
25954 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
25956 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
25957 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
25958 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
25959 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
25961 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
25964 * dwarf2out.c (is_trivial_indirect_ref): New function.
25965 (dwarf2out_late_global_decl): Do not generate a location
25966 attribute for variables that have a non-trivial DECL_VALUE_EXPR
25967 and that are not defined in the current unit.
25969 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
25971 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
25972 instead of a libcall for UNORDERED.
25974 2018-02-09 Tamar Christina <tamar.christina@arm.com>
25977 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
25978 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
25980 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25983 * config/s390/s390.c (s390_set_current_function): Invoke
25984 s390_indirect_branch_settings also if fndecl didn't change.
25986 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
25988 * config/rs6000/rs6000.md (blockage): Set length to zero.
25990 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
25992 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
25994 2018-02-09 Jakub Jelinek <jakub@redhat.com>
25997 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
25998 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
26002 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
26003 PARALLEL incoming that failed vt_get_decl_and_offset check.
26005 PR middle-end/84237
26006 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
26007 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
26009 (get_variable_section): For decls in named .bss* sections pass true as
26010 second argument to bss_initializer_p.
26012 2018-02-09 Marek Polacek <polacek@redhat.com>
26013 Jakub Jelinek <jakub@redhat.com>
26016 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
26017 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
26018 Sync some changes from cxx_fold_indirect_ref.
26020 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
26022 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
26024 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
26025 (BLOCK_INLINE_ENTRY_LABEL): New.
26026 (dwarf2out_var_location): Disregard inline entry markers.
26027 (inline_entry_data): New struct.
26028 (inline_entry_data_hasher): New hashtable type.
26029 (inline_entry_data_hasher::hash): New.
26030 (inline_entry_data_hasher::equal): New.
26031 (inline_entry_data_table): New variable.
26032 (add_high_low_attributes): Add DW_AT_entry_pc and
26033 DW_AT_GNU_entry_view attributes if a pending entry is found
26034 in inline_entry_data_table. Add old entry_pc attribute only
26035 if debug nonbinding markers are disabled.
26036 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
26037 markers are enabled.
26038 (block_within_block_p, dwarf2out_inline_entry): New.
26039 (dwarf2out_finish): Check that no entries remained in
26040 inline_entry_data_table.
26041 * final.c (reemit_insn_block_notes): Handle inline entry notes.
26042 (final_scan_insn, notice_source_line): Likewise.
26043 (rest_of_clean_state): Skip inline entry markers.
26044 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
26046 * gimple.c (gimple_build_debug_inline_entry): New.
26047 * gimple.h (enum gimple_debug_subcode): Add
26048 GIMPLE_DEBUG_INLINE_ENTRY.
26049 (gimple_build_debug_inline_entry): Declare.
26050 (gimple_debug_inline_entry_p): New.
26051 (gimple_debug_nonbind_marker_p): Adjust.
26052 * insn-notes.def (INLINE_ENTRY): New.
26053 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
26054 inline entry marker notes.
26055 (print_insn): Likewise.
26056 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
26057 (INSN_DEBUG_MARKER_KIND): Likewise.
26058 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
26059 * tree-inline.c (expand_call_inline): Build and insert
26060 debug_inline_entry stmt.
26061 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
26062 inline entry blocks early, if nonbind markers are enabled.
26063 (dump_scope_block): Dump fragment info.
26064 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
26065 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
26066 (gimple_build_debug_inline_entry): New.
26067 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
26068 Enable/disable inline entry points too.
26069 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
26070 (DEBUG_INSN): Describe inline entry markers.
26072 * common.opt (gvariable-location-views): New.
26073 (gvariable-location-views=incompat5): New.
26074 * config.in: Rebuilt.
26075 * configure: Rebuilt.
26076 * configure.ac: Test assembler for view support.
26077 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
26078 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
26079 * dwarf2out.c (var_loc_view): New typedef.
26080 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
26081 (dwarf2out_locviews_in_attribute): New.
26082 (dwarf2out_locviews_in_loclist): New.
26083 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
26084 (enum dw_line_info_opcode): Add LI_adv_address.
26085 (struct dw_line_info_table): Add view.
26086 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
26087 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
26088 (zero_view_p): New variable.
26089 (ZERO_VIEW_P): New macro.
26090 (output_asm_line_debug_info): New.
26091 (struct var_loc_node): Add view.
26092 (add_AT_view_list, AT_loc_list): New.
26093 (add_var_loc_to_decl): Add view param. Test it against last.
26094 (new_loc_list): Add view params. Record them.
26095 (AT_loc_list_ptr): Handle loc and view lists.
26096 (view_list_to_loc_list_val_node): New.
26097 (print_dw_val): Handle dw_val_class_view_list.
26098 (size_of_die): Likewise.
26099 (value_format): Likewise.
26100 (loc_list_has_views): New.
26101 (gen_llsym): Set vl_symbol too.
26102 (maybe_gen_llsym, skip_loc_list_entry): New.
26103 (dwarf2out_maybe_output_loclist_view_pair): New.
26104 (output_loc_list): Output view list or entries too.
26105 (output_view_list_offset): New.
26106 (output_die): Handle dw_val_class_view_list.
26107 (output_dwarf_version): New.
26108 (output_compilation_unit_header): Use it.
26109 (output_skeleton_debug_sections): Likewise.
26110 (output_rnglists, output_line_info): Likewise.
26111 (output_pubnames, output_aranges): Update version comments.
26112 (output_one_line_info_table): Output view numbers in asm comments.
26113 (dw_loc_list): Determine current endview, pass it to new_loc_list.
26114 Call maybe_gen_llsym.
26115 (loc_list_from_tree_1): Adjust.
26116 (add_AT_location_description): Create view list attribute if
26117 needed, check it's absent otherwise.
26118 (convert_cfa_to_fb_loc_list): Adjust.
26119 (maybe_emit_file): Call output_asm_line_debug_info for test.
26120 (dwarf2out_var_location): Reset views as needed. Precompute
26121 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
26122 attribute. Set view.
26123 (new_line_info_table): Reset next view.
26124 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
26125 (dwarf2out_source_line): Likewise. Output view resets and labels to
26126 the assembler, or select appropriate line info opcodes.
26127 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
26128 (optimize_string_length): Catch it. Adjust.
26129 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
26130 dw_val_class_view_list, and remove it if no longer needed.
26131 (hash_loc_list): Hash view numbers.
26132 (loc_list_hasher::equal): Compare them.
26133 (optimize_location_lists): Check whether a view list symbol is
26134 needed, and whether the locview attribute is present, and
26135 whether they match. Remove the locview attribute if no longer
26137 (index_location_lists): Call skip_loc_list_entry for test.
26138 (dwarf2out_finish): Call output_asm_line_debug_info for test.
26139 Use output_dwarf_version.
26140 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
26141 (struct dw_val_node): Add val_view_list.
26142 * final.c (SEEN_NEXT_VIEW): New.
26143 (set_next_view_needed): New.
26144 (clear_next_view_needed): New.
26145 (maybe_output_next_view): New.
26146 (final_start_function): Rename to...
26147 (final_start_function_1): ... this. Take pointer to FIRST,
26148 add SEEN parameter. Emit param bindings in the initial view.
26149 (final_start_function): Reintroduce SEEN-less interface.
26150 (final): Rename to...
26151 (final_1): ... this. Take SEEN parameter. Output final pending
26152 next view at the end.
26153 (final): Reintroduce seen-less interface.
26154 (final_scan_insn): Output pending next view before switching
26155 sections or ending a block. Mark the next view as needed when
26156 outputting variable locations. Notify debug backend of section
26157 changes, and of location view changes.
26158 (rest_of_handle_final): Adjust.
26159 * toplev.c (process_options): Autodetect value for debug variable
26160 location views option. Warn on incompat5 without -gdwarf-5.
26161 * doc/invoke.texi (gvariable-location-views): New.
26162 (gvariable-location-views=incompat5): New.
26163 (gno-variable-location-views): New.
26165 2018-02-08 David Malcolm <dmalcolm@redhat.com>
26167 PR tree-optimization/84136
26168 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
26169 that the result of find_edge is non-NULL.
26171 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
26174 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
26175 storing integer register in SImode. Fix cost of 256 and 512
26176 byte aligned SSE register store.
26178 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
26180 * config/i386/i386.c (ix86_multiplication_cost): Fix
26181 multiplication cost for TARGET_AVX512DQ.
26183 2018-02-08 Marek Polacek <polacek@redhat.com>
26185 PR tree-optimization/84238
26186 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
26189 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26191 PR tree-optimization/84265
26192 * tree-vect-stmts.c (vectorizable_store): Don't treat
26193 VMAT_CONTIGUOUS accesses as grouped.
26194 (vectorizable_load): Likewise.
26196 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26198 PR tree-optimization/81635
26199 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
26200 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
26201 (test_round_for_mask): New functions.
26202 (wide_int_cc_tests): Call test_round_for_mask.
26203 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
26204 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
26205 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
26206 range returned by get_range_info.
26208 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
26211 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
26212 * symtab.c: Include builtins.h
26213 (symtab_node::output_to_lto_symbol_table_p): Move here
26214 from lto-streamer-out.c:output_symbol_p.
26215 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
26216 (output_symbol_p): Move all logic to symtab.c
26217 (produce_symtab): Update.
26219 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26221 * config/s390/s390-opts.h (enum indirect_branch): Define.
26222 * config/s390/s390-protos.h (s390_return_addr_from_memory)
26223 (s390_indirect_branch_via_thunk)
26224 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
26225 (enum s390_indirect_branch_type): Define.
26226 * config/s390/s390.c (struct s390_frame_layout, struct
26227 machine_function): Remove.
26228 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
26229 (indirect_branch_table_label_no, indirect_branch_table_name):
26231 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
26232 (enum s390_indirect_branch_option): Define.
26233 (s390_return_addr_from_memory): New function.
26234 (s390_handle_string_attribute): New function.
26235 (s390_attribute_table): Add new attribute handler.
26236 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
26237 (s390_indirect_branch_via_thunk): New function.
26238 (s390_indirect_branch_via_inline_thunk): New function.
26239 (s390_function_ok_for_sibcall): When jumping via thunk disallow
26240 sibling call optimization for non z10 compiles.
26241 (s390_emit_call): Force indirect branch target to be a single
26242 register. Add r1 clobber for non-z10 compiles.
26243 (s390_emit_epilogue): Emit return jump via return_use expander.
26244 (s390_reorg): Handle JUMP_INSNs as execute targets.
26245 (s390_option_override_internal): Perform validity checks for the
26246 new command line options.
26247 (s390_indirect_branch_attrvalue): New function.
26248 (s390_indirect_branch_settings): New function.
26249 (s390_set_current_function): Invoke s390_indirect_branch_settings.
26250 (s390_output_indirect_thunk_function): New function.
26251 (s390_code_end): Implement target hook.
26252 (s390_case_values_threshold): Implement target hook.
26253 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
26255 * config/s390/s390.h (struct s390_frame_layout)
26256 (struct machine_function): Move here from s390.c.
26257 (TARGET_INDIRECT_BRANCH_NOBP_RET)
26258 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
26259 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
26260 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
26261 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
26262 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
26263 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
26264 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
26265 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
26266 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
26267 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
26268 (mnemonic attribute): Add values which aren't recognized
26270 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
26271 pattern for branch conversion. Fix mnemonic attribute.
26272 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
26273 indirect branch via thunk if requested.
26274 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
26275 ("*indirect_jump"): Disable for branch conversion using out of
26277 ("indirect_jump_via_thunk<mode>_z10")
26278 ("indirect_jump_via_thunk<mode>")
26279 ("indirect_jump_via_inlinethunk<mode>_z10")
26280 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
26281 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
26282 ("casesi_jump_via_inlinethunk<mode>_z10")
26283 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
26284 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
26285 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
26286 ("*indirect2_jump"): Disable for branch conversion.
26287 ("casesi_jump"): Turn into expander and expand patterns for branch
26289 ("return_use"): New expander.
26290 ("*return"): Emit return via thunk and rename it to ...
26291 ("*return<mode>"): ... this one.
26292 * config/s390/s390.opt: Add new options and and enum for the
26295 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
26297 * lra-constraints.c (match_reload): Unconditionally use
26298 gen_lowpart_SUBREG, rather than selecting between that
26299 and equivalent gen_rtx_SUBREG code.
26301 2018-02-08 Richard Biener <rguenther@suse.de>
26303 PR tree-optimization/84233
26304 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
26305 changed flag instead of boguously re-using phi_inserted.
26307 2018-02-08 Martin Jambor <mjambor@suse.cz>
26309 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
26310 static local variables.
26312 2018-02-08 Richard Biener <rguenther@suse.de>
26314 PR tree-optimization/84278
26315 * tree-vect-stmts.c (vectorizable_store): When looking for
26316 smaller vector types to perform grouped strided loads/stores
26317 make sure the mode is supported by the target.
26318 (vectorizable_load): Likewise.
26320 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
26322 * config/aarch64/aarch64.c (aarch64_components_for_bb):
26323 Increase LDP/STP opportunities by adding adjacent callee-saves.
26325 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
26327 PR rtl-optimization/84068
26328 PR rtl-optimization/83459
26329 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
26331 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
26333 PR tree-optimization/84224
26334 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
26335 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
26336 non-zero arguments.
26338 2018-02-07 Iain Sandoe <iain@codesourcery.com>
26341 * config/rs6000/altivec.md (*restore_world): Remove LR use.
26342 * config/rs6000/predicates.md (restore_world_operation): Adjust op
26343 count, remove one USE.
26345 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
26347 * doc/install.texi (Configuration): Document the
26348 --with-long-double-format={ibm,ieee} PowerPC configuration
26352 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
26353 Convert from define_expand to be define_insn_and_split. Rework
26354 float/double/_Float128 conversions to QI/HI/SImode to work with
26355 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
26356 conversions to QI/HImode types did a store and then a load to
26357 truncate the value. For conversions to VSX registers, don't split
26358 the insn, instead emit the code directly. Use the code iterator
26359 any_fix to combine signed and unsigned conversions.
26360 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
26361 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
26362 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
26363 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
26364 (fix_<mode>di2_hw): Likewise.
26365 (fixuns_<mode>di2_hw): Likewise.
26366 (fix_<mode>si2_hw): Likewise.
26367 (fixuns_<mode>si2_hw): Likewise.
26368 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
26369 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
26370 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
26371 fix<uns>_trunc<SFDF:mode>si2_p8.
26372 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
26373 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
26374 (fix<uns>_<mode>_mem): Likewise.
26375 (fctiw<u>z_<mode>_mem): Likewise.
26376 (fix<uns>_<mode>_mem): Likewise.
26377 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
26378 the register allocator from doing a direct move to the GPRs to do
26379 a store, and instead use the ISA 3.0 store byte/half-word from
26380 vector register instruction. For IEEE 128-bit floating point,
26381 also optimize stores of 32-bit ints.
26382 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
26384 2018-02-07 Alan Hayward <alan.hayward@arm.com>
26386 * genextract.c (push_pathstr_operand): New function to support
26388 (walk_rtx): Call push_pathstr_operand.
26389 (print_path): Support [a-zA-Z].
26391 2018-02-07 Richard Biener <rguenther@suse.de>
26393 PR tree-optimization/84037
26394 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
26395 (cse_and_gimplify_to_preheader): Declare.
26396 (vect_get_place_in_interleaving_chain): Likewise.
26397 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26399 (_loop_vec_info::~_loop_vec_info): Delete it.
26400 (cse_and_gimplify_to_preheader): New function.
26401 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
26402 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
26403 (vectorizable_load): Likewise. For grouped stores always base
26404 the IV on the first element.
26405 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
26406 condition before gimplifying.
26408 2018-02-07 Jakub Jelinek <jakub@redhat.com>
26410 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
26411 *DIV_EXPR and *MOD_EXPR.
26413 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
26416 * config/i386/i386.c (ix86_option_override_internal): Mask out
26417 the CF_SET bit when checking -fcf-protection.
26419 2018-02-07 Tom de Vries <tom@codesourcery.com>
26422 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
26425 2018-02-07 Richard Biener <rguenther@suse.de>
26427 PR tree-optimization/84204
26428 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
26431 PR tree-optimization/84205
26432 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
26433 special-case isl_ast_op_zdiv_r.
26435 PR tree-optimization/84223
26436 * graphite-scop-detection.c (gather_bbs::before_dom_children):
26437 Only add conditions from within the region.
26438 (gather_bbs::after_dom_children): Adjust.
26440 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
26443 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
26444 * config/avr/avr.md: Only post-reload split REG-REG moves if
26445 either register is GENERAL_REG_P.
26447 2018-02-07 Jakub Jelinek <jakub@redhat.com>
26449 PR tree-optimization/84235
26450 * tree-ssa-scopedtables.c
26451 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
26452 if the subtraction is performed in floating point type where NaNs are
26453 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
26454 build 1. Formatting fix.
26456 2018-02-06 Jakub Jelinek <jakub@redhat.com>
26459 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
26460 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
26461 and skip it regardless of bb boundaries. Use CALL_P macro,
26462 don't test INSN_P (insn) together with CALL_P or JUMP_P check
26463 unnecessarily, formatting fix.
26465 2018-02-06 Michael Collison <michael.collison@arm.com>
26467 * config/arm/thumb2.md:
26468 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
26469 (*thumb_mov_notscc): Ditto.
26471 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26474 * config/rs6000/rs6000.md (su code attribute): Use "u" for
26475 unsigned_fix, not "s".
26477 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26479 * configure.ac (gcc_fn_eh_frame_ro): New function.
26480 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
26481 correct .eh_frame permissions.
26482 * configure: Regenerate.
26484 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
26486 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
26487 irrelevant options.
26489 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26491 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26492 Display warning message for -mno-speculate-indirect-jumps.
26494 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
26496 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
26498 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
26500 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
26502 PR tree-optimization/84225
26503 * tree-eh.c (find_trapping_overflow): Only call
26504 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
26506 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26509 * config/i386/i386.c: Reimplement the check of possible options
26510 -mibt/-mshstk conbination. Change error messages.
26511 * doc/invoke.texi: Fix a typo: remove extra '='.
26513 2018-02-06 Marek Polacek <polacek@redhat.com>
26515 PR tree-optimization/84228
26516 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
26518 2018-02-06 Tamar Christina <tamar.christina@arm.com>
26521 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
26522 emitted arch directives.
26523 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
26524 __ARM_FEATURE_COPROC before changing architectures.
26526 2018-02-06 Richard Biener <rguenther@suse.de>
26528 * config/i386/i386.c (print_reg): Fix typo.
26529 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
26531 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
26533 * configure: Regenerate.
26535 2018-02-05 Martin Sebor <msebor@redhat.com>
26537 PR tree-optimization/83369
26538 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
26541 2018-02-05 Martin Liska <mliska@suse.cz>
26543 * doc/invoke.texi: Cherry-pick upstream r323995.
26545 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
26547 * ira.c (ira_init_register_move_cost): Adjust comment.
26549 2018-02-05 Martin Liska <mliska@suse.cz>
26551 PR gcov-profile/84137
26552 * doc/gcov.texi: Fix typo in documentation.
26554 2018-02-05 Martin Liska <mliska@suse.cz>
26556 PR gcov-profile/83879
26557 * doc/gcov.texi: Document necessity of --dynamic-list-data when
26558 using dlopen functionality.
26560 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
26562 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
26563 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
26564 _mm_maskz_range_ss, _mm_mask_range_round_ss,
26565 _mm_maskz_range_round_ss): New intrinsics.
26566 (__builtin_ia32_rangesd128_round)
26567 (__builtin_ia32_rangess128_round): Remove.
26568 (__builtin_ia32_rangesd128_mask_round,
26569 __builtin_ia32_rangess128_mask_round): New builtins.
26570 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
26571 __builtin_ia32_rangess128_round): Remove.
26572 (__builtin_ia32_rangesd128_mask_round,
26573 __builtin_ia32_rangess128_mask_round): New builtins.
26574 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
26575 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
26576 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
26577 "<round_saeonly_constraint>")): Changed to ...
26578 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
26579 "<round_saeonly_scalar_constraint>")): ... this.
26580 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
26581 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
26582 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
26583 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
26584 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
26586 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
26588 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
26590 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
26591 Remove all values except native, 8540 and 8548.
26593 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
26595 * config/i386/i386.c (ix86_output_function_return): Pass
26596 INVALID_REGNUM, instead of -1, as invalid register number to
26597 indirect_thunk_name and output_indirect_thunk.
26599 2018-02-02 Julia Koval <julia.koval@intel.com>
26601 * config.gcc: Add -march=icelake.
26602 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
26603 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
26604 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
26605 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
26606 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
26607 (processor_target_table): Add icelake.
26608 (ix86_option_override_internal): Handle new PTAs.
26609 (get_builtin_code_for_version): Handle icelake.
26610 (M_INTEL_COREI7_ICELAKE): New.
26611 (fold_builtin_cpu): Handle icelake.
26612 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
26613 * doc/invoke.texi: Add -march=icelake.
26615 2018-02-02 Julia Koval <julia.koval@intel.com>
26617 * config/i386/i386.c (ix86_option_override_internal): Change flags type
26618 to wide_int_bitmask.
26619 * wide-int-bitmask.h: New.
26621 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26624 * config/i386/i386.md: Replace Pmode with word_mode in
26625 builtin_setjmp_setup and builtin_longjmp to support x32.
26627 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
26631 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
26633 (rs6000_supported_cpu_names): New static variable.
26634 (linux_cpu_translation_table): Likewise.
26635 (elf_platform) <cpu>: Define new static variable and use it.
26636 Translate kernel AT_PLATFORM name to canonical name if needed.
26637 Error if platform name is unknown.
26639 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
26642 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
26644 2018-02-01 Jeff Law <law@redhat.com>
26647 * config/i386/i386.c (release_scratch_register_on_entry): Add new
26648 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
26649 the scratch if RELEASE_VIA_POP is false.
26650 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
26651 If we have to save a temporary register, decrement SIZE appropriately.
26652 Pass new arguments to release_scratch_register_on_entry.
26653 (ix86_adjust_stack_and_probe): Likewise.
26654 (ix86_emit_probe_stack_range): Pass new arguments to
26655 release_scratch_register_on_entry.
26657 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
26659 PR rtl-optimization/84157
26660 * combine.c (change_zero_ext): Use REG_P predicate in
26661 front of HARD_REGISTER_P predicate.
26663 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
26665 * config/avr/avr.c (avr_option_override): Move disabling of
26666 -fdelete-null-pointer-checks to...
26667 * common/config/avr/avr-common.c (avr_option_optimization_table):
26670 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26672 PR tree-optimization/81635
26673 * tree-data-ref.c (split_constant_offset_1): For types that
26674 wrap on overflow, try to use range info to prove that wrapping
26677 2018-02-01 Renlin Li <renlin.li@arm.com>
26680 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
26681 TAILCALL_ADDR_REGS.
26682 (aarch64_register_move_cost): Likewise.
26683 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
26684 TAILCALL_ADDR_REGS.
26685 (REG_CLASS_NAMES): Likewise.
26686 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
26687 TAILCALL_ADDR_REGS. Remove IP registers.
26688 * config/aarch64/aarch64.md (Ucs): Update register constraint.
26690 2018-02-01 Richard Biener <rguenther@suse.de>
26692 * domwalk.h (dom_walker::dom_walker): Add additional constructor
26693 for specifying RPO order and allow NULL for that.
26694 * domwalk.c (dom_walker::dom_walker): Likewise.
26695 (dom_walker::walk): Handle NULL RPO order.
26696 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
26698 (rewrite_update_dom_walker): Likewise.
26699 (mark_def_dom_walker): Likewise.
26701 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26703 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
26704 (aarch64_maybe_expand_sve_subreg_move): Declare.
26705 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
26706 * config/aarch64/predicates.md (aarch64_any_register_operand): New
26708 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
26709 that are semantically a reverse operation.
26710 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
26711 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
26712 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
26714 (aarch64_can_change_mode_class): For big-endian, forbid changes
26715 between two SVE modes if they have different element sizes.
26717 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26719 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
26720 the TImode handling for big-endian targets.
26722 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26724 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
26725 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
26727 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
26728 Remove BSWAP handing for big-endian targets and use the form of
26729 LD1RQ appropariate for the mode.
26731 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26733 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
26734 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
26735 duplicated element.
26737 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
26740 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
26741 check for operands that need to go through aarch64_sve_reload_be.
26743 2018-02-01 Jakub Jelinek <jakub@redhat.com>
26745 PR tree-optimization/81661
26746 PR tree-optimization/84117
26747 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
26748 * tree-eh.c: Include gimplify.h.
26749 (find_trapping_overflow, replace_trapping_overflow,
26750 rewrite_to_non_trapping_overflow): New functions.
26751 * tree-vect-loop.c: Include tree-eh.h.
26752 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
26753 * tree-data-ref.c: Include tree-eh.h.
26754 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
26756 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
26758 PR rtl-optimization/84123
26759 * combine.c (change_zero_ext): Check if hard register satisfies
26760 can_change_dest_mode before calling gen_lowpart_SUBREG.
26762 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
26765 * ira.c (ira_init_register_move_cost): Remove assert.
26767 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26769 PR rtl-optimization/84071
26770 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
26771 * doc/tm.texi: Regenerate.
26773 2018-01-31 Richard Biener <rguenther@suse.de>
26775 PR tree-optimization/84132
26776 * tree-data-ref.c (analyze_miv_subscript): Properly
26777 check whether evolution_function_is_affine_multivariate_p
26778 before calling gcd_of_steps_may_divide_p.
26780 2018-01-31 Julia Koval <julia.koval@intel.com>
26783 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26784 * config/i386/i386.md (rdpid_rex64) New.
26785 (rdpid): Make 32bit only.
26787 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
26790 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
26791 an IDENTIFIER_NODE for FUNCTION_TYPE's.
26793 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26796 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
26798 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
26800 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
26802 PR rtl-optimization/84071
26803 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
26804 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
26806 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
26808 * config/arc/arc.c (arc_handle_aux_attribute): New function.
26809 (arc_attribute_table): Add 'aux' attribute.
26810 (arc_in_small_data_p): Consider aux like variables.
26811 (arc_is_aux_reg_p): New function.
26812 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
26813 (arc_get_aux_arg): New function.
26814 (prepare_move_operands): Handle aux-register access.
26815 (arc_handle_aux_attribute): New function.
26816 * doc/extend.texi (ARC Variable attributes): Add subsection.
26818 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
26820 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
26821 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
26822 (arc_attribute_table): Add 'uncached' attribute.
26823 (arc_print_operand): Print '.di' flag for uncached memory
26825 (arc_in_small_data_p): Do not consider for small data the uncached
26827 (arc_is_uncached_mem_p): New function.
26828 * config/arc/predicates.md (compact_store_memory_operand): Check
26829 for uncached memory accesses.
26830 (nonvol_nonimm_operand): Likewise.
26831 * doc/extend.texi (ARC Type Attribute): New subsection.
26833 2018-01-31 Jakub Jelinek <jakub@redhat.com>
26836 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
26837 falign-loops=): Add Optimization flag.
26839 2018-01-30 Jeff Law <law@redhat.com>
26842 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
26843 INT_REGISTERS_SAVED. Check it prior to calling
26844 get_scratch_register_on_entry.
26845 (ix86_adjust_stack_and_probe): Similarly.
26846 (ix86_emit_probe_stack_range): Similarly.
26847 (ix86_expand_prologue): Corresponding changes.
26849 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26852 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
26853 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
26855 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
26858 * lra-constraints.c (curr_insn_transform): Process AND in the
26861 2018-01-30 Jakub Jelinek <jakub@redhat.com>
26863 PR rtl-optimization/83986
26864 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
26865 dependence against last_pending_memory_flush in addition to
26866 pending_jump_insns.
26868 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
26870 PR tree-optimization/81611
26871 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
26874 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26877 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
26880 2018-01-30 Richard Biener <rguenther@suse.de>
26881 Jakub Jelinek <jakub@redhat.com>
26883 PR tree-optimization/84111
26884 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
26885 inner loops added during recursion, as they don't have up-to-date
26888 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
26891 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
26892 (can_inline_edge_by_limits_p): ... here.
26893 (can_early_inline_edge_p, check_callers,
26894 update_caller_keys, update_callee_keys, recursive_inlining,
26895 add_new_edges_to_heap, speculation_useful_p,
26896 inline_small_functions,
26897 inline_small_functions, flatten_function,
26898 inline_to_all_callers_1): Update.
26900 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
26902 * profile-count.c (profile_count::combine_with_ipa_count): Handle
26905 2018-01-30 Richard Biener <rguenther@suse.de>
26907 PR tree-optimization/83008
26908 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
26909 invariant and constant vector uses in stmts when they need
26910 more than one stmt.
26912 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26915 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
26916 * configure: Regenerate.
26918 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26920 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
26922 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
26923 Use gen_rtx_REG rather than gen_lowpart.
26925 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26927 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
26928 rather than 0 when creating partial subregs.
26930 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
26932 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
26935 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
26938 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
26939 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
26940 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
26941 flags. This restores the settings used before the 2017-07-24.
26942 Turning off pre increment/decrement/modify allows IVOPTS to
26943 optimize DF/SF loops where the index is an int.
26945 2018-01-29 Richard Biener <rguenther@suse.de>
26946 Kelvin Nilsen <kelvin@gcc.gnu.org>
26949 * tree-vect-stmts.c (vectorizable_call): Don't call
26950 targetm.vectorize_builtin_md_vectorized_function if callee is
26953 2018-01-22 Carl Love <cel@us.ibm.com>
26955 * doc/extend.tex: Fix typo in second arg in
26956 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
26958 2018-01-29 Richard Biener <rguenther@suse.de>
26960 PR tree-optimization/84086
26961 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
26962 (flush_ssaname_freelist): When SSA names were released reset
26963 the SCEV hash table.
26965 2018-01-29 Richard Biener <rguenther@suse.de>
26967 PR tree-optimization/84057
26968 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
26969 removed paths when removing edges.
26971 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
26973 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
26974 -mfunction-return=@var{choice}.
26976 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
26978 PR diagnostic/84034
26979 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
26980 Handle CR like TAB.
26981 (layout::print_source_line): Likewise.
26982 (test_get_line_width_without_trailing_whitespace): Add test cases.
26984 2018-01-27 Jakub Jelinek <jakub@redhat.com>
26986 PR middle-end/84040
26987 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
26990 2018-01-26 Jim Wilson <jimw@sifive.com>
26992 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
26994 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
26997 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26999 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
27000 and CMP + SUB-immediate -> SUBS.
27002 2018-01-26 Martin Sebor <msebor@redhat.com>
27004 PR tree-optimization/83896
27005 * tree-ssa-strlen.c (get_string_len): Rename...
27006 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
27007 Avoid assuming length is constant.
27008 (handle_char_store): Use HOST_WIDE_INT for string length.
27010 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
27013 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
27014 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
27016 2018-01-26 Richard Biener <rguenther@suse.de>
27018 PR rtl-optimization/84003
27019 * dse.c (record_store): Only record redundant stores when
27020 the earlier store aliases at least all accesses the later one does.
27022 2018-01-26 Jakub Jelinek <jakub@redhat.com>
27024 PR rtl-optimization/83985
27025 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
27026 REG_CFA_RESTORE insns.
27027 (delete_unmarked_insns): Don't ignore separate shrink wrapping
27028 REG_CFA_RESTORE insns here.
27031 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
27032 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
27034 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27036 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
27037 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
27038 (arc_init): Likewise.
27039 (arc_override_options): Likewise.
27040 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
27042 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
27044 * config/arc/arc.h (TARGET_DBNZ): Define.
27045 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
27046 properly set the tune attribute.
27047 (dbnz): Use TARGET_DBNZ guard.
27048 * config/arc/arc.opt (mtune): Add core3 option.
27050 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27052 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
27053 recognize new pic like addresses.
27054 (arc_delegitimize_address): Clean up.
27056 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27058 * config/arc/arc-arches.def: Option mrf16 valid for all
27060 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
27061 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
27062 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
27063 * config/arc/arc-tables.opt: Regenerate.
27064 * config/arc/arc.c (arc_conditional_register_usage): Handle
27065 reduced register file case.
27066 (arc_file_start): Set must have build attributes.
27067 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
27068 mrf16 option value.
27069 * config/arc/arc.opt (mrf16): Add new option.
27070 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
27071 * config/arc/genmultilib.awk: Handle new mrf16 option.
27072 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
27073 * config/arc/t-multilib: Regenerate.
27074 * doc/invoke.texi (ARC Options): Document mrf16 option.
27076 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27078 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
27079 * config/arc/arc.c (arc_handle_secure_attribute): New function.
27080 (arc_attribute_table): Add 'secure_call' attribute.
27081 (arc_print_operand): Print secure call operand.
27082 (arc_function_ok_for_sibcall): Don't optimize tail calls when
27084 (arc_is_secure_call_p): New function. * config/arc/arc.md
27085 (call_i): Add support for sjli instruction.
27086 (call_value_i): Likewise.
27087 * config/arc/constraints.md (Csc): New constraint.
27089 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
27090 John Eric Martin <John.Martin@emmicro-us.com>
27092 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
27093 * config/arc/arc.c (_arc_jli_section): New struct.
27094 (arc_jli_section): New type.
27095 (rc_jli_sections): New static variable.
27096 (arc_handle_jli_attribute): New function.
27097 (arc_attribute_table): Add jli_always and jli_fixed attribute.
27098 (arc_file_end): New function.
27099 (TARGET_ASM_FILE_END): Define.
27100 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
27101 (arc_add_jli_section): New function.
27102 (jli_call_scan): Likewise.
27103 (arc_reorg): Call jli_call_scan.
27104 (arc_output_addsi): Remove 'S' from printing asm operand.
27105 (arc_is_jli_call_p): New function.
27106 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
27108 (movhi_insn): Likewise.
27109 (movsi_insn): Likewise.
27110 (movsi_set_cc_insn): Likewise.
27111 (loadqi_update): Likewise.
27112 (load_zeroextendqisi_update): Likewise.
27113 (load_signextendqisi_update): Likewise.
27114 (loadhi_update): Likewise.
27115 (load_zeroextendhisi_update): Likewise.
27116 (load_signextendhisi_update): Likewise.
27117 (loadsi_update): Likewise.
27118 (loadsf_update): Likewise.
27119 (movsicc_insn): Likewise.
27120 (bset_insn): Likewise.
27121 (bxor_insn): Likewise.
27122 (bclr_insn): Likewise.
27123 (bmsk_insn): Likewise.
27124 (bicsi3_insn): Likewise.
27125 (cmpsi_cc_c_insn): Likewise.
27126 (movsi_ne): Likewise.
27127 (movsi_cond_exec): Likewise.
27128 (clrsbsi2): Likewise.
27129 (norm_f): Likewise.
27139 (call_i): Remove 'S' asm letter, add jli instruction.
27140 (call_value_i): Likewise.
27141 * config/arc/arc.op (mjli-always): New option.
27142 * config/arc/constraints.md (Cji): New constraint.
27143 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
27145 (subsf3_fpx): Likewise.
27146 (mulsf3_fpx): Likewise.
27147 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
27149 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
27150 function attrbutes.
27151 * doc/invoke.texi (ARC): Document mjli-always option.
27153 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
27155 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
27156 avoid addition with 0 and use incw and decw where possible.
27158 2018-01-26 Richard Biener <rguenther@suse.de>
27160 PR tree-optimization/81082
27161 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
27162 association if it requires casting to unsigned.
27163 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
27164 from fold_plusminus_mult_expr to catch important cases late when
27165 range info is available.
27167 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27169 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
27170 * configure.ac (hidden_linkonce): New test.
27171 * configure: Regenerate.
27172 * config.in: Regenerate.
27174 2018-01-26 Julia Koval <julia.koval@intel.com>
27176 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
27177 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
27178 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
27179 _mm_mask_bitshuffle_epi64_mask): Fix type.
27180 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
27181 USI_FTYPE_V4DI_V4DI_USI): Remove.
27182 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
27183 __builtin_ia32_vpshufbitqmb256_mask,
27184 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
27185 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
27186 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
27188 2018-01-26 Alan Modra <amodra@gmail.com>
27191 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
27192 UNSPEC_VBPERMQ. Sort other unspecs.
27194 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
27196 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
27198 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
27200 PR middle-end/83055
27201 * predict.c (drop_profile): Do not push/pop cfun; update also
27203 (handle_missing_profiles): Fix logic looking for zero profiles.
27205 2018-01-25 Jakub Jelinek <jakub@redhat.com>
27207 PR middle-end/83977
27208 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
27209 on functions with #pragma omp declare simd or functions with simd
27211 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
27212 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
27213 Remove trailing \n from warning_at calls.
27215 2018-01-25 Tom de Vries <tom@codesourcery.com>
27218 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
27219 for neutered workers.
27221 2018-01-24 Joseph Myers <joseph@codesourcery.com>
27224 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
27225 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
27227 2018-01-24 Jeff Law <law@redhat.com>
27230 * i386.c (get_probe_interval): Move to earlier point.
27231 (ix86_compute_frame_layout): If -fstack-clash-protection and
27232 the frame is larger than the probe interval, then use pushes
27233 to save registers rather than reg->mem moves.
27234 (ix86_expand_prologue): Remove conditional for int_registers_saved
27237 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
27240 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
27241 min/max for never referenced object.
27243 2018-01-24 Jakub Jelinek <jakub@redhat.com>
27245 PR middle-end/83977
27246 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
27248 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
27249 attributes from DECL_ATTRIBUTES (decl) without affecting
27250 DECL_ATTRIBUTES (current_function_decl).
27251 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
27252 functions with non-NULL DECL_ABSTRACT_ORIGIN.
27254 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
27256 PR tree-optimization/83979
27257 * fold-const.c (fold_comparison): Use constant_boolean_node
27258 instead of boolean_{true,false}_node.
27260 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
27262 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
27265 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27267 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27268 Simplify the clause that sets the length attribute.
27269 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27270 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
27271 clause that sets the length attribute.
27272 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27274 2018-01-24 Tom de Vries <tom@codesourcery.com>
27277 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
27278 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
27279 Add strict parameter.
27280 (prevent_branch_around_nothing): Insert dummy insn between branch to
27281 label and label with no ptx insn inbetween.
27282 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
27284 2018-01-24 Tom de Vries <tom@codesourcery.com>
27287 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
27288 for neutered threads in warp.
27289 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
27291 2018-01-24 Richard Biener <rguenther@suse.de>
27293 PR tree-optimization/83176
27294 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
27297 2018-01-24 Richard Biener <rguenther@suse.de>
27299 PR tree-optimization/82819
27300 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
27301 code generating pluses that are no-ops in the target precision.
27303 2018-01-24 Richard Biener <rguenther@suse.de>
27305 PR middle-end/84000
27306 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
27308 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27310 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
27311 to merge probabilities.
27312 * predict.c (probably_never_executed): Also mark as cold functions
27313 with global 0 profile and guessed local profile.
27314 * profile-count.c (profile_probability::combine_with_count): New
27316 * profile-count.h (profile_probability::operator*,
27317 profile_probability::operator*=, profile_probability::operator/,
27318 profile_probability::operator/=): Reduce precision to adjusted
27319 and set value to guessed on contradictory divisions.
27320 (profile_probability::combine_with_freq): Remove.
27321 (profile_probability::combine_wiht_count): Declare.
27322 (profile_count::force_nonzero):: Set to adjusted.
27323 (profile_count::probability_in):: Set quality to adjusted.
27324 * tree-ssa-tail-merge.c (replace_block_by): Use
27325 combine_with_count.
27327 2018-01-23 Andrew Waterman <andrew@sifive.com>
27328 Jim Wilson <jimw@sifive.com>
27330 * config/riscv/riscv.c (riscv_stack_boundary): New.
27331 (riscv_option_override): Set riscv_stack_boundary. Handle
27332 riscv_preferred_stack_boundary_arg.
27333 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
27334 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
27335 (STACK_BOUNDARY): Set to riscv_stack_boundary.
27336 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
27337 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
27338 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
27340 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
27343 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
27344 of struct ix86_frame.
27345 (ix86_expand_epilogue): Likewise. Add a local variable for
27346 the reg_save_offset field in struct ix86_frame.
27348 2018-01-23 Bin Cheng <bin.cheng@arm.com>
27350 PR tree-optimization/82604
27351 * tree-loop-distribution.c (enum partition_kind): New enum item
27352 PKIND_PARTIAL_MEMSET.
27353 (partition_builtin_p): Support above new enum item.
27354 (generate_code_for_partition): Ditto.
27355 (compute_access_range): Differentiate cases that equality can be
27356 proven at all loops, the innermost loops or no loops.
27357 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
27358 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
27359 (finalize_partitions, distribute_loop): Don't fuse partition of
27360 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
27361 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
27362 parloop is enabled.
27364 2018-01-23 Martin Liska <mliska@suse.cz>
27366 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
27367 order to ignore the predictor.
27368 (PRED_POLYMORPHIC_CALL): Likewise.
27369 (PRED_RECURSIVE_CALL): Likewise.
27371 2018-01-23 Martin Liska <mliska@suse.cz>
27373 * tree-profile.c (tree_profiling): Print function header to
27374 aware reader which function we are working on.
27375 * value-prof.c (gimple_find_values_to_profile): Do not print
27376 not interesting value histograms.
27378 2018-01-23 Martin Liska <mliska@suse.cz>
27380 * profile-count.h (enum profile_quality): Add
27381 profile_uninitialized as the first value. Do not number values
27382 as they are zero based.
27383 (profile_count::verify): Update sanity check.
27384 (profile_probability::verify): Likewise.
27386 2018-01-23 Nathan Sidwell <nathan@acm.org>
27388 * doc/invoke.texi (ffor-scope): Deprecate.
27390 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27392 PR tree-optimization/83510
27393 * domwalk.c (set_all_edges_as_executable): New function.
27394 (dom_walker::dom_walker): Convert bool param
27395 "skip_unreachable_blocks" to enum reachability. Move setup of
27396 edge flags to set_all_edges_as_executable and only do it when
27397 reachability is REACHABLE_BLOCKS.
27398 * domwalk.h (enum dom_walker::reachability): New enum.
27399 (dom_walker::dom_walker): Convert bool param
27400 "skip_unreachable_blocks" to enum reachability.
27401 (set_all_edges_as_executable): New decl.
27402 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
27403 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
27405 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
27406 but converting true to REACHABLE_BLOCKS.
27407 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
27409 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
27410 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
27411 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
27413 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
27414 if check_all_array_refs will be called.
27416 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27418 * tree.c (selftest::test_location_wrappers): Add more test
27421 2018-01-23 David Malcolm <dmalcolm@redhat.com>
27423 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
27424 (selftest::test_bit_in_range): Likewise.
27426 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
27429 * doc/sourcebuild.texi (vect_float): Say that the selector
27430 only describes the situation when -funsafe-math-optimizations is on.
27431 (vect_float_strict): Document.
27433 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
27435 PR tree-optimization/83965
27436 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
27437 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
27438 instead of checking only for a reduction.
27439 (vect_recog_widen_sum_pattern): Likewise.
27441 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27443 * predict.c (probably_never_executed): Only use precise profile info.
27444 (compute_function_frequency): Skip after inlining hack since we now
27445 have quality checking.
27447 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
27449 * profile-count.h (profile_probability::very_unlikely,
27450 profile_probability::unlikely, profile_probability::even): Set
27451 precision to guessed.
27453 2018-01-23 Richard Biener <rguenther@suse.de>
27455 PR tree-optimization/83963
27456 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
27457 Properly terminate dominator walk when crossing the exit edge not
27458 when visiting its source block.
27460 2018-01-23 Jakub Jelinek <jakub@redhat.com>
27463 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
27464 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
27466 2018-01-22 Jakub Jelinek <jakub@redhat.com>
27468 PR tree-optimization/83957
27469 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
27470 semicolon after for body surrounded by braces.
27472 PR tree-optimization/83081
27473 * profile-count.h (profile_probability::split): New method.
27474 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
27475 Use profile_probability::split.
27476 (do_compare_rtx_and_jump): Fix adjustment of probabilities
27477 when splitting a single conditional jump into 2.
27479 2018-01-22 David Malcolm <dmalcolm@redhat.com>
27481 PR tree-optimization/69452
27482 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
27485 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27487 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
27488 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
27489 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
27491 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27493 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
27495 * config/rl78/rl78.md (movdi): New define_expand.
27496 * config/rl78/rl78.c (rl78_split_movdi): New function.
27498 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27501 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
27503 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
27504 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
27505 128-bit to produce an UNSPEC move to get the double word with the
27506 signbit and then a shift directly to do signbit.
27507 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
27508 implementation with a new version that just does either a direct
27509 move or a regular move. Move memory interface to separate insns.
27510 Move insns so they are next to the expander.
27511 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
27512 with signbit move. Split big and little endian case.
27513 (signbit<mode>2_dm_mem_le): Likewise.
27514 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
27515 (signbit<mode>2_dm2): Likewise.
27517 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27519 * config/rl78/rl78.md (anddi3): New define_expand.
27521 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27523 * config/rl78/rl78.md (umindi3): New define_expand.
27525 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27527 * config/rl78/rl78.md (smindi3): New define_expand.
27529 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27531 * config/rl78/rl78.md (smaxdi3): New define_expand.
27533 2018-01-22 Carl Love <cel@us.ibm.com>
27535 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
27536 LVX_V1TI): Add macro expansion.
27537 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
27538 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
27539 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
27540 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
27541 Change check to determine if the instruction is a byte reversing
27542 entry. Fix typo in comment.
27543 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
27544 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
27545 Add def_builtin calls for new builtins.
27546 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
27547 Add define_insn expansion.
27549 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27551 * config/rl78/rl78.md (umaxdi3): New define_expand.
27553 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
27555 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
27556 for non-QImode registers.
27558 2018-01-22 Richard Biener <rguenther@suse.de>
27560 PR tree-optimization/83963
27561 * graphite-scop-detection.c (scop_detection::get_sese): Delay
27562 including the loop exit block.
27563 (scop_detection::merge_sese): Likewise.
27564 (scop_detection::add_scop): Do it here instead.
27566 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27568 * doc/sourcebuild.texi (arm_softfloat): Document.
27570 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
27573 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
27574 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
27575 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
27577 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27578 David Edelsohn <dje.gcc@gmail.com>
27581 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27582 Change "crset eq" to "crset 2".
27583 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27584 (*call_indirect_aix<mode>_nospec): Likewise.
27585 (*call_value_indirect_aix<mode>_nospec): Likewise.
27586 (*call_indirect_elfv2<mode>_nospec): Likewise.
27587 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
27588 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
27589 change assembly output from . to $.
27590 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27591 (indirect_jump<mode>_nospec): Change assembly output from . to $.
27592 (*tablejump<mode>_internal1_nospec): Likewise.
27594 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
27597 * config/sh/sh_optimize_sett_clrt.cc:
27598 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
27600 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
27602 PR tree-optimization/83940
27603 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
27604 offset_dt to vect_constant_def rather than vect_unknown_def_type.
27605 (vect_check_load_store_mask): Add a mask_dt_out parameter and
27606 use it to pass back the definition type.
27607 (vect_check_store_rhs): Likewise rhs_dt_out.
27608 (vect_build_gather_load_calls): Add a mask_dt argument and use
27609 it instead of a call to vect_is_simple_use.
27610 (vectorizable_store): Update calls to vect_check_load_store_mask
27611 and vect_check_store_rhs. Use the dt returned by the latter instead
27612 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
27613 instead of calls to vect_is_simple_use. Pass the scalar rather
27614 than the vector operand to vect_is_simple_use when handling
27615 second and subsequent copies of an rhs value.
27616 (vectorizable_load): Update calls to vect_check_load_store_mask
27617 and vect_build_gather_load_calls. Use the cached mask_dt and
27618 gs_info.offset_dt instead of calls to vect_is_simple_use.
27620 2018-01-20 Jakub Jelinek <jakub@redhat.com>
27622 PR middle-end/83945
27623 * tree-emutls.c: Include gimplify.h.
27624 (lower_emutls_2): New function.
27625 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
27626 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
27627 it before further processing.
27630 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
27631 UINTVAL (trueop1) instead of INTVAL (op1).
27633 2018-01-19 Jakub Jelinek <jakub@redhat.com>
27637 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
27638 INCOMING_FRAME_SP_OFFSET if not defined.
27639 (scan_trace): Add ENTRY argument. If true and
27640 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
27641 emit a note to adjust the CFA offset.
27642 (create_cfi_notes): Adjust scan_trace callers.
27643 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
27644 INCOMING_FRAME_SP_OFFSET in the CIE.
27645 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
27646 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
27648 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
27649 * doc/tm.texi: Regenerated.
27651 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27653 PR rtl-optimization/83147
27654 * lra-constraints.c (remove_inheritance_pseudos): Use
27655 lra_substitute_pseudo_within_insn.
27657 2018-01-19 Tom de Vries <tom@codesourcery.com>
27658 Cesar Philippidis <cesar@codesourcery.com>
27661 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
27663 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
27666 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
27667 spaces for function labels.
27669 2018-01-19 Martin Liska <mliska@suse.cz>
27671 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
27672 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
27673 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
27674 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
27675 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
27676 (PRED_CONST_RETURN): Change from 69 to 65.
27677 (PRED_NULL_RETURN): Change from 91 to 71.
27678 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
27679 (PRED_LOOP_GUARD): Change from 66 to 73.
27681 2018-01-19 Martin Liska <mliska@suse.cz>
27683 * predict.c (predict_insn_def): Add new assert.
27684 (struct branch_predictor): Change type to signed integer.
27685 (test_prediction_value_range): Amend test to cover
27686 PROB_UNINITIALIZED.
27687 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
27688 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
27689 (PRED_LOOP_ITERATIONS_MAX): Likewise.
27690 (PRED_LOOP_IV_COMPARE): Likewise.
27691 * predict.h (PROB_UNINITIALIZED): Define new constant.
27693 2018-01-19 Martin Liska <mliska@suse.cz>
27695 * predict.c (dump_prediction): Add new format for
27696 analyze_brprob.py script which is enabled with -details
27698 * profile-count.h (precise_p): New function.
27700 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
27702 PR tree-optimization/83922
27703 * tree-vect-loop.c (vect_verify_full_masking): Return false if
27704 there are no statements that need masking.
27705 (vect_active_double_reduction_p): New function.
27706 (vect_analyze_loop_operations): Use it when handling phis that
27707 are not in the loop header.
27709 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
27711 PR tree-optimization/83914
27712 * tree-vect-loop.c (vectorizable_induction): Don't convert
27713 init_expr or apply the peeling adjustment for inductions
27714 that are nested within the vectorized loop.
27716 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27718 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
27721 2018-01-18 Jakub Jelinek <jakub@redhat.com>
27725 * function.h (gimplify_parameters): Add gimple_seq * argument.
27726 * function.c: Include gimple.h and options.h.
27727 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
27728 for the added local temporaries if needed.
27729 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
27730 if there are any parameter cleanups, wrap whole body into a
27731 try/finally with the cleanups.
27733 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27736 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27737 Use GET_MODE_CLASS for scalar floating point.
27739 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
27743 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
27744 Fix call of call_cgraph_insertion_hooks.
27746 2018-01-18 Martin Sebor <msebor@redhat.com>
27748 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
27750 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
27753 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
27756 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
27759 * common.opt: (-ffile-prefix-map): New option.
27760 * opts.c (common_handle_option): Defer it.
27761 * opts-global.c (handle_common_deferred_options): Handle it.
27762 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
27763 * file-prefix-map.h: New file.
27764 (remap_debug_filename, add_debug_prefix_map): ...here.
27765 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
27766 * final.c (debug_prefix_map, add_debug_prefix_map
27767 remap_debug_filename): Move to...
27768 * file-prefix-map.c: New file.
27769 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
27770 generalize, get rid of alloca(), use strrchr() instead of strchr().
27771 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
27772 Implement in terms of add_prefix_map().
27773 (remap_macro_filename, remap_debug_filename): Implement in term of
27775 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
27776 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
27777 * dbxout.c: Include file-prefix-map.h.
27778 * varasm.c: Likewise.
27779 * vmsdbgout.c: Likewise.
27780 * xcoffout.c: Likewise.
27781 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
27782 * doc/cppopts.texi (-fmacro-prefix-map): Document.
27783 * doc/invoke.texi (-ffile-prefix-map): Document.
27784 (-fdebug-prefix-map): Update description.
27786 2018-01-18 Martin Liska <mliska@suse.cz>
27788 * config/i386/i386.c (indirect_thunk_name): Document that also
27790 (output_indirect_thunk): Document why both instructions
27791 (pause and lfence) are generated.
27793 2018-01-18 Richard Biener <rguenther@suse.de>
27795 PR tree-optimization/83887
27796 * graphite-scop-detection.c
27797 (scop_detection::get_nearest_dom_with_single_entry): Remove.
27798 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
27799 (scop_detection::merge_sese): Re-implement with a flood-fill
27800 algorithm that properly finds a SESE region if it exists.
27802 2018-01-18 Jakub Jelinek <jakub@redhat.com>
27805 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
27806 pointer_diff optimizations use view_convert instead of convert.
27808 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27810 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
27811 Generate different code for -mno-speculate-indirect-jumps.
27812 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
27813 (*call_indirect_aix<mode>): Disable for
27814 -mno-speculate-indirect-jumps.
27815 (*call_indirect_aix<mode>_nospec): New define_insn.
27816 (*call_value_indirect_aix<mode>): Disable for
27817 -mno-speculate-indirect-jumps.
27818 (*call_value_indirect_aix<mode>_nospec): New define_insn.
27819 (*sibcall_nonlocal_sysv<mode>): Generate different code for
27820 -mno-speculate-indirect-jumps.
27821 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
27823 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
27825 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
27826 long double type, set the flags for noting the default long double
27827 type, even if we don't pass or return a long double type.
27829 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
27832 * ipa-inline.c (flatten_function): Do not overwrite final inlining
27835 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
27837 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
27838 support for merge[hl].
27839 (fold_mergehl_helper): New helper function.
27840 (tree-vector-builder.h): New #include for tree_vector_builder usage.
27841 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
27842 (altivec_vmrglw_direct): Add xxmrglw insn.
27844 2018-01-17 Andrew Waterman <andrew@sifive.com>
27846 * config/riscv/riscv.c (riscv_conditional_register_usage): If
27847 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
27849 2018-01-17 David Malcolm <dmalcolm@redhat.com>
27852 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
27853 call the lto_location_cache before reading the
27854 DECL_SOURCE_LOCATION of the types.
27856 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27857 Richard Sandiford <richard.sandiford@linaro.org>
27859 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
27860 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
27861 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
27862 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
27863 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
27865 * config/aarch64/constraints.md (aarch64_movti_operand):
27867 * config/aarch64/predicates.md (Uti): Add new constraint.
27869 2018-01-17 Carl Love <cel@us.ibm.com>
27871 * config/rs6000/vsx.md (define_expand xl_len_r,
27872 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
27873 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
27875 (define_expand, define_insn): Move the shift left from the
27876 define_insn to the define_expand for lxvl and stxvl instructions.
27877 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
27878 and XL_LEN_R definitions to PURE.
27880 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
27882 * config/i386/i386.c (indirect_thunk_name): Declare regno
27883 as unsigned int. Compare regno with INVALID_REGNUM.
27884 (output_indirect_thunk): Ditto.
27885 (output_indirect_thunk_function): Ditto.
27886 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
27887 in the call to output_indirect_thunk_function.
27889 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
27891 PR middle-end/83884
27892 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
27893 rather than the size of inner_type to determine the stack slot size
27894 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
27896 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
27899 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
27902 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
27904 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
27905 endian Linux systems to optionally enable multilibs for selecting
27906 the long double type if the user configured an explicit type.
27907 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
27908 have no long double multilibs if not defined.
27909 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
27910 warn if the user used -mabi={ieee,ibm}longdouble and we built
27911 multilibs for long double.
27912 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
27913 appropriate multilib option.
27914 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
27916 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
27917 for building long double multilibs.
27918 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
27920 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
27922 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
27925 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
27927 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
27930 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
27933 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
27936 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
27938 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
27939 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
27941 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
27943 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
27944 different rtl trees depending on TARGET_64BIT.
27945 (rs6000_gen_lvx): Likewise.
27947 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
27949 * config/visium/visium.md (nop): Tweak comment.
27950 (hazard_nop): Likewise.
27952 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27954 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
27955 -mspeculate-indirect-jumps.
27956 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
27957 for -mno-speculate-indirect-jumps.
27958 (*call_indirect_elfv2<mode>_nospec): New define_insn.
27959 (*call_value_indirect_elfv2<mode>): Disable for
27960 -mno-speculate-indirect-jumps.
27961 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
27962 (indirect_jump): Emit different RTL for
27963 -mno-speculate-indirect-jumps.
27964 (*indirect_jump<mode>): Disable for
27965 -mno-speculate-indirect-jumps.
27966 (*indirect_jump<mode>_nospec): New define_insn.
27967 (tablejump): Emit different RTL for
27968 -mno-speculate-indirect-jumps.
27969 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
27970 (tablejumpsi_nospec): New define_expand.
27971 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
27972 (tablejumpdi_nospec): New define_expand.
27973 (*tablejump<mode>_internal1): Disable for
27974 -mno-speculate-indirect-jumps.
27975 (*tablejump<mode>_internal1_nospec): New define_insn.
27976 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
27979 2018-01-16 Artyom Skrobov tyomitch@gmail.com
27981 * caller-save.c (insert_save): Drop unnecessary parameter. All
27984 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27985 Richard Biener <rguenth@suse.de>
27988 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
27989 return early, inline manually is_gimple_sizepos. Make sure if we
27990 call gimplify_expr we don't end up with a gimple constant.
27991 * tree.c (variably_modified_type_p): Don't return true for
27992 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
27993 * gimplify.h (is_gimple_sizepos): Remove.
27995 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27997 PR tree-optimization/83857
27998 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
27999 vectorizable_live_operation for pure SLP statements.
28000 (vectorizable_live_operation): Handle PHIs.
28002 2018-01-16 Richard Biener <rguenther@suse.de>
28004 PR tree-optimization/83867
28005 * tree-vect-stmts.c (vect_transform_stmt): Precompute
28006 nested_in_vect_loop_p since the scalar stmt may get invalidated.
28008 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28011 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
28012 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
28013 If off is not INTEGER_CST, issue a may not be aligned warning
28014 rather than isn't aligned. Use isn%'t rather than isn't.
28015 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
28017 <case MULT_EXPR>: Improve the case when bottom and one of the
28018 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
28019 operand, in that case check if the other operand is multiple of
28020 bottom divided by the INTEGER_CST operand.
28022 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28025 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
28026 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
28027 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
28028 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
28029 * config/pa/pa.c (pa_function_arg_advance): Likewise.
28030 (pa_function_arg, pa_arg_partial_bytes): Likewise.
28031 (pa_function_arg_size): New function.
28033 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28035 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
28036 in a separate statement.
28038 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
28040 PR tree-optimization/83847
28041 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
28042 group gathers and scatters.
28044 2018-01-16 Jakub Jelinek <jakub@redhat.com>
28046 PR rtl-optimization/86620
28047 * params.def (max-sched-ready-insns): Bump minimum value to 1.
28049 PR rtl-optimization/83213
28050 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
28051 to last if both are JUMP_INSNs.
28053 PR tree-optimization/83843
28054 * gimple-ssa-store-merging.c
28055 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
28056 store_immediate_info for bswap/nop orig_stores.
28058 2018-01-15 Andrew Waterman <andrew@sifive.com>
28060 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
28062 <UDIV>: Increase cost if !TARGET_DIV.
28064 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28066 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
28067 (define_attr "cr_logical_3op"): New.
28068 (cceq_ior_compare): Adjust.
28069 (cceq_ior_compare_complement): Adjust.
28070 (*cceq_rev_compare): Adjust.
28071 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
28072 (is_cracked_insn): Adjust.
28073 (insn_must_be_first_in_group): Adjust.
28074 * config/rs6000/40x.md: Adjust.
28075 * config/rs6000/440.md: Adjust.
28076 * config/rs6000/476.md: Adjust.
28077 * config/rs6000/601.md: Adjust.
28078 * config/rs6000/603.md: Adjust.
28079 * config/rs6000/6xx.md: Adjust.
28080 * config/rs6000/7450.md: Adjust.
28081 * config/rs6000/7xx.md: Adjust.
28082 * config/rs6000/8540.md: Adjust.
28083 * config/rs6000/cell.md: Adjust.
28084 * config/rs6000/e300c2c3.md: Adjust.
28085 * config/rs6000/e500mc.md: Adjust.
28086 * config/rs6000/e500mc64.md: Adjust.
28087 * config/rs6000/e5500.md: Adjust.
28088 * config/rs6000/e6500.md: Adjust.
28089 * config/rs6000/mpc.md: Adjust.
28090 * config/rs6000/power4.md: Adjust.
28091 * config/rs6000/power5.md: Adjust.
28092 * config/rs6000/power6.md: Adjust.
28093 * config/rs6000/power7.md: Adjust.
28094 * config/rs6000/power8.md: Adjust.
28095 * config/rs6000/power9.md: Adjust.
28096 * config/rs6000/rs64.md: Adjust.
28097 * config/rs6000/titan.md: Adjust.
28099 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28101 * config/i386/predicates.md (indirect_branch_operand): Rewrite
28102 ix86_indirect_branch_register logic.
28104 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28106 * config/i386/constraints.md (Bs): Update
28107 ix86_indirect_branch_register check. Don't check
28108 ix86_indirect_branch_register with GOT_memory_operand.
28110 * config/i386/predicates.md (GOT_memory_operand): Don't check
28111 ix86_indirect_branch_register here.
28112 (GOT32_symbol_operand): Likewise.
28114 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28116 * config/i386/predicates.md (constant_call_address_operand):
28117 Rewrite ix86_indirect_branch_register logic.
28118 (sibcall_insn_operand): Likewise.
28120 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28122 * config/i386/constraints.md (Bs): Replace
28123 ix86_indirect_branch_thunk_register with
28124 ix86_indirect_branch_register.
28126 * config/i386/i386.md (indirect_jump): Likewise.
28127 (tablejump): Likewise.
28128 (*sibcall_memory): Likewise.
28129 (*sibcall_value_memory): Likewise.
28130 Peepholes of indirect call and jump via memory: Likewise.
28131 * config/i386/i386.opt: Likewise.
28132 * config/i386/predicates.md (indirect_branch_operand): Likewise.
28133 (GOT_memory_operand): Likewise.
28134 (call_insn_operand): Likewise.
28135 (sibcall_insn_operand): Likewise.
28136 (GOT32_symbol_operand): Likewise.
28138 2018-01-15 Jakub Jelinek <jakub@redhat.com>
28140 PR middle-end/83837
28141 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
28142 type rather than type addr's type points to.
28143 (expand_omp_atomic_mutex): Likewise.
28144 (expand_omp_atomic): Likewise.
28146 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
28149 * config/i386/i386.c (output_indirect_thunk_function): Use
28150 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
28151 for __x86_return_thunk.
28153 2018-01-15 Richard Biener <rguenther@suse.de>
28155 PR middle-end/83850
28156 * expmed.c (extract_bit_field_1): Fix typo.
28158 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28161 * config/arm/iterators.md (VF): New mode iterator.
28162 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
28163 Remove integer-related logic from pattern.
28164 (neon_vabd<mode>_3): Likewise.
28166 2018-01-15 Jakub Jelinek <jakub@redhat.com>
28168 PR middle-end/82694
28169 * common.opt (fstrict-overflow): No longer an alias.
28170 (fwrapv-pointer): New option.
28171 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
28172 also for pointer types based on flag_wrapv_pointer.
28173 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
28174 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
28175 opts->x_flag_wrapv got set.
28176 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
28177 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
28178 POINTER_TYPE_OVERFLOW_UNDEFINED.
28179 * match.pd: Likewise in address comparison pattern.
28180 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
28182 2018-01-15 Richard Biener <rguenther@suse.de>
28185 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
28186 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
28187 Reset type names to their identifier if their TYPE_DECL doesn't
28188 have linkage (and thus is used for ODR and devirt).
28189 (save_debug_info_for_decl): Remove.
28190 (save_debug_info_for_type): Likewise.
28191 (add_tree_to_fld_list): Adjust.
28192 * tree-pretty-print.c (dump_generic_node): Make dumping of
28193 type names more robust.
28195 2018-01-15 Richard Biener <rguenther@suse.de>
28197 * BASE-VER: Bump to 8.0.1.
28199 2018-01-14 Martin Sebor <msebor@redhat.com>
28202 * builtins.c (check_access): Avoid warning when the no-warning bit
28205 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
28207 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
28208 * ira-color (allocno_hard_regs_compare): Likewise.
28210 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
28213 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
28214 Use .pushsection/.popsection.
28216 2018-01-14 Martin Sebor <msebor@redhat.com>
28219 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
28221 2018-01-14 Jakub Jelinek <jakub@redhat.com>
28223 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
28224 entry from extra_headers.
28225 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
28226 extra_headers, make the list bitwise identical to the i?86-*-* one.
28228 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28230 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
28231 -mcmodel=large with -mindirect-branch=thunk,
28232 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
28233 -mfunction-return=thunk-extern.
28234 * doc/invoke.texi: Document -mcmodel=large is incompatible with
28235 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
28236 -mfunction-return=thunk and -mfunction-return=thunk-extern.
28238 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28240 * config/i386/i386.c (print_reg): Print the name of the full
28241 integer register without '%'.
28242 (ix86_print_operand): Handle 'V'.
28243 * doc/extend.texi: Document 'V' modifier.
28245 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28247 * config/i386/constraints.md (Bs): Disallow memory operand for
28248 -mindirect-branch-register.
28250 * config/i386/predicates.md (indirect_branch_operand): Likewise.
28251 (GOT_memory_operand): Likewise.
28252 (call_insn_operand): Likewise.
28253 (sibcall_insn_operand): Likewise.
28254 (GOT32_symbol_operand): Likewise.
28255 * config/i386/i386.md (indirect_jump): Call convert_memory_address
28256 for -mindirect-branch-register.
28257 (tablejump): Likewise.
28258 (*sibcall_memory): Likewise.
28259 (*sibcall_value_memory): Likewise.
28260 Disallow peepholes of indirect call and jump via memory for
28261 -mindirect-branch-register.
28262 (*call_pop): Replace m with Bw.
28263 (*call_value_pop): Likewise.
28264 (*sibcall_pop_memory): Replace m with Bs.
28265 * config/i386/i386.opt (mindirect-branch-register): New option.
28266 * doc/invoke.texi: Document -mindirect-branch-register option.
28268 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28270 * config/i386/i386-protos.h (ix86_output_function_return): New.
28271 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
28272 set function_return_type.
28273 (indirect_thunk_name): Add ret_p to indicate thunk for function
28275 (output_indirect_thunk_function): Pass false to
28276 indirect_thunk_name.
28277 (ix86_output_indirect_branch_via_reg): Likewise.
28278 (ix86_output_indirect_branch_via_push): Likewise.
28279 (output_indirect_thunk_function): Create alias for function
28280 return thunk if regno < 0.
28281 (ix86_output_function_return): New function.
28282 (ix86_handle_fndecl_attribute): Handle function_return.
28283 (ix86_attribute_table): Add function_return.
28284 * config/i386/i386.h (machine_function): Add
28285 function_return_type.
28286 * config/i386/i386.md (simple_return_internal): Use
28287 ix86_output_function_return.
28288 (simple_return_internal_long): Likewise.
28289 * config/i386/i386.opt (mfunction-return=): New option.
28290 (indirect_branch): Mention -mfunction-return=.
28291 * doc/extend.texi: Document function_return function attribute.
28292 * doc/invoke.texi: Document -mfunction-return= option.
28294 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
28296 * config/i386/i386-opts.h (indirect_branch): New.
28297 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
28298 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
28299 with local indirect jump when converting indirect call and jump.
28300 (ix86_set_indirect_branch_type): New.
28301 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
28302 (indirectlabelno): New.
28303 (indirect_thunk_needed): Likewise.
28304 (indirect_thunk_bnd_needed): Likewise.
28305 (indirect_thunks_used): Likewise.
28306 (indirect_thunks_bnd_used): Likewise.
28307 (INDIRECT_LABEL): Likewise.
28308 (indirect_thunk_name): Likewise.
28309 (output_indirect_thunk): Likewise.
28310 (output_indirect_thunk_function): Likewise.
28311 (ix86_output_indirect_branch_via_reg): Likewise.
28312 (ix86_output_indirect_branch_via_push): Likewise.
28313 (ix86_output_indirect_branch): Likewise.
28314 (ix86_output_indirect_jmp): Likewise.
28315 (ix86_code_end): Call output_indirect_thunk_function if needed.
28316 (ix86_output_call_insn): Call ix86_output_indirect_branch if
28318 (ix86_handle_fndecl_attribute): Handle indirect_branch.
28319 (ix86_attribute_table): Add indirect_branch.
28320 * config/i386/i386.h (machine_function): Add indirect_branch_type
28321 and has_local_indirect_jump.
28322 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
28324 (tablejump): Likewise.
28325 (*indirect_jump): Use ix86_output_indirect_jmp.
28326 (*tablejump_1): Likewise.
28327 (simple_return_indirect_internal): Likewise.
28328 * config/i386/i386.opt (mindirect-branch=): New option.
28329 (indirect_branch): New.
28332 (thunk-inline): Likewise.
28333 (thunk-extern): Likewise.
28334 * doc/extend.texi: Document indirect_branch function attribute.
28335 * doc/invoke.texi: Document -mindirect-branch= option.
28337 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
28340 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
28342 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
28344 * ipa-inline.c (want_inline_small_function_p): Return false if
28345 inlining has already failed with CIF_FINAL_ERROR.
28346 (update_caller_keys): Call want_inline_small_function_p before
28348 (update_callee_keys): Likewise.
28350 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28352 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
28354 (rs6000_quadword_masked_address_p): Likewise.
28355 (quad_aligned_load_p): Likewise.
28356 (quad_aligned_store_p): Likewise.
28357 (const_load_sequence_p): Add comment to describe the outer-most loop.
28358 (mimic_memory_attributes_and_flags): New function.
28359 (rs6000_gen_stvx): Likewise.
28360 (replace_swapped_aligned_store): Likewise.
28361 (rs6000_gen_lvx): Likewise.
28362 (replace_swapped_aligned_load): Likewise.
28363 (replace_swapped_load_constant): Capitalize argument name in
28364 comment describing this function.
28365 (rs6000_analyze_swaps): Add a third pass to search for vector loads
28366 and stores that access quad-word aligned addresses and replace
28367 with stvx or lvx instructions when appropriate.
28368 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
28369 New function prototype.
28370 (rs6000_quadword_masked_address_p): Likewise.
28371 (rs6000_gen_lvx): Likewise.
28372 (rs6000_gen_stvx): Likewise.
28373 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
28374 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
28375 when memory address is aligned.
28376 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
28377 this split to select lvx instruction when memory address is aligned.
28378 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
28379 instruction when memory address is aligned.
28380 (*vsx_le_perm_load_v16qi): Likewise.
28381 (four unnamed splitters): Modify to select the stvx instruction
28382 when memory is aligned.
28384 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
28386 * predict.c (determine_unlikely_bbs): Handle correctly BBs
28387 which appears in the queue multiple times.
28389 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28390 Alan Hayward <alan.hayward@arm.com>
28391 David Sherwood <david.sherwood@arm.com>
28393 * tree-vectorizer.h (vec_lower_bound): New structure.
28394 (_loop_vec_info): Add check_nonzero and lower_bounds.
28395 (LOOP_VINFO_CHECK_NONZERO): New macro.
28396 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
28397 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
28398 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
28399 fields. Make seg_len the distance travelled, not including the
28401 (dr_direction_indicator): Declare.
28402 (dr_zero_step_indicator): Likewise.
28403 (dr_known_forward_stride_p): Likewise.
28404 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
28406 (runtime_alias_check_p): Allow runtime alias checks with
28408 (operator ==): Compare access_size and align.
28409 (prune_runtime_alias_test_list): Rework for new distinction between
28410 the access_size and seg_len.
28411 (create_intersect_range_checks_index): Likewise. Cope with polynomial
28413 (get_segment_min_max): New function.
28414 (create_intersect_range_checks): Use it.
28415 (dr_step_indicator): New function.
28416 (dr_direction_indicator): Likewise.
28417 (dr_zero_step_indicator): Likewise.
28418 (dr_known_forward_stride_p): Likewise.
28419 * tree-loop-distribution.c (data_ref_segment_size): Return
28420 DR_STEP * (niters - 1).
28421 (compute_alias_check_pairs): Update call to the dr_with_seg_len
28423 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
28424 (vect_preserves_scalar_order_p): New function, split out from...
28425 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
28426 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
28427 (vect_vfa_access_size): New function.
28428 (vect_vfa_align): Likewise.
28429 (vect_compile_time_alias): Take access_size_a and access_b arguments.
28430 (dump_lower_bound): New function.
28431 (vect_check_lower_bound): Likewise.
28432 (vect_small_gap_p): Likewise.
28433 (vectorizable_with_step_bound_p): Likewise.
28434 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
28435 depencies if the vectorization factor is 1. Convert the checks
28436 for nonzero steps into checks on the bounds of DR_STEP. Try using
28437 a bunds check for variable steps if the minimum required step is
28438 relatively small. Update calls to the dr_with_seg_len
28439 constructor and to vect_compile_time_alias.
28440 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
28442 (vect_loop_versioning): Call it.
28443 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
28445 (vect_estimate_min_profitable_iters): Account for any bounds checks.
28447 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28448 Alan Hayward <alan.hayward@arm.com>
28449 David Sherwood <david.sherwood@arm.com>
28451 * doc/sourcebuild.texi (vect_scatter_store): Document.
28452 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
28454 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
28456 * genopinit.c (main): Add supports_vec_scatter_store and
28457 supports_vec_scatter_store_cached to target_optabs.
28458 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
28459 IFN_MASK_SCATTER_STORE.
28460 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
28462 * internal-fn.h (internal_store_fn_p): Declare.
28463 (internal_fn_stored_value_index): Likewise.
28464 * internal-fn.c (scatter_store_direct): New macro.
28465 (expand_scatter_store_optab_fn): New function.
28466 (direct_scatter_store_optab_supported_p): New macro.
28467 (internal_store_fn_p): New function.
28468 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
28469 IFN_MASK_SCATTER_STORE.
28470 (internal_fn_mask_index): Likewise.
28471 (internal_fn_stored_value_index): New function.
28472 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
28473 for scatter stores.
28474 * optabs-query.h (supports_vec_scatter_store_p): Declare.
28475 * optabs-query.c (supports_vec_scatter_store_p): New function.
28476 * tree-vectorizer.h (vect_get_store_rhs): Declare.
28477 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
28478 true for scatter stores.
28479 (vect_gather_scatter_fn_p): Handle scatter stores too.
28480 (vect_check_gather_scatter): Consider using scatter stores if
28481 supports_vec_scatter_store_p.
28482 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
28483 scatter stores too.
28484 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28485 internal_fn_stored_value_index.
28486 (check_load_store_masking): Handle scatter stores too.
28487 (vect_get_store_rhs): Make public.
28488 (vectorizable_call): Use internal_store_fn_p.
28489 (vectorizable_store): Handle scatter store internal functions.
28490 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
28491 when deciding whether the end of the group has been reached.
28492 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
28493 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
28494 (mask_scatter_store<mode>): New insns.
28496 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28497 Alan Hayward <alan.hayward@arm.com>
28498 David Sherwood <david.sherwood@arm.com>
28500 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
28501 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
28502 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
28504 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
28505 Use vect_truncate_gather_scatter_offset if we can't treat the
28506 operation as a normal gather load or scatter store.
28507 (get_group_load_store_type): Take the gather_scatter_info
28508 as argument. Try using a gather load or scatter store for
28509 single-element groups.
28510 (get_load_store_type): Update calls to get_group_load_store_type
28511 and vect_use_strided_gather_scatters_p.
28513 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28514 Alan Hayward <alan.hayward@arm.com>
28515 David Sherwood <david.sherwood@arm.com>
28517 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
28518 optional tree argument.
28519 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
28521 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
28522 but continue to use the current value as a fallback.
28523 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
28524 to compare the updates.
28525 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
28526 (get_load_store_type): Use it when handling a strided access.
28527 (vect_get_strided_load_store_ops): New function.
28528 (vect_get_data_ptr_increment): Likewise.
28529 (vectorizable_load): Handle strided gather loads. Always pass
28530 a step to vect_create_data_ref_ptr and bump_vector_ptr.
28532 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28533 Alan Hayward <alan.hayward@arm.com>
28534 David Sherwood <david.sherwood@arm.com>
28536 * doc/md.texi (gather_load@var{m}): Document.
28537 (mask_gather_load@var{m}): Likewise.
28538 * genopinit.c (main): Add supports_vec_gather_load and
28539 supports_vec_gather_load_cached to target_optabs.
28540 * optabs-tree.c (init_tree_optimization_optabs): Use
28541 ggc_cleared_alloc to allocate target_optabs.
28542 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
28543 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
28545 * internal-fn.h (internal_load_fn_p): Declare.
28546 (internal_gather_scatter_fn_p): Likewise.
28547 (internal_fn_mask_index): Likewise.
28548 (internal_gather_scatter_fn_supported_p): Likewise.
28549 * internal-fn.c (gather_load_direct): New macro.
28550 (expand_gather_load_optab_fn): New function.
28551 (direct_gather_load_optab_supported_p): New macro.
28552 (direct_internal_fn_optab): New function.
28553 (internal_load_fn_p): Likewise.
28554 (internal_gather_scatter_fn_p): Likewise.
28555 (internal_fn_mask_index): Likewise.
28556 (internal_gather_scatter_fn_supported_p): Likewise.
28557 * optabs-query.c (supports_at_least_one_mode_p): New function.
28558 (supports_vec_gather_load_p): Likewise.
28559 * optabs-query.h (supports_vec_gather_load_p): Declare.
28560 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
28561 and memory_type field.
28562 (NUM_PATTERNS): Bump to 15.
28563 * tree-vect-data-refs.c: Include internal-fn.h.
28564 (vect_gather_scatter_fn_p): New function.
28565 (vect_describe_gather_scatter_call): Likewise.
28566 (vect_check_gather_scatter): Try using internal functions for
28567 gather loads. Recognize existing calls to a gather load function.
28568 (vect_analyze_data_refs): Consider using gather loads if
28569 supports_vec_gather_load_p.
28570 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
28571 (vect_get_gather_scatter_offset_type): Likewise.
28572 (vect_convert_mask_for_vectype): Likewise.
28573 (vect_add_conversion_to_patterm): Likewise.
28574 (vect_try_gather_scatter_pattern): Likewise.
28575 (vect_recog_gather_scatter_pattern): New pattern recognizer.
28576 (vect_vect_recog_func_ptrs): Add it.
28577 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
28578 internal_fn_mask_index and internal_gather_scatter_fn_p.
28579 (check_load_store_masking): Take the gather_scatter_info as an
28580 argument and handle gather loads.
28581 (vect_get_gather_scatter_ops): New function.
28582 (vectorizable_call): Check internal_load_fn_p.
28583 (vectorizable_load): Likewise. Handle gather load internal
28585 (vectorizable_store): Update call to check_load_store_masking.
28586 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
28587 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
28588 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
28589 (aarch64_gather_scale_operand_d): New predicates.
28590 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
28591 (mask_gather_load<mode>): New insns.
28593 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28594 Alan Hayward <alan.hayward@arm.com>
28595 David Sherwood <david.sherwood@arm.com>
28597 * optabs.def (fold_left_plus_optab): New optab.
28598 * doc/md.texi (fold_left_plus_@var{m}): Document.
28599 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
28600 * internal-fn.c (fold_left_direct): Define.
28601 (expand_fold_left_optab_fn): Likewise.
28602 (direct_fold_left_optab_supported_p): Likewise.
28603 * fold-const-call.c (fold_const_fold_left): New function.
28604 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
28605 * tree-parloops.c (valid_reduction_p): New function.
28606 (gather_scalar_reductions): Use it.
28607 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
28608 (vect_finish_replace_stmt): Declare.
28609 * tree-vect-loop.c (fold_left_reduction_fn): New function.
28610 (needs_fold_left_reduction_p): New function, split out from...
28611 (vect_is_simple_reduction): ...here. Accept reductions that
28612 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
28613 (vect_force_simple_reduction): Also store the reduction type in
28614 the assignment's STMT_VINFO_REDUC_TYPE.
28615 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
28616 (merge_with_identity): New function.
28617 (vect_expand_fold_left): Likewise.
28618 (vectorize_fold_left_reduction): Likewise.
28619 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
28620 scalar phi in place for it. Check for target support and reject
28621 cases that would reassociate the operation. Defer the transform
28622 phase to vectorize_fold_left_reduction.
28623 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
28624 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
28625 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
28627 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28629 * tree-if-conv.c (predicate_mem_writes): Remove redundant
28630 call to ifc_temp_var.
28632 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28633 Alan Hayward <alan.hayward@arm.com>
28634 David Sherwood <david.sherwood@arm.com>
28636 * target.def (legitimize_address_displacement): Take the original
28637 offset as a poly_int.
28638 * targhooks.h (default_legitimize_address_displacement): Update
28640 * targhooks.c (default_legitimize_address_displacement): Likewise.
28641 * doc/tm.texi: Regenerate.
28642 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
28643 as an argument, moving assert of ad->disp == ad->disp_term to...
28644 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
28645 Try calling targetm.legitimize_address_displacement before expanding
28646 the address rather than afterwards, and adjust for the new interface.
28647 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
28648 Match the new hook interface. Handle SVE addresses.
28649 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
28650 new hook interface.
28652 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28654 * Makefile.in (OBJS): Add early-remat.o.
28655 * target.def (select_early_remat_modes): New hook.
28656 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
28657 * doc/tm.texi: Regenerate.
28658 * targhooks.h (default_select_early_remat_modes): Declare.
28659 * targhooks.c (default_select_early_remat_modes): New function.
28660 * timevar.def (TV_EARLY_REMAT): New timevar.
28661 * passes.def (pass_early_remat): New pass.
28662 * tree-pass.h (make_pass_early_remat): Declare.
28663 * early-remat.c: New file.
28664 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
28666 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
28668 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28669 Alan Hayward <alan.hayward@arm.com>
28670 David Sherwood <david.sherwood@arm.com>
28672 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
28673 vfm1 with a bound_epilog parameter.
28674 (vect_do_peeling): Update calls accordingly, and move the prologue
28675 call earlier in the function. Treat the base bound_epilog as 0 for
28676 fully-masked loops and retain vf - 1 for other loops. Add 1 to
28677 this base when peeling for gaps.
28678 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
28679 with fully-masked loops.
28680 (vect_estimate_min_profitable_iters): Handle the single peeled
28681 iteration in that case.
28683 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28684 Alan Hayward <alan.hayward@arm.com>
28685 David Sherwood <david.sherwood@arm.com>
28687 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
28688 single-element interleaving even if the size is not a power of 2.
28689 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
28690 accesses for single-element interleaving if the group size is
28693 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28694 Alan Hayward <alan.hayward@arm.com>
28695 David Sherwood <david.sherwood@arm.com>
28697 * doc/md.texi (fold_extract_last_@var{m}): Document.
28698 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
28699 * optabs.def (fold_extract_last_optab): New optab.
28700 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
28701 * internal-fn.c (fold_extract_direct): New macro.
28702 (expand_fold_extract_optab_fn): Likewise.
28703 (direct_fold_extract_optab_supported_p): Likewise.
28704 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
28705 * tree-vect-loop.c (vect_model_reduction_cost): Handle
28706 EXTRACT_LAST_REDUCTION.
28707 (get_initial_def_for_reduction): Do not create an initial vector
28708 for EXTRACT_LAST_REDUCTION reductions.
28709 (vectorizable_reduction): Leave the scalar phi in place for
28710 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
28711 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
28712 epilogue code for EXTRACT_LAST_REDUCTION and defer the
28713 transform phase to vectorizable_condition.
28714 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
28716 (vect_finish_stmt_generation): ...here.
28717 (vect_finish_replace_stmt): New function.
28718 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
28719 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
28721 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
28723 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28724 Alan Hayward <alan.hayward@arm.com>
28725 David Sherwood <david.sherwood@arm.com>
28727 * doc/md.texi (extract_last_@var{m}): Document.
28728 * optabs.def (extract_last_optab): New optab.
28729 * internal-fn.def (EXTRACT_LAST): New internal function.
28730 * internal-fn.c (cond_unary_direct): New macro.
28731 (expand_cond_unary_optab_fn): Likewise.
28732 (direct_cond_unary_optab_supported_p): Likewise.
28733 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
28734 loops using EXTRACT_LAST.
28735 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
28736 (extract_last_<mode>): ...this optab.
28737 (vec_extract<mode><Vel>): Update accordingly.
28739 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28740 Alan Hayward <alan.hayward@arm.com>
28741 David Sherwood <david.sherwood@arm.com>
28743 * target.def (empty_mask_is_expensive): New hook.
28744 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
28745 * doc/tm.texi: Regenerate.
28746 * targhooks.h (default_empty_mask_is_expensive): Declare.
28747 * targhooks.c (default_empty_mask_is_expensive): New function.
28748 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
28749 if the target says that empty masks are expensive.
28750 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
28752 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
28754 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28755 Alan Hayward <alan.hayward@arm.com>
28756 David Sherwood <david.sherwood@arm.com>
28758 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
28759 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
28760 (vect_use_loop_mask_for_alignment_p): New function.
28761 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
28762 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
28763 niters_skip argument. Make sure that the first niters_skip elements
28764 of the first iteration are inactive.
28765 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
28766 Update call to vect_set_loop_masks_directly.
28767 (get_misalign_in_elems): New function, split out from...
28768 (vect_gen_prolog_loop_niters): ...here.
28769 (vect_update_init_of_dr): Take a code argument that specifies whether
28770 the adjustment should be added or subtracted.
28771 (vect_update_init_of_drs): Likewise.
28772 (vect_prepare_for_masked_peels): New function.
28773 (vect_do_peeling): Skip prologue peeling if we're using a mask
28774 instead. Update call to vect_update_inits_of_drs.
28775 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28777 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
28778 alignment. Do not include the number of peeled iterations in
28779 the minimum threshold in that case.
28780 (vectorizable_induction): Adjust the start value down by
28781 LOOP_VINFO_MASK_SKIP_NITERS iterations.
28782 (vect_transform_loop): Call vect_prepare_for_masked_peels.
28783 Take the number of skipped iterations into account when calculating
28785 * tree-vect-stmts.c (vect_gen_while_not): New function.
28787 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28788 Alan Hayward <alan.hayward@arm.com>
28789 David Sherwood <david.sherwood@arm.com>
28791 * doc/sourcebuild.texi (vect_fully_masked): Document.
28792 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
28793 default value to 0.
28794 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
28796 (vect_analyze_loop_2): ...here. Don't check the vectorization
28797 factor against the number of loop iterations if the loop is
28800 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28801 Alan Hayward <alan.hayward@arm.com>
28802 David Sherwood <david.sherwood@arm.com>
28804 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
28805 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
28806 (dump_groups): Update accordingly.
28807 (iv_use::mem_type): New member variable.
28808 (address_p): New function.
28809 (record_use): Add a mem_type argument and initialize the new
28811 (record_group_use): Add a mem_type argument. Use address_p.
28812 Remove obsolete null checks of base_object. Update call to record_use.
28813 (find_interesting_uses_op): Update call to record_group_use.
28814 (find_interesting_uses_cond): Likewise.
28815 (find_interesting_uses_address): Likewise.
28816 (get_mem_type_for_internal_fn): New function.
28817 (find_address_like_use): Likewise.
28818 (find_interesting_uses_stmt): Try find_address_like_use before
28819 calling find_interesting_uses_op.
28820 (addr_offset_valid_p): Use the iv mem_type field as the type
28821 of the addressed memory.
28822 (add_autoinc_candidates): Likewise.
28823 (get_address_cost): Likewise.
28824 (split_small_address_groups_p): Use address_p.
28825 (split_address_groups): Likewise.
28826 (add_iv_candidate_for_use): Likewise.
28827 (autoinc_possible_for_pair): Likewise.
28828 (rewrite_groups): Likewise.
28829 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
28830 (determine_group_iv_cost): Update after split of USE_ADDRESS.
28831 (get_alias_ptr_type_for_ptr_address): New function.
28832 (rewrite_use_address): Rewrite address uses in calls that were
28833 identified by find_address_like_use.
28835 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28836 Alan Hayward <alan.hayward@arm.com>
28837 David Sherwood <david.sherwood@arm.com>
28839 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
28841 * gimple-expr.h (is_gimple_addressable: Likewise.
28842 * gimple-expr.c (is_gimple_address): Likewise.
28843 * internal-fn.c (expand_call_mem_ref): New function.
28844 (expand_mask_load_optab_fn): Use it.
28845 (expand_mask_store_optab_fn): Likewise.
28847 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28848 Alan Hayward <alan.hayward@arm.com>
28849 David Sherwood <david.sherwood@arm.com>
28851 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
28852 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
28853 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
28854 (cond_umax@var{mode}): Document.
28855 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
28856 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
28857 (cond_umin_optab, cond_umax_optab): New optabs.
28858 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
28859 (COND_IOR, COND_XOR): New internal functions.
28860 * internal-fn.h (get_conditional_internal_fn): Declare.
28861 * internal-fn.c (cond_binary_direct): New macro.
28862 (expand_cond_binary_optab_fn): Likewise.
28863 (direct_cond_binary_optab_supported_p): Likewise.
28864 (get_conditional_internal_fn): New function.
28865 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
28866 Cope with reduction statements that are vectorized as calls rather
28868 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
28869 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
28870 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
28871 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
28872 (UNSPEC_COND_EOR): New unspecs.
28873 (optab): Add mappings for them.
28874 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
28875 (sve_int_op, sve_fp_op): New int attributes.
28877 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28878 Alan Hayward <alan.hayward@arm.com>
28879 David Sherwood <david.sherwood@arm.com>
28881 * optabs.def (while_ult_optab): New optab.
28882 * doc/md.texi (while_ult@var{m}@var{n}): Document.
28883 * internal-fn.def (WHILE_ULT): New internal function.
28884 * internal-fn.h (direct_internal_fn_supported_p): New override
28885 that takes two types as argument.
28886 * internal-fn.c (while_direct): New macro.
28887 (expand_while_optab_fn): New function.
28888 (convert_optab_supported_p): Likewise.
28889 (direct_while_optab_supported_p): New macro.
28890 * wide-int.h (wi::udiv_ceil): New function.
28891 * tree-vectorizer.h (rgroup_masks): New structure.
28892 (vec_loop_masks): New typedef.
28893 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
28894 and fully_masked_p.
28895 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
28896 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
28897 (vect_max_vf): New function.
28898 (slpeel_make_loop_iterate_ntimes): Delete.
28899 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
28900 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
28901 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
28902 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
28903 internal-fn.h, stor-layout.h and optabs-query.h.
28904 (vect_set_loop_mask): New function.
28905 (add_preheader_seq): Likewise.
28906 (add_header_seq): Likewise.
28907 (interleave_supported_p): Likewise.
28908 (vect_maybe_permute_loop_masks): Likewise.
28909 (vect_set_loop_masks_directly): Likewise.
28910 (vect_set_loop_condition_masked): Likewise.
28911 (vect_set_loop_condition_unmasked): New function, split out from
28912 slpeel_make_loop_iterate_ntimes.
28913 (slpeel_make_loop_iterate_ntimes): Rename to..
28914 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
28915 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
28916 (vect_do_peeling): Update call accordingly.
28917 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
28919 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28920 mask_compare_type, can_fully_mask_p and fully_masked_p.
28921 (release_vec_loop_masks): New function.
28922 (_loop_vec_info): Use it to free the loop masks.
28923 (can_produce_all_loop_masks_p): New function.
28924 (vect_get_max_nscalars_per_iter): Likewise.
28925 (vect_verify_full_masking): Likewise.
28926 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
28927 retries, and free the mask rgroups before retrying. Check loop-wide
28928 reasons for disallowing fully-masked loops. Make the final decision
28929 about whether use a fully-masked loop or not.
28930 (vect_estimate_min_profitable_iters): Do not assume that peeling
28931 for the number of iterations will be needed for fully-masked loops.
28932 (vectorizable_reduction): Disable fully-masked loops.
28933 (vectorizable_live_operation): Likewise.
28934 (vect_halve_mask_nunits): New function.
28935 (vect_double_mask_nunits): Likewise.
28936 (vect_record_loop_mask): Likewise.
28937 (vect_get_loop_mask): Likewise.
28938 (vect_transform_loop): Handle the case in which the final loop
28939 iteration might handle a partial vector. Call vect_set_loop_condition
28940 instead of slpeel_make_loop_iterate_ntimes.
28941 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
28942 (check_load_store_masking): New function.
28943 (prepare_load_store_mask): Likewise.
28944 (vectorizable_store): Handle fully-masked loops.
28945 (vectorizable_load): Likewise.
28946 (supportable_widening_operation): Use vect_halve_mask_nunits for
28948 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
28949 (vect_gen_while): New function.
28950 * config/aarch64/aarch64.md (umax<mode>3): New expander.
28951 (aarch64_uqdec<mode>): New insn.
28953 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28954 Alan Hayward <alan.hayward@arm.com>
28955 David Sherwood <david.sherwood@arm.com>
28957 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
28958 (reduc_xor_scal_optab): New optabs.
28959 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
28960 (reduc_xor_scal_@var{m}): Document.
28961 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
28962 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
28963 internal functions.
28964 * fold-const-call.c (fold_const_call): Handle them.
28965 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
28966 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
28967 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
28968 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
28969 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
28970 (UNSPEC_XORV): New unspecs.
28971 (optab): Add entries for them.
28972 (BITWISEV): New int iterator.
28973 (bit_reduc_op): New int attributes.
28975 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28976 Alan Hayward <alan.hayward@arm.com>
28977 David Sherwood <david.sherwood@arm.com>
28979 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
28980 * internal-fn.def (VEC_SHL_INSERT): New internal function.
28981 * optabs.def (vec_shl_insert_optab): New optab.
28982 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
28983 (duplicate_and_interleave): Likewise.
28984 * tree-vect-loop.c: Include internal-fn.h.
28985 (neutral_op_for_slp_reduction): New function, split out from
28986 get_initial_defs_for_reduction.
28987 (get_initial_def_for_reduction): Handle option 2 for variable-length
28988 vectors by loading the neutral value into a vector and then shifting
28989 the initial value into element 0.
28990 (get_initial_defs_for_reduction): Replace the code argument with
28991 the neutral value calculated by neutral_op_for_slp_reduction.
28992 Use gimple_build_vector for constant-length vectors.
28993 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
28994 but the first group_size elements have a neutral value.
28995 Use duplicate_and_interleave otherwise.
28996 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
28997 Update call to get_initial_defs_for_reduction. Handle SLP
28998 reductions for variable-length vectors by creating one vector
28999 result for each scalar result, with the elements associated
29000 with other scalar results stubbed out with the neutral value.
29001 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
29002 Require IFN_VEC_SHL_INSERT for double reductions on
29003 variable-length vectors, or SLP reductions that have
29004 a neutral value. Require can_duplicate_and_interleave_p
29005 support for variable-length unchained SLP reductions if there
29006 is no neutral value, such as for MIN/MAX reductions. Also require
29007 the number of vector elements to be a multiple of the number of
29008 SLP statements when doing variable-length unchained SLP reductions.
29009 Update call to vect_create_epilog_for_reduction.
29010 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
29011 and remove initial values.
29012 (duplicate_and_interleave): Make public.
29013 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
29014 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
29016 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29017 Alan Hayward <alan.hayward@arm.com>
29018 David Sherwood <david.sherwood@arm.com>
29020 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
29021 (can_duplicate_and_interleave_p): New function.
29022 (vect_get_and_check_slp_defs): Take the vector of statements
29023 rather than just the current one. Remove excess parentheses.
29024 Restriction rejectinon of vect_constant_def and vect_external_def
29025 for variable-length vectors to boolean types, or types for which
29026 can_duplicate_and_interleave_p is false.
29027 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
29028 (duplicate_and_interleave): New function.
29029 (vect_get_constant_vectors): Use gimple_build_vector for
29030 constant-length vectors and suitable variable-length constant
29031 vectors. Use duplicate_and_interleave for other variable-length
29032 vectors. Don't defer the update when inserting new statements.
29034 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29035 Alan Hayward <alan.hayward@arm.com>
29036 David Sherwood <david.sherwood@arm.com>
29038 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
29039 min_profitable_iters doesn't go negative.
29041 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29042 Alan Hayward <alan.hayward@arm.com>
29043 David Sherwood <david.sherwood@arm.com>
29045 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
29046 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
29047 * optabs.def (vec_mask_load_lanes_optab): New optab.
29048 (vec_mask_store_lanes_optab): Likewise.
29049 * internal-fn.def (MASK_LOAD_LANES): New internal function.
29050 (MASK_STORE_LANES): Likewise.
29051 * internal-fn.c (mask_load_lanes_direct): New macro.
29052 (mask_store_lanes_direct): Likewise.
29053 (expand_mask_load_optab_fn): Handle masked operations.
29054 (expand_mask_load_lanes_optab_fn): New macro.
29055 (expand_mask_store_optab_fn): Handle masked operations.
29056 (expand_mask_store_lanes_optab_fn): New macro.
29057 (direct_mask_load_lanes_optab_supported_p): Likewise.
29058 (direct_mask_store_lanes_optab_supported_p): Likewise.
29059 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
29061 (vect_load_lanes_supported): Likewise.
29062 * tree-vect-data-refs.c (strip_conversion): New function.
29063 (can_group_stmts_p): Likewise.
29064 (vect_analyze_data_ref_accesses): Use it instead of checking
29065 for a pair of assignments.
29066 (vect_store_lanes_supported): Take a masked_p parameter.
29067 (vect_load_lanes_supported): Likewise.
29068 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
29069 vect_store_lanes_supported and vect_load_lanes_supported.
29070 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
29071 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
29072 parameter. Don't allow gaps for masked accesses.
29073 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
29074 and vect_load_lanes_supported.
29075 (get_load_store_type): Take a masked_p parameter and update
29076 call to get_group_load_store_type.
29077 (vectorizable_store): Update call to get_load_store_type.
29078 Handle IFN_MASK_STORE_LANES.
29079 (vectorizable_load): Update call to get_load_store_type.
29080 Handle IFN_MASK_LOAD_LANES.
29082 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29083 Alan Hayward <alan.hayward@arm.com>
29084 David Sherwood <david.sherwood@arm.com>
29086 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
29088 * config/aarch64/aarch64-protos.h
29089 (aarch64_sve_struct_memory_operand_p): Declare.
29090 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
29091 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
29092 (VPRED, vpred): Handle SVE structure modes.
29093 * config/aarch64/constraints.md (Utx): New constraint.
29094 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
29095 (aarch64_sve_struct_nonimmediate_operand): New predicates.
29096 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
29097 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
29098 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
29099 structure modes. Split into pieces after RA.
29100 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
29101 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
29103 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
29104 SVE structure modes.
29105 (aarch64_classify_address): Likewise.
29106 (sizetochar): Move earlier in file.
29107 (aarch64_print_operand): Handle SVE register lists.
29108 (aarch64_array_mode): New function.
29109 (aarch64_sve_struct_memory_operand_p): Likewise.
29110 (TARGET_ARRAY_MODE): Redefine.
29112 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29113 Alan Hayward <alan.hayward@arm.com>
29114 David Sherwood <david.sherwood@arm.com>
29116 * target.def (array_mode): New target hook.
29117 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
29118 * doc/tm.texi: Regenerate.
29119 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
29120 * hooks.c (hook_optmode_mode_uhwi_none): New function.
29121 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
29122 targetm.array_mode.
29123 * stor-layout.c (mode_for_array): Likewise. Support polynomial
29126 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29127 Alan Hayward <alan.hayward@arm.com>
29128 David Sherwood <david.sherwood@arm.com>
29130 * fold-const.c (fold_binary_loc): Check the argument types
29131 rather than the result type when testing for a vector operation.
29133 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29135 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
29136 * doc/tm.texi: Regenerate.
29138 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29139 Alan Hayward <alan.hayward@arm.com>
29140 David Sherwood <david.sherwood@arm.com>
29142 * doc/invoke.texi (-msve-vector-bits=): Document new option.
29143 (sve): Document new AArch64 extension.
29144 * doc/md.texi (w): Extend the description of the AArch64
29145 constraint to include SVE vectors.
29146 (Upl, Upa): Document new AArch64 predicate constraints.
29147 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
29149 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
29150 (msve-vector-bits=): New option.
29151 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
29152 SVE when these are disabled.
29153 (sve): New extension.
29154 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
29155 modes. Adjust their number of units based on aarch64_sve_vg.
29156 (MAX_BITSIZE_MODE_ANY_MODE): Define.
29157 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
29158 aarch64_addr_query_type.
29159 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
29160 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
29161 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
29162 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
29163 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
29164 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
29165 (aarch64_simd_imm_zero_p): Delete.
29166 (aarch64_check_zero_based_sve_index_immediate): Declare.
29167 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
29168 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
29169 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
29170 (aarch64_sve_float_mul_immediate_p): Likewise.
29171 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
29172 rather than an rtx.
29173 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
29174 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
29175 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
29176 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
29177 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
29178 (aarch64_regmode_natural_size): Likewise.
29179 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
29180 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
29182 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
29183 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
29184 for VG and the SVE predicate registers.
29185 (V_ALIASES): Add a "z"-prefixed alias.
29186 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
29187 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
29188 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
29189 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
29190 (REG_CLASS_NAMES): Add entries for them.
29191 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
29192 and the predicate registers.
29193 (aarch64_sve_vg): Declare.
29194 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
29195 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
29196 (REGMODE_NATURAL_SIZE): Define.
29197 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
29199 * config/aarch64/aarch64.c: Include cfgrtl.h.
29200 (simd_immediate_info): Add a constructor for series vectors,
29201 and an associated step field.
29202 (aarch64_sve_vg): New variable.
29203 (aarch64_dbx_register_number): Handle VG and the predicate registers.
29204 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
29205 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
29206 (VEC_ANY_DATA, VEC_STRUCT): New constants.
29207 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
29208 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
29209 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
29210 (aarch64_get_mask_mode): New functions.
29211 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
29212 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
29213 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
29214 predicate modes and predicate registers. Explicitly restrict
29215 GPRs to modes of 16 bytes or smaller. Only allow FP registers
29216 to store a vector mode if it is recognized by
29217 aarch64_classify_vector_mode.
29218 (aarch64_regmode_natural_size): New function.
29219 (aarch64_hard_regno_caller_save_mode): Return the original mode
29221 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
29222 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
29223 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
29224 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
29226 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
29227 does not overlap dest if the function is frame-related. Handle
29229 (aarch64_split_add_offset): New function.
29230 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
29231 them aarch64_add_offset.
29232 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
29233 and update call to aarch64_sub_sp.
29234 (aarch64_add_cfa_expression): New function.
29235 (aarch64_expand_prologue): Pass extra temporary registers to the
29236 functions above. Handle the case in which we need to emit new
29237 DW_CFA_expressions for registers that were originally saved
29238 relative to the stack pointer, but now have to be expressed
29239 relative to the frame pointer.
29240 (aarch64_output_mi_thunk): Pass extra temporary registers to the
29242 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
29243 IP0 and IP1 values for SVE frames.
29244 (aarch64_expand_vec_series): New function.
29245 (aarch64_expand_sve_widened_duplicate): Likewise.
29246 (aarch64_expand_sve_const_vector): Likewise.
29247 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
29248 Handle SVE constants. Use emit_move_insn to move a force_const_mem
29249 into the register, rather than emitting a SET directly.
29250 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
29251 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
29252 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
29253 (offset_9bit_signed_scaled_p): New functions.
29254 (aarch64_replicate_bitmask_imm): New function.
29255 (aarch64_bitmask_imm): Use it.
29256 (aarch64_cannot_force_const_mem): Reject expressions involving
29257 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
29258 (aarch64_classify_index): Handle SVE indices, by requiring
29259 a plain register index with a scale that matches the element size.
29260 (aarch64_classify_address): Handle SVE addresses. Assert that
29261 the mode of the address is VOIDmode or an integer mode.
29262 Update call to aarch64_classify_symbol.
29263 (aarch64_classify_symbolic_expression): Update call to
29264 aarch64_classify_symbol.
29265 (aarch64_const_vec_all_in_range_p): New function.
29266 (aarch64_print_vector_float_operand): Likewise.
29267 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
29268 "vN" for FP registers with SVE modes. Handle (const ...) vectors
29269 and the FP immediates 1.0 and 0.5.
29270 (aarch64_print_address_internal): Handle SVE addresses.
29271 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
29272 (aarch64_regno_regclass): Handle predicate registers.
29273 (aarch64_secondary_reload): Handle big-endian reloads of SVE
29275 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
29276 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
29277 (aarch64_convert_sve_vector_bits): New function.
29278 (aarch64_override_options): Use it to handle -msve-vector-bits=.
29279 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
29280 rather than an rtx.
29281 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
29282 Handle SVE vector and predicate modes. Accept VL-based constants
29283 that need only one temporary register, and VL offsets that require
29284 no temporary registers.
29285 (aarch64_conditional_register_usage): Mark the predicate registers
29286 as fixed if SVE isn't available.
29287 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
29288 Return true for SVE vector and predicate modes.
29289 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
29290 rather than an unsigned int. Handle SVE modes.
29291 (aarch64_preferred_simd_mode): Update call accordingly. Handle
29293 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
29295 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
29296 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
29297 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
29298 (aarch64_sve_float_mul_immediate_p): New functions.
29299 (aarch64_sve_valid_immediate): New function.
29300 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
29301 Explicitly reject structure modes. Check for INDEX constants.
29302 Handle PTRUE and PFALSE constants.
29303 (aarch64_check_zero_based_sve_index_immediate): New function.
29304 (aarch64_simd_imm_zero_p): Delete.
29305 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
29306 vector modes. Accept constants in the range of CNT[BHWD].
29307 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
29308 ask for an Advanced SIMD mode.
29309 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
29310 (aarch64_simd_vector_alignment): Handle SVE predicates.
29311 (aarch64_vectorize_preferred_vector_alignment): New function.
29312 (aarch64_simd_vector_alignment_reachable): Use it instead of
29314 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
29315 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
29317 (MAX_VECT_LEN): Delete.
29318 (expand_vec_perm_d): Add a vec_flags field.
29319 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
29320 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
29321 (aarch64_evpc_ext): Don't apply a big-endian lane correction
29323 (aarch64_evpc_rev): Rename to...
29324 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
29325 (aarch64_evpc_rev_global): New function.
29326 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
29327 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
29329 (aarch64_evpc_sve_tbl): New function.
29330 (aarch64_expand_vec_perm_const_1): Update after rename of
29331 aarch64_evpc_rev. Handle SVE permutes too, trying
29332 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
29333 than aarch64_evpc_tbl.
29334 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
29335 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
29336 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
29337 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
29338 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
29339 (aarch64_expand_sve_vcond): New functions.
29340 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
29341 of aarch64_vector_mode_p.
29342 (aarch64_dwarf_poly_indeterminate_value): New function.
29343 (aarch64_compute_pressure_classes): Likewise.
29344 (aarch64_can_change_mode_class): Likewise.
29345 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
29346 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
29347 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
29348 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
29349 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
29350 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
29351 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
29352 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
29354 (Dn, Dl, Dr): Accept const as well as const_vector.
29355 (Dz): Likewise. Compare against CONST0_RTX.
29356 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
29357 of "vector" where appropriate.
29358 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
29359 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
29360 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
29361 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
29362 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
29363 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
29364 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
29365 (v_int_equiv): Extend to SVE modes.
29366 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
29368 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
29369 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
29370 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
29371 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
29372 (SVE_COND_FP_CMP): New int iterators.
29373 (perm_hilo): Handle the new unpack unspecs.
29374 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
29376 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
29377 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
29378 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
29379 (aarch64_equality_operator, aarch64_constant_vector_operand)
29380 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
29381 (aarch64_sve_nonimmediate_operand): Likewise.
29382 (aarch64_sve_general_operand): Likewise.
29383 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
29384 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
29385 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
29386 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
29387 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
29388 (aarch64_sve_float_arith_immediate): Likewise.
29389 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
29390 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
29391 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
29392 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
29393 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
29394 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
29395 (aarch64_sve_float_arith_operand): Likewise.
29396 (aarch64_sve_float_arith_with_sub_operand): Likewise.
29397 (aarch64_sve_float_mul_operand): Likewise.
29398 (aarch64_sve_vec_perm_operand): Likewise.
29399 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
29400 (aarch64_mov_operand): Accept const_poly_int and const_vector.
29401 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
29402 as well as const_vector.
29403 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
29404 in file. Use CONST0_RTX and CONSTM1_RTX.
29405 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
29406 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
29407 Use aarch64_simd_imm_zero.
29408 * config/aarch64/aarch64-sve.md: New file.
29409 * config/aarch64/aarch64.md: Include it.
29410 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
29411 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
29412 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
29413 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
29414 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
29415 (sve): New attribute.
29416 (enabled): Disable instructions with the sve attribute unless
29418 (movqi, movhi): Pass CONST_POLY_INT operaneds through
29419 aarch64_expand_mov_immediate.
29420 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
29421 CNT[BHSD] immediates.
29422 (movti): Split CONST_POLY_INT moves into two halves.
29423 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
29424 Split additions that need a temporary here if the destination
29425 is the stack pointer.
29426 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
29427 (*add<mode>3_poly_1): New instruction.
29428 (set_clobber_cc): New expander.
29430 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29432 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
29433 parameter and use it instead of GET_MODE_SIZE (innermode). Use
29434 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
29435 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
29436 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
29437 Change innermode from fixed_mode_size to machine_mode.
29438 (simplify_subreg): Update call accordingly. Handle a constant-sized
29439 subreg of a variable-length CONST_VECTOR.
29441 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29442 Alan Hayward <alan.hayward@arm.com>
29443 David Sherwood <david.sherwood@arm.com>
29445 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
29446 (add_offset_to_base): New function, split out from...
29447 (create_mem_ref): ...here. When handling a scale other than 1,
29448 check first whether the address is valid without the offset.
29449 Add it into the base if so, leaving the index and scale as-is.
29451 2018-01-12 Jakub Jelinek <jakub@redhat.com>
29454 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
29455 fold_for_warn before checking if arg2 is INTEGER_CST.
29457 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
29459 * config/rs6000/predicates.md (load_multiple_operation): Delete.
29460 (store_multiple_operation): Delete.
29461 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
29462 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
29463 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
29464 guarded by TARGET_STRING.
29465 (rs6000_output_load_multiple): Delete.
29466 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29467 OPTION_MASK_STRING / TARGET_STRING handling.
29468 (print_operand) <'N', 'O'>: Add comment that these are unused now.
29469 (const rs6000_opt_masks) <"string">: Change mask to 0.
29470 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
29471 (MASK_STRING): Delete.
29472 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
29474 (load_multiple): Delete.
29481 (store_multiple): Delete.
29488 (movmemsi_8reg): Delete.
29489 (corresponding unnamed define_insn): Delete.
29490 (movmemsi_6reg): Delete.
29491 (corresponding unnamed define_insn): Delete.
29492 (movmemsi_4reg): Delete.
29493 (corresponding unnamed define_insn): Delete.
29494 (movmemsi_2reg): Delete.
29495 (corresponding unnamed define_insn): Delete.
29496 (movmemsi_1reg): Delete.
29497 (corresponding unnamed define_insn): Delete.
29498 * config/rs6000/rs6000.opt (mno-string): New.
29499 (mstring): Replace by deprecation warning stub.
29500 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
29502 2018-01-12 Jakub Jelinek <jakub@redhat.com>
29504 * regrename.c (regrename_do_replace): If replacing the same
29505 reg multiple times, try to reuse last created gen_raw_REG.
29508 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
29509 main to workaround a bug in GDB.
29511 2018-01-12 Tom de Vries <tom@codesourcery.com>
29514 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
29516 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
29518 PR rtl-optimization/80481
29519 * ira-color.c (get_cap_member): New function.
29520 (allocnos_conflict_by_live_ranges_p): Use it.
29521 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
29522 (setup_slot_coalesced_allocno_live_ranges): Ditto.
29524 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
29527 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
29528 (*saddl_se_1): Ditto.
29529 (*ssubsi_1): Ditto.
29530 (*ssubl_se_1): Ditto.
29532 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29534 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
29535 rather than wi::to_widest for DR_INITs.
29536 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
29537 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
29538 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
29540 (vect_analyze_group_access_1): Note that here.
29542 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29544 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
29545 polynomial type sizes.
29547 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
29549 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
29550 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
29551 (gimple_add_tmp_var): Likewise.
29553 2018-01-12 Martin Liska <mliska@suse.cz>
29555 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
29556 (gimple_alloc_sizes): Likewise.
29557 (dump_gimple_statistics): Use PRIu64 in printf format.
29558 * gimple.h: Change uint64_t to int.
29560 2018-01-12 Martin Liska <mliska@suse.cz>
29562 * tree-core.h: Use uint64_t instead of int.
29563 * tree.c (tree_node_counts): Likewise.
29564 (tree_node_sizes): Likewise.
29565 (dump_tree_statistics): Use PRIu64 in printf format.
29567 2018-01-12 Martin Liska <mliska@suse.cz>
29569 * Makefile.in: As qsort_chk is implemented in vec.c, add
29570 vec.o to linkage of gencfn-macros.
29571 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
29572 passing the info to record_node_allocation_statistics.
29573 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
29575 * ggc-common.c (struct ggc_usage): Add operator== and use
29576 it in operator< and compare function.
29577 * mem-stats.h (struct mem_usage): Likewise.
29578 * vec.c (struct vec_usage): Remove operator< and compare
29579 function. Can be simply inherited.
29581 2018-01-12 Martin Jambor <mjambor@suse.cz>
29584 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
29585 * tree-ssa-math-opts.c: Include domwalk.h.
29586 (convert_mult_to_fma_1): New function.
29587 (fma_transformation_info): New type.
29588 (fma_deferring_state): Likewise.
29589 (cancel_fma_deferring): New function.
29590 (result_of_phi): Likewise.
29591 (last_fma_candidate_feeds_initial_phi): Likewise.
29592 (convert_mult_to_fma): Added deferring logic, split actual
29593 transformation to convert_mult_to_fma_1.
29594 (math_opts_dom_walker): New type.
29595 (math_opts_dom_walker::after_dom_children): New method, body moved
29596 here from pass_optimize_widening_mul::execute, added deferring logic
29598 (pass_optimize_widening_mul::execute): Moved most of code to
29599 math_opts_dom_walker::after_dom_children.
29600 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
29601 * config/i386/i386.c (ix86_option_override_internal): Added
29602 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
29604 2018-01-12 Richard Biener <rguenther@suse.de>
29607 * dwarf2out.c (gen_variable_die): Do not reset old_die for
29608 inline instance vars.
29610 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
29613 * config/rx/rx.c (rx_is_restricted_memory_address):
29614 Handle SUBREG case.
29616 2018-01-12 Richard Biener <rguenther@suse.de>
29618 PR tree-optimization/80846
29619 * target.def (split_reduction): New target hook.
29620 * targhooks.c (default_split_reduction): New function.
29621 * targhooks.h (default_split_reduction): Declare.
29622 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
29623 target requests first reduce vectors by combining low and high
29625 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
29626 (get_vectype_for_scalar_type_and_size): Export.
29627 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
29628 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
29629 * doc/tm.texi: Regenerate.
29630 * config/i386/i386.c (ix86_split_reduction): Implement
29631 TARGET_VECTORIZE_SPLIT_REDUCTION.
29633 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29636 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
29637 in PIC mode except for TARGET_VXWORKS_RTP.
29638 * config/sparc/sparc.c: Include cfgrtl.h.
29639 (TARGET_INIT_PIC_REG): Define.
29640 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
29641 (sparc_pic_register_p): New predicate.
29642 (sparc_legitimate_address_p): Use it.
29643 (sparc_legitimize_pic_address): Likewise.
29644 (sparc_delegitimize_address): Likewise.
29645 (sparc_mode_dependent_address_p): Likewise.
29646 (gen_load_pcrel_sym): Remove 4th parameter.
29647 (load_got_register): Adjust call to above. Remove obsolete stuff.
29648 (sparc_expand_prologue): Do not call load_got_register here.
29649 (sparc_flat_expand_prologue): Likewise.
29650 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
29651 (sparc_use_pseudo_pic_reg): New function.
29652 (sparc_init_pic_reg): Likewise.
29653 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
29654 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
29656 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
29658 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29659 Add item for branch_cost.
29661 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
29663 PR rtl-optimization/83565
29664 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
29665 not extend the result to a larger mode for rotate operations.
29666 (num_sign_bit_copies1): Likewise.
29668 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29671 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
29673 Use values-Xc.o for -pedantic.
29674 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
29676 2018-01-12 Martin Liska <mliska@suse.cz>
29679 * ipa-devirt.c (final_warning_record::grow_type_warnings):
29681 (possible_polymorphic_call_targets): Use it.
29682 (ipa_devirt): Likewise.
29684 2018-01-12 Martin Liska <mliska@suse.cz>
29686 * profile-count.h (enum profile_quality): Use 0 as invalid
29687 enum value of profile_quality.
29689 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
29691 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
29692 -mext-string options.
29694 2018-01-12 Richard Biener <rguenther@suse.de>
29696 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
29697 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
29698 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
29700 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
29702 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29704 * configure.ac (--with-long-double-format): Add support for the
29705 configuration option to change the default long double format on
29707 * config.gcc (powerpc*-linux*-*): Likewise.
29708 * configure: Regenerate.
29709 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
29710 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
29711 used without modification.
29713 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29715 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
29716 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
29717 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
29718 MISC_BUILTIN_SPEC_BARRIER.
29719 (rs6000_init_builtins): Likewise.
29720 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
29722 (speculation_barrier): New define_insn.
29723 * doc/extend.texi: Document __builtin_speculation_barrier.
29725 2018-01-11 Jakub Jelinek <jakub@redhat.com>
29728 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
29729 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
29730 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
29732 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
29733 integral modes instead of "ss" and "sd".
29734 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
29735 vectors with 32-bit and 64-bit elements.
29736 (vecdupssescalarmodesuffix): New mode attribute.
29737 (vec_dup<mode>): Use it.
29739 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
29742 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
29743 frame if argument is passed on stack.
29745 2018-01-11 Jakub Jelinek <jakub@redhat.com>
29748 * ree.c (combine_reaching_defs): Optimize also
29749 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
29750 reg2=any_extend(exp); reg1=reg2;, formatting fix.
29752 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
29754 PR middle-end/83189
29755 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
29757 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
29759 PR middle-end/83718
29760 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
29761 after they are computed.
29763 2018-01-11 Bin Cheng <bin.cheng@arm.com>
29765 PR tree-optimization/83695
29766 * gimple-loop-linterchange.cc
29767 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
29768 reset cached scev information after interchange.
29769 (pass_linterchange::execute): Remove call to scev_reset_htab.
29771 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29773 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
29774 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
29775 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
29776 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
29777 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
29778 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
29779 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
29780 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
29781 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
29782 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
29783 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
29784 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
29785 (V_lane_reg): Likewise.
29786 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
29788 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
29789 (vfmal_lane_low<mode>_intrinsic,
29790 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
29791 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
29792 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
29793 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
29794 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
29795 vfmsl_lane_high<mode>_intrinsic): New define_insns.
29797 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29799 * config/arm/arm-cpus.in (fp16fml): New feature.
29800 (ALL_SIMD): Add fp16fml.
29801 (armv8.2-a): Add fp16fml as an option.
29802 (armv8.3-a): Likewise.
29803 (armv8.4-a): Add fp16fml as part of fp16.
29804 * config/arm/arm.h (TARGET_FP16FML): Define.
29805 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
29807 * config/arm/arm-modes.def (V2HF): Define.
29808 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
29809 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
29810 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
29811 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
29812 vfmsl_low, vfmsl_high): New set of builtins.
29813 * config/arm/iterators.md (PLUSMINUS): New code iterator.
29814 (vfml_op): New code attribute.
29815 (VFMLHALVES): New int iterator.
29816 (VFML, VFMLSEL): New mode attributes.
29817 (V_reg): Define mapping for V2HF.
29818 (V_hi, V_lo): New mode attributes.
29819 (VF_constraint): Likewise.
29820 (vfml_half, vfml_half_selector): New int attributes.
29821 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
29823 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
29824 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
29826 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
29827 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
29828 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
29829 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
29831 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
29832 Document new effective target and option set.
29834 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29836 * config/arm/arm-cpus.in (armv8_4): New feature.
29837 (ARMv8_4a): New fgroup.
29838 (armv8.4-a): New arch.
29839 * config/arm/arm-tables.opt: Regenerate.
29840 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
29841 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
29842 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
29843 Add matching rules for -march=armv8.4-a and extensions.
29844 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
29846 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
29849 * config/rx/rx.md (BW): New mode attribute.
29850 (sync_lock_test_and_setsi): Add mode suffix to insn output.
29852 2018-01-11 Richard Biener <rguenther@suse.de>
29854 PR tree-optimization/83435
29855 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
29856 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
29857 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
29859 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29860 Alan Hayward <alan.hayward@arm.com>
29861 David Sherwood <david.sherwood@arm.com>
29863 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
29865 (aarch64_classify_address): Initialize it. Track polynomial offsets.
29866 (aarch64_print_address_internal): Use it to check for a zero offset.
29868 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29869 Alan Hayward <alan.hayward@arm.com>
29870 David Sherwood <david.sherwood@arm.com>
29872 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
29873 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
29874 Return a poly_int64 rather than a HOST_WIDE_INT.
29875 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
29876 rather than a HOST_WIDE_INT.
29877 * config/aarch64/aarch64.h (aarch64_frame): Protect with
29878 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
29879 hard_fp_offset, frame_size, initial_adjust, callee_offset and
29880 final_offset from HOST_WIDE_INT to poly_int64.
29881 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
29882 to_constant when getting the number of units in an Advanced SIMD
29884 (aarch64_builtin_vectorized_function): Check for a constant number
29886 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
29888 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
29889 attribute instead of GET_MODE_NUNITS.
29890 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
29891 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
29892 GET_MODE_SIZE for fixed-size registers.
29893 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
29894 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
29895 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
29896 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
29897 (aarch64_print_operand, aarch64_print_address_internal)
29898 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
29899 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
29900 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
29901 Handle polynomial GET_MODE_SIZE.
29902 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
29903 wider than SImode without modification.
29904 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
29905 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
29906 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
29907 passing and returning SVE modes.
29908 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
29909 rather than GEN_INT.
29910 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
29911 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
29912 (aarch64_allocate_and_probe_stack_space): Likewise.
29913 (aarch64_layout_frame): Cope with polynomial offsets.
29914 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
29915 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
29916 polynomial offsets.
29917 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
29918 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
29919 poly_int64 rather than a HOST_WIDE_INT.
29920 (aarch64_get_separate_components, aarch64_process_components)
29921 (aarch64_expand_prologue, aarch64_expand_epilogue)
29922 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
29923 (aarch64_anchor_offset): New function, split out from...
29924 (aarch64_legitimize_address): ...here.
29925 (aarch64_builtin_vectorization_cost): Handle polynomial
29926 TYPE_VECTOR_SUBPARTS.
29927 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
29929 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
29930 number of elements from the PARALLEL rather than the mode.
29931 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
29932 rather than GET_MODE_BITSIZE.
29933 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
29934 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
29935 (aarch64_expand_vec_perm_const_1): Handle polynomial
29936 d->perm.length () and d->perm elements.
29937 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
29938 Apply to_constant to d->perm elements.
29939 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
29940 polynomial CONST_VECTOR_NUNITS.
29941 (aarch64_move_pointer): Take amount as a poly_int64 rather
29943 (aarch64_progress_pointer): Avoid temporary variable.
29944 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
29945 the mode attribute instead of GET_MODE.
29947 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29948 Alan Hayward <alan.hayward@arm.com>
29949 David Sherwood <david.sherwood@arm.com>
29951 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
29952 x exists before using it.
29953 (aarch64_add_constant_internal): Rename to...
29954 (aarch64_add_offset_1): ...this. Replace regnum with separate
29955 src and dest rtxes. Handle the case in which they're different,
29956 including when the offset is zero. Replace scratchreg with an rtx.
29957 Use 2 additions if there is no spare register into which we can
29958 move a 16-bit constant.
29959 (aarch64_add_constant): Delete.
29960 (aarch64_add_offset): Replace reg with separate src and dest
29961 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
29962 Use aarch64_add_offset_1.
29963 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
29964 an rtx rather than an int. Take the delta as a poly_int64
29965 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
29966 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
29967 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
29968 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
29969 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
29970 and aarch64_add_sp.
29971 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
29972 aarch64_add_constant.
29974 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29976 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
29977 Use scalar_float_mode.
29979 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
29981 * config/aarch64/aarch64-simd.md
29982 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
29983 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
29984 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
29985 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
29986 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
29987 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
29988 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
29989 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
29990 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
29991 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
29993 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29996 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
29997 targ_options->x_arm_arch_string is non NULL.
29999 2018-01-11 Tamar Christina <tamar.christina@arm.com>
30001 * config/aarch64/aarch64.h
30002 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
30004 2018-01-11 Sudakshina Das <sudi.das@arm.com>
30007 * expmed.c (emit_store_flag_force): Swap if const op0
30008 and change VOIDmode to mode of op0.
30010 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
30012 PR rtl-optimization/83761
30013 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
30014 than bytes to mode_for_size.
30016 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30018 PR middle-end/83189
30019 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
30020 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
30023 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30025 PR middle-end/83575
30026 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
30027 when in layout mode.
30028 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
30029 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
30032 2018-01-10 Michael Collison <michael.collison@arm.com>
30034 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
30035 * config/aarch64/aarch64-option-extension.def: Add
30036 AARCH64_OPT_EXTENSION of 'fp16fml'.
30037 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30038 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
30039 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
30040 * config/aarch64/constraints.md (Ui7): New constraint.
30041 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
30042 (VFMLA_SEL_W): Ditto.
30045 (VFMLA16_LOW): New int iterator.
30046 (VFMLA16_HIGH): Ditto.
30047 (UNSPEC_FMLAL): New unspec.
30048 (UNSPEC_FMLSL): Ditto.
30049 (UNSPEC_FMLAL2): Ditto.
30050 (UNSPEC_FMLSL2): Ditto.
30051 (f16mac): New code attribute.
30052 * config/aarch64/aarch64-simd-builtins.def
30053 (aarch64_fmlal_lowv2sf): Ditto.
30054 (aarch64_fmlsl_lowv2sf): Ditto.
30055 (aarch64_fmlalq_lowv4sf): Ditto.
30056 (aarch64_fmlslq_lowv4sf): Ditto.
30057 (aarch64_fmlal_highv2sf): Ditto.
30058 (aarch64_fmlsl_highv2sf): Ditto.
30059 (aarch64_fmlalq_highv4sf): Ditto.
30060 (aarch64_fmlslq_highv4sf): Ditto.
30061 (aarch64_fmlal_lane_lowv2sf): Ditto.
30062 (aarch64_fmlsl_lane_lowv2sf): Ditto.
30063 (aarch64_fmlal_laneq_lowv2sf): Ditto.
30064 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
30065 (aarch64_fmlalq_lane_lowv4sf): Ditto.
30066 (aarch64_fmlsl_lane_lowv4sf): Ditto.
30067 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
30068 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
30069 (aarch64_fmlal_lane_highv2sf): Ditto.
30070 (aarch64_fmlsl_lane_highv2sf): Ditto.
30071 (aarch64_fmlal_laneq_highv2sf): Ditto.
30072 (aarch64_fmlsl_laneq_highv2sf): Ditto.
30073 (aarch64_fmlalq_lane_highv4sf): Ditto.
30074 (aarch64_fmlsl_lane_highv4sf): Ditto.
30075 (aarch64_fmlalq_laneq_highv4sf): Ditto.
30076 (aarch64_fmlsl_laneq_highv4sf): Ditto.
30077 * config/aarch64/aarch64-simd.md:
30078 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
30079 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
30080 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
30081 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
30082 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
30083 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
30084 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
30085 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
30086 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
30087 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
30088 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
30089 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
30090 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
30091 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
30092 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
30093 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
30094 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
30095 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
30096 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
30097 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
30098 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
30099 (vfmlsl_low_u32): Ditto.
30100 (vfmlalq_low_u32): Ditto.
30101 (vfmlslq_low_u32): Ditto.
30102 (vfmlal_high_u32): Ditto.
30103 (vfmlsl_high_u32): Ditto.
30104 (vfmlalq_high_u32): Ditto.
30105 (vfmlslq_high_u32): Ditto.
30106 (vfmlal_lane_low_u32): Ditto.
30107 (vfmlsl_lane_low_u32): Ditto.
30108 (vfmlal_laneq_low_u32): Ditto.
30109 (vfmlsl_laneq_low_u32): Ditto.
30110 (vfmlalq_lane_low_u32): Ditto.
30111 (vfmlslq_lane_low_u32): Ditto.
30112 (vfmlalq_laneq_low_u32): Ditto.
30113 (vfmlslq_laneq_low_u32): Ditto.
30114 (vfmlal_lane_high_u32): Ditto.
30115 (vfmlsl_lane_high_u32): Ditto.
30116 (vfmlal_laneq_high_u32): Ditto.
30117 (vfmlsl_laneq_high_u32): Ditto.
30118 (vfmlalq_lane_high_u32): Ditto.
30119 (vfmlslq_lane_high_u32): Ditto.
30120 (vfmlalq_laneq_high_u32): Ditto.
30121 (vfmlslq_laneq_high_u32): Ditto.
30122 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
30123 (AARCH64_FL_FOR_ARCH8_4): New.
30124 (AARCH64_ISA_F16FML): New ISA flag.
30125 (TARGET_F16FML): New feature flag for fp16fml.
30126 (doc/invoke.texi): Document new fp16fml option.
30128 2018-01-10 Michael Collison <michael.collison@arm.com>
30130 * config/aarch64/aarch64-builtins.c:
30131 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
30132 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30133 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
30134 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
30135 (AARCH64_ISA_SHA3): New ISA flag.
30136 (TARGET_SHA3): New feature flag for sha3.
30137 * config/aarch64/iterators.md (sha512_op): New int attribute.
30138 (CRYPTO_SHA512): New int iterator.
30139 (UNSPEC_SHA512H): New unspec.
30140 (UNSPEC_SHA512H2): Ditto.
30141 (UNSPEC_SHA512SU0): Ditto.
30142 (UNSPEC_SHA512SU1): Ditto.
30143 * config/aarch64/aarch64-simd-builtins.def
30144 (aarch64_crypto_sha512hqv2di): New builtin.
30145 (aarch64_crypto_sha512h2qv2di): Ditto.
30146 (aarch64_crypto_sha512su0qv2di): Ditto.
30147 (aarch64_crypto_sha512su1qv2di): Ditto.
30148 (aarch64_eor3qv8hi): Ditto.
30149 (aarch64_rax1qv2di): Ditto.
30150 (aarch64_xarqv2di): Ditto.
30151 (aarch64_bcaxqv8hi): Ditto.
30152 * config/aarch64/aarch64-simd.md:
30153 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
30154 (aarch64_crypto_sha512su0qv2di): Ditto.
30155 (aarch64_crypto_sha512su1qv2di): Ditto.
30156 (aarch64_eor3qv8hi): Ditto.
30157 (aarch64_rax1qv2di): Ditto.
30158 (aarch64_xarqv2di): Ditto.
30159 (aarch64_bcaxqv8hi): Ditto.
30160 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
30161 (vsha512h2q_u64): Ditto.
30162 (vsha512su0q_u64): Ditto.
30163 (vsha512su1q_u64): Ditto.
30164 (veor3q_u16): Ditto.
30165 (vrax1q_u64): Ditto.
30166 (vxarq_u64): Ditto.
30167 (vbcaxq_u16): Ditto.
30168 * config/arm/types.md (crypto_sha512): New type attribute.
30169 (crypto_sha3): Ditto.
30170 (doc/invoke.texi): Document new sha3 option.
30172 2018-01-10 Michael Collison <michael.collison@arm.com>
30174 * config/aarch64/aarch64-builtins.c:
30175 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
30176 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30177 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
30178 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
30179 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
30180 (AARCH64_ISA_SM4): New ISA flag.
30181 (TARGET_SM4): New feature flag for sm4.
30182 * config/aarch64/aarch64-simd-builtins.def
30183 (aarch64_sm3ss1qv4si): Ditto.
30184 (aarch64_sm3tt1aq4si): Ditto.
30185 (aarch64_sm3tt1bq4si): Ditto.
30186 (aarch64_sm3tt2aq4si): Ditto.
30187 (aarch64_sm3tt2bq4si): Ditto.
30188 (aarch64_sm3partw1qv4si): Ditto.
30189 (aarch64_sm3partw2qv4si): Ditto.
30190 (aarch64_sm4eqv4si): Ditto.
30191 (aarch64_sm4ekeyqv4si): Ditto.
30192 * config/aarch64/aarch64-simd.md:
30193 (aarch64_sm3ss1qv4si): Ditto.
30194 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
30195 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
30196 (aarch64_sm4eqv4si): Ditto.
30197 (aarch64_sm4ekeyqv4si): Ditto.
30198 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
30199 (sm3part_op): Ditto.
30200 (CRYPTO_SM3TT): Ditto.
30201 (CRYPTO_SM3PART): Ditto.
30202 (UNSPEC_SM3SS1): New unspec.
30203 (UNSPEC_SM3TT1A): Ditto.
30204 (UNSPEC_SM3TT1B): Ditto.
30205 (UNSPEC_SM3TT2A): Ditto.
30206 (UNSPEC_SM3TT2B): Ditto.
30207 (UNSPEC_SM3PARTW1): Ditto.
30208 (UNSPEC_SM3PARTW2): Ditto.
30209 (UNSPEC_SM4E): Ditto.
30210 (UNSPEC_SM4EKEY): Ditto.
30211 * config/aarch64/constraints.md (Ui2): New constraint.
30212 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
30213 * config/arm/types.md (crypto_sm3): New type attribute.
30214 (crypto_sm4): Ditto.
30215 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
30216 (vsm3tt1aq_u32): Ditto.
30217 (vsm3tt1bq_u32): Ditto.
30218 (vsm3tt2aq_u32): Ditto.
30219 (vsm3tt2bq_u32): Ditto.
30220 (vsm3partw1q_u32): Ditto.
30221 (vsm3partw2q_u32): Ditto.
30222 (vsm4eq_u32): Ditto.
30223 (vsm4ekeyq_u32): Ditto.
30224 (doc/invoke.texi): Document new sm4 option.
30226 2018-01-10 Michael Collison <michael.collison@arm.com>
30228 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
30229 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
30230 (AARCH64_FL_FOR_ARCH8_4): New.
30231 (AARCH64_FL_V8_4): New flag.
30232 (doc/invoke.texi): Document new armv8.4-a option.
30234 2018-01-10 Michael Collison <michael.collison@arm.com>
30236 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
30237 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
30238 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
30239 * config/aarch64/aarch64-option-extension.def: Add
30240 AARCH64_OPT_EXTENSION of 'sha2'.
30241 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
30242 (crypto): Disable sha2 and aes if crypto disabled.
30243 (crypto): Enable aes and sha2 if enabled.
30244 (simd): Disable sha2 and aes if simd disabled.
30245 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
30247 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
30248 (TARGET_SHA2): New feature flag for sha2.
30249 (TARGET_AES): New feature flag for aes.
30250 * config/aarch64/aarch64-simd.md:
30251 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
30252 conditional on TARGET_AES.
30253 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
30254 (aarch64_crypto_sha1hsi): Make pattern conditional
30256 (aarch64_crypto_sha1hv4si): Ditto.
30257 (aarch64_be_crypto_sha1hv4si): Ditto.
30258 (aarch64_crypto_sha1su1v4si): Ditto.
30259 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
30260 (aarch64_crypto_sha1su0v4si): Ditto.
30261 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
30262 (aarch64_crypto_sha256su0v4si): Ditto.
30263 (aarch64_crypto_sha256su1v4si): Ditto.
30264 (doc/invoke.texi): Document new aes and sha2 options.
30266 2018-01-10 Martin Sebor <msebor@redhat.com>
30268 PR tree-optimization/83781
30269 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
30272 2018-01-11 Martin Sebor <msebor@gmail.com>
30273 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30275 PR tree-optimization/83501
30276 PR tree-optimization/81703
30278 * tree-ssa-strlen.c (get_string_cst): Rename...
30279 (get_string_len): ...to this. Handle global constants.
30280 (handle_char_store): Adjust.
30282 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
30283 Jim Wilson <jimw@sifive.com>
30285 * config/riscv/riscv-protos.h (riscv_output_return): New.
30286 * config/riscv/riscv.c (struct machine_function): New naked_p field.
30287 (riscv_attribute_table, riscv_output_return),
30288 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
30289 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
30290 (riscv_compute_frame_info): Only compute frame->mask if not a naked
30292 (riscv_expand_prologue): Add early return for naked function.
30293 (riscv_expand_epilogue): Likewise.
30294 (riscv_function_ok_for_sibcall): Return false for naked function.
30295 (riscv_set_current_function): New.
30296 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
30297 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
30298 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
30299 * doc/extend.texi (RISC-V Function Attributes): New.
30301 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
30303 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
30304 check for 128-bit long double before checking TCmode.
30305 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
30306 128-bit long doubles before checking TFmode or TCmode.
30307 (FLOAT128_IBM_P): Likewise.
30309 2018-01-10 Martin Sebor <msebor@redhat.com>
30311 PR tree-optimization/83671
30312 * builtins.c (c_strlen): Unconditionally return zero for the empty
30314 Use -Warray-bounds for warnings.
30315 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
30316 for non-constant array indices with COMPONENT_REF, arrays of
30317 arrays, and pointers to arrays.
30318 (gimple_fold_builtin_strlen): Determine and set length range for
30319 non-constant character arrays.
30321 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
30323 PR middle-end/81897
30324 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
30327 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
30329 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
30331 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
30334 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
30335 VECTOR_MEM_ALTIVEC_OR_VSX_P.
30336 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
30337 indexed_or_indirect_operand predicate.
30338 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
30339 (*vsx_le_perm_load_v8hi): Likewise.
30340 (*vsx_le_perm_load_v16qi): Likewise.
30341 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
30342 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
30343 (*vsx_le_perm_store_v8hi): Likewise.
30344 (*vsx_le_perm_store_v16qi): Likewise.
30345 (eight unnamed splitters): Likewise.
30347 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
30349 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
30350 * config/rs6000/emmintrin.h: Likewise.
30351 * config/rs6000/mmintrin.h: Likewise.
30352 * config/rs6000/xmmintrin.h: Likewise.
30354 2018-01-10 David Malcolm <dmalcolm@redhat.com>
30357 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
30359 * tree.c (tree_nop_conversion): Return true for location wrapper
30361 (maybe_wrap_with_location): New function.
30362 (selftest::check_strip_nops): New function.
30363 (selftest::test_location_wrappers): New function.
30364 (selftest::tree_c_tests): Call it.
30365 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
30366 (maybe_wrap_with_location): New decl.
30367 (EXPR_LOCATION_WRAPPER_P): New macro.
30368 (location_wrapper_p): New inline function.
30369 (tree_strip_any_location_wrapper): New inline function.
30371 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
30374 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
30375 stack_realign_offset for the largest alignment of stack slot
30377 (ix86_find_max_used_stack_alignment): New function.
30378 (ix86_finalize_stack_frame_flags): Use it. Set
30379 max_used_stack_alignment if we don't realign stack.
30380 * config/i386/i386.h (machine_function): Add
30381 max_used_stack_alignment.
30383 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
30385 * config/arm/arm.opt (-mbranch-cost): New option.
30386 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
30389 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
30392 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
30393 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
30395 2018-01-10 Richard Biener <rguenther@suse.de>
30398 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
30399 early out so it also covers the case where we have a non-NULL
30402 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
30404 PR tree-optimization/83753
30405 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
30406 for non-strided grouped accesses if the number of elements is 1.
30408 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
30411 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
30412 * i386.h (TARGET_USE_GATHER): Define.
30413 * x86-tune.def (X86_TUNE_USE_GATHER): New.
30415 2018-01-10 Martin Liska <mliska@suse.cz>
30418 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
30419 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
30421 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
30422 CLEANUP_NO_PARTITIONING is not set.
30424 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
30426 * doc/rtl.texi: Remove documentation of (const ...) wrappers
30427 for vectors, as a partial revert of r254296.
30428 * rtl.h (const_vec_p): Delete.
30429 (const_vec_duplicate_p): Don't test for vector CONSTs.
30430 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
30431 * expmed.c (make_tree): Likewise.
30434 * common.md (E, F): Use CONSTANT_P instead of checking for
30436 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
30437 checking for CONST_VECTOR.
30439 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30441 PR middle-end/83575
30442 * predict.c (force_edge_cold): Handle in more sane way edges
30443 with no prediction.
30445 2018-01-09 Carl Love <cel@us.ibm.com>
30447 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
30449 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
30450 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
30451 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
30452 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
30453 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
30454 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
30455 * config/rs6000/rs6000-protos.h: Add extern defition for
30456 rs6000_generate_float2_double_code.
30457 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
30459 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
30460 (float2_v2df): Add define_expand.
30462 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
30465 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
30466 op_mode in the force_to_mode call.
30468 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30470 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
30471 instead of checking each element individually.
30472 (aarch64_evpc_uzp): Likewise.
30473 (aarch64_evpc_zip): Likewise.
30474 (aarch64_evpc_ext): Likewise.
30475 (aarch64_evpc_rev): Likewise.
30476 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
30477 instead of checking each element individually. Return true without
30479 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
30480 whether all selected elements come from the same input, instead of
30481 checking each element individually. Remove calls to gen_rtx_REG,
30482 start_sequence and end_sequence and instead assert that no rtl is
30485 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30487 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
30488 order of HIGH and CONST checks.
30490 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
30492 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
30493 if the destination isn't an SSA_NAME.
30495 2018-01-09 Richard Biener <rguenther@suse.de>
30497 PR tree-optimization/83668
30498 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
30500 (canonicalize_loop_form): ... here, renamed from ...
30501 (canonicalize_loop_closed_ssa_form): ... this and amended to
30502 swap successor edges for loop exit blocks to make us use
30503 the RPO order we need for initial schedule generation.
30505 2018-01-09 Joseph Myers <joseph@codesourcery.com>
30507 PR tree-optimization/64811
30508 * match.pd: When optimizing comparisons with Inf, avoid
30509 introducing or losing exceptions from comparisons with NaN.
30511 2018-01-09 Martin Liska <mliska@suse.cz>
30514 * asan.c (shadow_mem_size): Add gcc_assert.
30516 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
30518 Don't save registers in main().
30521 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
30522 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
30523 * config/avr/avr.c (avr_set_current_function): Don't error if
30524 naked, OS_task or OS_main are specified at the same time.
30525 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
30527 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
30529 * common/config/avr/avr-common.c (avr_option_optimization_table):
30530 Switch on -mmain-is-OS_task for optimizing compilations.
30532 2018-01-09 Richard Biener <rguenther@suse.de>
30534 PR tree-optimization/83572
30535 * graphite.c: Include cfganal.h.
30536 (graphite_transform_loops): Connect infinite loops to exit
30537 and remove fake edges at the end.
30539 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30541 * ipa-inline.c (edge_badness): Revert accidental checkin.
30543 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
30546 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
30547 symbols; not inline clones.
30549 2018-01-09 Jakub Jelinek <jakub@redhat.com>
30552 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
30553 hard registers. Formatting fixes.
30555 PR preprocessor/83722
30556 * gcc.c (try_generate_repro): Pass
30557 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
30558 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
30561 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
30562 Kito Cheng <kito.cheng@gmail.com>
30564 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
30565 (riscv_leaf_function_p): Delete.
30566 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
30568 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30570 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
30572 (do_ifelse): New function.
30573 (do_isel): New function.
30574 (do_sub3): New function.
30575 (do_add3): New function.
30576 (do_load_mask_compare): New function.
30577 (do_overlap_load_compare): New function.
30578 (expand_compare_loop): New function.
30579 (expand_block_compare): Call expand_compare_loop() when appropriate.
30580 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
30581 option description.
30582 (-mblock-compare-inline-loop-limit): New option.
30584 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30587 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
30588 Reverse order of second and third operands in first alternative.
30589 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
30590 of first and second elements in UNSPEC_VPERMR vector.
30591 (altivec_expand_vec_perm_le): Likewise.
30593 2018-01-08 Jeff Law <law@redhat.com>
30595 PR rtl-optimizatin/81308
30596 * tree-switch-conversion.c (cfg_altered): New file scoped static.
30597 (process_switch): If group_case_labels makes a change, then set
30599 (pass_convert_switch::execute): If a switch is converted, then
30600 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
30602 PR rtl-optimization/81308
30603 * recog.c (split_all_insns): Conditionally cleanup the CFG after
30606 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
30608 PR target/83663 - Revert r255946
30609 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
30610 generation for cases where splatting a value is not useful.
30611 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
30612 across a vec_duplicate and a paradoxical subreg forming a vector
30613 mode to a vec_concat.
30615 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30617 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
30618 -march=armv8.3-a variants.
30619 * config/arm/t-multilib: Likewise.
30620 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
30622 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30624 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
30626 (cceq_ior_compare_complement): Give it a name so I can use it, and
30627 change boolean_or_operator predicate to boolean_operator so it can
30628 be used to generate a crand.
30629 (eqne): New code iterator.
30630 (bd/bd_neg): New code_attrs.
30631 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
30632 a single define_insn.
30633 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
30634 decrement (bdnzt/bdnzf/bdzt/bdzf).
30635 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
30636 with the new names of the branch decrement patterns, and added the
30637 names of the branch decrement conditional patterns.
30639 2018-01-08 Richard Biener <rguenther@suse.de>
30641 PR tree-optimization/83563
30642 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
30645 2018-01-08 Richard Biener <rguenther@suse.de>
30647 PR middle-end/83713
30648 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
30650 2018-01-08 Richard Biener <rguenther@suse.de>
30652 PR tree-optimization/83685
30653 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
30654 references to abnormals.
30656 2018-01-08 Richard Biener <rguenther@suse.de>
30659 * dwarf2out.c (output_indirect_strings): Handle empty
30660 skeleton_debug_str_hash.
30661 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
30663 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
30665 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
30666 (emit_store_direct): Likewise.
30667 (arc_trampoline_adjust_address): Likewise.
30668 (arc_asm_trampoline_template): New function.
30669 (arc_initialize_trampoline): Use asm_trampoline_template.
30670 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
30671 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
30672 * config/arc/arc.md (flush_icache): Delete pattern.
30674 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
30676 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
30677 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
30680 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
30683 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
30684 by not USED_FOR_TARGET.
30685 (make_pass_resolve_sw_modes): Likewise.
30687 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
30689 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
30692 2018-01-08 Richard Biener <rguenther@suse.de>
30694 PR middle-end/83580
30695 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
30697 2018-01-08 Richard Biener <rguenther@suse.de>
30699 PR middle-end/83517
30700 * match.pd ((t * 2) / 2) -> t): Add missing :c.
30702 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
30704 PR middle-end/81897
30705 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
30706 basic blocks with a small number of successors.
30707 (convert_control_dep_chain_into_preds): Improve handling of
30709 (dump_predicates): Split apart into...
30710 (dump_pred_chain): ...here...
30711 (dump_pred_info): ...and here.
30712 (can_one_predicate_be_invalidated_p): Add debugging printfs.
30713 (can_chain_union_be_invalidated_p): Improve check for invalidation
30715 (uninit_uses_cannot_happen): Avoid unnecessary if
30716 convert_control_dep_chain_into_preds yielded nothing.
30718 2018-01-06 Martin Sebor <msebor@redhat.com>
30720 PR tree-optimization/83640
30721 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
30722 subtracting negative offset from size.
30723 (builtin_access::overlap): Adjust offset bounds of the access to fall
30724 within the size of the object if possible.
30726 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
30728 PR rtl-optimization/83699
30729 * expmed.c (extract_bit_field_1): Restrict the vector usage of
30730 extract_bit_field_as_subreg to cases in which the extracted
30731 value is also a vector.
30733 * lra-constraints.c (process_alt_operands): Test for the equivalence
30734 substitutions when detecting a possible reload cycle.
30736 2018-01-06 Jakub Jelinek <jakub@redhat.com>
30739 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
30740 by default if flag_selective_schedling{,2}. Formatting fixes.
30742 PR rtl-optimization/83682
30743 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
30744 if it has non-VECTOR_MODE element mode.
30745 (vec_duplicate_p): Likewise.
30747 PR middle-end/83694
30748 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
30749 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
30751 2018-01-05 Jakub Jelinek <jakub@redhat.com>
30754 * config/i386/i386-builtin.def
30755 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
30756 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
30757 Require also OPTION_MASK_ISA_AVX512F in addition to
30758 OPTION_MASK_ISA_GFNI.
30759 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
30760 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
30761 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
30762 to OPTION_MASK_ISA_GFNI.
30763 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
30764 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
30765 OPTION_MASK_ISA_AVX512BW.
30766 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
30767 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
30768 addition to OPTION_MASK_ISA_GFNI.
30769 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
30770 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
30771 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
30772 to OPTION_MASK_ISA_GFNI.
30773 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
30774 a requirement for all ISAs rather than any of them with a few
30776 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
30778 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
30779 bitmasks to be enabled with 3 exceptions, instead of requiring any
30780 enabled ISA with lots of exceptions.
30781 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
30782 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
30783 Change avx512bw in isa attribute to avx512f.
30784 * config/i386/sgxintrin.h: Add license boilerplate.
30785 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
30786 to __AVX512F__ and __AVX512VL to __AVX512VL__.
30787 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
30788 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
30790 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
30791 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
30792 temporarily sse2 rather than sse if not enabled already.
30795 * config/i386/sse.md (VI248_VLBW): Rename to ...
30796 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
30797 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
30798 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
30799 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
30800 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
30801 mode iterator instead of VI248_VLBW.
30803 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
30805 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
30806 (record_modified): Skip clobbers; add debug output.
30807 (param_change_prob): Use sreal frequencies.
30809 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
30811 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
30812 punt for user-aligned variables.
30814 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
30816 * tree-chrec.c (chrec_contains_symbols): Return true for
30819 2018-01-05 Sudakshina Das <sudi.das@arm.com>
30822 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
30823 of (x|y) == x for BICS pattern.
30825 2018-01-05 Jakub Jelinek <jakub@redhat.com>
30827 PR tree-optimization/83605
30828 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
30829 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
30832 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
30834 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
30835 * config/epiphany/rtems.h: New file.
30837 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30838 Uros Bizjak <ubizjak@gmail.com>
30841 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
30842 QIreg_operand instead of register_operand predicate.
30843 * config/i386/i386.c (ix86_rop_should_change_byte_p,
30844 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
30845 comments instead of -fmitigate[-_]rop.
30847 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30850 * cgraphunit.c (symbol_table::compile): Switch to text_section
30851 before calling assembly_start debug hook.
30852 * run-rtl-passes.c (run_rtl_passes): Likewise.
30855 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30857 * tree-vrp.c (extract_range_from_binary_expr_1): Check
30858 range_int_cst_p rather than !symbolic_range_p before calling
30859 extract_range_from_multiplicative_op_1.
30861 2018-01-04 Jeff Law <law@redhat.com>
30863 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
30864 redundant test in assertion.
30866 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30868 * doc/rtl.texi: Document machine_mode wrapper classes.
30870 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30872 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
30873 using tree_to_uhwi.
30875 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30877 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
30878 the VEC_PERM_EXPR fold to fail.
30880 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30883 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
30884 to switched_sections.
30886 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30889 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
30890 test for d.testing.
30892 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
30895 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
30896 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
30898 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30901 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
30902 is BLKmode and bitpos not zero or mode change is needed.
30904 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
30907 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
30910 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
30913 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
30914 instead of MULT rtx. Update all corresponding splitters.
30915 (*saddl_se): Ditto.
30916 (*ssub<modesuffix>): Ditto.
30917 (*ssubl_se): Ditto.
30918 (*cmp_sadd_di): Update split patterns.
30919 (*cmp_sadd_si): Ditto.
30920 (*cmp_sadd_sidi): Ditto.
30921 (*cmp_ssub_di): Ditto.
30922 (*cmp_ssub_si): Ditto.
30923 (*cmp_ssub_sidi): Ditto.
30924 * config/alpha/predicates.md (const23_operand): New predicate.
30925 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
30926 Look for ASHIFT, not MULT inner operand.
30927 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
30929 2018-01-04 Martin Liska <mliska@suse.cz>
30931 PR gcov-profile/83669
30932 * gcov.c (output_intermediate_file): Add version to intermediate
30934 * doc/gcov.texi: Document new field 'version' in intermediate
30935 file format. Fix location of '-k' option of gcov command.
30937 2018-01-04 Martin Liska <mliska@suse.cz>
30940 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
30942 2018-01-04 Jakub Jelinek <jakub@redhat.com>
30944 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
30946 2018-01-03 Martin Sebor <msebor@redhat.com>
30948 PR tree-optimization/83655
30949 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
30950 checking calls with invalid arguments.
30952 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30954 * tree-vect-stmts.c (vect_get_store_rhs): New function.
30955 (vectorizable_mask_load_store): Delete.
30956 (vectorizable_call): Return false for masked loads and stores.
30957 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
30958 instead of gimple_assign_rhs1.
30959 (vectorizable_load): Handle IFN_MASK_LOAD.
30960 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
30962 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30964 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
30966 (vectorizable_mask_load_store): ...here.
30967 (vectorizable_load): ...and here.
30969 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30971 * tree-vect-stmts.c (vect_build_all_ones_mask)
30972 (vect_build_zero_merge_argument): New functions, split out from...
30973 (vectorizable_load): ...here.
30975 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30977 * tree-vect-stmts.c (vect_check_store_rhs): New function,
30979 (vectorizable_mask_load_store): ...here.
30980 (vectorizable_store): ...and here.
30982 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30984 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
30986 (vectorizable_mask_load_store): ...here.
30988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30990 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
30991 (vect_model_store_cost): Take a vec_load_store_type instead of a
30993 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
30994 (vect_model_store_cost): Take a vec_load_store_type instead of a
30996 (vectorizable_mask_load_store): Update accordingly.
30997 (vectorizable_store): Likewise.
30998 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
31000 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31002 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
31003 IFN_MASK_LOAD calls here rather than...
31004 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
31006 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31007 Alan Hayward <alan.hayward@arm.com>
31008 David Sherwood <david.sherwood@arm.com>
31010 * expmed.c (extract_bit_field_1): For vector extracts,
31011 fall back to extract_bit_field_as_subreg if vec_extract
31014 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31015 Alan Hayward <alan.hayward@arm.com>
31016 David Sherwood <david.sherwood@arm.com>
31018 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
31019 they are variable or constant sized.
31020 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
31021 slots for constant-sized data.
31023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31024 Alan Hayward <alan.hayward@arm.com>
31025 David Sherwood <david.sherwood@arm.com>
31027 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
31028 handling COND_EXPRs with boolean comparisons, try to find a better
31029 basis for the mask type than the boolean itself.
31031 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31033 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
31034 is calculated and how it can be overridden.
31035 * genmodes.c (max_bitsize_mode_any_mode): New variable.
31036 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
31038 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
31041 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31042 Alan Hayward <alan.hayward@arm.com>
31043 David Sherwood <david.sherwood@arm.com>
31045 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
31046 Remove the mode argument.
31047 (aarch64_simd_valid_immediate): Remove the mode and inverse
31049 * config/aarch64/iterators.md (bitsize): New iterator.
31050 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
31051 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
31052 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
31053 aarch64_simd_valid_immediate.
31054 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
31055 (aarch64_reg_or_bic_imm): Likewise.
31056 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
31057 with an insn_type enum and msl with a modifier_type enum.
31058 Replace element_width with a scalar_mode. Change the shift
31059 to unsigned int. Add constructors for scalar_float_mode and
31060 scalar_int_mode elements.
31061 (aarch64_vect_float_const_representable_p): Delete.
31062 (aarch64_can_const_movi_rtx_p)
31063 (aarch64_simd_scalar_immediate_valid_for_move)
31064 (aarch64_simd_make_constant): Update call to
31065 aarch64_simd_valid_immediate.
31066 (aarch64_advsimd_valid_immediate_hs): New function.
31067 (aarch64_advsimd_valid_immediate): Likewise.
31068 (aarch64_simd_valid_immediate): Remove mode and inverse
31069 arguments. Rewrite to use the above. Use const_vec_duplicate_p
31070 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
31071 and aarch64_float_const_representable_p on the result.
31072 (aarch64_output_simd_mov_immediate): Remove mode argument.
31073 Update call to aarch64_simd_valid_immediate and use of
31074 simd_immediate_info.
31075 (aarch64_output_scalar_simd_mov_immediate): Update call
31078 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31079 Alan Hayward <alan.hayward@arm.com>
31080 David Sherwood <david.sherwood@arm.com>
31082 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
31083 (mode_nunits): Likewise CONST_MODE_NUNITS.
31084 * machmode.def (ADJUST_NUNITS): Document.
31085 * genmodes.c (mode_data::need_nunits_adj): New field.
31086 (blank_mode): Update accordingly.
31087 (adj_nunits): New variable.
31088 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
31090 (emit_mode_size_inline): Set need_bytesize_adj for all modes
31091 listed in adj_nunits.
31092 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
31093 listed in adj_nunits. Don't emit case statements for such modes.
31094 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
31095 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
31096 nothing if adj_nunits is nonnull.
31097 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
31098 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
31099 (emit_mode_fbit): Update use of print_maybe_const_decl.
31100 (emit_move_size): Likewise. Treat the array as non-const
31102 (emit_mode_adjustments): Handle adj_nunits.
31104 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31106 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
31107 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
31108 (VECTOR_MODES): Use it.
31109 (make_vector_modes): Take the prefix as an argument.
31111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31112 Alan Hayward <alan.hayward@arm.com>
31113 David Sherwood <david.sherwood@arm.com>
31115 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
31116 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
31117 for MODE_VECTOR_BOOL.
31118 * machmode.def (VECTOR_BOOL_MODE): Document.
31119 * genmodes.c (VECTOR_BOOL_MODE): New macro.
31120 (make_vector_bool_mode): New function.
31121 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
31123 * lto-streamer-in.c (lto_input_mode_table): Likewise.
31124 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
31126 * stor-layout.c (int_mode_for_mode): Likewise.
31127 * tree.c (build_vector_type_for_mode): Likewise.
31128 * varasm.c (output_constant_pool_2): Likewise.
31129 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
31130 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
31131 for MODE_VECTOR_BOOL.
31132 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
31133 of mode class checks.
31134 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
31135 instead of a list of mode class checks.
31136 (expand_vector_scalar_condition): Likewise.
31137 (type_for_widest_vector_mode): Handle BImode as an inner mode.
31139 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31140 Alan Hayward <alan.hayward@arm.com>
31141 David Sherwood <david.sherwood@arm.com>
31143 * machmode.h (mode_size): Change from unsigned short to
31145 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
31146 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
31147 or if measurement_type is not polynomial.
31148 (fixed_size_mode::includes_p): Check for constant-sized modes.
31149 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
31150 return a poly_uint16 rather than an unsigned short.
31151 (emit_mode_size): Change the type of mode_size from unsigned short
31152 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
31153 (emit_mode_adjustments): Cope with polynomial vector sizes.
31154 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31156 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31158 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
31159 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
31160 * caller-save.c (setup_save_areas): Likewise.
31161 (replace_reg_with_saved_mem): Likewise.
31162 * calls.c (emit_library_call_value_1): Likewise.
31163 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
31164 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
31165 (gen_lowpart_for_combine): Likewise.
31166 * convert.c (convert_to_integer_1): Likewise.
31167 * cse.c (equiv_constant, cse_insn): Likewise.
31168 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
31169 (cselib_subst_to_values): Likewise.
31170 * dce.c (word_dce_process_block): Likewise.
31171 * df-problems.c (df_word_lr_mark_ref): Likewise.
31172 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
31173 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
31174 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
31175 (rtl_for_decl_location): Likewise.
31176 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
31177 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
31178 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
31179 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
31180 (expand_expr_real_1): Likewise.
31181 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
31182 (pad_below): Likewise.
31183 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
31184 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
31185 * ira.c (get_subreg_tracking_sizes): Likewise.
31186 * ira-build.c (ira_create_allocno_objects): Likewise.
31187 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
31188 (ira_sort_regnos_for_alter_reg): Likewise.
31189 * ira-costs.c (record_operand_costs): Likewise.
31190 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
31191 (resolve_simple_move): Likewise.
31192 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
31193 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
31194 (lra_constraints): Likewise.
31195 (CONST_POOL_OK_P): Reject variable-sized modes.
31196 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
31197 (add_pseudo_to_slot, lra_spill): Likewise.
31198 * omp-low.c (omp_clause_aligned_alignment): Likewise.
31199 * optabs-query.c (get_best_extraction_insn): Likewise.
31200 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
31201 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
31202 (expand_mult_highpart, valid_multiword_target_p): Likewise.
31203 * recog.c (offsettable_address_addr_space_p): Likewise.
31204 * regcprop.c (maybe_mode_change): Likewise.
31205 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
31206 * regrename.c (build_def_use): Likewise.
31207 * regstat.c (dump_reg_info): Likewise.
31208 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
31209 (find_reloads, find_reloads_subreg_address): Likewise.
31210 * reload1.c (eliminate_regs_1): Likewise.
31211 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
31212 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
31213 (simplify_binary_operation_1, simplify_subreg): Likewise.
31214 * targhooks.c (default_function_arg_padding): Likewise.
31215 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
31216 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
31217 (verify_gimple_assign_ternary): Likewise.
31218 * tree-inline.c (estimate_move_cost): Likewise.
31219 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31220 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
31221 (get_address_cost_ainc): Likewise.
31222 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
31223 (vect_supportable_dr_alignment): Likewise.
31224 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31225 (vectorizable_reduction): Likewise.
31226 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
31227 (vectorizable_operation, vectorizable_load): Likewise.
31228 * tree.c (build_same_sized_truth_vector_type): Likewise.
31229 * valtrack.c (cleanup_auto_inc_dec): Likewise.
31230 * var-tracking.c (emit_note_insn_var_location): Likewise.
31231 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
31232 (ADDR_VEC_ALIGN): Likewise.
31234 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31235 Alan Hayward <alan.hayward@arm.com>
31236 David Sherwood <david.sherwood@arm.com>
31238 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
31240 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
31241 or if measurement_type is polynomial.
31242 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
31243 * combine.c (make_extraction): Likewise.
31244 * dse.c (find_shift_sequence): Likewise.
31245 * dwarf2out.c (mem_loc_descriptor): Likewise.
31246 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
31247 (extract_bit_field, extract_low_bits): Likewise.
31248 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
31249 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
31250 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
31251 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
31252 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
31253 * reload.c (find_reloads): Likewise.
31254 * reload1.c (alter_reg): Likewise.
31255 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
31256 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
31257 * tree-if-conv.c (predicate_mem_writes): Likewise.
31258 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
31259 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
31260 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
31261 * valtrack.c (dead_debug_insert_temp): Likewise.
31262 * varasm.c (mergeable_constant_section): Likewise.
31263 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
31265 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31266 Alan Hayward <alan.hayward@arm.com>
31267 David Sherwood <david.sherwood@arm.com>
31269 * expr.c (expand_assignment): Cope with polynomial mode sizes
31270 when assigning to a CONCAT.
31272 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31273 Alan Hayward <alan.hayward@arm.com>
31274 David Sherwood <david.sherwood@arm.com>
31276 * machmode.h (mode_precision): Change from unsigned short to
31278 (mode_to_precision): Return a poly_uint16 rather than an unsigned
31280 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
31281 or if measurement_type is not polynomial.
31282 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
31283 in which the mode is already known to be a scalar_int_mode.
31284 * genmodes.c (emit_mode_precision): Change the type of mode_precision
31285 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
31287 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31288 for GET_MODE_PRECISION.
31289 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31290 for GET_MODE_PRECISION.
31291 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
31293 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
31294 (expand_field_assignment, make_extraction): Likewise.
31295 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
31296 (get_last_value): Likewise.
31297 * convert.c (convert_to_integer_1): Likewise.
31298 * cse.c (cse_insn): Likewise.
31299 * expr.c (expand_expr_real_1): Likewise.
31300 * lra-constraints.c (simplify_operand_subreg): Likewise.
31301 * optabs-query.c (can_atomic_load_p): Likewise.
31302 * optabs.c (expand_atomic_load): Likewise.
31303 (expand_atomic_store): Likewise.
31304 * ree.c (combine_reaching_defs): Likewise.
31305 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
31306 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
31307 * tree.h (type_has_mode_precision_p): Likewise.
31308 * ubsan.c (instrument_si_overflow): Likewise.
31310 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31311 Alan Hayward <alan.hayward@arm.com>
31312 David Sherwood <david.sherwood@arm.com>
31314 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
31315 polynomial numbers of units.
31316 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
31317 (valid_vector_subparts_p): New function.
31318 (build_vector_type): Remove temporary shim and take the number
31319 of units as a poly_uint64 rather than an int.
31320 (build_opaque_vector_type): Take the number of units as a
31321 poly_uint64 rather than an int.
31322 * tree.c (build_vector_from_ctor): Handle polynomial
31323 TYPE_VECTOR_SUBPARTS.
31324 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
31325 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
31326 (build_vector_from_val): If the number of units is variable,
31327 use build_vec_duplicate_cst for constant operands and
31328 VEC_DUPLICATE_EXPR otherwise.
31329 (make_vector_type): Remove temporary is_constant ().
31330 (build_vector_type, build_opaque_vector_type): Take the number of
31331 units as a poly_uint64 rather than an int.
31332 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
31334 * cfgexpand.c (expand_debug_expr): Likewise.
31335 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
31336 (store_constructor, expand_expr_real_1): Likewise.
31337 (const_scalar_mask_from_tree): Likewise.
31338 * fold-const-call.c (fold_const_reduction): Likewise.
31339 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
31340 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
31341 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
31342 (fold_relational_const): Likewise.
31343 (native_interpret_vector): Likewise. Change the size from an
31344 int to an unsigned int.
31345 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
31346 TYPE_VECTOR_SUBPARTS.
31347 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
31348 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
31349 duplicating a non-constant operand into a variable-length vector.
31350 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
31351 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
31352 * ipa-icf.c (sem_variable::equals): Likewise.
31353 * match.pd: Likewise.
31354 * omp-simd-clone.c (simd_clone_subparts): Likewise.
31355 * print-tree.c (print_node): Likewise.
31356 * stor-layout.c (layout_type): Likewise.
31357 * targhooks.c (default_builtin_vectorization_cost): Likewise.
31358 * tree-cfg.c (verify_gimple_comparison): Likewise.
31359 (verify_gimple_assign_binary): Likewise.
31360 (verify_gimple_assign_ternary): Likewise.
31361 (verify_gimple_assign_single): Likewise.
31362 * tree-pretty-print.c (dump_generic_node): Likewise.
31363 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31364 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
31365 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
31366 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
31367 (vect_shift_permute_load_chain): Likewise.
31368 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
31369 (expand_vector_condition, optimize_vector_constructor): Likewise.
31370 (lower_vec_perm, get_compute_type): Likewise.
31371 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31372 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
31373 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
31374 (vect_recog_mask_conversion_pattern): Likewise.
31375 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
31376 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
31377 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31378 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
31379 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
31380 (vectorizable_shift, vectorizable_operation, vectorizable_store)
31381 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
31382 (supportable_widening_operation): Likewise.
31383 (supportable_narrowing_operation): Likewise.
31384 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
31386 * varasm.c (output_constant): Likewise.
31388 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31389 Alan Hayward <alan.hayward@arm.com>
31390 David Sherwood <david.sherwood@arm.com>
31392 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
31393 so that both the length == 3 and length != 3 cases set up their
31394 own permute vectors. Add comments explaining why we know the
31395 number of elements is constant.
31396 (vect_permute_load_chain): Likewise.
31398 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31399 Alan Hayward <alan.hayward@arm.com>
31400 David Sherwood <david.sherwood@arm.com>
31402 * machmode.h (mode_nunits): Change from unsigned char to
31404 (ONLY_FIXED_SIZE_MODES): New macro.
31405 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
31406 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
31407 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
31409 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
31410 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
31411 or if measurement_type is not polynomial.
31412 * genmodes.c (ZERO_COEFFS): New macro.
31413 (emit_mode_nunits_inline): Make mode_nunits_inline return a
31415 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
31416 Use ZERO_COEFFS when emitting initializers.
31417 * data-streamer.h (bp_pack_poly_value): New function.
31418 (bp_unpack_poly_value): Likewise.
31419 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
31420 for GET_MODE_NUNITS.
31421 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
31422 for GET_MODE_NUNITS.
31423 * tree.c (make_vector_type): Remove temporary shim and make
31424 the real function take the number of units as a poly_uint64
31425 rather than an int.
31426 (build_vector_type_for_mode): Handle polynomial nunits.
31427 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
31428 * emit-rtl.c (const_vec_series_p_1): Likewise.
31429 (gen_rtx_CONST_VECTOR): Likewise.
31430 * fold-const.c (test_vec_duplicate_folding): Likewise.
31431 * genrecog.c (validate_pattern): Likewise.
31432 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
31433 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
31434 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
31435 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
31436 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
31437 * rtlanal.c (subreg_get_info): Likewise.
31438 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31439 (vect_grouped_load_supported): Likewise.
31440 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
31441 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
31442 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
31443 (simplify_const_unary_operation, simplify_binary_operation_1)
31444 (simplify_const_binary_operation, simplify_ternary_operation)
31445 (test_vector_ops_duplicate, test_vector_ops): Likewise.
31446 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
31447 instead of CONST_VECTOR_NUNITS.
31448 * varasm.c (output_constant_pool_2): Likewise.
31449 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
31450 explicit-encoded elements in the XVEC for variable-length vectors.
31452 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31454 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
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 * coretypes.h (fixed_size_mode): Declare.
31461 (fixed_size_mode_pod): New typedef.
31462 * builtins.h (target_builtins::x_apply_args_mode)
31463 (target_builtins::x_apply_result_mode): Change type to
31464 fixed_size_mode_pod.
31465 * builtins.c (apply_args_size, apply_result_size, result_vector)
31466 (expand_builtin_apply_args_1, expand_builtin_apply)
31467 (expand_builtin_return): Update accordingly.
31469 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31471 * cse.c (hash_rtx_cb): Hash only the encoded elements.
31472 * cselib.c (cselib_hash_rtx): Likewise.
31473 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
31474 CONST_VECTOR encoding.
31476 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31477 Jeff Law <law@redhat.com>
31480 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
31481 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
31482 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
31483 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
31486 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
31487 explicitly probe *sp in a noreturn function if there were any callee
31488 register saves or frame pointer is needed.
31490 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31493 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
31494 BLKmode for ternary, binary or unary expressions.
31497 * var-tracking.c (delete_vta_debug_insn): New inline function.
31498 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
31499 insns from get_insns () to NULL instead of each bb separately.
31500 Use delete_vta_debug_insn. No longer static.
31501 (vt_debug_insns_local, variable_tracking_main_1): Adjust
31502 delete_vta_debug_insns callers.
31503 * rtl.h (delete_vta_debug_insns): Declare.
31504 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
31505 instead of variable_tracking_main.
31507 2018-01-03 Martin Sebor <msebor@redhat.com>
31509 PR tree-optimization/83603
31510 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
31511 arguments past the endof the argument list in functions declared
31512 without a prototype.
31513 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
31514 Avoid checking when arguments are null.
31516 2018-01-03 Martin Sebor <msebor@redhat.com>
31519 * doc/extend.texi (attribute const): Fix a typo.
31520 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
31521 issuing -Wsuggest-attribute for void functions.
31523 2018-01-03 Martin Sebor <msebor@redhat.com>
31525 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
31526 offset_int::from instead of wide_int::to_shwi.
31527 (maybe_diag_overlap): Remove assertion.
31528 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
31529 * gimple-ssa-sprintf.c (format_directive): Same.
31530 (parse_directive): Same.
31531 (sprintf_dom_walker::compute_format_length): Same.
31532 (try_substitute_return_value): Same.
31534 2018-01-03 Jeff Law <law@redhat.com>
31536 PR middle-end/83654
31537 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
31538 non-constant residual for zero at runtime and avoid probing in
31539 that case. Reorganize code for trailing problem to mirror handling
31542 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31544 PR tree-optimization/83501
31545 * tree-ssa-strlen.c (get_string_cst): New.
31546 (handle_char_store): Call get_string_cst.
31548 2018-01-03 Martin Liska <mliska@suse.cz>
31550 PR tree-optimization/83593
31551 * tree-ssa-strlen.c: Include tree-cfg.h.
31552 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
31553 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
31554 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
31556 (strlen_dom_walker::before_dom_children): Call
31557 gimple_purge_dead_eh_edges. Dump tranformation with details
31559 (strlen_dom_walker::before_dom_children): Update call by adding
31560 new argument cleanup_eh.
31561 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
31563 2018-01-03 Martin Liska <mliska@suse.cz>
31566 * cif-code.def (VARIADIC_THUNK): New enum value.
31567 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
31570 2018-01-03 Jan Beulich <jbeulich@suse.com>
31572 * sse.md (mov<mode>_internal): Tighten condition for when to use
31573 vmovdqu<ssescalarsize> for TI and OI modes.
31575 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31577 Update copyright years.
31579 2018-01-03 Martin Liska <mliska@suse.cz>
31582 * ipa-visibility.c (function_and_variable_visibility): Skip
31583 functions with noipa attribure.
31585 2018-01-03 Jakub Jelinek <jakub@redhat.com>
31587 * gcc.c (process_command): Update copyright notice dates.
31588 * gcov-dump.c (print_version): Ditto.
31589 * gcov.c (print_version): Ditto.
31590 * gcov-tool.c (print_version): Ditto.
31591 * gengtype.c (create_file): Ditto.
31592 * doc/cpp.texi: Bump @copying's copyright year.
31593 * doc/cppinternals.texi: Ditto.
31594 * doc/gcc.texi: Ditto.
31595 * doc/gccint.texi: Ditto.
31596 * doc/gcov.texi: Ditto.
31597 * doc/install.texi: Ditto.
31598 * doc/invoke.texi: Ditto.
31600 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31602 * vector-builder.h (vector_builder::m_full_nelts): Change from
31603 unsigned int to poly_uint64.
31604 (vector_builder::full_nelts): Update prototype accordingly.
31605 (vector_builder::new_vector): Likewise.
31606 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
31607 (vector_builder::operator ==): Likewise.
31608 (vector_builder::finalize): Likewise.
31609 * int-vector-builder.h (int_vector_builder::int_vector_builder):
31610 Take the number of elements as a poly_uint64 rather than an
31612 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
31613 from unsigned int to poly_uint64.
31614 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
31615 (vec_perm_indices::new_vector): Likewise.
31616 (vec_perm_indices::length): Likewise.
31617 (vec_perm_indices::nelts_per_input): Likewise.
31618 (vec_perm_indices::input_nelts): Likewise.
31619 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
31620 number of elements per input as a poly_uint64 rather than an
31621 unsigned int. Use the original encoding for variable-length
31622 vectors, rather than clamping each individual element.
31623 For the second and subsequent elements in each pattern,
31624 clamp the step and base before clamping their sum.
31625 (vec_perm_indices::series_p): Handle polynomial element counts.
31626 (vec_perm_indices::all_in_range_p): Likewise.
31627 (vec_perm_indices_to_tree): Likewise.
31628 (vec_perm_indices_to_rtx): Likewise.
31629 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
31630 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
31631 (tree_vector_builder::new_binary_operation): Handle polynomial
31632 element counts. Return false if we need to know the number
31633 of elements at compile time.
31634 * fold-const.c (fold_vec_perm): Punt if the number of elements
31635 isn't known at compile time.
31637 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31639 * vec-perm-indices.h (vec_perm_builder): Change element type
31640 from HOST_WIDE_INT to poly_int64.
31641 (vec_perm_indices::element_type): Update accordingly.
31642 (vec_perm_indices::clamp): Handle polynomial element_types.
31643 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
31644 (vec_perm_indices::all_in_range_p): Likewise.
31645 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
31647 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
31648 polynomial vec_perm_indices element types.
31649 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
31650 * fold-const.c (fold_vec_perm): Likewise.
31651 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
31652 * tree-vect-generic.c (lower_vec_perm): Likewise.
31653 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
31654 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
31655 element type to HOST_WIDE_INT.
31657 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31658 Alan Hayward <alan.hayward@arm.com>
31659 David Sherwood <david.sherwood@arm.com>
31661 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
31662 rather than an int. Use plus_constant.
31663 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
31664 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
31666 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31667 Alan Hayward <alan.hayward@arm.com>
31668 David Sherwood <david.sherwood@arm.com>
31670 * calls.c (emit_call_1, expand_call): Change struct_value_size from
31671 a HOST_WIDE_INT to a poly_int64.
31673 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31674 Alan Hayward <alan.hayward@arm.com>
31675 David Sherwood <david.sherwood@arm.com>
31677 * calls.c (load_register_parameters): Cope with polynomial
31678 mode sizes. Require a constant size for BLKmode parameters
31679 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
31680 forces a parameter to be padded at the lsb end in order to
31681 fill a complete number of words, require the parameter size
31682 to be ordered wrt UNITS_PER_WORD.
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 * reload1.c (spill_stack_slot_width): Change element type
31689 from unsigned int to poly_uint64_pod.
31690 (alter_reg): Treat mode sizes as polynomial.
31692 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31693 Alan Hayward <alan.hayward@arm.com>
31694 David Sherwood <david.sherwood@arm.com>
31696 * reload.c (complex_word_subreg_p): New function.
31697 (reload_inner_reg_of_subreg, push_reload): Use it.
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 * lra-constraints.c (process_alt_operands): Reject matched
31704 operands whose sizes aren't ordered.
31705 (match_reload): Refer to this check here.
31707 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31708 Alan Hayward <alan.hayward@arm.com>
31709 David Sherwood <david.sherwood@arm.com>
31711 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
31712 that the mode size is in the set {1, 2, 4, 8, 16}.
31714 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31715 Alan Hayward <alan.hayward@arm.com>
31716 David Sherwood <david.sherwood@arm.com>
31718 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
31719 Use plus_constant instead of gen_rtx_PLUS.
31721 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31722 Alan Hayward <alan.hayward@arm.com>
31723 David Sherwood <david.sherwood@arm.com>
31725 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
31726 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
31727 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
31728 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
31729 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
31730 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
31731 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
31732 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
31733 * config/i386/i386.c (ix86_push_rounding): ...this new function.
31734 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
31736 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
31737 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
31738 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
31739 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
31740 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
31741 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
31742 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
31743 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
31744 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
31745 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
31747 * expr.c (emit_move_resolve_push): Treat the input and result
31748 of PUSH_ROUNDING as a poly_int64.
31749 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
31750 (emit_push_insn): Likewise.
31751 * lra-eliminations.c (mark_not_eliminable): Likewise.
31752 * recog.c (push_operand): Likewise.
31753 * reload1.c (elimination_effects): Likewise.
31754 * rtlanal.c (nonzero_bits1): Likewise.
31755 * calls.c (store_one_arg): Likewise. Require the padding to be
31756 known at compile time.
31758 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31759 Alan Hayward <alan.hayward@arm.com>
31760 David Sherwood <david.sherwood@arm.com>
31762 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
31763 Use plus_constant instead of gen_rtx_PLUS.
31765 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31766 Alan Hayward <alan.hayward@arm.com>
31767 David Sherwood <david.sherwood@arm.com>
31769 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
31770 rather than an int.
31772 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31773 Alan Hayward <alan.hayward@arm.com>
31774 David Sherwood <david.sherwood@arm.com>
31776 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
31777 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
31778 via stack temporaries. Treat the mode size as polynomial too.
31780 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31781 Alan Hayward <alan.hayward@arm.com>
31782 David Sherwood <david.sherwood@arm.com>
31784 * expr.c (expand_expr_real_2): When handling conversions involving
31785 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
31786 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
31787 as a poly_uint64 too.
31789 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31790 Alan Hayward <alan.hayward@arm.com>
31791 David Sherwood <david.sherwood@arm.com>
31793 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
31795 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31796 Alan Hayward <alan.hayward@arm.com>
31797 David Sherwood <david.sherwood@arm.com>
31799 * combine.c (can_change_dest_mode): Handle polynomial
31800 REGMODE_NATURAL_SIZE.
31801 * expmed.c (store_bit_field_1): Likewise.
31802 * expr.c (store_constructor): Likewise.
31803 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
31804 and polynomial REGMODE_NATURAL_SIZE.
31805 (gen_lowpart_common): Likewise.
31806 * reginfo.c (record_subregs_of_mode): Likewise.
31807 * rtlanal.c (read_modify_subreg_p): Likewise.
31809 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31810 Alan Hayward <alan.hayward@arm.com>
31811 David Sherwood <david.sherwood@arm.com>
31813 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
31814 numbers of elements.
31816 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31817 Alan Hayward <alan.hayward@arm.com>
31818 David Sherwood <david.sherwood@arm.com>
31820 * match.pd: Cope with polynomial numbers of vector elements.
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 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
31827 in a POINTER_PLUS_EXPR.
31829 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31830 Alan Hayward <alan.hayward@arm.com>
31831 David Sherwood <david.sherwood@arm.com>
31833 * omp-simd-clone.c (simd_clone_subparts): New function.
31834 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
31835 (ipa_simd_modify_function_body): Likewise.
31837 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31838 Alan Hayward <alan.hayward@arm.com>
31839 David Sherwood <david.sherwood@arm.com>
31841 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
31842 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
31843 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
31844 (expand_vector_condition, vector_element): Likewise.
31845 (subparts_gt): New function.
31846 (get_compute_type): Use subparts_gt.
31847 (count_type_subparts): Delete.
31848 (expand_vector_operations_1): Use subparts_gt instead of
31849 count_type_subparts.
31851 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31852 Alan Hayward <alan.hayward@arm.com>
31853 David Sherwood <david.sherwood@arm.com>
31855 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
31856 (vect_compile_time_alias): ...this new function. Do the calculation
31857 on poly_ints rather than trees.
31858 (vect_prune_runtime_alias_test_list): Update call accordingly.
31860 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31861 Alan Hayward <alan.hayward@arm.com>
31862 David Sherwood <david.sherwood@arm.com>
31864 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
31866 (vect_schedule_slp_instance): Likewise.
31868 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31869 Alan Hayward <alan.hayward@arm.com>
31870 David Sherwood <david.sherwood@arm.com>
31872 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
31873 constant and extern definitions for variable-length vectors.
31874 (vect_get_constant_vectors): Note that the number of units
31875 is known to be constant.
31877 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31878 Alan Hayward <alan.hayward@arm.com>
31879 David Sherwood <david.sherwood@arm.com>
31881 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
31882 of units as polynomial. Choose between WIDE and NARROW based
31885 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31886 Alan Hayward <alan.hayward@arm.com>
31887 David Sherwood <david.sherwood@arm.com>
31889 * tree-vect-stmts.c (simd_clone_subparts): New function.
31890 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
31892 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31893 Alan Hayward <alan.hayward@arm.com>
31894 David Sherwood <david.sherwood@arm.com>
31896 * tree-vect-stmts.c (vectorizable_call): Treat the number of
31897 vectors as polynomial. Use build_index_vector for
31898 IFN_GOMP_SIMD_LANE.
31900 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31901 Alan Hayward <alan.hayward@arm.com>
31902 David Sherwood <david.sherwood@arm.com>
31904 * tree-vect-stmts.c (get_load_store_type): Treat the number of
31905 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
31906 for variable-length vectors.
31907 (vectorizable_mask_load_store): Treat the number of units as
31908 polynomial, asserting that it is constant if the condition has
31909 already been enforced.
31910 (vectorizable_store, vectorizable_load): Likewise.
31912 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31913 Alan Hayward <alan.hayward@arm.com>
31914 David Sherwood <david.sherwood@arm.com>
31916 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
31917 of units as polynomial. Punt if we can't tell at compile time
31918 which vector contains the final result.
31920 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31921 Alan Hayward <alan.hayward@arm.com>
31922 David Sherwood <david.sherwood@arm.com>
31924 * tree-vect-loop.c (vectorizable_induction): Treat the number
31925 of units as polynomial. Punt on SLP inductions. Use an integer
31926 VEC_SERIES_EXPR for variable-length integer reductions. Use a
31927 cast of such a series for variable-length floating-point
31930 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31931 Alan Hayward <alan.hayward@arm.com>
31932 David Sherwood <david.sherwood@arm.com>
31934 * tree.h (build_index_vector): Declare.
31935 * tree.c (build_index_vector): New function.
31936 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
31937 of units as polynomial, forcibly converting it to a constant if
31938 vectorizable_reduction has already enforced the condition.
31939 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
31940 to create a {1,2,3,...} vector.
31941 (vectorizable_reduction): Treat the number of units as polynomial.
31942 Choose vectype_in based on the largest scalar element size rather
31943 than the smallest number of units. Enforce the restrictions
31946 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31947 Alan Hayward <alan.hayward@arm.com>
31948 David Sherwood <david.sherwood@arm.com>
31950 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
31951 number of units as polynomial.
31953 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31954 Alan Hayward <alan.hayward@arm.com>
31955 David Sherwood <david.sherwood@arm.com>
31957 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
31958 * target.def (autovectorize_vector_sizes): Return the vector sizes
31959 by pointer, using vector_sizes rather than a bitmask.
31960 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
31961 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
31962 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
31964 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
31965 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
31966 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
31967 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
31968 * omp-general.c (omp_max_vf): Likewise.
31969 * omp-low.c (omp_clause_aligned_alignment): Likewise.
31970 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
31971 * tree-vect-loop.c (vect_analyze_loop): Likewise.
31972 * tree-vect-slp.c (vect_slp_bb): Likewise.
31973 * doc/tm.texi: Regenerate.
31974 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
31976 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
31977 the vector size as a poly_uint64 rather than an unsigned int.
31978 (current_vector_size): Change from an unsigned int to a poly_uint64.
31979 (get_vectype_for_scalar_type): Update accordingly.
31980 * tree.h (build_truth_vector_type): Take the size and number of
31981 units as a poly_uint64 rather than an unsigned int.
31982 (build_vector_type): Add a temporary overload that takes
31983 the number of units as a poly_uint64 rather than an unsigned int.
31984 * tree.c (make_vector_type): Likewise.
31985 (build_truth_vector_type): Take the number of units as a poly_uint64
31986 rather than an unsigned int.
31988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31989 Alan Hayward <alan.hayward@arm.com>
31990 David Sherwood <david.sherwood@arm.com>
31992 * target.def (get_mask_mode): Take the number of units and length
31993 as poly_uint64s rather than unsigned ints.
31994 * targhooks.h (default_get_mask_mode): Update accordingly.
31995 * targhooks.c (default_get_mask_mode): Likewise.
31996 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
31997 * doc/tm.texi: Regenerate.
31999 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32000 Alan Hayward <alan.hayward@arm.com>
32001 David Sherwood <david.sherwood@arm.com>
32003 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
32004 * omp-general.c (omp_max_vf): Likewise.
32005 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
32006 (expand_omp_simd): Handle polynomial safelen.
32007 * omp-low.c (omplow_simd_context): Add a default constructor.
32008 (omplow_simd_context::max_vf): Change from int to poly_uint64.
32009 (lower_rec_simd_input_clauses): Update accordingly.
32010 (lower_rec_input_clauses): Likewise.
32012 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32013 Alan Hayward <alan.hayward@arm.com>
32014 David Sherwood <david.sherwood@arm.com>
32016 * tree-vectorizer.h (vect_nunits_for_cost): New function.
32017 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
32018 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
32019 (vect_analyze_slp_cost): Likewise.
32020 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
32021 (vect_model_load_cost): Likewise.
32023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32024 Alan Hayward <alan.hayward@arm.com>
32025 David Sherwood <david.sherwood@arm.com>
32027 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
32028 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
32029 from an unsigned int * to a poly_uint64_pod *.
32030 (calculate_unrolling_factor): New function.
32031 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
32033 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32034 Alan Hayward <alan.hayward@arm.com>
32035 David Sherwood <david.sherwood@arm.com>
32037 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
32038 from an unsigned int to a poly_uint64.
32039 (_loop_vec_info::slp_unrolling_factor): Likewise.
32040 (_loop_vec_info::vectorization_factor): Change from an int
32042 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
32043 (vect_get_num_vectors): New function.
32044 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
32045 (vect_get_num_copies): Use vect_get_num_vectors.
32046 (vect_analyze_data_ref_dependences): Change max_vf from an int *
32047 to an unsigned int *.
32048 (vect_analyze_data_refs): Change min_vf from an int * to a
32050 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
32051 than an unsigned HOST_WIDE_INT.
32052 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
32053 (vect_analyze_data_ref_dependence): Change max_vf from an int *
32054 to an unsigned int *.
32055 (vect_analyze_data_ref_dependences): Likewise.
32056 (vect_compute_data_ref_alignment): Handle polynomial vf.
32057 (vect_enhance_data_refs_alignment): Likewise.
32058 (vect_prune_runtime_alias_test_list): Likewise.
32059 (vect_shift_permute_load_chain): Likewise.
32060 (vect_supportable_dr_alignment): Likewise.
32061 (dependence_distance_ge_vf): Take the vectorization factor as a
32062 poly_uint64 rather than an unsigned HOST_WIDE_INT.
32063 (vect_analyze_data_refs): Change min_vf from an int * to a
32065 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
32066 vfm1 as a poly_uint64 rather than an int. Make the same change
32067 for the returned bound_scalar.
32068 (vect_gen_vector_loop_niters): Handle polynomial vf.
32069 (vect_do_peeling): Likewise. Update call to
32070 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
32071 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
32073 * tree-vect-loop.c (vect_determine_vectorization_factor)
32074 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
32075 (vect_get_known_peeling_cost): Likewise.
32076 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
32077 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
32078 (vect_transform_loop): Likewise. Use the lowest possible VF when
32079 updating the upper bounds of the loop.
32080 (vect_min_worthwhile_factor): Make static. Return an unsigned int
32081 rather than an int.
32082 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
32083 polynomial unroll factors.
32084 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
32085 (vect_make_slp_decision): Likewise.
32086 (vect_supported_load_permutation_p): Likewise, and polynomial
32088 (vect_analyze_slp_cost): Handle polynomial vf.
32089 (vect_slp_analyze_node_operations): Likewise.
32090 (vect_slp_analyze_bb_1): Likewise.
32091 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
32092 than an unsigned HOST_WIDE_INT.
32093 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
32094 (vectorizable_load): Handle polynomial vf.
32095 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
32097 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
32099 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32100 Alan Hayward <alan.hayward@arm.com>
32101 David Sherwood <david.sherwood@arm.com>
32103 * match.pd: Handle bit operations involving three constants
32104 and try to fold one pair.
32106 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32108 * tree-vect-loop-manip.c: Include gimple-fold.h.
32109 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
32110 niters_maybe_zero parameters. Handle other cases besides a step of 1.
32111 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
32112 Add a path that uses a step of VF instead of 1, but disable it
32114 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
32115 and niters_no_overflow parameters. Update calls to
32116 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
32117 Create a new SSA name if the latter choses to use a ste other
32118 than zero, and return it via niters_vector_mult_vf_var.
32119 * tree-vect-loop.c (vect_transform_loop): Update calls to
32120 vect_do_peeling, vect_gen_vector_loop_niters and
32121 slpeel_make_loop_iterate_ntimes.
32122 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
32123 (vect_gen_vector_loop_niters): Update declarations after above changes.
32125 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
32127 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
32128 128-bit round to integer instructions.
32129 (ceil<mode>2): Likewise.
32130 (btrunc<mode>2): Likewise.
32131 (round<mode>2): Likewise.
32133 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32135 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
32136 unaligned VSX load/store on P8/P9.
32137 (expand_block_clear): Allow the use of unaligned VSX
32138 load/store on P8/P9.
32140 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32142 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
32144 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
32145 swap associated with both a load and a store.
32147 2018-01-02 Andrew Waterman <andrew@sifive.com>
32149 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
32150 * config/riscv/riscv.md (clear_cache): Use it.
32152 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
32154 * web.c: Remove out-of-date comment.
32156 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32158 * expr.c (fixup_args_size_notes): Check that any existing
32159 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
32160 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
32161 (emit_single_push_insn): ...here.
32163 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32165 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
32166 (const_vector_encoded_nelts): New function.
32167 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
32168 (const_vector_int_elt, const_vector_elt): Declare.
32169 * emit-rtl.c (const_vector_int_elt_1): New function.
32170 (const_vector_elt): Likewise.
32171 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
32172 of CONST_VECTOR_ELT.
32174 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32176 * expr.c: Include rtx-vector-builder.h.
32177 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
32178 directly on the tree encoding.
32179 (const_vector_from_tree): Likewise.
32180 * optabs.c: Include rtx-vector-builder.h.
32181 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
32182 sequence of "u" values.
32183 * vec-perm-indices.c: Include rtx-vector-builder.h.
32184 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
32185 directly on the vec_perm_indices encoding.
32187 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32189 * doc/rtl.texi (const_vector): Describe new encoding scheme.
32190 * Makefile.in (OBJS): Add rtx-vector-builder.o.
32191 * rtx-vector-builder.h: New file.
32192 * rtx-vector-builder.c: Likewise.
32193 * rtl.h (rtx_def::u2): Add a const_vector field.
32194 (CONST_VECTOR_NPATTERNS): New macro.
32195 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
32196 (CONST_VECTOR_DUPLICATE_P): Likewise.
32197 (CONST_VECTOR_STEPPED_P): Likewise.
32198 (CONST_VECTOR_ENCODED_ELT): Likewise.
32199 (const_vec_duplicate_p): Check for a duplicated vector encoding.
32200 (unwrap_const_vec_duplicate): Likewise.
32201 (const_vec_series_p): Check for a non-duplicated vector encoding.
32202 Say that the function only returns true for integer vectors.
32203 * emit-rtl.c: Include rtx-vector-builder.h.
32204 (gen_const_vec_duplicate_1): Delete.
32205 (gen_const_vector): Call gen_const_vec_duplicate instead of
32206 gen_const_vec_duplicate_1.
32207 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
32208 (gen_const_vec_duplicate): Use rtx_vector_builder.
32209 (gen_const_vec_series): Likewise.
32210 (gen_rtx_CONST_VECTOR): Likewise.
32211 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
32212 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
32213 Build a new vector rather than modifying a CONST_VECTOR in-place.
32214 (handle_special_swappables): Update call accordingly.
32215 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
32216 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
32217 Build a new vector rather than modifying a CONST_VECTOR in-place.
32218 (handle_special_swappables): Update call accordingly.
32220 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32222 * simplify-rtx.c (simplify_const_binary_operation): Use
32223 CONST_VECTOR_ELT instead of XVECEXP.
32225 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32227 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
32228 the selector elements to be different from the data elements
32229 if the selector is a VECTOR_CST.
32230 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
32231 ssizetype for the selector.
32233 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32235 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
32236 before testing each element individually.
32237 * tree-vect-generic.c (lower_vec_perm): Likewise.
32239 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32241 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
32242 * selftest-run-tests.c (selftest::run_tests): Call it.
32243 * vector-builder.h (vector_builder::operator ==): New function.
32244 (vector_builder::operator !=): Likewise.
32245 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
32246 (vec_perm_indices::all_from_input_p): New function.
32247 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
32248 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
32249 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
32250 instead of reading the VECTOR_CST directly. Detect whether both
32251 vector inputs are the same before constructing the vec_perm_indices,
32252 and update the number of inputs argument accordingly. Use the
32253 utility functions added above. Only construct sel2 if we need to.
32255 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32257 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
32258 the broadcast of the low byte.
32259 (expand_mult_highpart): Use an explicit encoding for the permutes.
32260 * optabs-query.c (can_mult_highpart_p): Likewise.
32261 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
32262 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32263 (vectorizable_bswap): Likewise.
32264 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
32265 explicit encoding for the power-of-2 permutes.
32266 (vect_permute_store_chain): Likewise.
32267 (vect_grouped_load_supported): Likewise.
32268 (vect_permute_load_chain): Likewise.
32270 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32272 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
32273 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
32274 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
32275 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
32276 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
32277 (vect_gen_perm_mask_any): Likewise.
32279 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32281 * int-vector-builder.h: New file.
32282 * vec-perm-indices.h: Include int-vector-builder.h.
32283 (vec_perm_indices): Redefine as an int_vector_builder.
32284 (auto_vec_perm_indices): Delete.
32285 (vec_perm_builder): Redefine as a stand-alone class.
32286 (vec_perm_indices::vec_perm_indices): New function.
32287 (vec_perm_indices::clamp): Likewise.
32288 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
32289 (vec_perm_indices::new_vector): New function.
32290 (vec_perm_indices::new_expanded_vector): Update for new
32291 vec_perm_indices class.
32292 (vec_perm_indices::rotate_inputs): New function.
32293 (vec_perm_indices::all_in_range_p): Operate directly on the
32294 encoded form, without computing elided elements.
32295 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
32296 encoding. Update for new vec_perm_indices class.
32297 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
32298 the given vec_perm_builder.
32299 (expand_vec_perm_var): Update vec_perm_builder constructor.
32300 (expand_mult_highpart): Use vec_perm_builder instead of
32301 auto_vec_perm_indices.
32302 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
32303 vec_perm_indices instead of auto_vec_perm_indices. Use a single
32304 or double series encoding as appropriate.
32305 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
32306 vec_perm_indices instead of auto_vec_perm_indices.
32307 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32308 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32309 (vect_permute_store_chain): Likewise.
32310 (vect_grouped_load_supported): Likewise.
32311 (vect_permute_load_chain): Likewise.
32312 (vect_shift_permute_load_chain): Likewise.
32313 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
32314 (vect_transform_slp_perm_load): Likewise.
32315 (vect_schedule_slp_instance): Likewise.
32316 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32317 (vectorizable_mask_load_store): Likewise.
32318 (vectorizable_bswap): Likewise.
32319 (vectorizable_store): Likewise.
32320 (vectorizable_load): Likewise.
32321 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
32322 vec_perm_indices instead of auto_vec_perm_indices. Use
32323 tree_to_vec_perm_builder to read the vector from a tree.
32324 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
32325 vec_perm_builder instead of a vec_perm_indices.
32326 (have_whole_vector_shift): Use vec_perm_builder and
32327 vec_perm_indices instead of auto_vec_perm_indices. Leave the
32328 truncation to calc_vec_perm_mask_for_shift.
32329 (vect_create_epilog_for_reduction): Likewise.
32330 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
32331 from auto_vec_perm_indices to vec_perm_indices.
32332 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
32333 instead of changing individual elements.
32334 (aarch64_vectorize_vec_perm_const): Use new_vector to install
32335 the vector in d.perm.
32336 * config/arm/arm.c (expand_vec_perm_d::perm): Change
32337 from auto_vec_perm_indices to vec_perm_indices.
32338 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
32339 instead of changing individual elements.
32340 (arm_vectorize_vec_perm_const): Use new_vector to install
32341 the vector in d.perm.
32342 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
32343 Update vec_perm_builder constructor.
32344 (rs6000_expand_interleave): Likewise.
32345 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
32346 (rs6000_expand_interleave): Likewise.
32348 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32350 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
32351 to qimode could truncate the indices.
32352 * optabs.c (expand_vec_perm_var): Likewise.
32354 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32356 * Makefile.in (OBJS): Add vec-perm-indices.o.
32357 * vec-perm-indices.h: New file.
32358 * vec-perm-indices.c: Likewise.
32359 * target.h (vec_perm_indices): Replace with a forward class
32361 (auto_vec_perm_indices): Move to vec-perm-indices.h.
32362 * optabs.h: Include vec-perm-indices.h.
32363 (expand_vec_perm): Delete.
32364 (selector_fits_mode_p, expand_vec_perm_var): Declare.
32365 (expand_vec_perm_const): Declare.
32366 * target.def (vec_perm_const_ok): Replace with...
32367 (vec_perm_const): ...this new hook.
32368 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
32369 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
32370 * doc/tm.texi: Regenerate.
32371 * optabs.def (vec_perm_const): Delete.
32372 * doc/md.texi (vec_perm_const): Likewise.
32373 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
32374 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
32375 expand_vec_perm for constant permutation vectors. Assert that
32376 the mode of variable permutation vectors is the integer equivalent
32377 of the mode that is being permuted.
32378 * optabs-query.h (selector_fits_mode_p): Declare.
32379 * optabs-query.c: Include vec-perm-indices.h.
32380 (selector_fits_mode_p): New function.
32381 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
32382 is defined, instead of checking whether the vec_perm_const_optab
32383 exists. Use targetm.vectorize.vec_perm_const instead of
32384 targetm.vectorize.vec_perm_const_ok. Check whether the indices
32385 fit in the vector mode before using a variable permute.
32386 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
32387 vec_perm_indices instead of an rtx.
32388 (expand_vec_perm): Replace with...
32389 (expand_vec_perm_const): ...this new function. Take the selector
32390 as a vec_perm_indices rather than an rtx. Also take the mode of
32391 the selector. Update call to shift_amt_for_vec_perm_mask.
32392 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
32393 Use vec_perm_indices::new_expanded_vector to expand the original
32394 selector into bytes. Check whether the indices fit in the vector
32395 mode before using a variable permute.
32396 (expand_vec_perm_var): Make global.
32397 (expand_mult_highpart): Use expand_vec_perm_const.
32398 * fold-const.c: Includes vec-perm-indices.h.
32399 * tree-ssa-forwprop.c: Likewise.
32400 * tree-vect-data-refs.c: Likewise.
32401 * tree-vect-generic.c: Likewise.
32402 * tree-vect-loop.c: Likewise.
32403 * tree-vect-slp.c: Likewise.
32404 * tree-vect-stmts.c: Likewise.
32405 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
32407 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
32408 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
32409 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
32410 (aarch64_vectorize_vec_perm_const): ...this new function.
32411 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32412 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32413 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
32414 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
32415 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32416 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32417 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
32419 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
32420 check for NEON modes.
32421 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
32422 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
32423 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
32424 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
32426 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
32427 the old VEC_PERM_CONST conditions.
32428 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
32429 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
32430 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
32431 (ia64_vectorize_vec_perm_const_ok): Merge into...
32432 (ia64_vectorize_vec_perm_const): ...this new function.
32433 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
32434 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
32435 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
32436 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
32437 * config/mips/mips.c (mips_expand_vec_perm_const)
32438 (mips_vectorize_vec_perm_const_ok): Merge into...
32439 (mips_vectorize_vec_perm_const): ...this new function.
32440 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
32441 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
32442 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
32443 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
32444 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
32445 (rs6000_expand_vec_perm_const): Delete.
32446 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
32448 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32449 (altivec_expand_vec_perm_const_le): Take each operand individually.
32450 Operate on constant selectors rather than rtxes.
32451 (altivec_expand_vec_perm_const): Likewise. Update call to
32452 altivec_expand_vec_perm_const_le.
32453 (rs6000_expand_vec_perm_const): Delete.
32454 (rs6000_vectorize_vec_perm_const_ok): Delete.
32455 (rs6000_vectorize_vec_perm_const): New function.
32456 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
32457 an element count and rtx array.
32458 (rs6000_expand_extract_even): Update call accordingly.
32459 (rs6000_expand_interleave): Likewise.
32460 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
32461 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
32462 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
32463 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
32464 (rs6000_expand_vec_perm_const): Delete.
32465 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
32466 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32467 (altivec_expand_vec_perm_const_le): Take each operand individually.
32468 Operate on constant selectors rather than rtxes.
32469 (altivec_expand_vec_perm_const): Likewise. Update call to
32470 altivec_expand_vec_perm_const_le.
32471 (rs6000_expand_vec_perm_const): Delete.
32472 (rs6000_vectorize_vec_perm_const_ok): Delete.
32473 (rs6000_vectorize_vec_perm_const): New function. Remove stray
32474 reference to the SPE evmerge intructions.
32475 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
32476 an element count and rtx array.
32477 (rs6000_expand_extract_even): Update call accordingly.
32478 (rs6000_expand_interleave): Likewise.
32479 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
32480 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
32482 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
32484 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32486 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
32487 vector mode and that that mode matches the mode of the data
32489 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
32490 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
32491 directly using expand_vec_perm_1 when forcing selectors into
32493 (expand_vec_perm_var): New function, split out from expand_vec_perm.
32495 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32497 * optabs-query.h (can_vec_perm_p): Delete.
32498 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
32499 * optabs-query.c (can_vec_perm_p): Split into...
32500 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
32501 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
32502 particular selector is valid.
32503 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32504 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32505 (vect_grouped_load_supported): Likewise.
32506 (vect_shift_permute_load_chain): Likewise.
32507 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
32508 (vect_transform_slp_perm_load): Likewise.
32509 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32510 (vectorizable_bswap): Likewise.
32511 (vect_gen_perm_mask_checked): Likewise.
32512 * fold-const.c (fold_ternary_loc): Likewise. Don't take
32513 implementations of variable permutation vectors into account
32514 when deciding which selector to use.
32515 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
32516 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
32517 with a false third argument.
32518 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
32519 to test whether the constant selector is valid and can_vec_perm_var_p
32520 to test whether a variable selector is valid.
32522 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32524 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
32525 * optabs-query.c (can_vec_perm_p): Likewise.
32526 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
32527 instead of vec_perm_indices.
32528 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
32529 (vect_gen_perm_mask_checked): Likewise,
32530 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
32531 (vect_gen_perm_mask_checked): Likewise,
32533 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
32535 * optabs-query.h (qimode_for_vec_perm): Declare.
32536 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
32537 (qimode_for_vec_perm): ...this new function.
32538 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
32540 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32542 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
32543 does not have a conditional at the top.
32545 2018-01-02 Richard Biener <rguenther@suse.de>
32547 * ipa-inline.c (big_speedup_p): Fix expression.
32549 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32552 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
32555 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
32559 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
32560 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
32561 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
32562 cond_taken_branch_cost 3->4.
32564 2018-01-01 Jakub Jelinek <jakub@redhat.com>
32566 PR tree-optimization/83581
32567 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
32568 TODO_cleanup_cfg if any changes have been made.
32570 PR middle-end/83608
32571 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
32572 convert_modes if target mode has the right side, but different mode
32575 PR middle-end/83609
32576 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
32577 last argument when extracting from CONCAT. If either from_real or
32578 from_imag is NULL, use expansion through memory. If result is not
32579 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
32580 the parts directly to inner mode, if even that fails, use expansion
32583 PR middle-end/83623
32584 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
32585 check for bswap in mode rather than HImode and use that in expand_unop
32588 Copyright (C) 2018 Free Software Foundation, Inc.
32590 Copying and distribution of this file, with or without modification,
32591 are permitted in any medium without royalty provided the copyright
32592 notice and this notice are preserved.