1 2018-12-07 Richard Sandiford <richard.sandiford@arm.com>
3 * config/aarch64/aarch64-sve.md (*mul<mode>3, *v<optab><mode>3):
4 Split the patterns after reload if we don't need the predicate
6 (*post_ra_mul<mode>3, *post_ra_v<optab><mode>3): New patterns.
8 2018-12-07 Richard Sandiford <richard.sandiford@arm.com>
10 * config/aarch64/iterators.md (SVE_UNPRED_FP_BINARY): New code
12 (sve_fp_op): Handle minus and mult.
13 * config/aarch64/aarch64-sve.md (*add<mode>3, *sub<mode>3)
14 (*mul<mode>3): Split the patterns after reload if we don't
15 need the predicate operand.
16 (*post_ra_<sve_fp_op><mode>3): New pattern.
18 2018-12-07 Bin Cheng <bin.cheng@linux.alibaba.com>
20 * profile-count.h (profile_count::oeprator>=): Fix typo by inverting
21 return condition when *this is precise zero.
23 2018-12-07 Jakub Jelinek <jakub@redhat.com>
26 * final.c (rest_of_handle_final): Don't call collect_fn_hard_reg_usage
27 for functions with naked attribute.
30 * tree-vrp.c (extract_range_from_binary_expr): For POINTER_PLUS_EXPR
31 with -fno-delete-null-pointer-checks, set_nonnull only if the pointer
32 is non-NULL and offset is known to have most significant bit clear.
33 * vr-values.c (vr_values::vrp_stmt_computes_nonzero): For ADDR_EXPR
34 of MEM_EXPR, return true if the MEM_EXPR has non-zero offset with
35 most significant bit clear. If offset does have most significant bit
36 set and -fno-delete-null-pointer-checks, don't return true even if
37 the base pointer is non-NULL.
39 2018-12-06 Alexandre Oliva <aoliva@redhat.com>
41 * cselib.c (cselib_record_sets): Skip strict low part sets
44 2018-12-06 Paul A. Clarke <pc@us.ibm.com>
47 * config/rs6000/smmintrin.h: New file.
48 * config.gcc: Add smmintrin.h to extra_headers for powerpc*-*-*.
50 2018-12-06 Paul A. Clarke <pc@us.ibm.com>
53 * config/rs6000/mmintrin.h (_mm_unpackhi_pi8): Fix for big-endian.
54 (_mm_unpacklo_pi8): Likewise.
55 (_mm_mulhi_pi16): Likewise.
56 (_mm_packs_pi16): Fix for big-endian. Use preferred API.
57 (_mm_packs_pi32): Likewise.
58 (_mm_packs_pu16): Likewise.
59 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix for big-endian.
60 (_mm_cvtss_si64): Likewise.
61 (_mm_cvtpi32x2_ps): Likewise.
62 (_mm_shuffle_ps): Likewise.
63 (_mm_movemask_pi8): Likewise.
64 (_mm_mulhi_pu16): Likewise.
65 (_mm_sad_pu8): Likewise.
66 (_mm_sad_pu8): Likewise.
67 (_mm_cvtpu16_ps): Fix for big-endian. Use preferred API.
68 (_mm_cvtpu8_ps): Likewise.
69 (_mm_movemask_ps): Better #else case for big-endian (no functional
71 (_mm_shuffle_pi16): Likewise.
72 * config/rs6000/emmintrin.h (_mm_movemask_pd): Fix for big-endian.
73 Better #else case for big-endian (no functional change).
74 (_mm_movemask_epi8): Likewise.
75 (_mm_shufflehi_epi16): Likewise.
76 (_mm_shufflelo_epi16): Likewise.
77 (_mm_shuffle_epi32): Likewise.
78 (_mm_mul_epu32): Fix for big-endian.
79 (_mm_bsrli_si128): Likewise.
80 (_mm_cvtps_pd): Better #else case for big endian.
81 (_mm_mulhi_epi16): Likewise.
82 (_mm_mul_epu32): Likewise.
83 (_mm_slli_si128): Likewise.
84 (_mm_sll_epi16): Likewise.
85 (_mm_sll_epi32): Likewise.
86 (_mm_sra_epi16): Likewise.
87 (_mm_sra_epi32): Likewise.
88 (_mm_srl_epi16): Likewise.
89 (_mm_srl_epi32): Likewise.
90 (_mm_mulhi_epu16): Likewise.
91 (_mm_sad_epu8): Likewise.
92 * config/rs6000/pmmintrin.h (_mm_hadd_ps): Fix for big-endian.
93 (_mm_sub_ps): Likewise.
94 * config/rs6000/mmintrin.h (_mm_cmpeq_pi8): Fix for 32-bit mode.
95 * gcc/config/rs6000/tmmintrin.h (_mm_alignr_epi8): Use ENDIAN
96 macros consistently (no functional changes).
97 (_mm_alignr_pi8): Likewise.
99 2018-12-06 Iain Sandoe <iain@sandoe.co.uk>
102 * config/darwin.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC) Remove, use the
104 * config/rs6000/darwin7.h (TARGET_WEAK_NOT_IN_ARCHIVE_TOC): New.
106 2018-12-06 Vladimir Makarov <vmakarov@redhat.com>
109 * ira.c (ira_init_register_move_cost): Use info from
110 hard_regno_mode_ok instead of contains_reg_of_mode.
111 * ira-costs.c (contains_reg_of_mode): Don't use cost from bigger
112 hard register class for some fixed hard registers.
114 2018-12-06 Segher Boessenkool <segher@kernel.crashing.org>
116 * doc/extend.texi (Using Assembly Language with C): Document asm inline.
117 (Size of an asm): Fix typo. Document asm inline.
118 * gimple-pretty-print.c (dump_gimple_asm): Handle asm inline.
119 * gimple.h (enum gf_mask): Add GF_ASM_INLINE.
120 (gimple_asm_set_volatile): Fix typo.
121 (gimple_asm_inline_p): New.
122 (gimple_asm_set_inline): New.
123 * gimplify.c (gimplify_asm_expr): Propagate the asm inline flag from
125 * ipa-icf-gimple.c (func_checker::compare_gimple_asm): Compare the
126 gimple_asm_inline_p flag, too.
127 * tree-core.h (tree_base): Document that protected_flag is ASM_INLINE_P
129 * tree-inline.c (estimate_num_insns): If gimple_asm_inline_p return
130 a minimum size for an asm.
131 * tree.h (ASM_INLINE_P): New.
133 2018-12-06 Segher Boessenkool <segher@kernel.crashing.org>
136 * doc/extend.texi (Basic Asm): Update grammar.
137 (Extended Asm): Update grammar.
139 2018-12-06 Iain Sandoe <iain@sandoe.co.uk>
142 * config/i386/darwin.h (STACK_BOUNDARY): Remove macro.
143 * config/i386/i386.c (ix86_compute_frame_layout): Ensure at least 128b
144 stack alignment in non-leaf functions.
146 2018-12-06 Jakub Jelinek <jakub@redhat.com>
149 * config/aarch64/aarch64.c (aarch64_print_address_internal): Don't
150 call output_operand_lossage on VOIDmode CONST_INTs. After
151 output_operand_lossage do return false.
153 2018-12-06 Richard Biener <rguenther@suse.de>
155 * df-problems.c (df_rd_local_compute): Use bitmap_release.
156 (df_live_free): Likewise.
157 (df_md_local_compute): Likewise.
158 (df_md_free): Release df_md_scratch bitmap.
159 * loop-invariant.c (calculate_loop_reg_pressure): Use
161 * sched-deps.c (true_dependency_cache, output_dependency_cache,
162 anti_dependency_cache, control_dependency_cache,
163 spec_dependency_cache): Use bitmap instead of bitmap_head *.
164 * sched-ebb.c (schedule_ebbs_init): Initialize non-GTY
165 dont_calc_deps as bitmap allocated from obstack not GC.
166 (schedule_ebbs_finish): Use bitmap_release.
167 * sched-rgn.c (schedule_insns): Initialize non-GTY
168 not_in_df as bitmap allocated from obstack not GC.
170 * sel-sched.c (_forced_ebb_heads): Remove premature optimization.
171 (sel_region_init): Allocate forced_ebb_heads.
172 (sel_region_finish): Free forced_ebb_heads.
174 2018-12-06 Richard Biener <rguenther@suse.de>
176 * bitmap.c (bitmap_head::crashme): Define.
177 * bitmap.h (bitmap_head): Add constexpr default constructor
178 poisoning the obstack member.
179 (bitmap_head::crashme): Declare.
180 (bitmap_release): New function clearing a bitmap and poisoning
182 * gengtype.c (main): Make it recognize CONSTEXPR.
183 * lra-constraints.c (lra_inheritance): Use bitmap_release
184 instead of bitmap_clear.
185 * ira.c (ira): Work around class-memaccess warning.
186 * regrename.c (create_new_chain): Likewise.
188 2018-12-06 Jakub Jelinek <jakub@redhat.com>
190 PR tree-optimization/85726
191 * generic-match-head.c (optimize_successive_divisions_p): New function.
192 * gimple-match-head.c (optimize_successive_divisions_p): Likewise.
193 * match.pd: Don't combine successive divisions if they aren't exact
194 and optimize_successive_divisions_p is false.
197 * tree-nested.c (convert_nonlocal_reference_stmt): Remove clobbers
198 for non-local automatic decls.
200 2018-12-05 David Edelsohn <dje.gcc@gmail.com>
202 * config/rs6000/aix72.h (ASM_DEFAULT_SPEC): Match Power7 processor
205 2018-12-05 Uros Bizjak <ubizjak@gmail.com>
207 * dwarf2out.c (field_byte_offset): Remove
208 #ifdef PCC_BITFIELD_TYPE_MATTERS check.
210 2018-12-05 Richard Biener <rguenther@suse.de>
211 Martin Sebor <msebor@redhat.com>
214 * calls.c (get_attr_nonstring_decl): Avoid setting *REF to
216 * gcc/gimple-low.c (lower_stmt): Fold builtin calls here.
217 * gimplify (maybe_fold_stmt): Avoid folding builtin calls.
219 2018-12-05 Iain Sandoe <iain@sandoe.co.uk>
221 * configure.ac (gcc_cv_otool): Set.
222 * configure: Regenerate.
224 2018-12-05 Uros Bizjak <ubizjak@gmail.com>
226 * config/i386/cygming.h (PCC_BITFIELD_TYPE_MATTERS): Remove.
227 (GROUP_BITFIELDS_BY_ALIGN): Ditto.
229 2018-12-05 Richard Sandiford <richard.sandiford@arm.com>
231 PR tree-optimization/88064
232 * tree-vect-patterns.c (vect_convert_input): Convert the result of
233 an existing cast if it has the right width but the wrong sign.
234 Do not test the signedness of the required result when
235 considering whether to split an existing cast; instead split to
236 a type with the same signedness as the source of the cast, then
237 convert it to the opposite signedness where necessary.
238 (vect_recog_over_widening_pattern): Handle sign changes between
239 the final PLUS_EXPR and the RSHIFT_EXPR.
240 (vect_recog_average_pattern): Use an unsigned operation when
241 truncating an addition, subtraction or multiplication. Cast the
242 result back to the "real" signedness before promoting.
244 2018-12-05 Richard Biener <rguenther@suse.de>
246 PR tree-optimization/86637
247 * tree-vectorizer.c (pass_slp_vectorize::execute): Reset
248 vect_location at the end.
250 2018-12-05 Jakub Jelinek <jakub@redhat.com>
252 PR tree-optimization/87360
253 * gimple-loop-jam.c (tree_loop_unroll_and_jam): On failure to analyze
254 data dependencies, don't return false, just continue. Formatting
256 (merge_loop_tree, bb_prevents_fusion_p, unroll_jam_possible_p,
257 fuse_loops): Formatting fixes.
260 * cfgexpand.c (expand_stack_vars): If asan_vec is empty, start with
261 aligning frame offset to ASAN_RED_ZONE_SIZE bytes.
263 2018-12-05 Xianmiao Qu <xianmiao_qu@c-sky.com>
265 * config.gcc (csky-*-linux-gnu*): Force .init_array support.
267 2018-12-05 Segher Boessenkool <segher@kernel.crashing.org>
269 * config/rs6000/rs6000.c (rs6000_function_arg): Only do the special
270 aggregate handling on actual AIX, not on somewhat similar systems.
271 (rs6000_arg_partial_bytes): Ditto.
273 2018-12-04 Vladimir Makarov <vmakarov@redhat.com>
275 PR rtl-optimization/88317
276 * lra-constraints.c (split_reg): Don't set up check_only_regs if
277 we are outside the inheritance pass.
279 2018-12-04 Jan Hubicka <hubicka@ucw.cz>
281 * ipa-prop.c (jump_function_useful_p): New.
282 (ipa_read_node_info): Do not allocated useless jump functions.
284 2018-12-04 David Malcolm <dmalcolm@redhat.com>
287 * spellcheck.c (selftest::test_suggestions): Add another assertion.
289 2018-12-04 Martin Sebor <msebor@redhat.com>
291 * doc/extend.texi (__builtin_has_attribute): Fix typo.
293 2018-12-04 Jakub Jelinek <jakub@redhat.com>
296 * config/rs6000/rs6000.c (ccr_bit): Return -1 instead of assertion
298 (print_operand): Use REG_P instead of GET_CODE == REG.
299 <case 'D'>: Also check CR_REGNO_P (REGNO (x)).
300 <case 't'>: Likewise. Remove GET_MODE check.
301 * config/rs6000/rs6000.md (scc patterns): Assert ccr_bit didn't
304 2018-12-04 Martin Sebor <msebor@redhat.com>
306 * doc/extend.texi (attribute aligned): Expand.
308 2018-12-04 David Edelsohn <dje.gcc@gmail.com>
311 * config/rs6000/rs6000.c (rs6000_function_arg): Don't pass aggregates
313 (rs6000_arg_partial_bytes): Same.
315 2018-12-04 Vladimir Makarov <vmakarov@redhat.com>
318 * ira-costs.c (exec): Try bigger class to use smaller register
321 2018-12-04 Michael Ploujnikov <michael.ploujnikov@oracle.com>
324 * ipa-cp.c (create_specialized_node): Track clone counters by
325 node assembler names.
326 (ipcp_driver): Change type of clone_num_suffixes key to const char*.
328 2018-12-04 Richard Biener <rguenther@suse.de>
330 * tree-vect-loop.c (vect_transform_loop): Fix epilogue iterations
331 for epilogue vectorization.
333 2018-12-04 Richard Biener <rguenther@suse.de>
335 PR tree-optimization/88301
336 * tree-vrp.c (register_edge_assert_for_2): Fix sign-conversion
337 issues in last commit.
339 2018-12-04 Richard Biener <rguenther@suse.de>
341 PR tree-optimization/88315
342 * tree-vect-loop.c (get_initial_defs_for_reduction): Simplify
343 and fix initialization vector for SAD and DOT_PROD SLP reductions.
345 2018-12-03 Sandra Loosemore <sandra@codesourcery.com>
348 * doc/extend.texi (Nonlocal gotos): New section.
350 2018-12-03 Uros Bizjak <ubizjak@gmail.com>
352 * config/i386/sse.md (vec_concatv2df): Change (v,xm,C)
353 alternative to (v,vm,c).
355 2018-12-03 Segher Boessenkool <segher@kernel.crashing.org>
357 * gcc/config.gcc (Unsupported targets): Add powerpc*-*-linux*paired*.
358 (powerpc*-*-linux*): Don't handle it here.
360 2018-12-03 Jeff Law <law@redhat.com>
362 * config/h8300/h8300.md (<code>qi3_1): Add gcc_unreachable to keep
365 * config/h8300/h8300.md (low byte masking peepholes): Only accept
366 nonimmediate_operand for the to-be-masked operand.
368 2018-12-03 Richard Biener <rguenther@suse.de>
370 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
371 allow expected function-pointer cast re-instantiation.
373 2018-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
375 * common/config/s390/s390-common.c (s390_option_init_struct):
376 Use section anchors by default.
377 * config/s390/s390.c (s390_check_symref_alignment): Handle
379 (TARGET_MAX_ANCHOR_OFFSET): Use short displacement.
380 * output.h (assemble_align): Pass `align' as unsigned int, so
381 that the value 0x80000000, which corresponds to `aligned(1 <<
382 28)', would pass the `align > BITS_PER_UNIT' check.
383 * varasm.c (assemble_align): Likewise.
385 2018-12-03 Julian Brown <julian@codesourcery.com>
387 * tree-pretty-print.c (dump_omp_clause): Make default case
390 2018-12-03 Richard Biener <rguenther@suse.de>
392 PR tree-optimization/88301
393 * tree-vrp.c (register_edge_assert_for_2): Handle conversions
394 that do not change the value by registering the same assert
397 2018-12-03 Richard Biener <rguenther@suse.de>
399 * tree-vect-loop.c (vect_transform_loop): Properly compute
400 upper bound for the epilogue when doing epilogue vectorization.
402 2018-12-03 Ilya Leoshkevich <iii@linux.ibm.com>
405 * cfgcleanup.c (class pass_postreload_jump): New pass.
406 (pass_postreload_jump::execute): Likewise.
407 (make_pass_postreload_jump): Likewise.
408 * passes.def: Add pass_postreload_jump before
410 * tree-pass.h (make_pass_postreload_jump): New pass.
412 2018-12-02 Uros Bizjak <ubizjak@gmail.com>
414 * config/i386/i386.c (inline_memory_move_cost): Correctly
415 compute number of TFmode GPR moves.
417 2018-12-02 Jakub Jelinek <jakub@redhat.com>
420 * config/i386/sse.md (*vec_concatv4sf_0, *vec_concatv4si_0): New insns.
422 2018-12-02 Jeff Law <law@redhat.com>
424 * config/h8300/h8300.md (call, call_value): Drop mode from
425 operand containing number of bytes pushed.
426 (call_insn, call_value_insn): Use mode iterator for number
427 of bytes pushed operand. Make pattern name vary appropriately.
429 2018-12-02 Jan Hubicka <jh@suse.cz>
431 * bb-reorer.c (find_rarely_executed_basic_blocks_and_crossin_edges):
432 trust precise profiles.
434 2018-12-02 Jakub Jelinek <jakub@redhat.com>
436 * tree-nested.c (convert_nonlocal_omp_clauses,
437 convert_local_omp_clauses): Handle OMP_CLAUSE_IN_REDUCTION,
438 OMP_CLAUSE_TASK_REDUCTION and OMP_CLAUSE__SIMT_ clauses.
439 (convert_nonlocal_reference_stmt, convert_local_reference_stmt):
440 Convert clauses for GIMPLE_OMP_TASKGROUP.
442 * omp-low.c (check_omp_nesting_restrictions): Allow cancel or
443 cancellation point with taskgroup clause inside of taskloop. Consider
444 a taskloop construct without nogroup clause as implicit taskgroup for
445 diagnostics if cancel/cancellation point with taskgroup clause is
446 closely nested inside of taskgroup region.
449 * asan.c (asan_clear_shadow): Move assert that len is multiple of 4
450 to the start of the function.
451 (asan_emit_stack_protection): When emitting clearing sequence for
452 epilogue, align offset down to ASAN_RED_ZONE_SIZE granularity,
453 add last_size_aligned which is last_size padded to multiples of
454 ASAN_RED_ZONE_SIZE and emit asan_clear_shadow always on 4 byte
457 2018-12-01 Jakub Jelinek <jakub@redhat.com>
460 * asan.c (asan_redzone_buffer::flush_redzone_payload): Fix up
461 an off-by-one for BYTES_BIG_ENDIAN.
464 * combine.c (find_split_point): For invalid memory address
465 nonobj + obj + const, if reg + obj + const is valid addressing
466 mode, split at nonobj. Use if rather than else if for the
467 fallback. Comment fixes.
469 2018-11-30 Indu Bhagat <indu.bhagat@oracle.com>
471 * coverage.c (get_coverage_counts): Use from_function_decl for precise
473 * profile-count.c (profile_count::dump): Add handling for precise
475 * profile.c (compute_branch_probabilities): Rely on exec_counts instead
476 of profile_info to set profile_status of function.
477 (branch_prob): Do not set profile_status of function based on
478 profile_info. Done above based on exec_counts.
480 2018-11-30 Jakub Jelinek <jakub@redhat.com>
482 PR tree-optimization/88274
483 * tree-ssa-reassoc.c (optimize_range_tests_xor,
484 optimize_range_tests_diff): If type has smaller precision than
485 corresponding mode or if it has non-standard min/max, compute
486 everything in a standard type for the precision.
489 * params.def (PARAM_LOGICAL_OP_NON_SHORT_CIRCUIT): New param.
490 * tree-ssa-ifcombine.c (ifcombine_ifandif): If
491 --param logical-op-non-short-circuit is present, override
492 LOGICAL_OP_NON_SHORT_CIRCUIT value from the param.
493 * fold-const.c (fold_range_test, fold_truth_andor): Likewise.
495 2018-11-30 Jeff Law <law@redhat.com>
497 * optabs.c (expand_binop): Use "machine_mode" rather than
498 "enum machine mode" in most recent change.
500 2018-11-30 Wilco Dijkstra <wdijkstr@arm.com>
503 * builtins.c (expand_builtin_longjmp): Use a temporary when restoring
505 (expand_builtin_nonlocal_goto): Likewise.
507 2018-11-30 David Malcolm <dmalcolm@redhat.com>
509 * diagnostic-core.h (emit_diagnostic): New decl.
510 * diagnostic.c (emit_diagnostic): New overload, taking a
513 2018-11-30 David Malcolm <dmalcolm@redhat.com>
515 * pretty-print.c (class selftest::test_pretty_printer): New
516 subclass of pretty_printer.
517 (selftest::test_prefixes_and_wrapping): New test.
518 (selftest::pretty_print_c_tests): Call it.
520 2018-11-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
522 Minimize clone counter memory usage in create_virtual_clone.
523 * cgraph.h (clone_function_name): Add a variant that takes a
525 * cgraph.h (cgraph_node::create_virtual_clone): Add a new
526 argument: num_suffix.
527 * cgraphclones.c (cgraph_node::create_virtual_clone): Pass
528 num_suffix to clone_function_name.
529 (clone_function_name): Add a variant that takes a tree decl.
530 * ipa-cp.c (create_specialized_node): Keep track of clone
531 counters in clone_num_suffixes hash map.
532 (ipcp_driver): Free the counter hash map.
533 * ipa-hsa.c (process_hsa_functions): Creates at most one hsa
536 2018-11-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
537 Make function assembly more independent.
539 This is achieved by having clone_function_name assign unique clone
540 numbers for each function independently.
542 * cgraphclones.c: Replaced clone_fn_id_num with clone_fn_ids;
544 (clone_function_name_numbered): Use clone_fn_ids.
546 2018-11-30 Vladimir Makarov <vmakarov@redhat.com>
548 PR rtl-optimization/88179
549 * lra-constraints.c (address_eliminator::address_eliminator):
550 Don't eleminate regs in illegitimate address.
552 2018-11-30 David Malcolm <dmalcolm@redhat.com>
554 PR preprocessor/88257
555 * input.c (get_substring_ranges_for_loc): Fix indentation.
556 Bulletproof against getting a different files back from
557 linemap_resolve_location and expand_location_to_spelling_point.
559 2018-11-30 Alexander Monakov <amonakov@ispras.ru>
561 PR gcov-profile/88279
562 * tree-profile.c (init_ic_make_global_vars): Call
563 decl_default_tls_model for variable and not it's type.
565 2018-11-30 Sam Tebbs <sam.tebbs@arm.com>
567 * config/arm/arm.h (TARGET_DOTPROD): Add TARGET_VFP5 constraint.
569 2018-11-30 Martin Liska <mliska@suse.cz>
572 * asan.c (asan_shadow_cst): Remove, partially transform
573 into flush_redzone_payload.
574 (RZ_BUFFER_SIZE): New.
575 (struct asan_redzone_buffer): New.
576 (asan_redzone_buffer::emit_redzone_byte): Likewise.
577 (asan_redzone_buffer::flush_redzone_payload): Likewise.
578 (asan_redzone_buffer::flush_if_full): Likewise.
579 (asan_emit_stack_protection): Use asan_redzone_buffer class
580 that is responsible for proper aligned stores and flushing
581 of shadow memory payload.
582 * asan.h (ASAN_MIN_RED_ZONE_SIZE): New.
583 (asan_var_and_redzone_size): Likewise.
584 * cfgexpand.c (expand_stack_vars): Use smaller alignment
585 (ASAN_MIN_RED_ZONE_SIZE) in order to make shadow memory
586 for automatic variables more compact.
588 2018-11-30 Alan Modra <amodra@gmail.com>
590 * config/rs6000/predicates.md (easy_fp_constant): Avoid long
591 dependent insn sequences.
592 * config/rs6000/rs6000.c (num_insns_constant): Support long
594 * config/rs6000/rs6000.md (mov<mode>_softfloat <FMOVE128>): Adjust
597 2018-11-30 Alan Modra <amodra@gmail.com>
599 * config/rs6000/rs6000.c (num_insns_constant_gpr): Renamed from
600 num_insns_constant_wide. Make static. Revise comment.
601 (num_insns_constant_multi): New function.
602 (num_insns_constant): Formatting. Correct CONST_WIDE_INT
603 calculation. Simplify and extract code common to both
604 CONST_INT and CONST_DOUBLE. Add gcc_unreachable for unhandled
606 * config/rs6000/rs6000-protos.h (num_insns_const_wide): Delete.
608 2018-11-30 Richard Biener <rguenther@suse.de>
610 PR tree-optimization/88274
611 * tree-vrp.c (ranges_from_anti_range): Fix handling of
614 2018-11-29 Andi Kleen <ak@linux.intel.com>
616 * config/i386/i386.c (current_fentry_section): Add.
617 (x86_function_profiler): Handle fentry section.
618 (ix86_attribute_table): Add fentry section.
619 * config/i386/i386.opt: Add -mfentry-section.
620 * doc/extend.texi: Document fentry_section attribute.
621 * doc/invoke.texi: Document -mfentry-section.
623 2018-11-29 Andi Kleen <ak@linux.intel.com>
625 * config/i386/i386.c (x86_print_call_or_nop): Handle nop name.
626 (current_fentry_name): Add.
627 (x86_function_profiler): Handle fentry_name attribute.
628 (ix86_handle_fentry_name): Add.
629 (ix86_attribute_table): Add fentry_name.
630 * config/i386/i386.opt: Add -mfentry-name
631 * doc/extend.texi: Document fentry_name.
632 * doc/invoke.texi: Document minstrument-return.
634 2018-11-29 Andi Kleen <ak@linux.intel.com>
636 * config/i386/i386-opts.h (enum instrument_return): Add.
637 * config/i386/i386.c (output_return_instrumentation): Add.
638 (ix86_output_function_return): Call output_return_instrumentation.
639 (ix86_output_call_insn): Call output_return_instrumentation.
640 * config/i386/i386.opt: Add -minstrument-return=.
641 * doc/invoke.texi (-minstrument-return): Document.
643 2018-11-29 Eric Botcazou <ebotcazou@adacore.com>
646 * config/sparc/sparc-modes.def: Minor tweak.
647 * config/sparc/sparc.c: Minor reordering.
648 (sparc_pass_by_reference): Move around.
649 (traverse_record_type): Change offset from HOST_WIDE_INT to int.
650 (classify_registers): Likewise for bitpos.
651 (function_arg_slotno): Remove dead test and tweak comments.
652 <MODE_RANDOM>: Remove useless assertion and test whether the
653 parameter is named in order to pass it in FP registers. Return
654 the regno for floating-point vector types.
655 (compute_int_layout): Change bitpos from HOST_WIDE_INT to int.
656 (compute_fp_layout): Likewise.
657 (count_registers): Likewise.
658 (assign_int_registers): Likewise.
659 (assign_fp_registers): Likewise.
660 (assign_registers): Likewise.
661 (function_arg_record_value): Change size from HOST_WIDE_INT to int
662 and use CEIL_NWORDS to compute the number of registers.
663 (function_arg_union_value): Minor tweaks.
664 (function_arg_vector_value): Add slotno and named parameters, use
665 CEIL_NWORDS to compute the number of registers.
666 (sparc_function_arg_1): Rework handling of vector types. Change
667 size from HOST_WIDE_INT to int.
668 (sparc_arg_partial_bytes): Rework handling of 32-bit ABI and deal
669 with vector types for the 64-bt ABI.
670 (sparc_function_arg_advance): Likewise.
671 (sparc_return_in_memory): Add reference to -fpcc-struct-return.
672 (sparc_struct_value_rtx): Return NULL_RTX instead of 0.
673 (sparc_function_value_1): Rework handling of vector types. Change
674 size from HOST_WIDE_INT to int.
676 2018-11-29 Jakub Jelinek <jakub@redhat.com>
679 * config/i386/sse.md (*<sse>_movmsk<ssemodesuffix><avxsizesuffix>_lt,
680 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_lt,
681 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_shift,
682 *<sse>_movmsk<ssemodesuffix><avxsizesuffix>_zext_shift,
683 *<sse2_avx2>_pmovmskb_lt, *<sse2_avx2>_pmovmskb_zext_lt): New
684 define_insn_and_split patterns.
688 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
689 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
690 *<sse4_1_avx2>_pblendvb_lt): Change define_insn into
691 define_insn_and_split.
693 2018-11-29 Peter Bergner <bergner@linux.ibm.com>
696 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
697 -mabi=ieeelongdouble without both -mpopcntd and -mvsx.
699 2018-11-29 Uros Bizjak <ubizjak@gmail.com>
701 * config/i386/i386.c (inline_memory_move_cost):
702 Check "in" for 2 in MMX_CLASS_P case.
703 * config/i386/mmx.md (*mov<mode>_internal): Correct
704 TARGET_INTER_UNIT_MOVES_FROM_VEC and TARGET_INTER_UNIT_MOVES_TO_VEC
705 alternatives in preferred_for_speed attribute calculation.
707 2018-11-29 Martin Sebor <msebor@redhat.com>
711 * doc/extend.texi (attribute constructor): Clarify.
713 2018-11-29 Martin Liska <mliska@suse.cz>
716 * builtins.c (expand_movstr): Fix thinko introduced
717 when switching to the new enum.
719 2018-11-29 qing zhao <qing.zhao@oracle.com>
721 * cif-code.def (EXTERN_LIVE_ONLY_STATIC): New CIF code.
722 * common.opt: Add -flive-patching flag.
723 * doc/invoke.texi: Document -flive-patching.
724 * flag-types.h (enum live_patching_level): New enum.
725 * ipa-inline.c (can_inline_edge_p): Disable external functions from
726 inlining when flag_live_patching is LIVE_PATCHING_INLINE_ONLY_STATIC.
727 * opts.c (control_options_for_live_patching): New function.
728 (finish_options): Make flag_live_patching incompatible with flag_lto.
729 Control IPA optimizations based on different levels of
732 2018-11-29 Giuliano Belinassi <giuliano.belinassi@usp.br>
734 * match.pd (sinh (atanh (x))): New simplification rules.
735 (cosh (atanh (x))): Likewise.
737 2018-11-29 Jakub Jelinek <jakub@redhat.com>
740 * config/i386/sse.md (ssebytemode): Add V16SI, V8SI and V4SI entries.
741 (ssefltmodesuffix, ssefltvecmode): New define_mode_attrs.
742 (*<sse4_1>_blendv<ssemodesuffix><avxsizesuffix>_lt,
743 *<sse4_1>_blendv<ssefltmodesuffix><avxsizesuffix>_ltint,
744 *<sse4_1_avx2>_pblendvb_lt): New define_insns.
747 * tree.h (build_uniform_cst, uniform_integer_cst_p): Declare.
748 * tree.c (build_uniform_cst, uniform_integer_cst_p): New functions.
749 * match.pd (define_predicates): Add uniform_integer_cst_p.
750 (cmp @0 INTEGER_CST@1, cmp (convert?@2 @0) INTEGER_CST@1): Adjust
751 so that it works also for vector comparisons with uniform constants
752 with INTEGER_CST element.
755 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): For
756 vec_add and vec_sub builtins, perform PLUS_EXPR or MINUS_EXPR
757 in unsigned_type_for instead of vector integral type where overflow
760 2018-11-29 Michael Ploujnikov <michael.ploujnikov@oracle.com>
762 There can be at most one .resolver clone per function
763 * config/rs6000/rs6000.c (make_resolver_func): Generate
764 resolver symbol with clone_function_name instead of
765 clone_function_name_numbered.
767 2018-11-29 Richard Biener <rguenther@suse.de>
769 PR tree-optimization/88243
770 * tree-vect-patterns.c (vect_mark_pattern_stmts): Set the def
771 type of all pattern-sequence stmts to vect_internal_def.
773 2018-11-29 Andre Vieira <andre.simoesdiasvieira@arm.com>
776 * config/arm/arm-cpus.in (armv7-r): Add FP16conv configurations.
777 (cortex-r7, cortex-r8): Update default and add new configuration.
778 * doc/invoke.texi (armv7-r): Add two new vfp options.
779 (nofp.dp): Add cortex-r7 and cortex-r8 to the list of targets that
782 2018-11-29 Alan Modra <amodra@gmail.com>
784 * config/rs6000/rs6000.c (rs6000_emit_move): Disable long
785 double split for targets other than Darwin.
787 2018-11-29 Alan Modra <amodra@gmail.com>
789 * config/rs6000/rs6000.md (movdi_internal64): Correct lengths.
791 2018-11-29 Alan Modra <amodra@gmail.com>
793 * config/rs6000/predicates.md (easy_fp_constant): Remove code
794 dealing with integers in const_double. Assert on unexpected
795 modes. Delete superfluous ABI_V4 test.
796 * config/rs6000/rs6000.c (rs6000_emit_move): Comment fixes.
797 Don't call easy_fp_constant for Pmode.
798 (rs6000_hash_constant): Remove code dealing with integers in
800 (rs6000_legitimate_constant_p): Likewise.
801 (output_toc): Formatting, use CONST_DOUBLE_P.
803 2018-11-29 Alan Modra <amodra@gmail.com>
805 * config/rs6000/constraints.md (G, H): Comment on purpose of
806 constraint. Correct mode comments and "G" mode test.
807 * config/rs6000/rs6000.md (movdi_internal32): Remove "GH" from
808 alternative handling "F". Add length attr. Formatting.
809 (mov<mode>_softfloat <FMOVE128>): Delete "GH" from
810 alternative, and move "F" to separate alternative. Correct
813 2018-11-29 Alan Modra <amodra@gmail.com>
815 * config/rs6000/predicates.md (easy_vector_constant): Don't call
818 2018-11-29 Alan Modra <amodra@gmail.com>
820 * config.in (HAVE_AS_PLTSEQ): Add.
821 * config/rs6000/predicates.md (indirect_call_operand): New.
822 * config/rs6000/rs6000-protos.h (rs6000_pltseq_template),
823 (rs6000_sibcall_sysv): Declare.
824 * config/rs6000/rs6000.c (init_cumulative_args): Set cookie
825 CALL_LONG for -fno-plt.
826 (print_operand <T, z, 0>): Handle UNSPEC_PLTSEQ.
827 (rs6000_indirect_call_template_1): Emit .reloc directives for
829 (rs6000_pltseq_template): New function.
830 (rs6000_longcall_ref): Add arg parameter. Use PLT16 insns if
831 relocs supported by assembler. Move SYMBOL_REF test to callers.
832 (rs6000_call_aix): Adjust rs6000_longcall_ref call. Package
833 insns in UNSPEC_PLTSEQ, preserving original func_desc.
834 (rs6000_call_sysv): Likewise.
835 (rs6000_sibcall_sysv): New function.
836 * config/rs6000/rs6000.h (HAVE_AS_PLTSEQ): Provide default.
837 * config/rs6000/rs6000.md (UNSPEC_PLTSEQ, UNSPEC_PLT16_HA,
838 UNSPEC_PLT16_LO): New.
839 (pltseq_tocsave, pltseq_plt16_ha, pltseq_plt16_lo, pltseq_mtctr): New.
840 (call_indirect_nonlocal_sysv): Don't differentiate zero from non-zero
841 cookie in constraints. Test explicitly for flags in length attr.
842 Handle unspec operand 1.
843 (call_value_indirect_nonlocal_sysv): Likewise.
844 (call_indirect_aix, call_value_indirect_aix): Handle unspec operand 1.
845 (call_indirect_elfv2, call_value_indirect_elfv2): Likewise.
846 (sibcall, sibcall_value): Use rs6000_sibcall_sysv.
847 (sibcall_indirect_nonlocal_sysv): New pattern.
848 (sibcall_value_indirect_nonlocal_sysv): Likewise.
849 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv): Remove indirect
851 * configure.ac: Check for gas plt sequence marker support.
852 * configure: Regenerate.
854 2018-11-29 Alan Modra <amodra@gmail.com>
856 * config/rs6000/predicates.md (unspec_tls): New.
857 * config/rs6000/rs6000-protos.h (rs6000_call_template),
858 (rs6000_sibcall_template): Update prototype.
859 (rs6000_longcall_ref): Delete.
860 (rs6000_call_sysv): Declare.
861 * config/rs6000/rs6000.c (edit_tls_call_insn): New function.
862 (global_tlsarg): New variable.
863 (rs6000_legitimize_tls_address): Rewrite __tls_get_addr call
865 (print_operand): Extract UNSPEC_TLSGD address operand.
866 (rs6000_call_template, rs6000_sibcall_template): Remove arg
867 parameter, extract from second call operand instead.
868 (rs6000_longcall_ref): Make static, localize vars.
869 (rs6000_call_aix): Rename parameter to reflect new usage. Take
870 tlsarg from global_tlsarg. Don't create unused rtl or nop insns.
871 (rs6000_sibcall_aix): Rename parameter to reflect new usage. Take
872 tlsarg from global_tlsarg.
873 (rs6000_call_sysv): New function.
874 * config/rs6000/rs6000.md: Adjust rs6000_call_template and
875 rs6000_sibcall_template throughout.
876 (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix, tls_gd_call_sysv): Delete.
877 (tls_ld_aix, tls_ld_sysv, tls_ld_call_aix, tls_ld_call_sysv): Delete.
878 (tls_gdld_nomark): New insn.
879 (tls_gd): Swap operand order. Simplify mode selection.
880 (tls_gd_high, tls_gd_low): Swap operand order.
881 (tls_ld): Remove const_int 0 vector element from UNSPEC_TLSLD.
882 Simplify mode selection.
883 (tls_ld_high, tls_ld_low): Similarly adjust UNSPEC_TLSLD.
884 (call, call_value): Don't assert for second call operand.
885 Use rs6000_call_sysv.
887 2018-11-29 Alan Modra <amodra@gmail.com>
889 * config/rs6000/darwin.md (call_indirect_nonlocal_darwin64),
890 (call_nonlocal_darwin64, call_value_indirect_nonlocal_darwin64),
891 (call_value_nonlocal_darwin64): Remove constraints from second call
892 arg, the rounded_stack_size_rtx arg.
893 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix),
894 (tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv, tls_ld_call_aix),
895 (tls_ld_call_sysv, call_local32, call_local64, call_value_local32),
896 (call_value_local64, call_indirect_nonlocal_sysv),
897 (call_nonlocal_sysv, call_nonlocal_sysv_secure),
898 (call_value_indirect_nonlocal_sysv, call_value_nonlocal_sysv),
899 (call_value_nonlocal_sysv_secure, call_local_aix),
900 (call_value_local_aix, call_nonlocal_aix, call_value_nonlocal_aix),
901 (call_indirect_aix, call_value_indirect_aix, call_indirect_elfv2),
902 (call_value_indirect_elfv2, sibcall_local32, sibcall_local64),
903 (sibcall_value_local32, sibcall_value_local64, sibcall_aix),
904 (sibcall_value_aix): Likewise.
906 2018-11-29 Alan Modra <amodra@gmail.com>
908 * config/rs6000/rs6000.md (TLSmode): Delete mode iterator. Replace
909 with P throughout except for call mems which should use SI.
910 (tls_abi_suffix, tls_sysv_suffix, tls_insn_suffix): Delete mode
911 attributes. Replace with bits, mode and ptrload respectively.
913 2018-11-29 Alan Modra <amodra@gmail.com>
915 * config/rs6000/rs6000-protos.h (rs6000_indirect_call_template),
916 (rs6000_indirect_sibcall_template): Declare.
917 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
918 (rs6000_indirect_call_template, rs6000_indirect_sibcall_template):
920 * config/rs6000/rs6000.md (call_indirect_nonlocal_sysv),
921 (call_value_indirect_nonlocal_sysv, sibcall_nonlocal_sysv),
922 (call_indirect_aix, call_value_indirect_aix): Use
923 rs6000_indirect_call_template and rs6000_indirect_sibcall_template.
924 call_indirect_elfv2, call_value_indirect_elfv2): Likewise, and
925 handle both speculation and non-speculation cases.
926 (call_indirect_aix_nospec, call_value_indirect_aix_nospec): Delete.
927 (call_indirect_elfv2_nospec, call_value_indirect_elfv2_nospec): Delete.
929 2018-11-29 Alan Modra <amodra@gmail.com>
931 * config/rs6000/rs6000-protos.h (rs6000_call_template): Declare.
932 (rs6000_sibcall_template): Declare.
933 (macho_call_template): Rename from output_call.
934 * config/rs6000/rs6000.c (rs6000_call_template_1): New function.
935 (rs6000_call_template, rs6000_sibcall_template): Likewise.
936 (macho_call_template): Rename from output_call.
937 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv),
938 (tls_gd_call_aix, tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv),
939 (tls_ld_call_aix, tls_ld_call_sysv, call_nonlocal_sysv),
940 (call_nonlocal_sysv_secure, call_value_nonlocal_sysv),
941 (call_value_nonlocal_sysv_secure, call_nonlocal_aix),
942 (call_value_nonlocal_aix): Use rs6000_call_template and update
943 occurrences of output_call to macho_call_template.
944 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv, sibcall_aix),
945 (sibcall_value_aix): Use rs6000_sibcall_template.
947 2018-11-28 Aaron Sawdey <acsawdey@linux.ibm.com>
949 * config/rs6000/rs6000-string.c (expand_block_clear): Change how
950 we determine if unaligned vsx is ok.
952 2018-11-28 Jan Hubicka <jh@suse.cz>
954 * predict.c (determine_unlikely_bbs): Forward declare; also determine
955 edges that are always known to be taken because it is only likely
956 edge out of given BB.
957 (tree_estimate_probability): Call before profile guessing to get bit
960 2018-11-28 Jan Hubicka <jh@suse.cz>
962 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Handle
963 profile_probability::always better.
965 2018-11-28 Jan Hubicka <jh@suse.cz>
967 * profile-count.h (profile_count::split): Give better result when
968 splitting profile_probability::always.
970 2018-11-28 Vladimir Makarov <vmakarov@redhat.com>
973 * ira-costs.c (scan_one_insn): Process subregs when updating costs
974 for pseudos and allocnos from insn.
976 2018-11-28 David Edelsohn <dje.gcc@gmail.com>
978 * config/rs6000/aix72.h: Update to match aix71.h changes.
980 2018-11-28 Jeff Law <law@redhat.com>
982 * config/h8300/constraints.md: Add "C" constraint for call insns.
983 * config/h8300/h8300.md (call, call_value): Turn into a define_expand
984 and define_insn pair. Move invalid call targets into a register in
985 the expander and fix constraints in the matching pattern.
986 * config/h8300/predicates.md (call_expander_operand): Renamed from
987 call_insn_operand. Reject things we shouldn't be trying to handle.
988 (call_insn_operand): New predicate for use by the call/call_value
990 (small_call_insn_operand): Update appropriately.
992 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
994 * config/aarch64/aarch64.c (aarch64_process_target_attr): Replace
995 calls to strtok with strtok_r.
997 2018-11-28 Richard Biener <rguenther@suse.de>
999 PR tree-optimization/88223
1000 * tree-ssa-sccvn.c (vn_reference_lookup_3): When skipping
1001 over a stored-same value may-alias store make sure to consider
1002 partial overlaps which are valid when TBAA reasonings do not
1003 apply and byte-granular overlaps are possible at all.
1005 2018-11-28 Richard Biener <rguenther@suse.de>
1007 PR tree-optimization/88217
1008 * vr-values.c (vr_values::extract_range_from_phi_node): Make
1009 sure to handle results > +INF and < -INF correctly when
1010 trying to drop down to +INF - 1 or -INF + 1.
1012 2018-11-28 Alan Modra <amodra@gmail.com>
1014 * xcoffout.c (do_block): Signed/unsigned warning fix.
1016 2018-11-28 Richard Biener <rguenther@suse.de>
1018 PR tree-optimization/79351
1019 * tree-ssa-sccvn.c (vn_reference_lookup_3): For assignments from
1020 empty CONSTRUCTORs ensure the store is at a constant position.
1022 2018-11-28 Richard Biener <rguenther@suse.de>
1024 PR tree-optimization/88229
1025 * tree-ssa.c (non_rewritable_mem_ref_base): Check DECL_SIZE_UNIT
1026 is an INTEGER_CST before accessing it so.
1028 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
1030 * doc/invoke.texi (-mtune=): Remove obsolete CPU names.
1032 2018-11-28 Jakub Jelinek <jakub@redhat.com>
1035 * config/i386/i386.c (ix86_expand_sse_movcc): Handle DFmode and
1036 SFmode using sse4_1_blendvs[sd] with TARGET_SSE4_1. Formatting fixes.
1037 * config/i386/sse.md (sse4_1_blendv<ssemodesuffix>): New pattern.
1039 2018-11-27 Martin Liska <mliska@suse.cz>
1041 * gcov.c (generate_results): Append current_working_directory
1044 2018-11-27 Mike Gulick <mgulick@mathworks.com>
1046 PR preprocessor/83173
1047 * input.c (dump_location_info): Dump reason and included_from
1048 fields from line_map_ordinary struct. Fix indentation when
1049 location > 5 digits.
1050 * diagnostic-show-locus.c (num_digits, num_digits): Move to
1051 diagnostic.c to allow it to be utilized by input.c.
1052 * diagnostic.c (num_digits, selftest::test_num_digits): Moved
1054 (selftest::diagnostic_c_tests): Run selftest::test_num_digits.
1055 * diagnostic.h (num_digits): Add extern definition.
1057 2018-11-27 Fredrik Noring <noring@nocrew.org>
1059 * config/mips/mips.c (mips_reorg_process_insns)
1060 (mips_option_override): Handle `-mfix-r5900'.
1061 * config/mips/mips.h (ASM_SPEC): Add `mfix-r5900' and
1063 * config/mips/mips.opt (mfix-r5900): New option.
1064 * doc/invoke.texi: Document the `r5900' processor name, and
1065 `-mfix-r5900' and `-mno-fix-r5900' options.
1067 2018-11-27 Jakub Jelinek <jakub@redhat.com>
1070 * config/rs6000/rs6000.c (print_operand) <case 'D'>: Use
1071 output_operand_lossage instead of gcc_assert.
1072 <case 't'>: Likewise.
1073 <case 'z'>: Likewise.
1074 <case 'V'>: Use output_operand_lossage instead of gcc_unreachable.
1076 2018-11-27 Jeff Law <law@redhat.com>
1078 * config/riscv/riscv (riscv_block_mvoe_straight): Use RETURN_BEGIN
1079 in call to move_by_pieces.
1081 * config/sh/sh-mem.c (expand_block_move): Use RETURN_BEGIN in call
1084 * config/lm32/lm32.c (lm32_block_move_inline): Use RETURN_BEGIN in
1085 call to move_by_pieces.
1087 * config/mips/mips.c (mips_block_move_straight): Use RETURN_BEGIN
1088 in call to move_by_pieces.
1090 * config/microblaze/microblaze.c (microblaze_block_move_straight): Use
1091 RETURN_BEGIN in call to move_by_pieces.
1092 (microblaze_expand_block_move): Likewise.
1094 2018-11-27 Tamar Christina <tamar.christina@arm.com>
1096 * doc/md.texi (xorsign): Document it.
1098 2018-11-27 Martin Liska <mliska@suse.cz>
1100 * config/gnu-user.h (TARGET_F951_OPTIONS): New.
1101 * gcc.c (find_fortran_preinclude_file): New function
1102 to handle Fortran pre-include.
1104 2018-11-27 Martin Liska <mliska@suse.cz>
1106 * asan.c (asan_emit_stack_protection): Use new enum values
1107 instead of int constants.
1108 * builtins.c (expand_builtin_memory_copy_args): Replace int
1109 type with memop_ret enum type.
1110 (expand_builtin_mempcpy_args): Likewise.
1111 (expand_builtin_memcpy): Use new enum values
1112 instead of int constants. Likewise.
1113 (expand_builtin_mempcpy): Likewise.
1114 (expand_movstr): Likewise.
1115 (expand_builtin_strcpy_args): Likewise.
1116 (expand_builtin_stpcpy_1): Likewise.
1117 (expand_builtin_strncpy): Likewise.
1118 (expand_builtin_memset_args): Likewise.
1119 * expr.c (move_by_pieces_d::finish_endp): Rename to ...
1120 (move_by_pieces_d::finish_retmode): ... this.
1121 (move_by_pieces): Change last argument type to memop_ret.
1122 (store_by_pieces): Use new enum values
1123 instead of int constants.
1124 (emit_block_move_hints): Likewise.
1125 (emit_push_insn): Likewise.
1126 (store_expr): Likewise.
1127 * expr.h (store_by_pieces): Change int to newly added enum
1129 * rtl.h (enum memop_ret): Define.
1130 (move_by_pieces): Use the enum type.
1132 2018-11-27 Alan Modra <amodra@gmail.com>
1134 * config/rs6000/aix71.h (ASM_SPEC): Don't select default -maix64
1136 (ASM_CPU_SPEC): Do so here. Rewrite using if .. else if .. specs
1137 form. Error on missing -mcpu case.
1138 * config/rs6000/driver-rs6000.c (asm_names <_AIX>): Update NULL case.
1139 (asm_names <!_AIX>): Add missing cpus. Update NULL case. Apply
1140 PR63177 fix for -mcpu=power8 and -mcpu=powerpc64le.
1141 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Rewrite using if ..
1142 else if .. specs form. Error on missing -mcpu case. Don't output
1143 duplicate -maltivec. Apply PR63177 fix for -mcpu=powerpc64le.
1145 2018-11-26 David Malcolm <dmalcolm@redhat.com>
1147 * dump-context.h (dump_context::dump_loc): Convert 1st param from
1148 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
1149 const dump_location_t & to const dump_user_location_t &.
1150 (dump_context::dump_loc_immediate): Convert 2nd param from
1151 const dump_location_t & to const dump_user_location_t &.
1152 (dump_context::dump_gimple_stmt): Convert 1st param from
1153 dump_flags_t to const dump_metadata_t &.
1154 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
1155 2nd param from const dump_location_t & to
1156 const dump_user_location_t &.
1157 (dump_context::dump_gimple_expr): Convert 1st param from
1158 dump_flags_t to const dump_metadata_t &.
1159 (dump_context::dump_gimple_expr_loc): Likewise; convert
1160 2nd param from const dump_location_t & to
1161 const dump_user_location_t &.
1162 (dump_context::dump_generic_expr): Convert 1st param from
1163 dump_flags_t to const dump_metadata_t &.
1164 (dump_context::dump_generic_expr_loc): Likewise; convert
1165 2nd param from const dump_location_t & to
1166 const dump_user_location_t &.
1167 (dump_context::dump_printf_va): Convert 1st param from
1168 dump_flags_t to const dump_metadata_t &.
1169 (dump_context::dump_printf_loc_va): Likewise; convert
1170 2nd param from const dump_location_t & to
1171 const dump_user_location_t &.
1172 (dump_context::dump_dec): Convert 1st param from
1173 dump_flags_t to const dump_metadata_t &.
1174 (dump_context::dump_symtab_node): Likewise.
1175 (dump_context::begin_scope): Split out 2nd param into
1176 user and impl locations.
1177 (dump_context::ensure_pending_optinfo): Add metadata param.
1178 (dump_context::begin_next_optinfo): Replace dump_location_t param
1179 with metadata and user location.
1180 * dumpfile.c (dump_context::dump_loc): Convert 1st param from
1181 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
1182 const dump_location_t & to const dump_user_location_t &.
1183 (dump_context::dump_loc_immediate): Convert 2nd param from
1184 const dump_location_t & to const dump_user_location_t &.
1185 (dump_context::dump_gimple_stmt): Convert 1st param from
1186 dump_flags_t to const dump_metadata_t &.
1187 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
1188 2nd param from const dump_location_t & to
1189 const dump_user_location_t &.
1190 (dump_context::dump_gimple_expr): Convert 1st param from
1191 dump_flags_t to const dump_metadata_t &.
1192 (dump_context::dump_gimple_expr_loc): Likewise; convert
1193 2nd param from const dump_location_t & to
1194 const dump_user_location_t &.
1195 (dump_context::dump_generic_expr): Convert 1st param from
1196 dump_flags_t to const dump_metadata_t &.
1197 (dump_context::dump_generic_expr_loc): Likewise; convert
1198 2nd param from const dump_location_t & to
1199 const dump_user_location_t &.
1200 (dump_context::dump_printf_va): Convert 1st param from
1201 dump_flags_t to const dump_metadata_t &.
1202 (dump_context::dump_printf_loc_va): Likewise; convert
1203 2nd param from const dump_location_t & to
1204 const dump_user_location_t &.
1205 (dump_context::dump_dec): Convert 1st param from
1206 dump_flags_t to const dump_metadata_t &.
1207 (dump_context::dump_symtab_node): Likewise.
1208 (dump_context::begin_scope): Split out 2nd param into
1209 user and impl locations.
1210 (dump_context::ensure_pending_optinfo): Add metadata param.
1211 (dump_context::begin_next_optinfo): Replace dump_location_t param
1212 with metadata and user location.
1213 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
1214 const dump_metadata_t &.
1215 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
1216 const dump_location_t & to const dump_user_location_t &.
1217 (dump_gimple_expr): Convert 1st param from dump_flags_t to
1218 const dump_metadata_t &.
1219 (dump_gimple_expr_loc): Likewise; convert 2nd param from
1220 const dump_location_t & to const dump_user_location_t &.
1221 (dump_generic_expr): Convert 1st param from dump_flags_t to
1222 const dump_metadata_t &.
1223 (dump_generic_expr_loc): Likewise; convert 2nd param from
1224 const dump_location_t & to const dump_user_location_t &.
1225 (dump_printf): Convert 1st param from dump_flags_t to
1226 const dump_metadata_t &.
1227 (dump_printf_loc): Likewise; convert 2nd param from
1228 const dump_location_t & to const dump_user_location_t &.
1229 (dump_dec): Convert 1st param from dump_flags_t to
1230 const dump_metadata_t &.
1231 (dump_symtab_node): Likewise.
1232 (dump_begin_scope): Split out 2nd param into user and impl
1234 (selftest::assert_impl_location_eq): New function.
1235 (ASSERT_IMPL_LOCATION_EQ): New macro.
1236 (selftest::test_impl_location): Update to use
1237 ASSERT_IMPL_LOCATION_EQ.
1238 (selftest::test_capture_of_dump_calls): Convert "loc" to
1239 dump_user_location_t. Add ASSERT_IMPL_LOCATION_EQ throughout,
1240 verifying line numbers of dump emissions.
1241 * dumpfile.h (class dump_metadata_t): New class.
1242 (dump_printf): Convert 1st param from dump_flags_t to
1243 const dump_metadata_t &.
1244 (dump_printf_loc): Likewise; convert 2nd param from
1245 const dump_location_t & to const dump_user_location_t &.
1246 (dump_generic_expr_loc): Likewise.
1247 (dump_generic_expr): Convert 1st param from dump_flags_t to
1248 const dump_metadata_t &.
1249 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
1250 const dump_location_t & to const dump_user_location_t &.
1251 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
1252 const dump_metadata_t &.
1253 (dump_gimple_expr_loc): Likewise; convert 2nd param from
1254 const dump_location_t & to const dump_user_location_t &.
1255 (dump_gimple_expr): Convert 1st param from dump_flags_t to
1256 const dump_metadata_t &.
1257 (dump_symtab_node): Likewise.
1258 (dump_dec): Likewise.
1259 (dump_begin_scope): Split out 2nd param into user and impl
1261 (auto_dump_scope::auto_dump_scope): Split "loc" param into a user
1262 location and impl_location, and capture the impl_location.
1263 (AUTO_DUMP_SCOPE): Rename param from LOC to USER_LOC.
1264 * loop-unroll.c (report_unroll): Update for changes to
1265 dump_printf_loc and dump_printf.
1266 * opt-problem.cc (opt_problem::opt_problem): Update for change to
1268 * optinfo-emit-json.cc
1269 (selftest::test_building_json_from_dump_calls): Convert "loc" from
1270 dump_location_t to dump_user_location_t.
1271 * optinfo.cc (optinfo::emit_for_opt_problem): Update for change in
1273 * profile.c (compute_branch_probabilities): Update for change to
1275 * selftest.h (ASSERT_STR_CONTAINS_AT): New macro.
1276 * tree-vect-slp.c (vect_print_slp_tree): Update for change to
1279 2018-11-27 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1281 * doc/sourcebuild.texi: Document check_effective_target_int_eq_float
1282 and check_effective_target_ptr_eq_long.
1284 2018-11-27 Alan Modra <amodra@gmail.com>
1286 * config.gcc (powerpc*-*-freebsd*, powerpc-*-netbsd*),
1287 (powerpc-*-eabisimaltivec*, powerpc-*-eabisim*, powerpc-*-elf*),
1288 (powerpc-*-eabialtivec*, powerpc-*-eabi*, powerpc-*-rtems*),
1289 (powerpc-wrs-vxworks*, powerpc-*-lynxos*, powerpcle-*-elf*),
1290 (powerpcle-*-eabisim*, powerpcle-*-eabi*): Add gnu-user.h to tm_file.
1291 * config/rs6000/freebsd.h (CPLUSPLUS_CPP_SPEC),
1292 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
1293 (ASM_APP_ON, ASM_APP_OFF): Don't define.
1294 * config/rs6000/freebsd64.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
1295 * config/rs6000/lynx.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
1296 * config/rs6000/linux64.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
1297 * config/rs6000/netbsd.h (CPLUSPLUS_CPP_SPEC),
1298 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
1299 * config/rs6000/rtems.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
1300 * config/rs6000/sysv4.h (GNU_USER_TARGET_CRTI): Redefine.
1301 (GNU_USER_TARGET_CRTN): Redefine.
1302 (CC1_SPEC): Use GNU_USER_TARGET_CC1_SPEC.
1303 (LIB_LINUX_SPEC): Use GNU_USER_TARGET_LIB_SPEC.
1304 (CRTOFFLOADBEGIN, CRTOFFLOADEND): Don't define.
1305 (STARTFILE_LINUX_SPEC): Define as GNU_USER_TARGET_STARTFILE_SPEC.
1306 (ENDFILE_LINUX_SPEC): Define as GNU_USER_TARGET_ENDFILE_SPEC.
1307 (UCLIBC_DYNAMIC_LINKER, CHOOSE_DYNAMIC_LINKER): Don't define.
1308 (LINK_EH_SPEC): Don't define.
1310 2018-11-26 Martin Sebor <msebor@redhat.com>
1313 * expr.c (string_constant): Handle top-level decls of all character
1314 types and subobjects of narrow character type.
1316 2018-11-27 Alan Modra <amodra@gmail.com>
1318 * config.gcc (powerpc*-*-linux*): Add linux.h to tm_file.
1319 * config/rs6000/linux.h (TARGET_OS_CPP_BUILTINS): Use
1320 GNU_USER_TARGET_OS_CPP_BUILTINS.
1321 (RS6000_ABI_NAME): Define.
1322 * config/rs6000/linux64.h (TARGET_OS_CPP_BUILTINS): Use
1323 GNU_USER_TARGET_OS_CPP_BUILTINS.
1324 (MUSL_DYNAMIC_LINKER32): Undef before defining.
1325 (UCLIBC_DYNAMIC_LINKER32, UCLIBC_DYNAMIC_LINKER64): Don't define.
1326 (CHOOSE_DYNAMIC_LINKER): Don't define.
1327 (GNU_USER_DYNAMIC_LINKER32, GNU_USER_DYNAMIC_LINKER64): Don't define.
1328 * config/rs6000/sysv4.h (MUSL_DYNAMIC_LINKER): Undef before defining.
1329 (CHOOSE_DYNAMIC_LINKER, GNU_USER_DYNAMIC_LINKER): Only define when
1330 not already defined.
1331 (CPP_OS_LINUX_SPEC): Remove defines and asserts handled by
1332 TARGET_OS_CPP_BUILTINS.
1334 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1336 * doc/sourcebuild.texi: Document check_effective_target_newlib_nano_io.
1338 2018-11-26 Jakub Jelinek <jakub@redhat.com>
1341 * config/i386/i386.c (def_builtin2): If tcode == VOID_FTYPE_UINT64
1342 and !TARGET_64BIT, return NULL_TREE.
1345 * dwarf2out.c (secname_for_decl): For functions with
1346 DECL_SECTION_NAME if in_cold_section_p, try to return
1347 current_function_section's name if it is a named section.
1349 2018-11-26 Maya Rashish <coypu@sdf.org>
1352 * config.host (*-*-netbsd*): Use x-netbsd and host-netbsd.o.
1353 * config/x-netbsd: New file.
1354 * config/host-netbsd: New file.
1356 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1358 * doc/sourcebuild.texi: Document check_effective_target_size20plus.
1359 Clarify documentation for check_effective_target_size32plus.
1361 2018-11-26 Aaron Sawdey <acsawdey@linux.ibm.com>
1363 * config/rs6000/rs6000-string.c (expand_cmp_vec_sequence): Rename
1364 and modify expand_strncmp_vec_sequence.
1365 (emit_final_compare_vec): Rename and modify emit_final_str_compare_vec.
1366 (generate_6432_conversion): New function.
1367 (expand_block_compare): Add support for vsx.
1368 (expand_block_compare_gpr): New function.
1369 * config/rs6000/rs6000.opt (rs6000_block_compare_inline_limit): Increase
1370 default limit to 63 because of more compact vsx code.
1372 2018-11-26 Uros Bizjak <ubizjak@gmail.com>
1375 * config/i386/i386.c (dbx_register_map): Use IGNORED_DWARF_REGNUM
1376 for registers for which no debug information can be generated.
1377 Use INVALID_REGNUM for non-existent registers.
1378 (dbx64_register_map): Ditto.
1379 (svr4_dbx_register_map): Ditto.
1381 2018-11-26 Sandra Loosemore <sandra@codesourcery.com>
1384 * doc/invoke.texi (Optimize Options): Document that
1385 -freorder-functions also uses "hot" and "cold" attributes.
1387 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1389 * config/aarch64/aarch64.c (aarch64_mangle_type): Fix typo in comment.
1391 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1393 * config/arm/types.md (mul64): Rename to...
1394 (widen_mul64): ... This.
1395 * config/arm/arm-generic.md: Rename mul64 to widen_mul64.
1396 * config/arm/cortex-a15.md: Likewise.
1397 * config/arm/cortex-a5.md: Likewise.
1398 * config/arm/cortex-a53.md: Likewise.
1399 * config/arm/cortex-a57.md: Likewise.
1400 * config/arm/cortex-a7.md: Likewise.
1401 * config/arm/cortex-m4.md: Likewise.
1402 * config/arm/exynos-m1.md: Likewise.
1403 * config/arm/marvell-pj4.md: Likewise.
1404 * config/arm/xgene1.md: Likewise.
1406 2018-11-26 Richard Biener <rguenther@suse.de>
1408 PR tree-optimization/88182
1409 * tree-vect-loop.c (vectorizable_reduction): Pick up single
1410 correct reduc_def_info.
1411 * tree-vect-slp.c (vect_analyze_slp_instance): Set
1412 STMT_VINFO_REDUC_DEF of the first stmt.
1414 2018-11-26 Andreas Krebbel <krebbel@linux.ibm.com>
1416 * doc/invoke.texi: Document z14/arch12 -march option.
1418 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1421 * doc/invoke.texi (Option Summary): Add -Wmissing-noreturn.
1422 (Warning Options): Likewise.
1424 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1427 * doc/extend.texi (Common Function Attributes): Clarify that
1428 functions with "const" attribute can read const global variables.
1430 2018-11-26 Alan Modra <amodra@gmail.com>
1432 * config/rs6000/rs6000.h (OBJECT_PEF, TARGET_MACOS): Delete.
1434 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
1437 * doc/extend.texi (Common Variable Attributes): Use preferred
1438 placement of type attributes in examples, plus whitespace fixes.
1439 (Type Attributes): Clarify why placement of attributes
1440 immediately after struct/union/enum keyword is preferred.
1441 (Common Type Attributes): Use preferred placement of type
1442 attributes in examples, plus more whitespace fixes.
1444 2018-11-25 Paul Koning <ni1d@arrl.net>
1446 * config/pdp11/pdp11.h (TARGET_HAS_NO_HW_DIVIDE): Define.
1448 2018-11-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1450 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
1451 attributes): Document d_runtime.
1453 2018-11-25 Vladimir Makarov <vmakarov@redhat.com>
1456 * ira-costs.c (record_operand_costs): Initiate register move cost
1459 2018-11-23 Jeff Law <law@redhat.com>
1461 PR rtl-optimization/87468
1462 * tree-ssa-threadupdate.c (create_block_for_threading): Clear
1463 EDGE_IGNORE on all outgoing edges of the duplicate block.
1465 2018-11-23 Vladimir Makarov <vmakarov@redhat.com>
1468 * ira-costs.c (record_operand_costs): Use bigger hard reg class if
1469 its mode does not fit to the original class.
1471 2018-11-23 Martin Sebor <msebor@redhat.com>
1473 PR tree-optimization/87756
1474 * expr.c (string_constant): Synthesize a string literal from
1475 the address of a constant character.
1476 * tree.c (build_string_literal): Add an argument.
1477 * tree.h (build_string_literal): Same.
1479 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
1481 * config/aarch64/aarch64-cores.def: Define emag.
1482 * config/aarch64/aarch64-tune.md: Regenerated with emag.
1483 * config/aarch64/aarch64.c (emag_tunings): New struct.
1484 * doc/invoke.texi: Document mtune value.
1486 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
1487 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
1489 * config/aarch64/aarch64.c (xgene1_tunings): Optimize Xgene1 tunings
1491 * config/aarch64/aarch64.c (xgene1_tunings): Add Xgene1 specific
1493 * config/aarch64/aarch64.c (xgene1_addrcost_table): Correct the post-modify
1495 * config/arm/aarch-cost-tables.h (xgene1_extra_costs): Update the cost table
1498 2018-11-23 Richard Biener <rguenther@suse.de>
1500 PR tree-optimization/88149
1501 * tree-vect-slp.c (vect_slp_analyze_node_operations): Detect
1502 the case where there are two different def types for the
1503 same operand at different operand position in the same stmt.
1505 2018-23-11 Mihail Ionescu <mihail.ionescu@arm.com>
1507 * config/arm/arm.c (arm_expand_compare_and_swap): Simplify and call
1508 gen_atomic_compare_swap_1.
1509 (arm_evpc_neon_vuzp): Likewise gen_neon_vuzp_internal.
1510 (arm_evpc_neon_vtrn): Likewise gen_neon_vtrn_internal.
1511 (arm_evpc_neon_vext): Likewise gen_neon_vext_internal.
1512 (arm_evpc_neon_vzip): Likewise gen_neon_vzip_internal.
1513 (arm_evpc_neon_vrev): Replace the function pointer and simplify the mode
1515 * config/arm/arm.md (neon_vext<mode>),
1516 (neon_vrev64<mode>, neon_vrev32<mode>),
1517 (neon_vrev16<mode>, neon_vtrn<mode>_internal),
1518 (neon_vzip<mode>_internal, neon_vuzp<mode>_internal): Add an '@'character
1519 before the pattern name.
1520 * config/arm/sync.md:
1521 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1),
1522 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): Likewise.
1524 2018-11-23 Jakub Jelinek <jakub@redhat.com>
1526 PR tree-optimization/86614
1527 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return early
1528 if TREE_NO_WARNING is set on ref.ref.
1530 2018-11-22 Sandra Loosemore <sandra@codesourcery.com>
1531 Alan Coopersmith <alan.coopersmith@oracle.com>
1534 * doc/extend.texi (Designated Inits): Clarify handling of multiple
1535 initializers for unions.
1537 2018-11-22 Jan Hubicka <jh@suse.cz>
1540 * ipa-devirt.c (type_variants_equivalent_p): Drop warn and warned
1541 parameters; do not warn here.
1542 (odr_subtypes_equivalent_p): Likewise.
1543 (warn_odr): Fix typo.
1544 (warn_types_mismatch): Do not output confused warnings on integer types.
1545 (odr_types_equivalent_p): Update.
1547 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1549 * config/i386/i386.c (ix86_option_override_internal,
1550 ix86_can_inline_p, classify_argument, construct_container,
1551 ix86_expand_prologue, ix86_expand_split_stack_prologue,
1552 ix86_expand_carry_flag_compare, expand_set_or_movmem_via_loop,
1553 expand_setmem_epilogue_via_loop, promote_duplicated_reg,
1554 ix86_expand_set_or_movmem, ix86_init_builtins_va_builtins_abi):
1557 * config/i386/i386.c (ix86_option_override_internal): For
1558 stack_protector_guard related options, use opts_set->x_ instead
1559 of global_options_set. and prefix options with opts->x_ . Move
1560 defaults for offset and reg into else block.
1562 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
1564 * config/i386/i386.c (ix86_check_avx_upper_register):
1565 Return true for all SSE registers with mode bitsize > 128.
1567 2018-11-22 Vladimir Makarov <vmakarov@redhat.com>
1569 PR rtl-optimization/87718
1570 * ira-costs.c: Remove trailing white-spaces.
1571 (record_operand_costs): Add a special treatment for moves
1572 involving a hard register.
1574 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
1576 * config/i386/i386.c (ix86_avx_emit_vzeroupper): Remove.
1577 (ix86_emit_mode_set) <case AVX_U128>: Emit vzeroupper here.
1579 2018-11-22 Martin Liska <mliska@suse.cz>
1581 * common/config/i386/i386-common.c (processor_names): Add
1582 static assert and add missing "znver2".
1583 (ix86_get_valid_option_values): Add checking assert for null
1584 values and add "native" value if feasible.
1585 * config/i386/i386.h: Do not declare size of processor_names.
1586 * common/config/i386/i386-common.c:
1587 * config/i386/i386.c: Add static assert for size
1588 of processor_cost_table.
1590 2018-11-22 Thomas Preud'homme <thomas.preudhomme@linaro.org>
1592 * target-insns.def (stack_protect_combined_set): Define new standard
1594 (stack_protect_combined_test): Likewise.
1595 * cfgexpand.c (stack_protect_prologue): Try new
1596 stack_protect_combined_set pattern first.
1597 * function.c (stack_protect_epilogue): Try new
1598 stack_protect_combined_test pattern first.
1599 * config/arm/arm.c (require_pic_register): Add pic_reg and compute_now
1600 parameters to control which register to use as PIC register and force
1601 reloading PIC register respectively. Insert in the stream of insns if
1603 (legitimize_pic_address): Expose above new parameters in prototype and
1604 adapt recursive calls accordingly. Use pic_reg if non null instead of
1606 (arm_load_pic_register): Add pic_reg parameter and use it if non null.
1607 (arm_legitimize_address): Adapt to new legitimize_pic_address
1609 (thumb_legitimize_address): Likewise.
1610 (arm_emit_call_insn): Adapt to require_pic_register prototype change.
1611 (arm_expand_prologue): Adapt to arm_load_pic_register prototype change.
1612 (thumb1_expand_prologue): Likewise.
1613 * config/arm/arm-protos.h (legitimize_pic_address): Adapt to prototype
1615 (arm_load_pic_register): Likewise.
1616 * config/arm/predicated.md (guard_addr_operand): New predicate.
1617 (guard_operand): New predicate.
1618 * config/arm/arm.md (movsi expander): Adapt to legitimize_pic_address
1620 (builtin_setjmp_receiver expander): Adapt to thumb1_expand_prologue
1622 (stack_protect_combined_set): New expander..
1623 (stack_protect_combined_set_insn): New insn_and_split pattern.
1624 (stack_protect_set_insn): New insn pattern.
1625 (stack_protect_combined_test): New expander.
1626 (stack_protect_combined_test_insn): New insn_and_split pattern.
1627 (arm_stack_protect_test_insn): New insn pattern.
1628 * config/arm/thumb1.md (thumb1_stack_protect_test_insn): New insn pattern.
1629 * config/arm/unspecs.md (UNSPEC_SP_SET): New unspec.
1630 (UNSPEC_SP_TEST): Likewise.
1631 * doc/md.texi (stack_protect_combined_set): Document new standard
1633 (stack_protect_set): Clarify that the operand for guard's address is
1635 (stack_protect_combined_test): Document new standard pattern name.
1636 (stack_protect_test): Clarify that the operand for guard's address is
1639 2018-11-22 Richard Biener <rguenther@suse.de>
1641 PR tree-optimization/88148
1642 * tree-ssa-loop-niter.c (simplify_replace_tree): Get optional
1643 valueization callback parameter and handle it.
1644 * tree-ssa-loop-niter.h (simplify_replace_tree): Export.
1645 * tree-ssa-sccvn.c (process_bb): Eliminate in loop niter trees.
1647 2018-11-22 Richard Biener <rguenther@suse.de>
1651 * lto-streamer-out.c (lto_is_streamable): Allow CALL_EXPRs
1652 which can appear in size expressions.
1653 * tree-streamer-in.c (unpack_ts_base_value_fields): Stream
1654 CALL_EXPR_BY_DESCRIPTOR.
1655 (streamer_read_tree_bitfields): Stream CALL_EXPR_IFN.
1656 * tree-streamer-out.c (pack_ts_base_value_fields): Stream
1657 CALL_EXPR_BY_DESCRIPTOR.
1658 (streamer_write_tree_bitfields): Stream CALL_EXPR_IFN.
1662 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
1665 2018-11-22 Richard Biener <rguenther@suse.de>
1667 PR tree-optimization/88069
1668 * tree-ssa-sccvn.c (visit_phi): Tweak previous fix to not
1669 apply to default defs.
1671 2018-11-22 Jakub Jelinek <jakub@redhat.com>
1675 * config/i386/i386.c (ix86_option_override_internal): Default
1676 ix86_stack_protector_guard to SSP_TLS only if TARGET_THREAD_SSP_OFFSET
1678 * config/i386/i386.md (stack_protect_set, stack_protect_set_<mode>,
1679 stack_protect_test, stack_protect_test_<mode>): Use empty condition
1680 instead of TARGET_SSP_TLS_GUARD.
1682 2018-11-22 Martin Liska <mliska@suse.cz>
1685 * config/i386/i386.c (ix86_option_override_internal): Error
1686 about usage -mabi=ms and -fsanitize=thread.
1688 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1691 2013-10-26 Vladimir Makarov <vmakarov@redhat.com>
1694 2013-10-25 Vladimir Makarov <vmakarov@redhat.com>
1696 * lra-spills.c (lra_final_code_change): Remove useless move insns.
1698 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
1701 * function.c (expand_function_end): Do not emit extra blockage insn.
1703 2018-11-21 Lokesh Janghel <lokeshjanghel91@gmail.com>
1706 * config/i386/i386.c (function_value_ms_64): Return AX_REG instead
1707 of FIRST_SSE_REG for 4 or 8 byte modes.
1709 2018-11-21 Jan Hubicka <jh@suse.cz>
1712 * tree.c (fld_decl_context): Break out from ...
1713 (free_lang_data_in_decl): ... here; free TREE_PUBLIC, TREE_PRIVATE
1714 DECL_ARTIFICIAL of TYPE_DECL; do not free TREE_TYPE of TYPE_DECL.
1715 (fld_incomplete_type_of): Build copy of TYP_DECL.
1716 * ipa-devirt.c (free_enum_values): Rename to ...
1717 (free_odr_warning_data): ... this one; free also duplicated TYPE_DECLs
1718 and TREE_TYPEs of TYPE_DECLs.
1719 (get_odr_type): Initialize odr_vtable_hash if needed.
1721 2018-11-21 Alexandre Oliva <oliva@adacore.com>
1723 * final.c (compute_discriminator): Declare. Renamed from...
1724 (maybe_set_discriminator): ... this. Set and return a local.
1725 (override_discriminator): New.
1726 (final_scan_insn_1): Set it.
1727 (notice_source_line): Adjust. Always set discriminator.
1729 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1732 * config/aarch64/atomics.md (@aarch64_compare_and_swap<mode>): Use
1733 rIJ constraint for aarch64_plus_operand rather than rn.
1735 2018-11-21 Renlin Li <renlin.li@arm.com>
1738 * explow.h (get_dynamic_stack_size): Declare it as external.
1739 * explow.c (record_new_stack_level): Remove function static attribute.
1740 * function.c (assign_stack_local_1): Dynamically align the stack slot
1741 addr for parameter copy on the stack.
1743 2018-11-21 Richard Biener <rguenther@suse.de>
1746 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
1747 * Makefile.in (bitmap.o-warn): Remove again.
1749 2018-11-20 Jeff Law <law@redhat.com>
1751 PR tree-optimization/88069
1752 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
1753 degenerate virtual PHIs.
1755 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1758 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
1759 nonimmediate_operand with register_operand.
1760 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
1761 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
1762 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
1763 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
1764 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
1765 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
1766 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
1767 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
1768 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
1769 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
1770 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
1771 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
1772 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
1773 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
1774 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
1775 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
1776 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
1777 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
1778 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
1779 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
1780 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
1781 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
1782 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
1783 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
1784 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
1785 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
1786 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
1788 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
1790 * read-rtl.c (apply_subst_iterator): Handle define_split and
1791 define_insn_and_split.
1793 2018-11-21 Jakub Jelinek <jakub@redhat.com>
1795 PR rtl-optimization/87817
1796 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
1797 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
1798 in the pattern to avoid triggering UB when operands[2] is zero.
1799 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
1800 (*tbm_bextri_<mode>): ... this.
1802 2018-11-21 Tom de Vries <tdevries@suse.de>
1805 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
1808 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
1810 * configure.ac: Add check for Binutils to determine whether vector
1811 load/store alignments hints are being supported.
1812 * config.in: Regenerate.
1813 * configure: Regenerate.
1814 * config/s390/s390.c (print_operand): Support new output
1816 * config/s390/s390.md ("movti"): Append alignment hint output
1817 using the new output modifier 'A'.
1818 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
1819 ("*vec_ti_to_v1ti"): Likewise.
1821 2018-11-20 Martin Sebor <msebor@redhat.com>
1823 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
1825 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1828 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
1829 compare ENUM values.
1831 2018-11-20 Jakub Jelinek <jakub@redhat.com>
1833 PR tree-optimization/87895
1834 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
1835 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
1836 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
1837 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
1838 has any preds and always use EDGE_PRED (, 0).
1840 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
1843 * mode-switching.c (create_pre_exit): After reload, always split the
1844 fallthrough edge to the exit block.
1846 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1848 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
1849 * tree.c (build_array_type_1): Forward declare.
1850 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
1851 (fld_type_variant): Likewise.
1852 (fld_simplified_types): New hash.
1853 (fld_process_array_type): New function.
1854 (fld_incomplete_type_of): Handle array and enumeration types.
1855 (fld_simplified_type): Handle simplification of arrays.
1856 (free_lang_data): Allocate and free simplified types hash.
1858 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1861 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
1862 when sybtype already violates ODR.
1863 (get_odr_type): Do not ICE when insert is false and type duplicate
1864 is not registered yet.
1865 (register_odr_type): Be sure to register subtypes first.
1867 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
1869 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
1870 DImode parts of the target operand.
1872 2018-11-20 Nathan Sidwell <nathan@acm.org>
1875 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
1877 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1880 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
1882 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
1884 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1887 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
1888 * ipa.c (possible_inline_candidate_p): Break out from ..
1889 (process_references): ... here ; drop before_inlining_p;
1890 cleanup handling of alises.
1891 (walk_polymorphic_call_targets): Likewise.
1892 (symbol_table::remove_unreachable_nodes): Likewise.
1893 * passes.c (pass_data_ipa_remove_symbols): New structure.
1894 (pass_ipa_remove_symbols): New pass.
1895 (make_pass_ipa_remove_symbols): New function.
1896 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
1897 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
1899 2018-11-20 Richard Biener <rguenther@suse.de>
1901 * tree-vect-stmts.c (vectorizable_condition): Do not get
1902 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
1903 pointless vect_is_simple_use calls.
1905 2018-11-20 Richard Biener <rguenther@suse.de>
1907 PR tree-optimization/88074
1908 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
1911 2018-11-20 Richard Biener <rguenther@suse.de>
1913 PR tree-optimization/88069
1914 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
1915 virtual PHI arguments.
1917 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
1920 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
1922 * rtl.h (contains_constant_pool_address_p): New function.
1923 * rtlanal.c (contains_constant_pool_address_p): Likewise.
1925 2018-11-20 Richard Biener <rguenther@suse.de>
1928 * alias.c (component_uses_parent_alias_set_from): Remove
1929 alias-set zero and TYPE_TYPELESS_STORAGE case both already
1930 handled in other ways.
1932 2018-11-20 Richard Biener <rguenther@suse.de>
1934 PR tree-optimization/88087
1935 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
1937 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
1940 2018-11-20 Richard Biener <rguenther@suse.de>
1943 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
1945 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
1948 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
1950 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
1953 * doc/invoke.texi (Link Options): Mention shared libraries
1954 in documentation for the -l option. Simplify discussion and
1955 point to the system linker documentation for details.
1957 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
1959 PR rtl-optimization/88033
1960 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
1961 to itself. Use HARD_REGISTER_NUM_P.
1963 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1965 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
1966 of the loop if vf is non-constant.
1968 2018-11-19 David Malcolm <dmalcolm@redhat.com>
1970 PR tree-optimization/87025
1971 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
1972 immediately after creating the scope optinfo.
1973 (selftest::test_pr87025): New function.
1974 (selftest::dumpfile_c_tests): Call it.
1975 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
1976 that we're not popping the top-level records array.
1977 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
1978 not changing the kind of a "scope" optinfo.
1980 2018-11-19 David Malcolm <dmalcolm@redhat.com>
1982 PR tree-optimization/87025
1983 * dump-context.h: Include "optinfo.h".
1984 (class optrecord_json_writer): New forward decl.
1985 (dump_context::forcibly_enable_optinfo_p): Delete.
1986 (dump_context::optinfo_enabled_p): New member function.
1987 (dump_context::optimization_records_enabled_p): New member
1989 (dump_context::set_json_writer): New member function.
1990 (dump_context::emit_optinfo): New member function.
1991 (dump_context::m_forcibly_enable_optinfo): Delete.
1992 (dump_context::m_json_writer): New member data.
1993 * dumpfile.c (dump_context::set_json_writer): New member function.
1994 (dump_context::finish_any_json_writer): New member function.
1995 (dump_context::end_scope): Replace call to
1996 optimization_records_maybe_pop_dump_scope with call to
1997 m_json_writer->pop_scope.
1998 (dump_context::optinfo_enabled_p): New member function.
1999 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
2000 to dump_context::emit_optinfo.
2001 (dump_context::emit_optinfo): New member function.
2002 (temp_dump_context::temp_dump_context): Replace
2003 m_forcibly_enable_optinfo with call to set_json_writer.
2004 (temp_dump_context::~temp_dump_context): Clean up any json writer.
2005 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
2007 (the_json_writer): Delete.
2008 (optimization_records_start): Delete.
2009 (optimization_records_finish): Delete.
2010 (optimization_records_enabled_p): Delete, in favor of
2011 dump_context::optimization_records_enabled_p.
2012 (optimization_records_maybe_record_optinfo): Delete.
2013 (optimization_records_maybe_pop_dump_scope): Delete.
2014 * optinfo-emit-json.h: Include "json.h". Delete forward
2016 (optimization_records_start): Delete.
2017 (optimization_records_finish): Delete.
2018 (optimization_records_enabled_p): Delete.
2019 (optimization_records_maybe_record_optinfo): Delete.
2020 (optimization_records_maybe_pop_dump_scope): Delete.
2021 (class optrecord_json_writer): Move here from
2022 optinfo-emit-json.cc.
2023 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
2024 to optinfo::emit with call to dump_context::emit_optinfo.
2025 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
2026 (optinfo_enabled_p): Delete, in favor of
2027 dump_context::optinfo_enabled_p.
2028 (optinfo_wants_inlining_info_p): Update for conversion o
2029 optimization_records_enabled_p to a member function of
2031 * optinfo.h (optinfo_enabled_p): Delete, in favor of
2032 dump_context::optinfo_enabled_p.
2033 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
2034 * toplev.c: Include "dump-context.h".
2035 (compile_file): Replace call to optimization_records_finish with
2036 dump_context::finish_any_json_writer.
2037 (do_compile): Replace call to optimization_records_start with
2038 conditionally creating a optrecord_json_writer for the
2041 2018-11-19 Jakub Jelinek <jakub@redhat.com>
2043 PR tree-optimization/88071
2044 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
2045 of false as last argument to gsi_remove.
2046 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
2047 false as last argument to gsi_replace.
2050 * omp-expand.c: Don't include debug.h.
2051 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
2052 from innermost outer parallel, task, teams or target that has a
2053 child_fn set, or, if there is no such outer region, use
2054 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
2055 whether a suitable BLOCK is found or not.
2056 (expand_parallel_call, expand_teams_call): Don't call
2057 adjust_context_and_scope here.
2058 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
2059 (expand_omp_taskreg, expand_omp_target): Likewise. Call
2060 adjust_context_and_scope.
2061 * dwarf2out.c (dwarf2out_early_global_decl): For
2062 decl_function_context recurse instead of calling dwarf2out_decl.
2064 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
2066 PR rtl-optimization/85925
2067 * rtl.h (word_register_operation_p): New predicate.
2068 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
2069 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
2070 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
2071 (num_sign_bit_copies1): Likewise.
2073 2018-11-19 Richard Biener <rguenther@suse.de>
2076 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
2079 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
2081 * compare-elim.c (struct comparison): Add not_in_a field.
2082 (is_not): New static function.
2083 (strip_not): Likewise.
2084 (conforming_compare): Handle a NOT in the first operand.
2085 (can_eliminate_compare): Likewise.
2086 (find_comparison_dom_walker::before_dom_children): Likewise.
2087 (try_eliminate_compare): Likewise.
2088 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
2089 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
2091 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
2093 * doc/extend.texi (Common Type Attributes): Fix typos.
2095 2018-11-19 Martin Liska <mliska@suse.cz>
2097 PR gcov-profile/88045
2098 * coverage.c (coverage_begin_function): Add assert.
2100 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
2103 * doc/extend.texi (Common Function Attributes): Document that
2104 no_instrument_function applies to -p and -pg, too.
2105 * doc/invoke.texi (Instrumentation Options): Add cross-references
2106 to docs for -p, -pg, and -finstrument-functions.
2108 2018-11-18 Alan Modra <amodra@gmail.com>
2110 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
2111 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
2112 (GNU_USER_TARGET_CRTN): Define.
2113 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
2114 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
2116 2018-11-18 Alan Modra <amodra@gmail.com>
2118 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
2119 !HAVE_LD_PIE variant.
2120 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
2122 2018-11-17 Nick Clifton <nickc@redhat.com>
2123 Sandra Loosemore <sandra@codesourcery.com>
2126 * doc/invoke.texi (Overall Options): Document that --help and
2127 --help=<value> options cannot be combined.
2129 2018-11-17 Jim Wilson <jimw@sifive.com>
2131 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
2132 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
2133 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
2134 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
2135 to step2 if frame_pointer_needed and step1 is 0.
2137 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
2140 * doc/cpp.texi (System Headers): Add note about implicit
2141 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
2143 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
2146 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
2147 -fno-implicit-templates.
2149 2018-11-17 Jakub Jelinek <jakub@redhat.com>
2151 PR tree-optimization/87546
2152 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
2153 min_precision variable, initially set it to orig_precision, only does
2154 something if op_type's precision is <= min_precision and update
2155 min_precision whenever calling set_op.
2157 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
2160 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
2162 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
2164 * doc/invoke.texi (Option Summary): Fix whitespace and line
2165 breaks in @gccoptlist environments.
2166 (Warning Options): Likewise.
2167 (Optimize Options): Likewise.
2168 (PowerPC SPE Options): Likewise.
2169 (RS/6000 and PowerPC Options): Likewise.
2171 2018-11-16 Jeff Law <law@redhat.com>
2173 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
2176 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
2179 * doc/invoke.texi (Optimize Options): Update options enabled by
2180 fprofile-generate, -fprofile-use, and -fauto-profile.
2182 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2185 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
2186 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
2187 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
2189 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
2190 * targhooks.c (default_print_patchable_function_entry): Assert
2191 asm_int_op does not return a NULL string.
2192 * varasm.c (integer_asm_op): Return the op for a partial int type
2193 when the requested size does not correspond to an integer type.
2194 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
2195 * doc/tm.texi: Regenerate.
2197 2018-11-12 Jason Merrill <jason@redhat.com>
2199 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
2201 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
2203 * config/rs6000/constraints.md (wF constraint): Remove power9
2204 fusion documentation. Just document wF's use for power8 fusion.
2205 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
2206 (fusion_gpr_addis): Delete power9 fusion support. Change power8
2207 fusion support to require the upper 12 bits to be all 0's or all
2209 (fusion_gpr_mem_load): Add comment.
2210 (fusion_addis_mem_combo_load): Remove power9 fusion support.
2211 (fusion_addis_mem_combo_store): Delete.
2212 (fusion_offsettable_mem_operand): Delete.
2213 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
2215 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
2216 (POWERPC_MASKS): Delete power9 fusion.
2217 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
2218 (fusion_p9_p): Delete.
2219 (expand_fusion_p9_load): Delete.
2220 (expand_fusion_p9_store): Delete.
2221 (emit_fusion_p9_load): Delete.
2222 (emit_fusion_p9_store): Delete.
2223 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
2225 (rs6000_option_override_internal): Set power8 fusion based on
2226 whether we are tuning for power8. Delete power9 fusion support.
2227 (rs6000_opt_masks): Delete -mpower9-fusion switch.
2228 (emit_fusion_load): Rename emit_fusion_load_store to
2229 emit_fusion_load, and drop fusion store support. Update callers.
2230 (emit_fusion_load_store): Likewise.
2231 (emit_fusion_gpr_load): Likewise.
2232 (fusion_p9_p): Delete.
2233 (expand_fusion_p9_load): Delete.
2234 (expand_fusion_p9_store): Delete.
2235 (emit_fusion_p9_load): Delete.
2236 (emit_fusion_p9_store): Delete.
2237 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
2238 (GPR_FUSION): Delete.
2239 (FPR_FUSION): Delete.
2240 (power9 fusion peephole2s): Delete.
2241 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
2242 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
2243 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
2244 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
2245 (fusion_p9_<mode>_constant): Delete.
2246 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
2247 power9 fusion switch.
2248 * doc/md.texi (PowerPC constraints): Update wF constraint
2249 documentation for power8 fusion only.
2251 2018-11-16 Jakub Jelinek <jakub@redhat.com>
2253 PR rtl-optimization/87475
2254 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
2255 CROSSING_JUMP_P insns.
2256 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
2258 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
2261 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
2262 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
2263 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
2265 2018-11-16 Jakub Jelinek <jakub@redhat.com>
2268 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
2269 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
2271 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
2273 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
2274 location_t parameter.
2275 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
2276 (switch_decision_tree::do_jump_if_equal): Likewise.
2277 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
2278 of switch statement to emit_case_nodes.
2279 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
2280 set it on the newly built GIMPLE comparison statement.
2281 (switch_decision_tree::do_jump_if_equal): Likewise.
2282 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
2283 in calls to do_jump_if_equal as well as recursive calls.
2285 2018-11-16 Andi Kleen <ak@linux.intel.com>
2287 * config/i386/i386.md: Allow memory operands to ptwrite.
2289 2018-11-16 Michael Matz <matz@suse.de>
2291 * system.h (PRsa): New macro.
2292 (SIZE_AMOUNT): Cast number to uint64_t.
2293 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
2294 (pool_usage::dump_footer): Likewise and also use PRIu64.
2295 * bitmap.h (bitmap_usage::dump): Likewise.
2296 * ggc-common.c (ggc_usage::dump): Likewise.
2297 * ggc-page.c (ggc_print_statistics): Likewise.
2298 * input.c (dump_line_table_statistics): Likewise.
2299 * mem-stats.h (mem_usage::dump): Likewise.
2300 (mem_usage::dump_footer): Likewise.
2301 * rtl.c (dump_rtx_statistics): Likewise.
2302 * tree-cfg.c (dump_cfg_stats): Likewise.
2303 * tree-dfa.c (dump_dfa_stats): Likewise.
2304 * tree-phinodes.c (phinodes_print_statistics): Likewise.
2305 * tree-ssanames (ssanames_print_statistics): Likewise.
2306 * vec.c (vec_usage::dump): Likewise.
2307 (vec_usage::dump_footer): Likewise.
2309 2018-11-16 Richard Biener <rguenther@suse.de>
2311 PR tree-optimization/88011
2312 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
2313 replacing set_value_range_to_undefined and
2314 set_value_range_to_varying with method calls.
2316 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
2318 * config/s390/s390.md
2319 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
2321 (*r<noxa>sbg_<mode>_sll): Likewise.
2322 (*r<noxa>sbg_<mode>_srl): Likewise.
2323 (*r<noxa>sbg_sidi_srl): New pattern.
2325 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
2327 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
2328 Default to TARGET_VXWORKS_RTP.
2329 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
2330 of TARGET_VXWORKS_RTP.
2331 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
2332 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
2333 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
2335 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
2338 * doc/extend.texi (Common Type Attributes): Make it explicit
2339 that attribute "packed" can apply to C++ classes.
2341 2018-11-15 Martin Sebor <msebor@redhat.com>
2345 * tree.c (type_argument_type): New function.
2346 * tree.h (type_argument_type): Declare it.
2347 * gcc/doc/extend.texi (alloc_align): Update and clarify.
2348 (alloc_size, nonnull, sentinel): Same.
2350 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
2351 Kwok Cheung Yeung <kcy@codesourcery.com>
2353 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
2354 int_mode_for_size fails.
2355 (vectorizable_load): Likewise.
2357 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2359 * doc/ux.texi (Group logically-related diagnostics): Move
2360 discussion of auto_diagnostic_group into this new subsection.
2361 Give an example of where this grouping is used.
2363 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2366 * Makefile.in (OBJS): Move json.o to...
2367 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
2368 * common.opt (fdiagnostics-format=): New option.
2369 (diagnostics_output_format): New enum.
2370 * diagnostic-format-json.cc: New file.
2371 * diagnostic.c (default_diagnostic_final_cb): New function, taken
2372 from start of diagnostic_finish.
2373 (diagnostic_initialize): Initialize final_cb to
2374 default_diagnostic_final_cb.
2375 (diagnostic_finish): Move "being treated as errors" messages to
2376 default_diagnostic_final_cb. Call any final_cb.
2377 (default_diagnostic_finalizer): Add diagnostic_t param.
2378 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
2379 diagnostic_finalizer callback.
2380 * diagnostic.h (enum diagnostics_output_format): New enum.
2381 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
2382 (struct diagnostic_context): Add "final_cb".
2383 (default_diagnostic_finalizer): Add diagnostic_t param.
2384 (diagnostic_output_format_init): New decl.
2385 * doc/invoke.texi (-fdiagnostics-format): New option.
2386 * dwarf2out.c (gen_producer_string): Ignore
2387 OPT_fdiagnostics_format_.
2388 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
2389 * lto-wrapper.c (append_diag_options): Ignore it.
2390 * opts.c (common_handle_option): Handle it.
2392 2018-11-15 David Malcolm <dmalcolm@redhat.com>
2394 PR tree-optimization/88015
2395 * graphite-isl-ast-to-gimple.c
2396 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
2398 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
2400 2018-11-15 Richard Biener <rguenther@suse.de>
2403 * gimple.c (gimple_call_flags): Union flags from decl, type
2405 * trans-mem.c (is_tm_pure_call): Simplify.
2407 2018-11-15 Richard Biener <rguenther@suse.de>
2409 PR tree-optimization/88031
2410 * tree-vect-loop.c (vectorizable_reduction): Move check
2411 for multiple types earlier so we get the expected dump.
2412 Simplify calls to vectorizable_condition.
2413 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
2414 * tree-vect-stmts.c (vectorizable_condition): Instead of
2415 reduc_def and reduc_index take just a flag. Simplify
2416 code-generation now that we can rely on the defs being set up.
2417 (vectorizable_comparison): Remove unused argument.
2419 2018-11-15 Richard Biener <rguenther@suse.de>
2421 PR tree-optimization/88030
2422 * tree-complex.c (need_eh_cleanup): New global.
2423 (update_complex_assignment): Mark blocks that need EH update.
2424 (expand_complex_comparison): Likewise.
2425 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
2426 perform EH cleanup and schedule CFG cleanup if that did anything.
2428 2018-11-15 Jakub Jelinek <jakub@redhat.com>
2430 PR rtl-optimization/88018
2431 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
2432 on the presence of fallthru edge, rather than if it is a USE or not.
2434 2018-11-15 Richard Biener <rguenther@suse.de>
2437 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
2438 analyze_subscript_affine_affine properly.
2440 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
2442 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
2443 (GLIBC_DYNAMIC_LINKER): Define.
2444 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
2446 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2448 PR tree-optimization/84648
2449 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
2450 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
2451 by calling adjust_cond_for_loop_until_wrap.
2453 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
2456 * doc/extend.texi (Common Function Attributes): Clarify linker
2457 restrictions on "aligned" attribute.
2458 (Common Variable Attributes): Likewise. Mention that linker
2459 restrictions don't apply to stack-allocated variables.
2461 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
2463 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
2466 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
2470 * doc/invoke.texi (Optimize Options): Remove bad example about
2471 interaction between -flto and -O. Replace it with a note that
2472 you need to compile with -O and not just link. Copy-edit -flto
2473 discussion to reduce verbiage and improve flow.
2475 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
2478 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
2479 also suppress many optimizations. Alphabetize option lists for
2480 -O1, -O2, and -Os. Add list of options disabled with -Og, and
2481 correct documentation for those options to say that.
2482 * opts.c (default_options_table): Sort table by level and option
2483 name, to make it easier to correlate to the manual.
2485 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
2487 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
2488 printing of the names of segment registers.
2490 2018-11-14 Nathan Sidwell <nathan@acm.org>
2494 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
2497 2018-11-14 David Malcolm <dmalcolm@redhat.com>
2499 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
2501 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2503 P1236R1 - Signed integers are two's complement
2504 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
2507 * hash-map.h (hash_map::iterator::reference_pair): New class.
2508 (hash_map::iterator::operator*): Return it rather than std::pair.
2510 2018-11-14 Jeff Law <law@redhat.com>
2512 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
2513 iff the operand is a constant.
2515 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
2517 * gimple-ssa-evrp-analyze.c
2518 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
2519 ignore_equivs_equal_p to equal_p.
2520 * ipa-cp.c (meet_with_1): Use equal_p instead of
2521 ignore_equivs_equal_p.
2522 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
2523 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
2524 (value_range::operator==): Remove.
2525 (value_range::operator!=): Remove.
2526 (vrp_prop::visit_stmt): Use equal_p.
2527 * tree-vrp.h (value_range): Remove operator==, operator!=,
2528 ignore_equivs_equal_p.
2529 * vr-values.c (update_value_range): Use equal_p.
2531 2018-11-14 Michael Matz <matz@suse.de>
2534 * gimplify.c (collect_fallthrough_labels): Add new argument,
2535 return location via that, don't modify statements.
2536 (warn_implicit_fallthrough_r): Adjust call, don't use
2537 statement location directly.
2539 2018-11-14 Richard Biener <rguenther@suse.de>
2542 * tree-data-ref.c (split_constant_offset): Add wrapper
2543 allocating a cache hash-map.
2544 (split_constant_offset_1): Cache results of expanding
2545 expressions from SSA def stmts.
2547 2018-11-14 Richard Biener <rguenther@suse.de>
2550 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
2552 (lambda_vector_mult_const): Likewise.
2553 (lambda_matrix_right_hermite): Use lambda_int temporaries.
2555 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
2556 Jackson Woodruff <jackson.woodruff@arm.com>
2558 PR 71026/tree-optimization
2559 * match.pd: Simplify floating point comparisons.
2561 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2563 PR rtl-optimization/87817
2564 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
2565 last argument has low 8 bits clear, fold to 0.
2567 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
2569 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
2570 MAX_OFILE_ALIGNMENT as needed.
2572 2018-11-14 Richard Biener <rguenther@suse.de>
2574 PR tree-optimization/88019
2575 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
2576 COND_EXPR arguments to non-trapping overflow.
2578 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
2580 * config/arc/arc.h (reg_class): Reorder registers classes, remove
2581 unused register classes.
2582 (REG_CLASS_NAMES): Likewise.
2583 (REG_CLASS_CONTENTS): Likewise.
2584 (FIXED_REGISTERS): Make lp_count fixed.
2585 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
2586 (PROGRAM_COUNTER_REGNO): Remove.
2587 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
2588 register classes, use constants for register numbers, remove
2589 reg_class_contents references.
2590 (arc_process_double_reg_moves): Add asserts.
2591 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
2592 lra_in_progress predicate.
2593 (arc_init_reg_tables): Remove unused register classes.
2594 (arc_register_move_cost): Likewise.
2595 (arc_preferred_reload_class): Likewise.
2596 (hwloop_optimize): Update rtx patterns involving lp_count register.
2597 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
2598 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
2599 CHEAP_CORE_REGS. Former one will be used for LRA.
2600 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
2601 one will be used for LRA.
2602 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
2603 one will be used for LRA.
2604 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
2605 Former one will be used for LRA.
2606 ("f"): Delete constraint.
2609 ("l"): Change it from register constraint to constraint.
2610 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
2611 (movhi_insn): Likewise.
2612 (movsi_insn): Update pattern.
2615 (stack_tie): Remove 'b' constraint letter.
2617 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
2618 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
2619 (R57_REG, R59_REG, PCL_REG): Likewise.
2620 (ILINK1_REGNUM): Renamed to ILINK1_REG.
2621 (ILINK2_REGNUM): Renamed to ILINK2_REG.
2623 (SP_REGS): Likewise.
2624 (Rcw): Remove unused reg classes.
2625 * config/arc/predicates.md (dest_reg_operand): Just default on
2626 register_operand predicate.
2627 (mpy_dest_reg_operand): Likewise.
2628 (move_dest_operand): Use macros instead of constants.
2629 * config/arc/arc.opt (mlra): Switch to lra as default.
2631 2018-11-14 Richard Biener <rguenther@suse.de>
2633 PR tree-optimization/87974
2634 * tree-vect-loop.c (vectorizable_reduction): When computing
2635 the vectorized reduction PHI vector type ignore constant
2638 2018-11-14 Jakub Jelinek <jakub@redhat.com>
2640 PR tree-optimization/87977
2641 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
2642 stmt, build a new one and replace the old one with it. Formatting fix.
2643 Call release_ssa_name (x) if !has_other_use and !delete_div.
2644 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
2645 verify lhs of stmt is still def.
2647 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2649 PR rtl-optimization/87507
2650 * lower-subreg.c (operand_for_swap_move_operator): New function.
2651 (simple_move): Strip simple operators.
2652 (find_pseudo_copy): Likewise.
2653 (resolve_operand_for_swap_move_operator): New function.
2654 (resolve_simple_move): Strip simple operators and swap operands.
2656 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2658 PR tree-optimization/87898
2659 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
2660 (ipa_simd_modify_function_body): Remove debug stmts where the first
2661 argument was changed into a non-decl.
2663 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
2665 PR rtl-optimization/87899
2666 * lra-lives.c (start_living): Update white space in comment.
2667 (enum point_type): New.
2668 (sparseset_contains_pseudos_p): New function.
2669 (update_pseudo_point): Likewise.
2670 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
2671 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
2672 handling. Move early exit after adding conflicts.
2673 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
2674 if regno is already live. Remove all handling of program points.
2675 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
2676 after adding conflicts. Remove all handling of program points and
2677 ignore_reg_for_conflicts.
2678 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
2679 and do not guard call to mark_pseudo_live.
2680 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
2681 and do not guard call to mark_pseudo_dead.
2682 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
2683 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
2684 Use new function update_pseudo_point. Handle register copies by
2685 removing the source register from the live set. Handle INOUT operands.
2686 Update to the next program point using the unused_set, dead_set and
2688 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
2690 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2692 * builtins.c: Replace "source_location" with "location_t".
2693 * diagnostic-show-locus.c: Likewise.
2694 * diagnostic.c: Likewise.
2695 * dumpfile.c: Likewise.
2696 * gcc-rich-location.h: Likewise.
2697 * genmatch.c: Likewise.
2698 * gimple.h: Likewise.
2699 * gimplify.c: Likewise.
2700 * input.c: Likewise.
2701 * input.h: Likewise. Eliminate the typedef.
2702 * omp-expand.c: Likewise.
2703 * selftest.h: Likewise.
2704 * substring-locations.h (get_source_location_for_substring):
2706 (get_location_within_string): ...this.
2707 * tree-cfg.c: Replace "source_location" with "location_t".
2708 * tree-cfgcleanup.c: Likewise.
2709 * tree-diagnostic.c: Likewise.
2710 * tree-into-ssa.c: Likewise.
2711 * tree-outof-ssa.c: Likewise.
2712 * tree-parloops.c: Likewise.
2713 * tree-phinodes.c: Likewise.
2714 * tree-phinodes.h: Likewise.
2715 * tree-ssa-loop-ivopts.c: Likewise.
2716 * tree-ssa-loop-manip.c: Likewise.
2717 * tree-ssa-phiopt.c: Likewise.
2718 * tree-ssa-phiprop.c: Likewise.
2719 * tree-ssa-threadupdate.c: Likewise.
2720 * tree-ssa.c: Likewise.
2721 * tree-ssa.h: Likewise.
2722 * tree-vect-loop-manip.c: Likewise.
2724 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
2726 * doc/extend.texi: Fix typo in the weakref description.
2728 2018-11-13 Richard Biener <rguenther@suse.de>
2730 PR tree-optimization/86991
2731 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
2732 group building until we have successfully detected the SLP
2734 (vect_is_simple_reduction): Remove fixup code here.
2736 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2738 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
2739 (dump_gimple_stmt): Use it.
2740 (dump_gimple_stmt_loc): Likewise.
2741 (dump_gimple_expr): Likewise.
2742 (dump_gimple_expr_loc): Likewise.
2743 (dump_generic_expr): Likewise.
2744 (dump_generic_expr_loc): Likewise.
2745 (dump_printf): Likewise.
2746 (dump_printf_loc): Likewise.
2747 (dump_dec): Likewise.
2748 (dump_dec): Likewise.
2749 (dump_hex): Likewise.
2750 (dump_symtab_node): Likewise.
2751 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2752 Guard dump call with dump_enabled_p.
2753 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
2754 * graphite-optimize-isl.c (optimize_isl): Likewise.
2755 * graphite.c (graphite_transform_loops): Likewise.
2756 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
2757 * tree-parloops.c (parallelize_loops): Likewise.
2758 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
2759 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
2760 (vect_prune_runtime_alias_test_list): Likewise.
2761 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
2762 (vect_estimate_min_profitable_iters): Likewise.
2763 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
2764 (vect_build_slp_tree_2): Likewise.
2765 (vect_supported_load_permutation_p): Likewise.
2766 (vect_slp_analyze_operations): Likewise.
2767 (vect_slp_analyze_bb_1): Likewise.
2768 (vect_slp_bb): Likewise.
2769 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
2770 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
2771 (pass_slp_vectorize::execute): Likewise.
2772 (increase_alignment): Likewise.
2774 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2777 * ipa-inline.c (report_inline_failed_reason): Guard calls to
2778 cl_target_option_print_diff and cl_optimization_print_diff with
2781 2018-11-13 David Malcolm <dmalcolm@redhat.com>
2783 * doc/invoke.texi (-fsave-optimization-record): Note that the
2784 output is compressed.
2785 * optinfo-emit-json.cc: Include <zlib.h>.
2786 (optrecord_json_writer::write): Compress the output.
2788 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
2790 * tree-vrp.c (value_range_base::dump): Dump type.
2791 Do not use INF nomenclature for 1-bit types.
2792 (dump_value_range): Group all variants to common dumping code.
2793 (debug): New overloaded functions for value_ranges.
2794 (value_range_base::dump): Remove no argument version.
2795 (value_range::dump): Same.
2797 2018-11-13 Richard Biener <rguenther@suse.de>
2799 PR tree-optimization/87931
2800 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
2801 nested cycles we support to latch computations vectorizable_reduction
2804 2018-11-13 Martin Liska <mliska@suse.cz>
2806 PR tree-optimization/87885
2807 * cfghooks.c (account_profile_record): Rename
2809 (profile_record_check_consistency): ... this.
2810 Calculate missing num_mismatched_freq_in.
2811 (profile_record_account_profile): New function
2812 that calculates time and size of a function.
2813 * cfghooks.h (struct profile_record): Remove
2815 (struct cfg_hooks): Remove after_pass flag.
2816 (account_profile_record): Rename to ...
2817 (profile_record_check_consistency): ... this.
2818 (profile_record_account_profile): New.
2819 * cfgrtl.c (rtl_account_profile_record): Remove
2821 * passes.c (check_profile_consistency): Do only
2823 (account_profile): Calculate size and time of
2825 (pass_manager::dump_profile_report): Reformat
2827 (execute_one_ipa_transform_pass): Call
2828 consistency check before clean upand call account_profile
2829 after a clean up is done.
2830 (execute_one_pass): Call check_profile_consistency and
2831 account_profile instead of using after_pass flag..
2832 * tree-cfg.c (gimple_account_profile_record): Likewise.
2834 2018-11-13 Martin Liska <mliska@suse.cz>
2837 * config/i386/i386.c (ix86_option_override_internal): Error
2838 about usage -mabi=ms and -fsanitize={,kernel-}address.
2840 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
2842 * config/aarch64/aarch64.c
2843 (aarch64_vectorize_preferred_vector_alignment): Change return type to
2845 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
2846 alignment being a poly int.
2847 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
2848 return type to poly_uint64.
2849 * target.def (default_preferred_vector_alignment): Likewise.
2850 * targhooks.c (default_preferred_vector_alignment): Likewise.
2851 * targhooks.h (default_preferred_vector_alignment): Likewise.
2852 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
2853 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
2855 (vect_update_misalignment_for_peel): Likewise.
2856 (vect_enhance_data_refs_alignment): Likewise.
2857 (vect_find_same_alignment_drs): Likewise.
2858 (vect_duplicate_ssa_name_ptr_info): Likewise.
2859 (vect_setup_realignment): Likewise.
2860 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2862 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
2863 mask with a compile time variable vector alignment.
2864 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
2866 (vect_do_peeling): Exit early if vector alignment is not constant.
2867 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
2869 (vectorizable_store): Likewise.
2870 (vectorizable_load): Likweise.
2871 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
2873 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
2875 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2878 2018-11-13 Richard Biener <rguenther@suse.de>
2880 PR tree-optimization/87962
2881 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
2882 detect outer reduction for disqualifying in-loop uses.
2884 2018-11-13 Richard Biener <rguenther@suse.de>
2886 PR tree-optimization/87967
2887 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
2888 for constants for the scalar loop.
2890 2018-11-13 Alan Modra <amodra@gmail.com>
2892 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
2893 integer constants when -mcmodel=medium.
2895 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2897 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
2898 * config/arc/elf.h (PROFILE_HOOK): Define.
2899 * config/arc/linux.h (PROFILE_HOOK): Likewise.
2901 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2903 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
2904 jump table data in the text section.
2906 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2908 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
2909 to fit the eh_return pattern.
2910 * config/arc/arc.md (eh_return): Define.
2911 (VUNSPEC_ARC_EH_RETURN): Likewise.
2912 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
2914 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
2917 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2919 * common/config/arc/arc-common.c (arc_option_optimization_table):
2920 Millicode optimization is default on for size optimizations.
2921 * config/arc/arc-protos.h (arc_check_multi): New function.
2922 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
2923 (ENTER_LEAVE_START_REG): Define.
2924 (ENTER_LEAVE_END_REG): Likewise.
2925 (arc_override_options): Disable millicode when long calls option
2927 (arc_frame_info): Change it from int to bool.
2928 (arc_compute_frame_size): Clean up.
2929 (arc_save_restore): Remove.
2930 (frame_save_reg): New function.
2931 (frame_restore_reg): Likewise.
2932 (arc_enter_leave_p): Likewise.
2933 (arc_save_callee_saves): Likewise.
2934 (arc_restore_callee_saves): Likewise.
2935 (arc_save_callee_enter): Likewise.
2936 (arc_restore_callee_leave): Likewise.
2937 (arc_save_callee_milli): Likewise.
2938 (arc_restore_callee_milli): Likewise.
2939 (arc_expand_prologue): Reimplement to emit enter/leave
2941 (arc_expand_epilogue): Likewise.
2942 (arc_check_multi): New function.
2943 * config/arc/arc.md (push_multi_fp): New pattern.
2944 (push_multi_fp_blink): Likewise.
2945 (pop_multi_fp): Likewise.
2946 (pop_multi_fp_blink): Likewise.
2947 (pop_multi_fp_ret): Likewise.
2948 (pop_multi_fp_blink_ret): Likewise.
2949 * config/arc/arc.opt (mmillicode): Update option.
2950 (mcode-density-frame): New option.
2951 * config/arc/predicates.md (push_multi_operand): New predicate.
2952 (pop_multi_operand): Likewise.
2953 * doc/invoke.texi (ARC): Update ARC options information.
2956 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2958 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
2959 * config/arc/arc.c (operands_ok_ldd_std): New function.
2960 (mem_ok_for_ldd_std): Likewise.
2961 (gen_operands_ldd_std): Likewise.
2962 * config/arc/arc.md: Add peephole2 rules for std/ldd.
2964 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
2966 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
2967 test on stack_usage_file.
2968 (lang_dependent_init): Do not open the .su file if generating LTO.
2970 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2972 PR rtl-optimization/87918
2973 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
2974 simplify_gen_relational rather than simplify_gen_binary.
2976 2018-11-13 Richard Biener <rguenther@suse.de>
2978 * tree-ssanames.h (set_range_info): Use value_range_base.
2979 (get_range_info): Likewise.
2980 * tree-ssanames.c (set_range_info): Likewise.
2981 (get_range_info): Likewise.
2982 * tree-vrp.c (value_range_base::union_helper): Split
2983 out common parts of value_range[_base]::union_.
2984 (value_range_base::union_): Update.
2985 (value_range::union_): Likewise.
2986 (determine_value_range_1): Use value_range_base.
2987 (determine_value_range): Likewise.
2988 * tree-vrp.h (value_range_base::union_helper): Move ...
2989 (value_range::union_helper): ... from here.
2991 2018-11-13 Alan Modra <amodra@gmail.com>
2993 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
2995 (rs6000_secondary_reload_gpr): Don't call find_replacement.
2997 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
3000 * doc/invoke.texi (Optimize Options): Clarify that the
3001 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
3002 apply to data cache size.
3004 2018-11-13 Alan Modra <amodra@gmail.com>
3006 * config/rs6000/predicates.md (logical_const_operand),
3007 (logical_operand): Correct comment.
3008 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
3010 2018-11-13 Alan Modra <amodra@gmail.com>
3012 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
3014 2018-11-13 Alan Modra <amodra@gmail.com>
3016 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
3018 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
3019 covered by alternative.
3020 (movcc_internal1): Ignore h for register preference.
3021 (mov<mode>_hardfloat64): Likewise.
3022 (mov<mode>_softfloat): Ignore c, l, h for register preference.
3024 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
3026 PR preprocessor/47823
3027 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
3030 * doc/extend.texi (Pragmas): Note additional pragmas documented
3033 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3037 * tree-scalar-evolution.c (expression_expensive_p): Make
3038 BUILTIN POPCOUNT as expensive when backend does not define it.
3040 2018-11-12 Fredrik Noring <noring@nocrew.org>
3042 * config.gcc: Update with-llsc defaults for MIPS r5900.
3044 2018-11-12 Martin Liska <mliska@suse.cz>
3046 PR gcov-profile/87442
3047 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
3049 * doc/invoke.texi: Document them.
3050 * tree-profile.c (parse_profile_filter): New.
3051 (parse_profile_file_filtering): Likewise.
3052 (release_profile_file_filtering): Likewise.
3053 (include_source_file_for_profile): Likewise.
3054 (tree_profiling): Filter source files based on the
3055 newly added options.
3057 2018-11-12 Martin Liska <mliska@suse.cz>
3060 * doc/extend.texi: Add missing values for __builtin_cpu_is and
3061 __builtin_cpu_supports for x86 target.
3063 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
3066 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
3068 (Arithmetic): Likewise for "lo_sum".
3070 2018-11-12 Renlin Li <renlin.li@arm.com>
3073 * dse.c (get_stored_val): Add check for compile-time constantness
3076 2018-11-12 Sudakshina Das <sudi.das@arm.com>
3078 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
3079 (ARMv8_5a): New fgroup.
3080 (armv8.5-a): New arch.
3081 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
3083 * config/arm/arm-tables.opt: Regenerate.
3084 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
3085 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
3086 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
3087 Add matching rules for -march=armv8.5-a and extensions.
3088 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
3089 Add sb and predres to all armv8-a except armv8.5-a.
3091 2018-11-12 Richard Biener <rguenther@suse.de>
3093 * tree-vrp.h (value_range[_base]::set): Make public. Provide
3094 overload for single value.
3095 (value_range[_base]::set_nonnull): New.
3096 (value_range[_base]::set_null): Likewise.
3097 (value_range): Document bitmap copying behavior, mark
3098 copy constructor and assignment operator deleted.
3099 (value_range::move): New.
3100 (value_range::set_and_canonicalize): Default bitmap to zero.
3101 (set_value_range_to_nonnull): Remove.
3102 (set_value_range_to_null): Likewise.
3103 (set_value_range): Likewise.
3104 (set_value_range_to_value): Likewise.
3105 (extract_range_from_unary_expr): Work on value_range_base.
3106 (extract_range_from_binary_expr_1): Likewise. Rename to...
3107 (extract_range_from_binary_expr): ... this.
3108 * tree-vrp.c (value_range::update): Clear equiv bitmap
3110 (value_range::move): New, move equiv bitmap.
3111 (value_range_base::set_undefined): Avoid assignment.
3112 (value_range::set_undefined): Likewise.
3113 (value_range_base::set_varying): Likewise.
3114 (value_range::set_varying): Likewise.
3115 (set_value_range): Remove.
3116 (value_range_base::set): New overload for value.
3117 (value_range::set): Likewise.
3118 (set_value_range_to_nonnull): Remove.
3119 (value_range_base::set_nonnull): New.
3120 (value_range::set_nonnull): Likewise.
3121 (set_value_range_to_null): Remove.
3122 (value_range_base::set_null): New.
3123 (value_range::set_null): Likewise.
3124 (range_is_null): Work on value_range_base.
3125 (range_is_nonnull): Likewise.
3126 (ranges_from_anti_range): Likewise.
3127 (extract_range_into_wide_ints): Likewise.
3128 (extract_range_from_multiplicative_op): Likewise.
3129 (extract_range_from_binary_expr): Likewise. Update for API changes.
3130 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
3132 (value_range::intersect_helper): Avoid copy and assignment.
3133 (value_range::union_helper): Likewise.
3134 (determine_value_range_1): Adjust.
3135 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3136 Avoid assignment by using move.
3137 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
3138 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3140 * tree-ssanames.c (get_range_info): Likewise.
3141 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
3142 * vr-values.c (vr_values::get_value_range): Adjust.
3143 (vr_values::update_value_range): Likewise.
3144 (symbolic_range_based_on_p): Work on value_range_base.
3145 (vr_values::extract_range_from_binary_expr): Use value_range_base.
3146 (vr_values::extract_range_from_unary_expr): Likewise.
3147 (vr_values::extract_range_from_cond_expr): Avoid assignment.
3148 (vr_values::extract_range_from_comparison): Adjust.
3149 (vr_values::check_for_binary_op_overflow): Use value_range_base.
3150 (vr_values::extract_range_basic): Adjust.
3151 (vr_values::adjust_range_with_scev): Likewise.
3152 (vr_values::vrp_visit_assignment_or_call): Likewise.
3153 (vr_values::get_vr_for_comparison): Change API to avoid
3154 assignment and copy construction.
3155 (vr_values::compare_name_with_value): Adjust accordingly.
3156 (vr_values::compare_names): Likewise.
3157 (vr_values::extract_range_from_phi_node): Avoid assignment and
3158 bogus in-place modify of equiv bitmap.
3159 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
3160 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
3161 for extract_range_from_unary_expr API change.
3162 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
3164 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
3166 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
3168 2018-11-12 Richard Biener <rguenther@suse.de>
3170 * tree-vrp.h (value_range_base::symbolic_p,
3171 value_range_base::constant_p, value_range_base::zero_p,
3172 value_range_base::singleton_p): Move from value_range.
3173 (value_range::dump): Add.
3174 * gimple-ssa-evrp-analyze.c
3175 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
3176 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
3177 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3179 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
3180 (value_range::constant_p): Likewise.
3181 (value_range::singleton_p): Likewise.
3182 (value_range_base::dump): Add.
3183 (set_value_range_to_undefined): Remove.
3184 (set_value_range_to_varying): Likewise.
3185 (range_int_cst_p): Take value_range_base argument.
3186 (range_int_cst_singleton_p): Likewise.
3187 (value_range_constant_singleton): Likewise.
3188 (vrp_set_zero_nonzero_bits): Likewise.
3189 (extract_range_from_multiplicative_op): Use set_varying.
3190 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
3191 (extract_range_from_unary_expr): Likewise.
3192 (dump_value_range_base): Change to overload of dump_value_range.
3193 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
3194 (vrp_prop::visit_stmt): Likewise.
3195 (value_range::intersect_helper): Likewise.
3196 (value_range::union_helper): Likewise.
3197 (determine_value_range_1): Likewise.
3199 2018-11-12 Richard Biener <rguenther@suse.de>
3201 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
3202 (set_value_range_to_null): Likewise.
3203 * vr-values.c (vr_values::extract_range_from_comparison):
3204 Clear equiv for constant singleton ranges.
3206 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
3208 * config/i386/sse.md: Combine VFIXUPIMM* patterns
3209 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3210 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3211 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
3212 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3213 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3214 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
3216 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3219 * doc/extend.texi (Common Type Attributes): For the align type
3220 attribute, copy language about decreasing alignment from the
3221 corresponding variable attribute.
3223 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3225 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
3228 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3231 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
3232 -frtti and -fno-rtti code.
3234 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
3237 * doc/extend.texi (Other Builtins): Document probability associated
3238 with __builtin_expect.
3240 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
3243 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
3244 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
3245 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
3246 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
3248 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3250 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
3252 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
3254 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
3256 2018-11-11 Richard Biener <rguenther@suse.de>
3258 * tree-vrp.h (class value_range_base): New base class for
3259 value_range containing all but the m_equiv member.
3260 (dump_value_range_base): Add.
3261 (range_includes_zero_p): Work on value_range_base.
3262 * tree-vrp.c (value_range_base::set): Split out base handling
3264 (value_range::set): this.
3265 (value_range::set_equiv): New.
3266 (value_range_base::value_range_base): New constructors.
3267 (value_range_base::check): Split out base handling from...
3268 (value_range::check): this.
3269 (value_range::equal_p): Refactor in terms of
3270 ignore_equivs_equal_p which is now member of the base.
3271 (value_range_base::set_undefined): New.
3272 (value_range_base::set_varying): Likewise.
3273 (value_range_base::dump):Split out base handling from...
3274 (value_range::dump): this.
3275 (value_range_base::set_and_canonicalize): Split out base handling
3277 (value_range::set_and_canonicalize): this.
3278 (value_range_base::union_): New.
3279 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
3281 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
3282 instead of value_range everywhere.
3283 (ipcp_vr_lattice::print): Use dump_value_range_base.
3284 (ipcp_vr_lattice::meet_with): Adjust.
3285 (ipcp_vr_lattice::meet_with_1): Likewise.
3286 (ipa_vr_operation_and_type_effects): Likewise.
3287 (propagate_vr_across_jump_function): Likewise.
3288 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
3289 (ipa_get_value_range): Likewise.
3290 (ipa_set_jfunc_vr): Likewise.
3291 (ipa_compute_jump_functions_for_edge): Likewise.
3293 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
3296 * doc/invoke.texi (Optimize Options): Add @opindex entries
3297 for the positive forms of -fno-xxx and -mno-xxx options
3298 that were lacking them.
3300 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
3302 * combine.c (make_more_copies): Only make an intermediate copy if the
3303 dest of a move is a pseudo.
3305 2018-11-09 Maya Rashish <coypu@sdf.org>
3308 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
3309 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
3311 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
3315 * doc/invoke.texi (Optimize Options): Clarify default behavior
3316 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
3318 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
3320 PR tree-optimization/87940
3321 * expr.c (string_constant): Don't strip NOPS in subexpressions.
3322 Fold PLUS_EXPR correctly.
3324 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
3327 * config/s390/s390.c (s390_safe_relative_long_p): New function.
3328 (annotate_constant_pool_refs): Skip insns which support
3329 relative addressing.
3330 (annotate_constant_pool_refs_1): New helper function.
3331 (find_constant_pool_ref): Skip insns which support relative
3333 (find_constant_pool_ref_1): New helper function.
3334 (replace_constant_pool_ref): Skip insns which support
3335 relative addressing.
3336 (replace_constant_pool_ref_1): New helper function.
3337 (s390_mainpool_start): Adapt to the new signature.
3338 (s390_mainpool_finish): Likewise.
3339 (s390_chunkify_start): Likewise.
3340 (s390_chunkify_finish): Likewise.
3341 (pass_s390_early_mach::execute): Likewise.
3342 (s390_prologue_plus_offset): Likewise.
3343 (s390_emit_prologue): Likewise.
3344 (s390_emit_epilogue): Likewise.
3346 2018-11-09 Jakub Jelinek <jakub@redhat.com>
3348 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
3349 but unsupported lastprivate with conditional modifier.
3351 2018-11-09 Jeff Law <law@redhat.com>
3353 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
3354 unused argument better. Add gcc_unreachable to silence warning.
3356 2018-11-09 Martin Sebor <msebor@redhat.com>
3359 * attribs.c (has_attribute): New helper function.
3360 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
3361 * attribs.h (decls_mismatched_attributes): Declare.
3362 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
3363 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
3364 * common.opt (-Wattribute-alias): Take an argument.
3365 (-Wno-attribute-alias): New option.
3366 * doc/extend.texi (Common Function Attributes): Document copy.
3367 (Common Variable Attributes): Same.
3368 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
3369 (-Wattribute-alias): Document new option argument.
3371 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
3373 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
3374 matching criteria. Remove unused array initializer.
3376 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
3377 Jinsong Ji <jji@us.ibm.com>
3379 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
3380 constraints by introducing a new temporary.
3381 (_mm_cvtss_si64): Likewise.
3383 2018-11-09 Martin Liska <mliska@suse.cz>
3385 * common.opt: Add -fipa-stack-alignment flag.
3386 * doc/invoke.texi: Document it.
3387 * final.c (rest_of_clean_state): Guard stack
3388 shrinking with flag.
3390 2018-11-09 Martin Liska <mliska@suse.cz>
3392 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
3394 (ipa_discover_variable_flags): ... this.
3395 * common.opt: Come up with new flag -fipa-reference-addressable.
3396 * doc/invoke.texi: Document it.
3397 * ipa-reference.c (propagate): Call the renamed fn.
3398 * ipa-visibility.c (whole_program_function_and_variable_visibility):
3400 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
3402 (ipa_discover_variable_flags): ... this. Discover
3403 non-addressable variables only with the newly added flag.
3404 * opts.c: Enable the newly added flag with -O1 and higher
3407 2018-11-09 David Malcolm <dmalcolm@redhat.com>
3409 * json.cc (selftest::test_writing_literals): Fix comment.
3411 2018-11-09 Martin Liska <mliska@suse.cz>
3413 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
3414 string to a stack buffer.
3415 (aarch64_parse_cpu): Likewise.
3416 (aarch64_parse_tune): Likewise.
3418 2018-11-09 Richard Biener <rguenther@suse.de>
3420 PR tree-optimization/87953
3421 * tree-vect-loop.c (vectorizable_reduction): For analysis
3422 always pass ops[0] to vectorizable_condition.
3424 2018-11-09 Stafford Horne <shorne@gmail.com>
3425 Richard Henderson <rth@twiddle.net>
3426 Joel Sherrill <joel@rtems.org>
3428 * common/config/or1k/or1k-common.c: New file.
3429 * config/or1k/*: New.
3430 * config.gcc (or1k*-*-*): New.
3431 * configure.ac (or1k*-*-*): New test for openrisc tls.
3432 * configure: Regenerated.
3433 * doc/install.texi: Document OpenRISC triplets.
3434 * doc/invoke.texi: Document OpenRISC arguments.
3435 * doc/md.texi: Document OpenRISC.
3437 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
3439 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
3440 (arm7tdmi-s): Delete CPU.
3441 (arm710t): Add aliases for arm720t and arm740t.
3442 (arm720t, arm740t): Delete CPUs.
3443 (arm920t): Add aliases for arm920, arm922t and arm940t.
3444 (arm920, arm922t, arm940t): Delete CPUs.
3445 (arm10tdmi): Add alias for arm1020t.
3446 (arm1020t): Delete CPU.
3447 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
3448 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
3449 (arm10e): Add aliases for arm1020e and arm1022e.
3450 (arm1020e, arm1022e): Delete CPU.
3451 * config/arm/arm.md (generic_sched): Remove entries that are now
3453 (generic_vfp): Likewise.
3454 * config/arm/arm1020e.md: Simplify tuning selection based on alias
3456 * config/arm/arm-tune.md: Regenerated.
3457 * config/arm/arm-tables.opt: Regenerated.
3459 2018-11-09 Richard Biener <rguenther@suse.de>
3461 PR tree-optimization/87621
3462 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
3463 op with only phi inputs.
3464 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
3465 (ch_base::copy_headers): Run CSE on copied loop headers.
3466 (pass_ch_vect::process_loop_p): Simplify.
3468 2018-11-09 Alexandre Oliva <oliva@adacore.com>
3470 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
3471 for non-w64 x86_64 biarch.
3473 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
3475 PR rtl-optimization/86438
3476 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
3477 of in_b for the compare if in_b is SET_DEST.
3480 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
3481 non-toplevel UNSPEC.
3483 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3485 * tree-vrp.c (value_range::check): Do not access internals
3487 (value_range::singleton_p): Same.
3488 (value_range::type): Same.
3489 (vrp_finalize): Use value_range API.
3491 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3493 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
3495 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3497 * vr-values.c (vr_values::get_value_range): Use value_range API
3498 instead of piecing together ranges.
3499 (vr_values::update_value_range): Same.
3501 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
3503 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
3504 of performing ad-hoc calculations.
3505 * tree-ssanames.c (set_range_info): New overloaded function
3506 accepting value_range &.
3507 (get_range_info): Same.
3508 * tree-ssanames.h (set_range_info_raw): Remove.
3509 (set_range_info): New prototype.
3510 (get_range_info): Same.
3511 * tree-vrp.h (value_range::null_p): Rename to zero_p.
3512 * tree-vrp.c (value_range::null_p): Same.
3514 2018-11-09 Jan Hubicka <jh@suse.cz>
3516 * tree.c (fld_type_variant_equal_p): Test user align flag.
3517 (flt_type_variant): Copy user align flag.
3518 (fld_incomplete_type_of): Clear it.
3520 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3522 * config/arm/neon.md (div<mode>3): New pattern.
3524 2018-11-08 Andi Kleen <ak@linux.intel.com>
3526 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
3527 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
3528 (ix86_handle_option): Handle OPT_mptwrite.
3529 * config/i386/cpuid.h (bit_PTWRITE): Add.
3530 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
3531 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
3532 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
3533 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
3534 (ix86_option_override_internal): Handle PTA_PTWRITE.
3535 (ix86_valid_target_attribute_inner_p): Define ptwrite.
3536 (def_builtin2): Force UINT64 to be 64bit only.
3537 * config/i386/i386.h (TARGET_PTWRITE): Add.
3538 (TARGET_PTWRITE_P): Add.
3540 * config/i386/i386.md: Define ptwrite.
3541 * config/i386/i386.opt: Add -mptwrite.
3542 * config/i386/immintrin.h (_ptwrite64): Add.
3544 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
3545 * doc/invoke.texi: Document -mptwrite.
3547 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
3549 PR rtl-optimization/87600
3550 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
3551 * lra-constraints.c (process_alt_operands): Skip illegal hard
3552 register usage. Prefer reloading non hard register operands.
3554 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
3557 * doc/invoke.texi (Optimize Options): Clarify default behavior
3558 for -fno-sched-interblock and -fno-sched-spec.
3560 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
3562 * collect2.c (linker_select): Add USE_LLD_LD.
3563 (ld_suffixes): Add ld.lld.
3564 (main): Handle -fuse-ld=lld.
3565 * common.opt (-fuse-ld=lld): New option.
3566 * doc/invoke.texi (-fuse-ld=lld): Document.
3567 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
3569 2018-11-08 Paul Koning <ni1d@arrl.net>
3571 * config/pdp11/constraints.md: Add "Z" series constraints for use
3572 with pre-dec and post-inc addressing.
3573 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
3574 (pdp11_expand_operands): Add int argument (word count).
3575 (pdp11_sp_frame_offset): Delete.
3576 (pdp11_cmp_length): New function.
3577 (pushpop_regeq): New function.
3578 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
3580 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
3582 (pdp11_initial_elimination_offset): Ditto.
3583 (pdp11_expand_operands): Add word count argument. Bugfixes.
3584 (output_move_multiple): Change how pointer adjustment is done.
3585 (pdp11_gen_int_label): Correct format.
3586 (output_ascii): Ditto.
3587 (pdp11_asm_output_var): Add code for DEC assembler case.
3588 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
3590 (legitimate_const_double_p): Ditto.
3591 (pdp11_register_move_cost): Adjust for new register classes.
3592 (pdp11_regno_reg_class): Ditto.
3593 (expand_block_move): Delete.
3594 (pushpop_regeq): New function.
3595 (pdp11_legitimate_address_p): Bugfix in check for constant
3597 (pdp11_sp_frame_offset): Delete.
3598 (pdp11_reg_save_size): New helper function for new frame layout.
3599 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
3600 (pdp11_expand_shift): Bugfix in check for constant shift count.
3601 (pdp11_shift_length): Ditto.
3602 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
3603 (pdp11_cmp_length): New function.
3604 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
3605 some compile options.
3606 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
3607 (CALL_USED_REGISTERS): Ditto.
3608 (ELIMINABLE_REGS): Ditto.
3609 (REGISTER_NAMES): Ditto.
3610 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
3612 (REG_CLASS_NAMES): Ditto.
3613 (REG_CLASS_CONTENTS): Ditto. Also remove
3614 HARD_FRAME_POINTER_REGNUM.
3615 (CPU_REG_CLASS): New macro.
3616 (CLASS_MAX_NREGS): Adjust for new register classes.
3617 (FUNCTION_PROFILER): Make no-op.
3618 (may_call_alloca): Remove unused declaration.
3619 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
3620 (ASM_OUTPUT_SKIP): Fix format.
3621 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
3622 (HARD_FRAME_POINTER_REGNUM): Remove.
3624 (*rts): Rename. Remove epilogue related checks.
3625 (cmpsi, cmpdi): New insn.
3626 (cbranch<mode>4): Change to apply to SI and DI modes as well.
3627 (mov<mode>): Change constraints to enforce that push/pop
3628 destination cannot use the same register as source.
3629 (*mov<mode><cc_cc>): Ditto.
3630 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
3631 at assembly output rather than as RTL expander.
3632 (zero_extendqihi2): Bugfix in check for same registers.
3633 (adddi3_nocc): Bugfix in check for constant operand.
3634 (addsi3_nocc): Ditto.
3635 (subdi3_nocc): Ditto.
3636 (subsi3_nocc): Ditto.
3637 (negdi2_nocc): Copy input to pdp11_expand_operands.
3638 (negsi2_nocc): Ditto.
3639 (bswap2_nocc): Ditto.
3640 * config/pdp11/pdp11.opt (mlra): Fix documentation.
3641 * config/pdp11/t-pdp11: Use -Os.
3643 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
3645 * config/arm/parsecpu.awk (/alias/): New parsing rule.
3646 (/begin cpu/): Check that the cpu name hasn't been previously defined.
3647 (gen_comm_data): Print out CPU alias tables.
3648 (check_cpu): Match aliases when checking the CPU name.
3649 * config/arm/arm-protos.h (cpu_alias): New structure.
3650 (cpu_option): Add entry for aliases.
3651 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
3652 strongarm1100 and strongarm1110.
3653 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
3654 (config/arm/arm-generic.md): Remove redundant references to
3655 strongarm110, strongarm1100 and strongarm1110.
3656 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
3657 Scan aliases for additional hints.
3658 (arm_parse_cpu_option_name): Also match a cpu name against the list
3660 * config/arm/arm-tables.opt: Regenerated.
3661 * config/arm/arm-tune.md: Regenerated.
3663 2018-11-08 Jakub Jelinek <jakub@redhat.com>
3665 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
3666 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
3667 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3668 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
3669 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
3670 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
3671 * gengtype.c (open_base_files): Add omp-general.h.
3672 * gimple.c (gimple_build_omp_critical):
3673 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
3674 gimple_omp_taskgroup_set_clauses.
3675 (gimple_build_omp_atomic_load): Add mo argument, call
3676 gimple_omp_atomic_set_memory_order.
3677 (gimple_build_omp_atomic_store): Likewise.
3678 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
3679 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
3681 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
3682 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
3683 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
3684 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
3685 different value for GF_OMP_ATOMIC_NEED_VALUE.
3686 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
3688 (struct gimple_statement_omp_single_layout): And remove here.
3689 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
3690 than gimple_statement_omp_single_layout.
3691 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
3693 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
3694 (gimple_omp_subcode): Formatting fix.
3695 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
3696 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
3697 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
3698 gimple_omp_teams_host, gimple_omp_teams_set_host,
3699 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
3700 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
3701 gimple_omp_taskgroup_set_clauses): New inline functions.
3702 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
3703 (gimple_build_omp_atomic_store): Likewise.
3704 (gimple_omp_atomic_seq_cst_p): Remove.
3705 (gimple_omp_atomic_memory_order): New function.
3706 (gimple_omp_atomic_set_seq_cst): Remove.
3707 (gimple_omp_atomic_set_memory_order): New function.
3708 (gimple_build_omp_taskgroup): Add clauses argument.
3709 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
3710 (dump_gimple_omp_task): Print taskwait with depend clauses.
3711 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
3712 dump_omp_atomic_memory_order.
3713 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
3714 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
3715 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
3716 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
3717 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
3718 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
3719 ORT_UNTIED_TASKLOOP enumerators.
3720 (enum gimplify_defaultmap_kind): New.
3721 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
3722 target_map_pointers_as_0len_arrays members, add defaultmap.
3723 (new_omp_context): Initialize defaultmap member.
3724 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3725 (maybe_fold_stmt): Don't fold even in host teams regions.
3726 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
3727 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
3728 ctx->omp_firstprivatize_variable.
3729 (omp_add_variable): Don't add private/firstprivate for VLAs in
3731 (omp_default_clause): Print "taskloop" rather than "task" if
3733 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3734 Handle new defaultmap clause kinds.
3735 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
3736 iterator to be lastprivate or private. Fix up diagnostics if linear
3737 is used on collapse>1 simd iterator.
3738 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
3739 (gimplify_omp_depend): New function.
3740 (gimplify_scan_omp_clauses): Add shared clause on parallel for
3741 combined parallel master taskloop{, simd} if taskloop has
3742 firstprivate, lastprivate or reduction clause. Handle
3743 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
3744 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
3745 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
3746 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
3747 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
3749 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
3750 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
3751 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
3752 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3753 (gimplify_omp_task): Handle taskwait with depend clauses.
3754 (gimplify_omp_for): Add shared clause on parallel for combined
3755 parallel master taskloop{, simd} if taskloop has firstprivate,
3756 lastprivate or reduction clause. Use ORT_TASKLOOP or
3757 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
3758 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
3759 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
3760 __for_end and __for_range temporaries on OMP_PARALLEL for
3761 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
3762 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
3763 sandwiched in between two taskloops.
3764 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
3765 instead of ctx->omp_firstprivatize_variable.
3766 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
3767 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
3768 host teams, use gimplify_and_return_first etc. for body like
3769 for target or target data constructs, and at the end call
3770 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
3771 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
3772 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
3773 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
3774 gimple_omp_atomic_set_seq_cst calls.
3775 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
3776 case, handle taskgroup clauses.
3777 * lto-streamer-out.c (hash_tree): Handle
3778 OMP_CLAUSE_{TASK,IN}_REDUCTION.
3779 * Makefile.in (GTFILES): Add omp-general.h.
3780 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
3781 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
3782 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
3783 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
3784 BUILT_IN_GOMP_LOOP_DOACROSS_START,
3785 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
3786 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3787 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
3788 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
3789 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
3790 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
3791 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
3792 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
3793 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
3794 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
3795 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
3796 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
3797 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
3798 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
3799 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
3800 * omp-expand.c (workshare_safe_to_combine_p): Return false for
3801 non-worksharing loops.
3802 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
3803 (determine_parallel_type): Don't combine parallel with worksharing
3804 which has _reductemp_ clause.
3805 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
3806 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
3807 if there is nonmonotonic modifier or if there is no modifier and no
3808 ordered clause. For dynamic and guided schedule without monotonic
3809 and nonmonotonic modifier, default to nonmonotonic.
3810 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
3811 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
3813 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
3814 there are any reduction clauses.
3815 (expand_taskwait_call): New function.
3816 (expand_teams_call): New function.
3817 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
3818 expand_teams_call for it. Formatting fix. Handle taskwait with
3820 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
3821 of worksharing loops with task reductions.
3822 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
3823 expansion of worksharing loops with task reductions.
3824 (expand_omp_sections): Handle expansion of sections with task
3826 (expand_omp_synch): For host teams call expand_omp_taskreg.
3827 (omp_memory_order_to_memmodel): New function.
3828 (expand_omp_atomic_load, expand_omp_atomic_store,
3829 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
3830 instead of gimple_omp_atomic_seq_cst_p.
3831 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
3832 depend clauses as a standalone directive.
3833 * omp-general.c (enum omp_requires): New variable.
3834 (omp_extract_for_data): Initialize have_reductemp member. Allow
3835 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
3836 GT_EXPR loops depending on incr sign. Formatting fixes.
3837 * omp-general.h (struct omp_for_data): Add have_reductemp member.
3838 (enum omp_requires): New enum.
3839 (omp_requires_mask): Declare.
3840 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
3842 * omp-low.c (struct omp_context): Add task_reductions and
3843 task_reduction_map fields.
3844 (is_host_teams_ctx): New function.
3845 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
3846 (use_pointer_for_field): Use is_global_var instead of
3847 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
3849 (build_outer_var_ref): Ignore taskgroup outer contexts.
3850 (delete_omp_context): Release task_reductions and task_reduction_map.
3851 (scan_sharing_clauses): Don't add any fields for reduction clause on
3852 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
3853 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
3854 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
3855 Handle OMP_CLAUSE_NONTEMPORAL.
3856 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
3858 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
3859 clauses with task modifier.
3860 (scan_omp_task): Handle taskwait with depend clauses.
3861 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
3862 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
3863 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
3864 (scan_omp_for): Fix comment formatting.
3865 (scan_omp_teams): Handle host teams constructs.
3866 (check_omp_nesting_restrictions): Allow teams with no outer
3867 OpenMP context. Adjust diagnostics for teams strictly nested into
3868 some explicit OpenMP construct other than target. Allow OpenMP atomics
3869 inside of simd regions.
3870 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
3871 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
3872 taskreg_nesting_level while scanning host teams construct.
3873 (task_reduction_read): New function.
3874 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
3875 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
3876 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
3877 second argument create_tmp_var if it is NULL. Don't ignore shared
3878 clauses in is_host_teams_ctx contexts. Handle
3879 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
3881 (lower_reduction_clauses): Ignore reduction clauses with task
3882 modifier. Remove second argument create_tmp_var if it is NULL.
3883 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
3884 (lower_send_clauses): Ignore reduction clauses with task modifier.
3885 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
3886 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
3887 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
3888 rely that it is the last stmt in body so far. Ignore outer taskgroup
3890 (omp_task_reductions_find_first, omp_task_reduction_iterate,
3891 lower_omp_task_reductions): New functions.
3892 (lower_omp_sections): Handle reduction clauses with taskgroup
3893 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
3894 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
3895 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
3897 (lower_omp_taskgroup): Handle taskgroup reductions.
3898 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
3899 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
3900 (lower_depend_clauses): If there are any
3901 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3902 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
3903 seen, assume lowering is done already and return early. Set kind
3904 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
3905 (lower_omp_taskreg): Handle reduction clauses with task modifier on
3906 parallel construct. Handle reduction clause on taskloop construct.
3907 Handle taskwait with depend clauses.
3908 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
3909 for host teams constructs.
3910 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
3911 nontemporal and _reductemp_ clause entries.
3912 (omp_clause_code_name): Likewise.
3913 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
3914 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
3915 * tree-core.h (enum omp_clause_code): Add
3916 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
3917 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
3918 (struct tree_base): Add omp_atomic_memory_order field into union.
3919 Remove OMP_ATOMIC_SEQ_CST comment.
3920 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3921 and OMP_CLAUSE_DEPEND_DEPOBJ.
3922 (struct tree_omp_clause): Add subcode.defaultmap_kind.
3923 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
3924 OpenMP constructs with body and clauses operands.
3925 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
3926 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
3927 (OMP_TASKGROUP_CLAUSES): Define.
3928 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
3929 OMP_CLAUSE__LOOPTEMP_.
3930 (OMP_ATOMIC_SEQ_CST): Remove.
3931 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
3932 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
3933 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
3934 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
3935 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
3936 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
3937 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
3938 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
3939 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
3940 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
3942 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
3943 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
3944 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
3945 (convert_local_omp_clauses): Likewise. Remove useless test.
3946 * tree-parloops.c (create_call_for_reduction_1): Pass
3947 OMP_MEMORY_ORDER_RELAXED as new argument to
3948 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
3949 * tree-pretty-print.c (dump_omp_iterators): New function.
3950 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
3951 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
3952 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
3953 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
3954 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
3955 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
3956 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
3957 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
3958 (dump_omp_atomic_memory_order): New function.
3959 (dump_generic_node): Use it. Print taskgroup clauses. Print
3960 taskwait with depend clauses.
3961 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
3962 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
3963 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
3964 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
3965 write_ts_omp_clause_tree_pointers): Likewise.
3967 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3970 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
3971 "-missed", "-note", and "-all" sub-options.
3972 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
3974 (can_early_inline_edge_p): Likewise.
3975 (want_early_inline_function_p): Likewise.
3976 (want_inline_self_recursive_call_p): Likewise.
3977 (recursive_inlining): Likewise.
3978 (inline_small_functions): Likewise.
3979 (flatten_function): Likewise.
3980 (ipa_inline): Likewise.
3981 (inline_always_inline_functions): Likewise.
3982 (early_inline_small_functions): Likewise.
3983 (early_inliner): Likewise.
3984 * tree-inline.c (expand_call_inline): Likewise.
3986 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3988 * pretty-print.c (pp_format): Handle %f.
3989 (selftest::test_pp_format): Add test of %f.
3990 * pretty-print.h (pp_double): New macro.
3992 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3994 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
3995 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
3996 (dump_pretty_printer::decode_format): Implement "%C" for
3998 (selftest::test_capture_of_dump_calls): Rename "where" to
3999 "stmt_loc". Convert test_decl to a function decl and set its
4000 location. Add a symbol_table_test RAII instance and a
4001 cgraph_node, using it to test "%C" and dump_symtab_node.
4003 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
4006 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
4008 2018-11-08 David Malcolm <dmalcolm@redhat.com>
4010 * cgraph.c: Include "selftest.h".
4011 (saved_symtab): New variable.
4012 (selftest::symbol_table_test::symbol_table_test): New ctor.
4013 (selftest::symbol_table_test::~symbol_table_test): New dtor.
4014 (selftest::test_symbol_table_test): New test.
4015 (selftest::cgraph_c_tests): New.
4016 * cgraph.h (saved_symtab): New decl.
4017 (selftest::symbol_table_test): New class.
4018 * selftest-run-tests.c (selftest::run_tests): Call
4019 selftest::cgraph_c_tests.
4020 * selftest.h (selftest::cgraph_c_tests): New decl.
4022 2018-11-08 Richard Biener <rguenther@suse.de>
4024 * tree-data-ref.h (lambda_int): New typedef.
4025 (lambda_vector_gcd): Adjust.
4026 (lambda_vector_new): Likewise.
4027 (lambda_matrix_new): Likewise.
4028 * tree-data-ref.c (print_lambda_vector): Adjust.
4030 2018-11-08 Richard Biener <rguenther@suse.de>
4032 PR tree-optimization/87929
4033 * tree-complex.c (expand_complex_comparison): Clean EH.
4035 2018-11-08 Martin Liska <mliska@suse.cz>
4037 * doc/extend.texi: Reword.
4038 * predict.c (expr_expected_value_1): Likewise.
4040 2018-11-08 Richard Biener <rguenther@suse.de>
4042 PR tree-optimization/87913
4043 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
4044 of extreme values to ordered comparisons.
4046 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4049 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
4051 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4053 * doc/invoke.texi: Remove leading dash from @opindex entries
4054 throughout the file.
4056 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
4059 * doc/invoke.texi (Option Summary): Add -e and --entry.
4060 (Link Options): Likewise.
4062 2018-11-07 Nathan Sidwell <nathan@acm.org>
4065 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
4066 --disable-checking bootstrap.
4068 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
4070 * configure: Regenerated.
4072 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4075 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
4076 to the mode of the widest field iff the widest field has mode class
4077 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
4080 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
4082 * common.opt: Add -Wattribute-warning.
4083 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
4084 * expr.c (expand_expr_real_1): Add new attribute to warning_at
4085 call to allow user configure behavior of "warning" attribute.
4087 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
4089 * target.def: Put @: after every vs., e.g., and i.e. where it is
4090 followed by whitespace.
4091 * doc/extend.texi: Ditto.
4092 * doc/fragments.texi: Ditto.
4093 * doc/gimple.texi: Ditto.
4094 * doc/implement-c.texi: Ditto.
4095 * doc/install.texi: Ditto.
4096 * doc/invoke.texi: Ditto.
4097 * doc/md.texi: Ditto.
4098 * doc/plugins.texi: Ditto.
4099 * doc/rtl.texi: Ditto.
4100 * doc/sourcebuild.texi: Ditto.
4101 * doc/tm.texi.in: Ditto.
4102 * doc/ux.texi: Ditto.
4103 * doc/tm.texi: Regenerate.
4105 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4107 * config/arm/arm-cpus.in (ares): New entry.
4108 * config/arm/arm-tables.opt: Regenerate.
4109 * config/arm/arm-tune.md: Likewise.
4110 * doc/invoke.texi (ARM Options): Document ares.
4112 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4114 * config/aarch64/aarch64-cores.def (ares): Define.
4115 * config/aarch64/aarch64-tune.md: Regenerate.
4116 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
4118 2018-11-07 Jan Hubicka <jh@suse.cz>
4120 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
4121 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
4122 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
4125 2018-11-07 Richard Biener <rguenther@suse.de>
4127 PR tree-optimization/87914
4128 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
4130 (vectorizable_reduction): Handle shifts and rotates by dispatching
4131 to vectorizable_shift.
4132 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
4133 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
4135 (vectorizable_shift): Export and handle being called as
4137 (vect_analyze_stmt): Call vectorizable_shift after
4138 vectorizable_reduction.
4139 * tree-vectorizer.h (vectorizable_shift): Declare.
4141 2018-11-07 Jan Hubicka <jh@suse.cz>
4143 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
4144 than const decls in TREE_VALUE of enum.
4145 (dump_type_inheritance_graph): Improve duplicate dumping.
4146 (free_enum_values): New.
4147 (build_type_inheritance_graph): Use it.
4148 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
4149 which are not main variants or not ODR types.
4150 (verify_type_variant): Expect variants to have no TYPE_VALUES.
4152 2018-11-07 Richard Biener <rguenther@suse.de>
4154 * ipa-inline.c (want_inline_small_function_p): Compute
4155 big_speedup_p lazily and last.
4157 2018-11-07 Jan Hubicka <jh@suse.cz>
4159 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
4160 building incomplete variant of complete type.
4161 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
4162 variant of complete type.
4164 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4166 * config/mips/mips.c: Fix typo in documentation of
4167 mips_loongson_ext2_prefetch_cookie.
4168 (mips_option_override): fix brain twister logical.
4169 * config/mips/mips.h: Fix typo in documentation of
4170 ISA_HAS_CTZ_CTO and define pattern.
4171 * config/mips/mips.md (prefetch): Hoist EXT2 above
4173 (prefetch_indexed): Hoist EXT2 above the EXT block.
4175 2018-11-07 Jan Hubicka <jh@suse.cz>
4177 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
4178 return and parameter types of function and method types.
4179 (free_lang_data_in_cgraph): Update.
4181 2018-11-07 Martin Liska <mliska@suse.cz>
4183 PR rtl-optimization/87868
4184 * postreload-gcse.c (eliminate_partially_redundant_load): Set
4185 threshold to max_count if we would overflow.
4186 * profile-count.h: Make max_count a public constant.
4188 2018-11-07 Martin Liska <mliska@suse.cz>
4190 * mem-stats.h: Fix GNU coding style.
4192 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4194 * config/mips/gs264e.md: New.
4195 * config/mips/mips-cpus.def: Define gs264e.
4196 * config/mips/mips-tables.opt: Regenerate.
4197 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
4199 (mips_issue_rate): Add support for gs264e.
4200 (mips_multipass_dfa_lookahead): Likewise.
4201 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
4202 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
4203 (MIPS_ASE_MSA_SPEC): New.
4204 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
4205 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
4206 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
4207 * config/mips/mips.md: Include gs264e.md.
4208 (processor): Add gs264e.
4209 * config/mips/mips.opt (MSA): Use Mask instead of Var.
4210 * doc/invoke.texi: Add gs264e to supported architectures.
4212 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4214 * config/mips/gs464e.md: New.
4215 * config/mips/mips-cpus.def: Define gs464e.
4216 * config/mips/mips-tables.opt: Regenerate.
4217 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
4219 (mips_issue_rate): Add support for gs464e.
4220 (mips_multipass_dfa_lookahead): Likewise.
4221 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
4222 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
4223 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
4224 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
4225 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
4226 * config/mips/mips.md: Include gs464e.md.
4227 (processor): Add gs464e.
4228 * doc/invoke.texi: Add gs464e to supported architectures.
4230 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4232 * config/mips/loongson3a.md: Rename to ...
4233 * config/mips/gs464.md: ... here.
4234 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
4235 as an alias of gs464 processor.
4236 * config/mips/mips-tables.opt: Regenerate.
4237 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
4238 instead of PROCESSOR_LOONGSON_3A.
4239 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
4241 (mips_option_override): Enable MMI and EXT for gs464.
4242 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
4243 Rename TUNE_LOONGSON_3A to TUNE_GS464.
4244 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
4245 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
4246 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
4248 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
4249 (processor): Add gs464;
4250 * doc/invoke.texi: Add gs464 to supported architectures.
4252 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4254 * config/mips/mips-protos.h
4255 (mips_loongson_ext2_prefetch_cookie): New prototype.
4256 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
4257 (mips_option_override): Enable TARGET_LOONGSON_EXT when
4258 TARGET_LOONGSON_EXT2 is true.
4259 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
4260 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
4261 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
4262 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
4263 TARGET_LOONGSON_EXT2.
4264 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
4265 (define_insn "ctz<mode>2"): New insn pattern.
4266 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
4267 (define_insn "prefetch_indexed_<mode>"): Include
4268 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
4269 * config/mips/mips.opt (-mloongson-ext2): Add option.
4270 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
4272 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4274 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
4275 __mips_loongson_ext.
4276 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
4277 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
4279 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
4280 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
4281 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
4282 instead of TARGET_LOONGSON_3A.
4283 * config/mips/mips.opt (-mloongson-ext): Add option.
4284 * gcc/doc/invoke.texi (-mloongson-ext): Document.
4286 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
4288 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
4289 * config/mips/loongson.md: Move to ...
4290 * config/mips/loongson-mmi.md: here; Adjustment.
4291 * config/mips/loongson.h: Move to ...
4292 State as deprecated. Include loongson-mmiintrin.h for back
4293 compatibility and warning.
4294 * config/mips/loongson-mmiintrin.h: ... here.
4295 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
4296 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
4297 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
4298 (mips_option_override): Make sure MMI use hard float;
4299 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
4300 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
4301 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
4302 TARGET_LOONGSON_VECTORS.
4303 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
4304 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
4305 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
4306 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
4308 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
4309 TARGET_LOONGSON_VECTORS.
4310 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
4311 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
4312 (Loongson MMI patterns): Include loongson-mmi.md instead of
4314 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
4315 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
4317 2018-11-07 Richard Biener <rguenther@suse.de>
4320 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
4321 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
4323 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
4325 PR rtl-optimization/87874
4326 * lra.c (lra_substitute_pseudo): Do not create a subreg for
4329 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
4331 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
4332 if not in indexed or indirect form.
4333 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
4334 (bswap<mode>2_store): Ditto.
4336 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
4338 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
4341 2018-11-06 Richard Biener <rguenther@suse.de>
4343 PR tree-optimization/86850
4344 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
4345 instead of src.m_vec.
4347 2018-11-06 Jan Hubicka <jh@suse.cz>
4349 * tree.c (fld_simplified_type_name): Break out form ...
4350 (free_lang_data_in_type): ... here.
4351 (fld_type_variant_equal_p): Use it.
4353 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4355 * config/default-d.c: Include memmodel.h.
4357 * config/sol2-d.c: New file.
4358 * config/t-sol2 (sol2-d.o): New rule.
4359 * config.gcc <*-*-solaris2*>: Set d_target_objs,
4360 target_has_targetdm.
4362 2018-11-06 Jan Hubicka <jh@suse.cz>
4364 * tree.c (fld_type_variant): Also copy alignment; be sure that
4365 new variant is equal.
4367 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
4370 * config/s390/s390.md: Add relative_long attribute.
4372 2018-11-06 Jan Hubicka <jh@suse.cz>
4374 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
4375 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
4376 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
4377 stream TYPE_NEEDS_CONSTRUCTING.
4378 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
4379 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
4381 2018-11-06 Richard Biener <rguenther@suse.de>
4383 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
4385 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
4387 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
4389 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
4390 REG_CLASS_CONTENTS[GEN_REGS].
4391 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
4393 2018-11-06 Jan Hubicka <jh@suse.cz>
4395 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
4397 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4400 * config/sol2.h (ASAN_CC1_SPEC): Define.
4401 (LD_WHOLE_ARCHIVE_OPTION): Define.
4402 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
4403 (ASAN_REJECT_SPEC): Provide default.
4404 (LIBASAN_EARLY_SPEC): Define.
4405 (LIBTSAN_EARLY_SPEC): Define.
4406 (LIBLSAN_EARLY_SPEC): Define.
4407 * config/i386/sol2.h (CC1_SPEC): Redefine.
4408 (ASAN_REJECT_SPEC): Define.
4410 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
4411 (TARGET_ASAN_SHADOW_OFFSET): Define.
4412 (sparc_asan_shadow_offset): New function.
4413 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
4414 (ASAN_REJECT_SPEC): Define.
4416 2018-11-06 Jan Hubicka <jh@suse.cz>
4418 * tree.c (fld_type_variant): Copy canonical type.
4419 (fld_incomplete_type_of): Check that canonical types looks sane;
4420 copy canonical type.
4421 (verify_type): Accept when incomplete type has complete canonical type.
4423 2018-11-06 Jan Hubicka <jh@suse.cz>
4425 * tree.c (free_lang_data): Reset overwite_assembler_name,
4426 print_xnode, print_decl, print_type and print_identifier of
4429 2018-11-06 Richard Biener <rguenther@suse.de>
4431 PR tree-optimization/87889
4432 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
4433 Do nothing if old and new arg are the same
4435 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
4438 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
4439 attributes for operands 3 and 4.
4441 2018-11-06 Richard Biener <rguenther@suse.de>
4444 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
4445 matching bitfield insertion.
4447 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
4449 * auto-inc-dec.c: Include valtrack.h. Improve comments.
4450 (reg_next_debug_use): New.
4451 (attempt_change): Propagate adjusted expression into affected
4453 (merge_in_block): Track uses in debug insns.
4454 (pass_inc_dec::execute): Allocate and release
4457 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
4459 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
4460 (_mm512_fixupimm_round_pd): Update parameters and builtin.
4461 (_mm512_maskz_fixupimm_round_pd): Ditto.
4462 (_mm512_fixupimm_round_ps): Ditto.
4463 (_mm512_maskz_fixupimm_round_ps): Ditto.
4464 (_mm_fixupimm_round_sd): Ditto.
4465 (_mm_maskz_fixupimm_round_sd): Ditto.
4466 (_mm_fixupimm_round_ss): Ditto.
4467 (_mm_maskz_fixupimm_round_ss): Ditto.
4468 (_mm512_fixupimm_pd): Ditto.
4469 (_mm512_maskz_fixupimm_pd): Ditto.
4470 (_mm512_fixupimm_ps): Ditto.
4471 (_mm512_maskz_fixupimm_ps): Ditto.
4472 (_mm_fixupimm_sd): Ditto.
4473 (_mm_maskz_fixupimm_sd): Ditto.
4474 (_mm_fixupimm_ss): Ditto.
4475 (_mm_maskz_fixupimm_ss): Ditto.
4476 (_mm512_mask_fixupimm_round_pd): Update builtin.
4477 (_mm512_mask_fixupimm_round_ps): Ditto.
4478 (_mm_mask_fixupimm_round_sd): Ditto.
4479 (_mm_mask_fixupimm_round_ss): Ditto.
4480 (_mm512_mask_fixupimm_pd): Ditto.
4481 (_mm512_mask_fixupimm_ps): Ditto.
4482 (_mm_mask_fixupimm_sd): Ditto.
4483 (_mm_mask_fixupimm_ss): Ditto.
4484 * config/i386/avx512vlintrin.h:
4485 (_mm256_fixupimm_pd): Update parameters and builtin.
4486 (_mm256_maskz_fixupimm_pd): Ditto.
4487 (_mm256_fixupimm_ps): Ditto.
4488 (_mm256_maskz_fixupimm_ps): Ditto.
4489 (_mm_fixupimm_pd): Ditto.
4490 (_mm_maskz_fixupimm_pd): Ditto.
4491 (_mm_fixupimm_ps): Ditto.
4492 (_mm_maskz_fixupimm_ps): Ditto.
4493 (_mm256_mask_fixupimm_pd): Update builtin.
4494 (_mm256_mask_fixupimm_ps): Ditto.
4495 (_mm_mask_fixupimm_pd): Ditto.
4496 (_mm_mask_fixupimm_ps): Ditto.
4497 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
4498 * config/i386/i386-builtin.def: Update builtin definitions.
4499 * config/i386/i386.c: Handle new builtin types and remove useless ones.
4500 * config/i386/sse.md: Update VFIXUPIMM* patterns.
4501 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
4502 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
4503 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
4504 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
4505 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
4506 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
4507 * config/i386/subst.md:
4508 (round_saeonly_sd_mask_operand4): Add new subst_attr.
4509 (round_saeonly_sd_mask_op4): Ditto.
4510 (round_saeonly_expand_operand5): Ditto.
4511 (round_saeonly_expand): Update.
4513 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
4515 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
4517 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
4519 PR rtl-optimization/87871
4520 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
4522 2018-11-05 Paul Koning <ni1d@arrl.net>
4524 * doc/sourcebuild.texi (target attributes): Document new "inf"
4525 effective target keyword.
4527 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4529 * config/s390/s390.c (s390_register_move_cost): Increase costs for
4530 moves involving the CC reg.
4532 2018-11-05 Richard Biener <rguenther@suse.de>
4534 PR tree-optimization/87873
4535 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
4537 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
4538 * tree-vect-loop.c (vect_transform_loop): When splitting the
4539 loop exit also create forwarder PHIs for constants.
4540 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
4541 Handle constant to_arg, add extra checking we match up the correct
4544 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4546 * config/s390/s390.md: QImode and HImode for load on condition.
4548 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
4550 * config/s390/predicates.md: Fix typo.
4551 * config/s390/s390.md: Allow immediates for load on condition.
4553 2018-11-05 Martin Liska <mliska@suse.cz>
4555 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
4556 * bitmap.h (struct bitmap_usage): Likewise.
4557 * ggc-common.c (SCALE): Remove.
4559 (struct ggc_usage): Use SIZE_AMOUNT. And update
4561 * ggc-page.c (SCALE): Remove.
4562 (STAT_LABEL): Remove.
4563 (ggc_print_statistics): Use SIZE_AMOUNT.
4564 * gimple.h (SCALE): Remove.
4566 * input.c (ONE_K): Remove.
4569 (STAT_LABEL): Likewise.
4570 (FORMAT_AMOUNT): Likewise.
4571 (dump_line_table_statistics): Use SIZE_AMOUNT.
4572 * mem-stats.h (struct mem_usage): Likewise.
4573 * rtl.c (dump_rtx_statistics): Likewise.
4574 (rtx_alloc_counts): Change type to size_t.
4575 (rtx_alloc_sizes): Likewise.
4576 (rtx_count_cmp): New.
4577 (dump_rtx_statistics): Sort first based on counts.
4578 * tree.c (tree_nodes_cmp): New.
4579 (tree_codes_cmp): New.
4580 (dump_tree_statistics): Sort first based on counts.
4581 * system.h (ONE_K): New.
4583 (SIZE_SCALE): Likewise.
4584 (SIZE_LABEL): Likewise.
4585 (SIZE_AMOUNT): Likewise.
4586 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
4587 * tree-dfa.c (dump_dfa_stats): Likewise.
4588 * tree-phinodes.c (phinodes_print_statistics): Likewise.
4589 * tree-ssanames.c (ssanames_print_statistics): Likewise.
4590 * tree.c (dump_tree_statistics): Likewise.
4591 * vec.c (struct vec_usage): Likewise.
4592 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
4593 trigger a -Werror=format-overflow with
4594 --enable-gather-detailed-stats.
4596 2018-11-05 Martin Liska <mliska@suse.cz>
4598 * mem-stats.h (mem_alloc_description::release_instance_overhead):
4600 * vec.c (struct vec_usage): Register m_element_size.
4601 (vec_prefix::register_overhead): New arguments: elements and
4603 (vec_prefix::release_overhead): Subtract elements.
4604 * vec.h (struct vec_prefix): Change signature.
4605 (va_heap::reserve): Pass proper arguments.
4606 (va_heap::release): Likewise.
4608 2018-11-05 Martin Liska <mliska@suse.cz>
4610 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
4614 2018-11-05 Richard Biener <rguenther@suse.de>
4616 * tree-scalar-evolution.h (final_value_replacement_loop): Update
4618 * tree-scalar-evolution.c (final_value_replacement_loop): Return
4619 whether anything was done.
4620 (scev_const_prop): Remove constant propagation part, fold
4622 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
4623 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
4626 2018-11-05 Jakub Jelinek <jakub@redhat.com>
4628 PR tree-optimization/87859
4629 * gimple-ssa-store-merging.c (struct merged_store_group): Add
4630 only_constants and first_nonmergeable_order members.
4631 (merged_store_group::merged_store_group): Initialize them.
4632 (merged_store_group::do_merge): Clear only_constants member if
4633 adding something other than INTEGER_CST store.
4634 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
4635 stores with order >= first_nonmergeable_order. Use
4636 merged_store->only_constants instead of always recomputing it.
4637 Set merged_store->first_nonmergeable_order if we've skipped any
4638 stores. Attempt to merge overlapping INTEGER_CST stores that
4639 we would otherwise skip.
4642 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
4644 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
4647 * config/i386/emmintrin.h (__v16qs): New to cope with option
4649 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
4650 (_mm_cmplt_epi8): Likewise.
4651 (_mm_cmpgt_epi8): Likewise.
4653 2018-11-05 Richard Biener <rguenther@suse.de>
4655 PR rtl-optimization/87852
4656 * fwprop.c (use_killed_between): Only consider single-defs of the
4657 use whose definition statement dominates the use.
4659 2018-11-05 Martin Liska <mliska@suse.cz>
4662 * doc/invoke.texi: Remove options that are
4663 not disabled with -Os.
4665 2018-11-05 Martin Liska <mliska@suse.cz>
4668 * doc/extend.texi: Update constrain about the last argument
4669 of __builtin_expect_with_probability.
4671 2018-11-05 Martin Liska <mliska@suse.cz>
4674 * predict.c (expr_expected_value_1): Verify
4675 that last argument is a real constants and emit
4678 2018-11-05 Martin Liska <mliska@suse.cz>
4680 PR gcov-profile/77698
4681 * ipa-profile.c (ipa_profile): Adjust hotness threshold
4684 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4686 PR tree-optimization/86572
4687 * builtins.c (c_strlen): Handle negative offsets in a safe way.
4689 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4691 PR tree-optimization/87672
4692 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
4693 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
4695 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
4698 * cfgexpand.c (pass_expand::execute): Move the call to
4699 finish_eh_generation in front of the call to expand_stack_alignment.
4701 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
4703 * common/config/i386/i386-common.c (processor_alias_table): Add
4705 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
4706 (case ${target}): Add znver2.
4707 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
4708 -march=native recognize znver2 processors.
4709 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
4710 * config/i386/i386.c (m_znver2): New definition.
4711 (m_ZNVER): New definition.
4712 (m_AMD_MULTIPLE): Includes m_znver2.
4713 (processor_cost_table): Add znver2 entry.
4714 (processor_target_table): Add znver2 entry.
4715 (get_builtin_code_for_version): Set priority for
4717 (processor_model): Add M_AMDFAM17H_ZNVER2.
4718 (arch_names_table): Ditto.
4719 (ix86_reassociation_width): Include znver2.
4720 * config/i386/i386.h (TARGET_znver2): New definition.
4721 (struct ix86_size_cost): Add TARGET_ZNVER2.
4722 (enum processor_type): Add PROCESSOR_ZNVER2.
4723 * config/i386/i386.md (define_attr "cpu"): Add znver2.
4724 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
4725 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
4726 (ix86_adjust_cost): Add znver2.
4727 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
4728 * gcc/doc/extend.texi: Add details about znver2.
4729 * gcc/doc/invoke.texi: Add details about znver2.
4731 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
4735 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
4738 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4740 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
4741 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
4743 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
4745 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
4746 shrink-wrapping when -mtrack-speculation.
4748 2018-11-02 Richard Biener <rguenther@suse.de>
4750 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
4751 (pop_cost_one_pair): Do not free pair.
4752 (pop_best_coalesce): Likewise.
4753 (create_coalesce_list): Initialize obstack.
4754 (delete_coalesce_list): Free obstack.
4755 (find_coalesce_pair): Obstack-allocate coalesce pairs.
4756 (add_cost_one_coalesce): Likewise.
4757 (struct live_track): Remove bitmap pointer indirections.
4758 (new_live_track): Adjust.
4759 (delete_live_track): Likewise.
4760 (live_track_remove_partition): Likewise.
4761 (live_track_add_partition): Likewise.
4762 (live_track_live_p): Likewise.
4763 (live_track_process_def): Likewise.
4764 (live_track_clear_base_vars): Likewise.
4766 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4768 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
4769 * configure: Regenerate.
4770 * config.in: Regenerate.
4771 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
4772 HAVE_AS_REGISTER_PSEUDO_OP guard.
4773 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
4774 !HAVE_AS_REGISTER_PSEUDO_OP support.
4775 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
4778 2018-11-02 Richard Biener <rguenther@suse.de>
4780 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
4783 2018-11-02 Richard Biener <rguenther@suse.de>
4785 PR tree-optimization/87776
4786 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
4787 executable when iterating but running into rpo-vn-max-loop-depth
4788 and not eliding the iteration.
4790 2018-11-30 Jan Hubicka <jh@suse.cz>
4792 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
4795 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
4797 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
4799 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
4800 Combine with rs6000_address_for_fpconvert.
4801 (rs6000_address_for_fpconvert): Combine with
4802 rs6000_force_indexed_or_indirect_mem.
4803 (rs6000_expand_vector_init): Change function call from
4804 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4805 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
4806 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
4807 (floatsi<mode>2_lfiwax_mem): Ditto.
4808 (floatunssi<mode>2_lfiwzx): Ditto.
4809 (floatunssi<mode>2_lfiwzx_mem): Ditto.
4810 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
4811 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
4812 (fix_trunc<mode>si2_stfiwx): Ditto.
4813 (fixuns_trunc<mode>si2_stfiwx): Ditto.
4814 (float_<mode>si2_hw): Ditto.
4815 (floatuns_<mode>si2_hw): Ditto.
4816 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
4817 (vsx_splat_<mode>): Ditto.
4819 2018-11-01 Joseph Myers <joseph@codesourcery.com>
4821 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
4822 at end of assembler input text.
4823 * configure: Regenerate.
4825 2018-11-01 Jakub Jelinek <jakub@redhat.com>
4827 PR tree-optimization/87826
4828 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
4829 negative or larger or equal to type's precision.
4831 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
4833 * opts.c (default_options_table): Do not enable
4834 OPT_fdelayed_branch at -Og.
4835 * doc/invoke.texi (-fdelayed-branch): Document it.
4837 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4839 * optabs-libfuncs.c (build_libfunc_function_visibility):
4840 New, split out from...
4841 (build_libfunc_function): ... here.
4842 (init_one_libfunc_visibility): New, split out from ...
4843 (init_one_libfunc): ... here.
4845 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
4846 scratch register need not be early-clobber. Document the reason
4847 why we cannot use ST<OP>.
4849 2018-10-31 Joseph Myers <joseph@codesourcery.com>
4852 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
4853 line for second argument of AC_DEFINE_UNQUOTED.
4854 * doc/install.texi (Tools/packages necessary for modifying GCC):
4855 Update to autoconf 2.69 and automake 1.15.1.
4856 * aclocal.m4, config.in, configure: Regenerate.
4858 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
4860 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
4863 2018-10-31 Martin Liska <mliska@suse.cz>
4866 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
4867 Add new argument invalid_extension.
4868 (aarch64_get_all_extension_candidates): New function.
4869 (aarch64_rewrite_selected_cpu): Add NULL to function call.
4870 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
4872 (aarch64_get_all_extension_candidates): New function.
4873 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
4874 argument invalid_extension.
4875 (aarch64_parse_cpu): Likewise.
4876 (aarch64_print_hint_for_extensions): New function.
4877 (aarch64_validate_mcpu): Provide hint about invalid extension.
4878 (aarch64_validate_march): Likewise.
4879 (aarch64_handle_attr_arch): Pass new argument.
4880 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
4881 (aarch64_handle_attr_isa_flags): Likewise.
4883 2018-10-31 Richard Biener <rguenther@suse.de>
4887 * tree-outof-ssa.c (insert_backedge_copies): Restrict
4888 copy generation to useful cases. Place the copy before
4889 the definition of the backedge value when possible.
4891 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4893 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
4894 * config/arc/arc.c (arc_active_insn): New function.
4895 (check_store_cacheline_hazard): Likewise.
4896 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
4897 (arc_override_options): Disable delay slot scheduler for older
4899 (arc_store_addr_hazard_p): New implementation, old one renamed to
4901 (arc_store_addr_hazard_internal_p): Renamed.
4902 (arc_reorg): Don't combine into brcc instructions which are part
4903 of hardware hazard solution.
4904 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
4905 (tune_arc700): Likewise.
4906 * config/arc/arc.opt (arc7xx): New tune value.
4907 * config/arc/arc700.md: Improve A7 scheduler.
4909 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4911 * config/arc/arc.c (arc_override_options): Remove
4912 TARGET_COMPACT_CASESI.
4913 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
4914 (CASE_VECTOR_MODE): Likewise.
4915 (CASE_VECTOR_PC_RELATIVE): Likewise.
4916 (CASE_VECTOR_SHORTEN_MODE): Likewise.
4917 (CASE_VECTOR_SHORTEN_MODE1): Delete.
4918 (ADDR_VEC_ALIGN): Update.
4919 (ASM_OUTPUT_CASE_LABEL): Undefine.
4920 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
4921 (TARGET_BI_BIH): Define.
4922 (DEFAULT_BRANCH_INDEX): Likewise.
4923 * config/arc/arc.md (casesi): Rework to accept BI/BIH
4924 instructions, remove compact_casesi use case.
4925 (casesi_compact_jump): Remove.
4926 (casesi_dispatch): New pattern.
4927 * config/arc/arc.opt: Add mbranch-index option. Deprecate
4928 compact_casesi option.
4929 * doc/invoke.texi: Document mbranch-index option.
4931 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4933 * config/arc/arc.c (arc_get_tp): Remove function.
4934 (arc_emit_call_tls_get_addr): Likewise.
4935 (arc_call_tls_get_addr): New function.
4936 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
4937 * config/arc/arc.md (tls_load_tp_soft): Remove.
4938 (tls_gd_get_addr): Likewise.
4940 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4942 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
4943 (mulsi3_600_lib): Remove pattern.
4944 (umulsi3_highpart_600_lib_le): Likewise.
4945 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
4946 (umulsidi3): Remove call to umulsidi3_600_lib.
4947 (umulsidi3_600_lib): Remove pattern.
4948 (peephole2): Remove peephole using the above deprecated patterns.
4950 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
4953 * config/arm/arm.c (arm_option_check_internal): Disable the combined
4954 use of -mslow-flash-data and -mword-relocations.
4955 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
4956 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
4958 * doc/invoke.texi (-mword-relocations): Mention conflict with
4960 (-mslow-flash-data): Reciprocally.
4962 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4964 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
4965 16-byte modes held in GP registers to use an even regno.
4967 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
4968 (aarch64_atomic_ldop_supported_p): Remove.
4969 (aarch64_gen_atomic_ldop): Remove.
4970 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
4971 Fully expand LSE operations here.
4972 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
4973 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
4974 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
4975 and use ATOMIC_LDOP instead; use register_operand for the input;
4976 drop the split and emit insns directly.
4977 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
4978 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
4979 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
4981 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
4982 (aarch64_gen_atomic_ldop): Don't call it.
4983 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
4984 Use aarch64_reg_or_zero.
4985 (aarch64_atomic_exchange<ALLI>): Likewise.
4986 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
4987 operand 0; use aarch64_reg_or_zero for input; merge ...
4988 (@aarch64_atomic_swp<ALLI>): ... this and remove.
4990 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
4991 (aarch64_split_compare_and_swap): Use it.
4992 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
4993 test oldval against the proper predicate.
4994 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
4995 Use nonmemory_operand for expected.
4996 (cas_short_expected_pred): New.
4997 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
4998 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
4999 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
5000 (aarch64_plushi_operand): New.
5002 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
5003 Force oldval into the rval register for TARGET_LSE; emit the compare
5004 during initial expansion so that it may be deleted if unused.
5005 (aarch64_gen_atomic_cas): Remove.
5006 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
5007 Change =&r to +r for operand 0; use match_dup for operand 2;
5008 remove is_weak and mod_f operands as unused. Drop the split
5010 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
5011 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
5012 (@aarch64_atomic_cas<GPI>): Similarly.
5014 2018-10-31 Richard Biener <rguenther@suse.de>
5016 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
5019 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
5021 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
5022 * config/aarch64/aarch64.md: Include saphira.md
5023 * config/aarch64/saphira.md: New file for pipeline description.
5025 2018-10-30 Martin Sebor <msebor@redhat.com>
5028 * gimple-ssa-sprintf.c (format_directive): Use %G to include
5030 (sprintf_dom_walker::compute_format_length):
5031 Avoid setting POSUNDER4K here.
5032 (get_destination_size): Handle null argument values.
5033 (get_user_idx_format): New function.
5034 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
5035 functions, including user-defined with attribute format printf.
5036 Use %G to include inlining context.
5037 Set POSUNDER4K here.
5039 2018-10-30 Jan Hubicka <jh@suse.cz>
5041 * params.def (lto-partitions): Bump from 32 to 128.
5043 2018-10-30 Jan Hubicka <jh@suse.cz>
5046 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
5048 (free_lang_data_in_type): Forward declare.
5049 (fld_type_variant_equal_p): New function.
5050 (fld_type_variant): New function
5051 (fld_incomplete_types): New hash.
5052 (fld_incomplete_type_of): New function
5053 (fld_simplfied-type): New function.
5054 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
5055 (free_lang_data): Allocate and free fld_incomplete_type; update call
5056 of free_lang_data_in_decl.
5058 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
5060 * gcov.c (output_lines): Remove duplicate line.
5062 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5064 * config/rs6000/rs6000.md (bswapdi2): Force address into register
5065 if not in indexed or indirect form.
5066 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
5067 (bswapdi2_store): Ditto.
5068 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
5070 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
5071 Prototype for helper function.
5073 2018-10-30 Martin Sebor <msebor@redhat.com>
5075 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
5076 (target, pragma GCC optimize, pragma GCC target): Ditto.
5078 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
5080 * doc/extend.texi: Fix prototype and description of
5081 __builtin_expect_with_probability.
5083 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
5085 * cgraph.h (clone_function_name_1): Replaced by new
5086 clone_function_name_numbered that takes name as string; for
5087 privatize_symbol_name_1 use only.
5088 (clone_function_name): Renamed to
5089 clone_function_name_numbered to be explicit about numbering.
5090 (clone_function_name): New two-argument function that does
5091 not number its output.
5092 (clone_function_name): New three-argument function that
5093 takes a number to append to its output.
5094 * cgraphclones.c (duplicate_thunk_for_node):
5095 (clone_function_name_1): Renamed.
5096 (clone_function_name_numbered): Two new functions.
5097 (clone_function_name): Improved documentation.
5098 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
5099 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
5100 * final.c (final_scan_insn_1): Use the new clone_function_name
5102 * multiple_target.c (create_dispatcher_calls): Ditto.
5103 (create_target_clone): Ditto.
5104 * omp-expand.c (grid_expand_target_grid_body): Ditto.
5105 * omp-low.c (create_omp_child_function_name): Ditto.
5106 * omp-simd-clone.c (simd_clone_create): Ditto.
5107 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
5108 new clone_function_name without numbering.
5110 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
5112 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
5113 Assert that the allocation size is not zero.
5115 2018-10-30 Richard Biener <rguenther@suse.de>
5117 PR tree-optimization/87800
5118 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
5119 non-induction or reduction PHIs.
5121 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
5123 * config/aarch64/falkor-tag-collision-avoidance.c
5124 (execute_tag_collision_avoidance): Call df_note_add_problem.
5126 2018-10-30 Martin Liska <mliska@suse.cz>
5128 * doc/extend.texi: Fix typo in documentation
5129 of __builtin_expect_with_probability.
5131 2018-10-29 David Malcolm <dmalcolm@redhat.com>
5134 * input.c (get_substring_ranges_for_loc): Detect if
5135 linemap_resolve_location gives us a NULL map, and reject
5138 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
5140 * config.gcc (xstormy16-*-elf): Set tm_d_file.
5142 2018-10-29 David Malcolm <dmalcolm@redhat.com>
5143 Martin Sebor <msebor@redhat.com>
5144 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
5146 * doc/ux.texi (Quoting): New subsection, adapted from material at
5147 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
5148 MartinSebor and ManuelLopezIbanez.
5149 (Fix-it hints): Note that fix-it hints shouldn't be marked for
5152 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
5155 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
5158 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
5160 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
5162 2018-10-29 Paul Koning <ni1d@arrl.net>
5164 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
5166 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
5168 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
5169 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
5170 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
5171 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
5172 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
5173 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
5174 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
5175 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
5176 Change 'vector' to '__vector'.
5177 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
5178 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
5179 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
5180 _mm_avg_pu16): Likewise. And, whitespace corrections.
5182 2018-10-29 Richard Biener <rguenther@suse.de>
5184 PR tree-optimization/87785
5185 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
5188 2018-10-29 Olivier Hainque <hainque@adacore.com>
5190 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
5192 2018-10-29 Olivier Hainque <hainque@adacore.com>
5194 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
5195 ports configuration macro, defaults to "gnu".
5196 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
5197 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
5199 2018-10-29 Olivier Hainque <hainque@adacore.com>
5201 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
5203 2018-10-29 Richard Biener <rguenther@suse.de>
5205 PR tree-optimization/87790
5206 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
5207 (vect_make_slp_decision): Adjust.
5208 (vect_slp_analyze_bb_1): Likewise.
5209 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
5212 2018-10-29 Richard Biener <rguenther@suse.de>
5214 PR tree-optimization/87785
5215 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
5217 (vect_build_slp_tree): Likewise.
5218 (vect_gather_slp_loads): New function.
5219 (vect_analyze_slp_instance): Gather loads separately from the
5222 2018-10-29 Martin Liska <mliska@suse.cz>
5224 * Makefile.in: Make dependency to json.o.
5225 * doc/gcov.texi: Document new JSON format, remove
5226 old intermediate format documentation.
5227 * gcov.c (struct function_info): Come up with m_name and
5229 (function_info::function_info): Initialize it.
5230 (function_info::~function_info): Release it.
5231 (main): Rename flag_intermediate_format to flag_json_format.
5232 (print_usage): Describe --json-format.
5233 (process_args): Set flag_json_format.
5234 (output_intermediate_line): Remove.
5235 (output_intermediate_json_line): Likewise.
5236 (get_gcov_intermediate_filename): Return new extension
5238 (output_intermediate_file): Implement JSON emission.
5239 (output_json_intermediate_file): Implement JSON emission.
5240 (generate_results): Use ::get_name for function name.
5241 Handle JSON output file.
5242 (read_graph_file): Use ::get_name instead of cplus_demangle.
5243 (read_count_file): Likewise.
5244 (solve_flow_graph): Likewise.
5245 (add_line_counts): Likewise.
5246 (accumulate_line_counts): Use new flag_json_format.
5247 (output_function_details): Use ::get_name instead of cplus_demangle.
5248 (output_lines): Likewise.
5249 * json.cc (test_writing_literals): Add new tests.
5250 * json.h (class literal): Add new boolean constructor.
5252 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
5254 PR rtl-optimization/87701
5255 PR rtl-optimization/87780
5256 * combine.c (make_more_copies): Rewrite.
5258 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
5260 * doc/generic.texi (ABSU_EXPR): Document.
5261 * match.pd (absu(x)*absu(x) -> x*x): Handle.
5262 (absu(absu(X)) -> absu(X)): Likewise.
5263 (absu(-X) -> absu(X)): Likewise.
5264 (absu(X) where X is nonnegative -> X): Likewise.
5266 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
5268 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
5269 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
5270 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
5271 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
5272 (s-tm-texi): Also check timestamp on d-target.def.
5273 (generated_files): Add TM_D_H and d-target-hooks-def.h.
5274 (build/genhooks.o): Also depend on D_TARGET_DEF.
5275 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
5277 * config/aarch64/aarch64-d.c: New file.
5278 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
5280 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
5282 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
5283 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
5284 * config/arm/arm-d.c: New file.
5285 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
5286 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
5287 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5288 * config/arm/t-arm (arm-d.o): New rule.
5289 * config/default-d.c: New file.
5290 * config/glibc-d.c: New file.
5291 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5292 * config/i386/i386-d.c: New file.
5293 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
5294 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
5295 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5296 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
5297 * config/i386/t-i386 (i386-d.o): New rule.
5298 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5299 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5300 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
5301 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5302 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
5303 * config/mips/mips-d.c: New file.
5304 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
5305 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
5306 * config/mips/t-mips (mips-d.o): New rule.
5307 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5308 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5309 * config/powerpcspe/powerpcspe-d.c: New file.
5310 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
5312 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
5313 Support GNU D by using 0 as the language type.
5314 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
5315 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
5316 * config/riscv/riscv-d.c: New file.
5317 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
5319 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
5320 * config/riscv/t-riscv (riscv-d.o): New rule.
5321 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5322 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
5323 * config/rs6000/rs6000-d.c: New file.
5324 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
5326 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
5327 Support GNU D by using 0 as the language type.
5328 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
5329 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
5330 * config/s390/s390-d.c: New file.
5331 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
5332 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
5333 * config/s390/t-s390 (s390-d.o): New rule.
5334 * config/sparc/sparc-d.c: New file.
5335 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
5337 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
5338 * config/sparc/t-sparc (sparc-d.o): New rule.
5339 * config/t-glibc (glibc-d.o): New rule.
5340 * configure: Regenerated.
5341 * configure.ac (tm_d_file): New variable.
5342 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
5343 * doc/contrib.texi (Contributors): Add self for the D frontend.
5344 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
5345 * doc/install.texi (Configuration): Mention libphobos as an option for
5346 --enable-shared. Mention d as an option for --enable-languages.
5347 (Testing): Mention check-d as a target.
5348 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
5349 name suffixes. Mention d as a -x option.
5350 * doc/sourcebuild.texi (Top Level): Mention libphobos.
5351 * doc/standards.texi (Standards): Add section on D language.
5352 * doc/tm.texi: Regenerated.
5353 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
5354 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
5355 * dwarf2out.c (is_dlang): New function.
5356 (gen_compile_unit_die): Use DW_LANG_D for D.
5357 (declare_in_namespace): Return module die for D, instead of adding
5358 extra declarations into the namespace.
5359 (gen_namespace_die): Generate DW_TAG_module for D.
5360 (gen_decl_die): Handle CONST_DECLSs for D.
5361 (dwarf2out_decl): Likewise.
5362 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
5363 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
5364 kinds of aggregates.
5365 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
5366 * genhooks.c: Include d/d-target.def.
5368 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
5371 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
5372 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
5373 (STACK_DYNAMIC_OFFSET): Likewise.
5375 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
5378 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
5381 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
5383 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
5384 return type and other typos.
5386 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
5388 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
5389 a shorter sequence with fewer branches.
5390 (emit_final_str_compare_gpr): Ditto.
5392 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5394 * config/rs6000/tmmintrin.h: New file.
5395 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
5397 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5399 * config/rs6000/mmintrin.h: Enable 32bit compilation.
5400 * config/rs6000/xmmintrin.h: Likewise.
5402 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
5404 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
5406 2018-10-26 Richard Biener <rguenther@suse.de>
5408 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
5410 (vect_mark_slp_stmts_relevant): Likewise.
5411 (vect_detect_hybrid_slp_stmts): Likewise.
5412 (vect_bb_slp_scalar_cost): Likewise.
5413 (vect_remove_slp_scalar_calls): Likewise.
5415 2018-10-26 Jan Hubicka <jh@suse.cz>
5417 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
5418 (warn_types_mismatch): Fix walk of DECL_NAME.
5419 (odr_types_equivalent_p): Fix overactive assert.
5421 2018-10-26 Richard Biener <rguenther@suse.de>
5423 PR tree-optimization/87105
5424 * tree-vectorizer.h (_slp_tree::refcnt): New member.
5425 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
5427 (vect_create_new_slp_node): Initialize refcnt to one.
5429 (scalar_stmts_set_t, bst_fail): Remove.
5430 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
5431 (vect_build_slp_tree): Add bst_map argument and lookup
5432 already created SLP nodes.
5433 (vect_print_slp_tree): Handle a SLP graph, print SLP node
5435 (vect_slp_rearrange_stmts): Handle a SLP graph.
5436 (vect_analyze_slp_instance): Adjust and free SLP nodes from
5437 the CSE map. Fix indenting.
5438 (vect_schedule_slp_instance): Add short-cut.
5440 2018-10-26 Martin Liska <mliska@suse.cz>
5443 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
5444 addr_expr and not with pointers.
5446 2018-10-26 Jan Hubicka <jh@suse.cz>
5448 * tree.c (free_lang_data_in_type): Only check main variants.
5449 * ipa-devirt.c (warn_odr): Make static.
5450 (types_same_for_odr): Drop strict variant.
5451 (types_odr_comparable): Likewise.
5452 (odr_or_derived_type_p): Look for main variants.
5453 (odr_name_hasher::equal): Cleanup comment.
5454 (odr_subtypes_equivalent): Add warn and warned arguments; check main
5456 (type_variants_equivalent_p): break out from ...
5457 (odr_types_equivalent): ... here; go for main variants where needed.
5458 (warn_odr): ... here; turn static.
5459 (warn_types_mismatch): Compare mangled names of main variants.
5460 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
5461 (type_with_linkage_p): Sanity check that we look at main variant.
5462 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
5463 * tree.h (types_same_for_odr): Drop strict argument.
5465 2018-10-26 Richard Biener <rguenther@suse.de>
5467 PR tree-optimization/87746
5468 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
5469 Simplify and fix WRT strided store groups with size not
5470 equal to step in element count.
5471 (vect_analyze_group_access_1): Dump the whole group.
5473 2018-10-25 Carl Love <cel@us.ibm.com>
5475 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
5476 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
5477 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
5478 precicion entry for each overloaded builtin.
5479 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
5480 VSCEDPUO): Rename overloaded name.
5481 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
5482 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
5483 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
5484 define_expand for xscmpexqp instruction.
5485 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
5487 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5488 Jinsong Ji <jji@us.ibm.com>
5490 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
5491 function with vec_sl.
5492 (_mm_slli_epi32): Likewise.
5493 (_mm_slli_epi64): Likewise.
5494 (_mm_srai_epi16): Replace deprecated function with vec_sra.
5495 (_mm_srai_epi32): Likewise.
5496 (_mm_srli_epi16): Replace deprecated function with vec_sr.
5497 (_mm_srli_epi32): Likewise.
5498 (_mm_srli_epi64): Likewise.
5499 (_mm_sll_epi16): Replace deprecated function with vec_sl.
5500 (_mm_sll_epi32): Likewise.
5501 (_mm_sll_epi64): Likewise.
5502 (_mm_sra_epi16): Replace deprecated function with vec_sra.
5503 (_mm_sra_epi32): Likewise.
5504 (_mm_srl_epi16): Replace deprecated function with vec_sr.
5505 (_mm_srl_epi32): Likewise.
5506 (_mm_srl_epi64): Likewise.
5508 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5509 Jinsong Ji <jji@us.ibm.com>
5511 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
5512 comparison operators with vec_cmp* for compatibility due to
5513 unfortunate history; clean up formatting and use types more
5515 (_mm_sll_epi32): Likewise.
5516 (_mm_sll_epi64): Likewise.
5517 (_mm_srl_epi16): Likewise.
5518 (_mm_srl_epi32): Likewise.
5519 (_mm_srl_epi64): Likewise.
5521 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
5522 Jinsong Ji <jji@us.ibm.com>
5524 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
5525 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
5526 __vector __bool int. Use vec_cmpgt in preference to deprecated
5527 function vec_vcmpgtfp.
5528 (_mm_max_ps): Likewise.
5530 2018-10-25 Jeff Law <law@redhat.com>
5532 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
5535 2018-10-25 Martin Sebor <msebor@redhat.com>
5537 * doc/extend.texi (aligned): Expand attribute description.
5538 (Alignment): Rename section. Discuss function arguments.
5540 2018-10-25 Jan Hubicka <jh@suse.cz>
5542 * ipa-devirt.c (main_odr_variant): Remove.
5543 (hash_odr_name, types_same_for_odr, types_odr_comparable,
5544 odr_name_hasher::equal, odr_subtypes_equivalent_p):
5545 Drop use of main_odr_variant.
5546 (add_type_duplicate): Silence confused warnings on integer types.
5547 (get_odr_type): Always look for main variant.
5548 (register_odr_type): Simplify.
5550 2018-10-25 Richard Biener <rguenther@suse.de>
5552 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
5553 Initialize ng to silence error with release checking bootstrap.
5555 2018-10-25 Richard Biener <rguenther@suse.de>
5557 * tree-if-conv.c: Include tree-ssa-sccvn.h.
5558 (tree_if_conversion): Run CSE on the if-converted loop body.
5560 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
5562 * config/s390/constraints.md (ZL): New constraint.
5563 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
5565 * config/s390/s390.md (movdi_larl): Remove.
5566 (movdi_64): Add the LARL alternative.
5568 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
5571 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
5572 (RTX_CODE_HWINT_P): New macro.
5573 (rtx_code_size): Use RTX_CODE_HWINT_P ().
5575 2018-10-25 Jan Hubicka <jh@suse.cz>
5577 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
5580 2018-10-25 Richard Biener <rguenther@suse.de>
5582 PR tree-optimization/87665
5583 PR tree-optimization/87745
5584 * tree-vectorizer.h (get_earlier_stmt): Remove.
5585 (get_later_stmt): Pick up UID from the original non-pattern stmt.
5587 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
5589 * options.texi (Deprecated): Move list to Var section.
5591 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
5592 Jinsong Ji <jji@us.ibm.com>
5594 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
5595 __vector long to __vector long long.
5596 (_mm_cvtpd_ps): Likewise.
5597 (_mm_cvttpd_epi32): Likewise.
5598 (_mm_cvtpi32_pd): Likewise.
5599 (_mm_unpackhi_epi64): Likewise.
5600 (_mm_unpacklo_epi64): Likewise.
5602 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
5604 PR rtl-optimization/87720
5605 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
5607 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
5609 * gimple-ssa-isolate-paths.c
5610 (find_implicit_erroneous_behavior): Do not change code if the
5611 pass is running for warnings only.
5612 (find_explicit_erroneous_behavior): Likewise.
5614 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
5616 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
5617 Define as rs6000_mangle_decl_assembler_name.
5618 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
5619 long double to IEEE long double, switch the names of the long
5620 double built-in functions to be <func>f128 instead of <func>l.
5622 2018-10-24 Martin Sebor <msebor@redhat.com>
5624 * doc/extend.texi (nonnull): List no-argument form. Reference
5625 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
5627 2018-10-24 Richard Biener <rguenther@suse.de>
5629 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
5631 2018-10-24 Martin Liska <mliska@suse.cz>
5633 PR tree-optimization/84436
5634 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
5636 (switch_conversion::contains_linear_function_p): New.
5637 (switch_conversion::build_one_array): Support linear
5638 transformation on input.
5639 * tree-switch-conversion.h (struct switch_conversion): Add
5640 contains_linear_function_p declaration.
5642 2018-10-24 Richard Biener <rguenther@suse.de>
5644 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
5645 if its argument is CONSTANT_CLASS_P.
5647 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5649 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
5650 it is wrong for forward declarations.
5652 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
5654 * config/s390/s390.c (s390_check_qrst_address): Add the missing
5655 SYMBOL_REF_P () check.
5657 2018-10-24 Richard Biener <rguenther@suse.de>
5659 PR tree-optimization/87105
5660 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
5661 dump classification.
5662 (vect_analyze_data_ref_accesses): Handle duplicate loads and
5663 stores by splitting the affected group after the fact.
5664 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
5665 fail the SLP build because of size constraints.
5667 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5669 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
5670 * configure: Regenerate.
5671 * config.in: Regenerate.
5672 * varasm.c (mergeable_string_section): Use readonly_data_section
5673 if linker doesn't support SHF_MERGE with alignment > 8.
5674 (mergeable_constant_section): Likewise.
5676 2018-10-24 Richard Biener <rguenther@suse.de>
5678 PR tree-optimization/84013
5679 * tree-ssa-structalias.c (struct msdi_data): New struct for
5680 marshalling data to walk_stmt_load_store_ops.
5681 (maybe_set_dependence_info): Refactor as callback for
5682 walk_stmt_load_store_ops.
5683 (compute_dependence_clique): Set restrict info on all stmt kinds.
5685 2018-10-24 Martin Liska <mliska@suse.cz>
5687 * cgraph.c (cgraph_node::dump):
5688 Remove reduntant dumps and make tp_first_run dump more compact.
5690 2018-10-24 Richard Biener <rguenther@suse.de>
5692 PR tree-optimization/87665
5693 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
5696 2018-10-12 Jeff Law <law@redhat.com>
5698 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
5701 2018-10-23 Richard Biener <rguenther@suse.de>
5703 * tree-vrp.c (add_assert_info): Guard dump_printf with
5705 * gimple-ssa-evrp-analyze.c
5706 (evrp_range_analyzer::record_ranges_from_incoming_edge):
5707 Use value_range::ignore_equivs_equal_p.
5709 2018-10-23 Richard Biener <rguenther@suse.de>
5711 PR tree-optimization/87105
5712 PR tree-optimization/87608
5713 * passes.def (pass_all_early_optimizations): Add early phi-opt
5715 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
5716 addition to debug stmts.
5717 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
5718 and abs replacement early.
5719 * tree-cfg.c (gimple_empty_block_p): Likewise.
5721 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
5724 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
5725 specified to configure.
5726 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
5728 2018-10-23 Richard Biener <rguenther@suse.de>
5730 PR tree-optimization/87700
5731 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
5733 2018-10-23 Jakub Jelinek <jakub@redhat.com>
5736 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
5737 second argument from __mmask16 to __mmask8.
5738 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
5739 _mm_mask_packs_epi32): Likewise.
5740 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
5742 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
5744 2018-10-23 Richard Biener <rguenther@suse.de>
5746 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
5748 2018-10-23 Richard Biener <rguenther@suse.de>
5750 PR tree-optimization/86144
5751 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
5752 over simd attribute.
5754 2018-10-23 Richard Biener <rguenther@suse.de>
5756 PR tree-optimization/87693
5757 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
5758 the case we do not find the taken edge.
5760 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
5761 Jinsong Ji <jji@us.ibm.com>
5763 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
5764 (_mm_store_pd): Use unaligned vector type for pointer cast.
5765 (_mm_maskmoveu_si128): Likewise.
5766 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
5767 (_mm_store_ps): Use unaligned vector type for pointer cast.
5769 2018-10-22 Paul Koning <ni1d@arrl.net>
5771 * symtab.c (symtab_node::increase_alignment): Correct max
5774 2018-10-22 Yury Gribov <tetra2005@gmail.com>
5776 PR tree-optimization/87633
5777 * match.pd: Do not generate unordered integer comparisons.
5779 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5781 PR rtl-optimization/87600
5782 * combine.c: Add include of expr.h.
5783 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
5784 register to a pseudo.
5785 (make_more_copies): New function, add a copy to a new pseudo after
5786 the moves from hard registers into pseudos.
5787 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
5788 later. Call make_more_copies.
5790 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
5792 * lra-constraints.c (process_alt_operands): New local array,
5793 matching_early_clobber. Check matching_early_clobber before
5794 decrementing reject, and set matching_early_clobber after.
5796 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
5799 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
5800 call output_addr_const and hope for the best.
5802 2018-10-22 Richard Biener <rguenther@suse.de>
5804 * gimple-ssa-evrp-analyze.c
5805 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
5806 smarter about what ranges to use.
5807 * tree-vrp.c (add_assert_info): Dump here.
5808 (register_edge_assert_for_2): Instead of here at multiple but
5811 * gcc.dg/tree-ssa/evrp12.c: New testcase.
5812 * gcc.dg/predict-6.c: Adjust.
5813 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
5814 * gcc.dg/tree-ssa/vrp02.c: Likewise.
5815 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
5817 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
5818 Richard Biener <rguenther@suse.de>
5820 * bitmap.h: Update data structure documentation, including a
5821 description of bitmap views as either linked-lists or splay trees.
5822 (struct bitmap_element_def): Update comments for splay tree bitmaps.
5823 (struct bitmap_head_def): Likewise.
5824 (bitmap_list_view, bitmap_tree_view): New prototypes.
5825 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
5827 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
5828 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
5829 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
5830 released bitmap element here.
5831 (bitmap_element_free): Remove.
5832 (bitmap_elt_clear_from): Work on splay tree bitmaps.
5833 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
5834 this function similar ones such that linked-list bitmap implementation
5835 functions are grouped.
5836 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
5837 and moved for grouping.
5838 (bitmap_list_insert_element_after): Renamed from
5839 bitmap_elt_insert_after, and moved for grouping.
5840 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
5841 (bitmap_tree_link_left, bitmap_tree_link_right,
5842 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
5843 bitmap_tree_link_element, bitmap_tree_unlink_element,
5844 bitmap_tree_find_element): New functions for splay-tree bitmap
5846 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
5847 Renamed and moved, see above entries.
5848 (bitmap_tree_listify_from): New function to convert part of a splay
5849 tree bitmap to a linked-list bitmap.
5850 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
5851 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
5852 (bitmap_find_bit): Remove.
5853 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
5854 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
5855 Handle splay tree bitmaps.
5856 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
5857 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
5858 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
5859 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
5860 bitmap_intersect_compl_p, bitmap_ior_and_compl,
5861 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
5862 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
5863 corresponding changes to use linked-list specific bitmap_element
5864 manipulation functions as applicable for efficiency.
5865 (bitmap_tree_to_vec): New function.
5866 (debug_bitmap_elt_file): New function split out from ...
5867 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
5868 (bitmap_print): Likewise.
5870 PR tree-optimization/63155
5871 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
5873 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
5876 2018-10-22 Martin Liska <mliska@suse.cz>
5878 PR tree-optimization/87686
5880 2018-08-29 Martin Liska <mliska@suse.cz>
5882 * tree-switch-conversion.c (switch_conversion::expand):
5883 Strenghten assumption about gswitch statements.
5885 2018-10-22 Martin Liska <mliska@suse.cz>
5887 * ipa-icf.c (sem_item::compare_attributes): Remove.
5888 (sem_item::compare_referenced_symbol_properties): Use
5889 attribute_list_equal instead.
5890 (sem_function::equals_wpa): Likewise.
5891 * ipa-icf.h: Remove compare_attributes.
5893 2018-10-22 Richard Biener <rguenther@suse.de>
5896 * mem-stats.h (mem_usage::operator==): Fix pasto.
5898 2018-10-22 Richard Biener <rguenther@suse.de>
5900 PR tree-optimization/87640
5901 * tree-vrp.c (set_value_range_with_overflow): Decompose
5903 (extract_range_from_binary_expr_1): Adjust.
5905 2018-10-22 Martin Jambor <mjambor@suse.cz>
5907 * tree-eh.h (stmt_could_throw_p): Add function parameter.
5908 (stmt_can_throw_external): Likewise.
5909 (stmt_can_throw_internal): Likewise.
5910 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
5911 (lower_eh_constructs_2): Likewise.
5912 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
5913 (stmt_can_throw_external): Likewise.
5914 (stmt_can_throw_internal): Likewise.
5915 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
5916 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
5917 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
5918 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
5919 (pass_lower_eh_dispatch::execute): Pass cfun to
5920 stmt_can_throw_external.
5921 (cleanup_empty_eh): Likewise.
5922 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
5923 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
5924 stmt_can_throw_external instead of pushing it to cfun.
5925 (symbol_table::create_edge): Likewise.
5926 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
5927 stmt_can_throw_internal.
5928 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
5929 to stmt_could_throw_p.
5930 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
5931 stmt_can_throw_internal.
5932 (pass_store_merging::execute): Likewise.
5933 * gimple-ssa-strength-reduction.c
5934 (find_candidates_dom_walker::before_dom_children): Pass cfun to
5936 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
5937 stmt_can_throw_internal.
5938 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
5939 to stmt_can_throw_external.
5940 (check_stmt): Pass cfun to stmt_could_throw_p.
5941 (check_stmt): Pass cfun to stmt_can_throw_external.
5942 (pass_nothrow::execute): Likewise.
5943 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
5944 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
5945 stmt_can_throw_internal.
5946 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
5947 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
5948 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
5949 * tree-complex.c (expand_complex_libcall): Pass cfun to
5950 stmt_could_throw_p and to stmt_can_throw_internal.
5951 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
5952 * tree-inline.c (copy_edges_for_bb): Likewise.
5953 (maybe_move_debug_stmts_to_successors): Likewise.
5954 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
5956 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
5957 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
5958 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
5959 stmt_can_throw_internal.
5960 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
5961 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
5963 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
5964 stmt_can_throw_internal.
5965 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
5966 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
5968 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
5969 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
5970 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5971 (convert_mult_to_fma_1): Likewise.
5972 (convert_to_divmod): Likewise.
5973 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
5974 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
5975 * tree-ssa-propagate.c
5976 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
5977 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
5978 (maybe_optimize_range_tests): Likewise.
5979 (linearize_expr_tree): Likewise.
5980 (reassociate_bb): Likewise.
5981 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
5982 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
5983 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
5984 (handle_char_store): Likewise.
5985 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
5986 stmt_can_throw_internal.
5987 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
5989 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5990 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
5991 (vectorizable_simd_clone_call): Likewise.
5992 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
5993 (gimple_stringop_fixed_value): Likewise.
5995 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
5997 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
5998 literal pool references.
5999 (s390_check_qrst_address): Adapt to the new behavior of
6000 s390_loadrelative_operand_p ().
6002 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6005 * config/i386/sse.md (*andnot<mode>3_bcst): New.
6007 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6010 * config/i386/sse.md (*<code><mode>3_bcst): New.
6012 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
6015 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
6016 V4DI, V16SI and V8DI.
6017 (*sub<mode>3<mask_name>_bcst): New.
6018 (*add<mode>3<mask_name>_bcst): Likewise.
6020 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
6021 Jinsong Ji <jji@us.ibm.com>
6023 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
6024 __m64 with __vector unsigned long long for compatibility.
6025 (_mm_movemask_epi8): Likewise.
6026 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
6027 (_mm_cvttps_pi32): Likewise.
6028 (_mm_cvtpi32_ps): Likewise.
6029 (_mm_cvtps_pi16): Likewise.
6030 (_mm_loadh_pi): Likewise.
6031 (_mm_storeh_pi): Likewise.
6032 (_mm_movehl_ps): Likewise.
6033 (_mm_movelh_ps): Likewise.
6034 (_mm_loadl_pi): Likewise.
6035 (_mm_storel_pi): Likewise.
6036 (_mm_movemask_ps): Likewise.
6037 (_mm_shuffle_pi16): Likewise.
6039 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6042 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
6043 __builtin_ia32_vfnmsubpd512_mask.
6044 (_mm512_mask_fnmsub_round_pd): Likewise.
6045 (_mm512_fnmsub_pd): Likewise.
6046 (_mm512_mask_fnmsub_pd): Likewise.
6047 (_mm512_maskz_fnmsub_round_pd): Use
6048 __builtin_ia32_vfnmsubpd512_maskz.
6049 (_mm512_maskz_fnmsub_pd): Likewise.
6050 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
6051 (_mm512_mask_fnmsub_round_ps): Likewise.
6052 (_mm512_fnmsub_ps): Likewise.
6053 (_mm512_mask_fnmsub_ps): Likewise.
6054 (_mm512_maskz_fnmsub_round_ps): Use
6055 __builtin_ia32_vfnmsubps512_maskz.
6056 (_mm512_maskz_fnmsub_ps): Likewise.
6057 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
6058 __builtin_ia32_vfnmsubpd256_mask.
6059 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
6060 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
6061 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
6062 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
6063 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
6064 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
6065 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
6066 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
6067 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
6068 __builtin_ia32_vfnmsubpd.
6069 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
6070 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
6071 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
6072 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
6073 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
6074 * config/i386/i386-builtin.def: Add
6075 __builtin_ia32_vfnmsubpd256_mask,
6076 __builtin_ia32_vfnmsubpd256_maskz,
6077 __builtin_ia32_vfnmsubpd128_mask,
6078 __builtin_ia32_vfnmsubpd128_maskz,
6079 __builtin_ia32_vfnmsubps256_mask,
6080 __builtin_ia32_vfnmsubps256_maskz,
6081 __builtin_ia32_vfnmsubps128_mask,
6082 __builtin_ia32_vfnmsubps128_maskz,
6083 __builtin_ia32_vfnmsubpd512_mask,
6084 __builtin_ia32_vfnmsubpd512_maskz,
6085 __builtin_ia32_vfnmsubps512_mask,
6086 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
6087 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
6088 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
6089 __builtin_ia32_vfnmsubpd256.
6090 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
6091 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
6092 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
6094 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
6096 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
6098 (fmai_vmfnmsub_<mode><round_name>): Likewise.
6100 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6103 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
6104 __builtin_ia32_vfnmaddpd512_mask.
6105 (_mm512_mask_fnmadd_round_pd): Likewise.
6106 (_mm512_fnmadd_pd): Likewise.
6107 (_mm512_mask_fnmadd_pd): Likewise.
6108 (_mm512_maskz_fnmadd_round_pd): Use
6109 __builtin_ia32_vfnmaddpd512_maskz.
6110 (_mm512_maskz_fnmadd_pd): Likewise.
6111 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
6112 (_mm512_mask_fnmadd_round_ps): Likewise.
6113 (_mm512_fnmadd_ps): Likewise.
6114 (_mm512_mask_fnmadd_ps): Likewise.
6115 (_mm512_maskz_fnmadd_round_ps): Use
6116 __builtin_ia32_vfnmaddps512_maskz.
6117 (_mm512_maskz_fnmadd_ps): Likewise.
6118 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
6119 __builtin_ia32_vfnmaddpd256_mask.
6120 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
6121 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
6122 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
6123 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
6124 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
6125 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
6126 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
6127 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
6128 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
6129 __builtin_ia32_vfnmaddpd.
6130 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
6131 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
6132 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
6133 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
6134 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
6135 * config/i386/i386-builtin.def: Add
6136 __builtin_ia32_vfnmaddpd256_mask,
6137 __builtin_ia32_vfnmaddpd256_maskz,
6138 __builtin_ia32_vfnmaddpd128_mask,
6139 __builtin_ia32_vfnmaddpd128_maskz,
6140 __builtin_ia32_vfnmaddps256_mask,
6141 __builtin_ia32_vfnmaddps256_maskz,
6142 __builtin_ia32_vfnmaddps128_mask,
6143 __builtin_ia32_vfnmaddps128_maskz,
6144 __builtin_ia32_vfnmaddpd512_mask,
6145 __builtin_ia32_vfnmaddpd512_maskz,
6146 __builtin_ia32_vfnmaddps512_mask,
6147 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
6148 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
6149 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
6150 __builtin_ia32_vfnmaddpd256.
6151 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
6152 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
6153 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
6155 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
6157 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
6159 (fmai_vmfnmadd_<mode><round_name>): Likewise.
6161 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6164 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
6165 __builtin_ia32_vfmsubpd512_mask.
6166 (_mm512_mask_fmsub_round_pd): Likewise.
6167 (_mm512_fmsub_pd): Likewise.
6168 (_mm512_mask_fmsub_pd): Likewise.
6169 (_mm512_maskz_fmsub_round_pd): Use
6170 __builtin_ia32_vfmsubpd512_maskz.
6171 (_mm512_maskz_fmsub_pd): Likewise.
6172 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
6173 (_mm512_mask_fmsub_round_ps): Likewise.
6174 (_mm512_fmsub_ps): Likewise.
6175 (_mm512_mask_fmsub_ps): Likewise.
6176 (_mm512_maskz_fmsub_round_ps): Use
6177 __builtin_ia32_vfmsubps512_maskz.
6178 (_mm512_maskz_fmsub_ps): Likewise.
6179 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
6180 __builtin_ia32_vfmsubpd256_mask.
6181 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
6182 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
6183 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
6184 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
6185 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
6186 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
6187 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
6188 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
6189 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
6190 __builtin_ia32_vfmsubpd.
6191 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
6192 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
6193 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
6194 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
6195 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
6196 * config/i386/i386-builtin.def: Add
6197 __builtin_ia32_vfmsubpd256_mask,
6198 __builtin_ia32_vfmsubpd256_maskz,
6199 __builtin_ia32_vfmsubpd128_mask,
6200 __builtin_ia32_vfmsubpd128_maskz,
6201 __builtin_ia32_vfmsubps256_mask,
6202 __builtin_ia32_vfmsubps256_maskz,
6203 __builtin_ia32_vfmsubps128_mask,
6204 __builtin_ia32_vfmsubps128_maskz,
6205 __builtin_ia32_vfmsubpd512_mask,
6206 __builtin_ia32_vfmsubpd512_maskz,
6207 __builtin_ia32_vfmsubps512_mask,
6208 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
6209 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
6210 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
6211 __builtin_ia32_vfmsubpd256.
6212 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
6213 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
6214 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
6216 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
6218 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
6220 (fmai_vmfmsub_<mode><round_name>): Likewise.
6222 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6224 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
6225 Remove plus. Renamed to ...
6226 (*sub<mode>3<mask_name>_bcst): This.
6227 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
6228 (*add<mode>3<mask_name>_bcst): This.
6230 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6233 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
6235 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
6238 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
6239 (_mm_or_epi32): Likewise.
6240 (_mm256_xor_epi32): Likewise.
6241 (_mm_xor_epi32): Likewise.
6242 (_mm256_or_epi64): Likewise.
6243 (_mm_or_epi64): Likewise.
6244 (_mm256_xor_epi64): Likewise.
6245 (_mm_xor_epi64): Likewise.
6247 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
6250 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
6252 2018-10-20 Jakub Jelinek <jakub@redhat.com>
6255 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
6257 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
6259 * doc/ux.texi: Move @section directly after @node.
6261 2018-10-19 Jakub Jelinek <jakub@redhat.com>
6265 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
6266 depend closely nested inside of loop with ordered clause with
6269 2018-10-19 David Malcolm <dmalcolm@redhat.com>
6271 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
6272 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
6273 * doc/ux.texi: New file.
6275 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
6277 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
6278 be the first CR field allocated.
6280 2018-10-19 Richard Biener <rguenther@suse.de>
6283 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
6284 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
6286 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
6289 * config/i386/sse.md
6290 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
6291 (*add<mode>3<mask_name>_bcst_2): Likewise.
6293 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
6295 * config/i386/sse.md
6296 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
6297 Replace nonimmediate_operand with register_operand.
6298 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
6300 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
6303 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
6305 PR rtl-optimization/87596
6306 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
6307 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
6308 for instructions in FROM..TO range.
6310 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
6312 * cfgexpand.c (expand_one_var): Use specific wording in error message
6313 for non-local frame variables.
6314 * stor-layout.c (layout_decl): Do not issue a warning for them.
6316 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
6318 * haifa-sched.c (priority): Add force_recompute parameter.
6319 (apply_replacement): Call priority () with force_recompute = true.
6320 (restore_pattern): Likewise.
6322 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6324 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
6325 HOST_BITS_PER_WIDE_INT.
6326 (test_vector_ops_duplicate): Likewise.
6328 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6331 * config/i386/sse.md (VF_AVX512): New.
6332 (avx512bcst): Likewise.
6333 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
6335 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
6337 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
6340 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
6342 * doc/invoke.texi (-dumpversion): Improve grammar.
6343 (-dumpfullversion): Make more consistent with -dumpversion.
6345 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
6347 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
6348 Set JUMP_LABEL to the jump insn.
6349 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
6350 Predict emitted jump and add label to jump insn.
6352 2018-10-18 David Malcolm <dmalcolm@redhat.com>
6354 PR tree-optimization/87562
6355 * input.c (get_substring_ranges_for_loc): Use
6356 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
6357 getting the linemap for the endpoint. Verify that it's either
6358 in the same linemap as the start point's spelling location, or
6359 at least in the same file.
6361 2018-10-18 Richard Biener <rguenther@suse.de>
6363 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
6364 feed width-specific load/store costs through ix86_vec_cost.
6365 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
6366 (k8_cost): Likewise.
6367 (bdver_cost): Likewise.
6368 (znver1_cost): Likewise.
6369 (btver1_cost): Likewise.
6370 (btver2_cost): Likewise.
6372 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6374 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
6375 to simplify subreg of vec_merge.
6377 2018-10-18 Richard Biener <rguenther@suse.de>
6379 * config/i386/i386.c: Fix costing of vector FMA.
6381 2018-10-18 Richard Biener <rguenther@suse.de>
6383 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
6385 (ix86_builtin_vectorization_cost): For vec_construct properly
6386 cost insertion into SSE regs.
6387 (...): Adjust calls to ix86_vec_cost.
6389 2018-10-18 Richard Biener <rguenther@suse.de>
6393 2018-02-07 Richard Biener <rguenther@suse.de>
6395 PR tree-optimization/84204
6396 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
6399 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
6402 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
6404 (test_vector_ops_duplicate): Add test for a scalar subreg of a
6405 VEC_MERGE of a VEC_DUPLICATE.
6407 2018-10-17 Joseph Myers <joseph@codesourcery.com>
6409 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
6410 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
6411 * doc/standards.texi (C Language): Document C2X.
6412 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
6413 (rl78_option_override): Handle "GNU C2X" language name.
6415 2018-10-17 Joseph Myers <joseph@codesourcery.com>
6417 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
6418 Document C17 as published in 2018.
6420 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
6423 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
6424 bail out if both sides do not have the same storage order.
6426 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6428 * bitmap.c (bitmap_head::dump): New.
6429 * bitmap.h (bitmap_head): Add dump().
6430 * gimple-ssa-evrp-analyze.c
6431 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
6432 (evrp_range_analyzer::set_ssa_range_info): Same.
6433 (evrp_range_analyzer::record_ranges_from_phis): Same.
6434 (evrp_range_analyzer::record_ranges_from_stmt): Same.
6435 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
6436 * gimple-ssa-sprintf.c (get_int_range): Same.
6437 (format_integer): Same.
6438 (sprintf_dom_walker::handle_gimple_call): Same.
6439 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
6440 (ipcp_vr_lattice::top_p): Same.
6441 (ipcp_vr_lattice::bottom_p): Same.
6442 (ipcp_vr_lattice::set_to_bottom): Same.
6443 (ipa_vr_operation_and_type_effects): Same.
6444 (propagate_vr_across_jump_function): Same.
6445 (ipcp_store_vr_results): Same.
6446 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
6447 (ipa_print_node_jump_functions_for_edge): Same.
6448 (ipa_get_value_range): Same.
6449 (ipa_compute_jump_functions_for_edge): Same.
6450 (ipa_write_jump_function): Same.
6451 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
6452 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
6454 * vr-values.c (set_value_range_to_nonnegative): Same.
6455 (set_value_range_to_truthvalue): Same.
6456 (vr_values::get_value_range): Same.
6457 (vr_values::set_defs_to_varying): Same.
6458 (vr_values::update_value_range): Same.
6459 (symbolic_range_based_on_p): Same.
6460 (vr_values::op_with_boolean_value_range_p): Same.
6461 (vr_values::extract_range_for_var_from_comparison_expr): Same.
6462 (vr_values::extract_range_from_ssa_name): Same.
6463 (vr_values::extract_range_from_binary_expr): Same.
6464 (vr_values::extract_range_from_unary_expr): Same.
6465 (vr_values::extract_range_from_cond_expr): Same.
6466 (vr_values::extract_range_from_comparison): Same.
6467 (vr_values::check_for_binary_op_overflow): Same.
6468 (vr_values::extract_range_basic): Same.
6469 (vr_values::extract_range_from_assignment): Same.
6470 (compare_ranges): Same.
6471 (compare_range_with_value): Same.
6472 (vr_values::adjust_range_with_scev): Same.
6473 (vrp_valueize): Same.
6474 (vrp_valueize_1): Same.
6475 (vr_values::get_vr_for_comparison): Same.
6476 (vr_values::compare_name_with_value): Same.
6477 (vr_values::compare_names): Same.
6478 (vr_values::vrp_evaluate_conditional): Same.
6479 (find_case_label_ranges): Same.
6480 (vr_values::vrp_visit_switch_stmt): Same.
6481 (vr_values::extract_range_from_phi_node): Same.
6482 (vr_values::simplify_div_or_mod_using_ranges): Same.
6483 (vr_values::simplify_bit_ops_using_ranges): Same.
6484 (test_for_singularity): Same.
6485 (range_fits_type_p): Same.
6486 (vr_values::simplify_cond_using_ranges_1): Same.
6487 (vr_values::simplify_switch_using_ranges): Same.
6488 (vr_values::simplify_float_conversion_using_ranges): Same.
6489 (vr_values::two_valued_val_range_p): Same.
6490 (vr_values::add_equivalence): Move to value_range::equiv_add.
6491 * vr-values.h (vr_values::add_equivalence): Remove.
6492 (VR_INITIALIZER): Remove.
6493 * tree-vrp.c (value_range::set): New.
6494 (value_range::equiv_add): New.
6495 (value_range::value_range): New.
6496 (value_range::deep_copy): New.
6497 (value_range::check): New.
6498 (value_range::equal_p): New.
6499 (value_range::ignore_equivs_equal_p): New.
6500 (value_range::operator==): New.
6501 (value_range::operator!=): New.
6502 (value_range::symbolic_p): New.
6503 (value_range::numeric_p): New.
6504 (value_range::set_undefined): New.
6505 (value_range::set_varying): New.
6506 (value_range::may_contain_p): New.
6507 (value_range::equiv_clear): New.
6508 (value_range::singleton_p): New.
6509 (value_range::intersect): New.
6510 (value_range::dump): New.
6511 (value_range::set_and_canonicalize): New.
6512 (set_value_range): Adjust for value_range API.
6513 (set_value_range_to_undefined): Same.
6514 (set_value_range_to_varying): Same.
6515 (set_and_canonicalize_value_range): Same.
6516 (set_value_range_to_nonnull): Same.
6517 (set_value_range_to_null): Same.
6518 (range_is_null): Same.
6519 (range_is_nonnull): Same.
6520 (range_int_cst_p): Same.
6521 (range_int_cst_singleton_p): Same.
6522 (symbolic_range_p): Same.
6523 (range_includes_zero_p): Same.
6524 (value_range_constant_singleton): Same.
6525 (vrp_set_zero_nonzero_bits): Same.
6526 (ranges_from_anti_range): Same.
6527 (extract_range_into_wide_ints): Same.
6528 (extract_range_from_multiplicative_op): Same.
6529 (set_value_range_with_overflow): Same.
6530 (extract_range_from_binary_expr_1): Same.
6531 (extract_range_from_unary_expr): Same.
6532 (dump_value_range): Same.
6533 (debug_value_range): Same.
6534 (vrp_prop::check_array_ref): Same.
6535 (vrp_prop::check_mem_ref): Same.
6536 (vrp_prop::vrp_initialize): Same.
6537 (vrp_prop::visit_stmt): Same.
6538 (intersect_ranges): Same.
6539 (vrp_prop::visit_phi): Same.
6540 (vrp_prop::vrp_finalize): Same.
6541 (determine_value_range_1): Same.
6542 (determine_value_range): Same.
6543 (vrp_intersect_ranges_1): Rename to...
6544 (vrp_intersect_1): this.
6545 (vrp_intersect_ranges): Rename to...
6546 (value_range::intersect_helper): ...this.
6547 (vrp_meet_1): Rename to...
6548 (value_range::union_helper): ...this.
6549 (vrp_meet): Rename to...
6550 (value_range::union_): ...this.
6551 (copy_value_range): Remove.
6552 * tree-vrp.h (struct value_range): Rewrite into a proper class.
6553 (value_range::vrtype): New.
6554 (value_range::type): New.
6555 (value_range::equiv): New.
6556 (value_range::min): New.
6557 (value_range::max): New.
6558 (value_range::varying_p): New.
6559 (value_range::undefined_p): New.
6560 (value_range::null_p): New.
6561 (value_range::equiv_add): New.
6562 (copy_value_range): Remove.
6564 2018-10-17 David Malcolm <dmalcolm@redhat.com>
6566 * Makefile.in (SELFTEST_TARGETS): New.
6567 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
6568 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
6569 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
6571 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
6572 (selftest-c++-gdb, selftest-c++-valgrind): Move to
6574 * configure: Regenerate.
6575 * configure.ac (selftest_languages): New.
6577 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6579 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
6580 overflow wraps argument.
6581 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
6582 wide_int_range_multiplicative_op.
6583 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
6584 overflow wraps argument.
6585 (wide_int_range_multiplicative_op): Same.
6586 (wide_int_range_lshift): Same.
6587 (wide_int_range_div): Same.
6588 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
6589 (wide_int_range_lshift): Same.
6590 (wide_int_range_div): Same.
6592 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
6594 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
6595 use sign as argument.
6596 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
6597 wide_int_range_shift_undefined_p.
6599 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6601 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
6603 (@despeculate_copy<ALLI_TI:mode>): ... This.
6604 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
6607 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6609 * config.gcc: Obsolete *-*-solaris2.10*.
6610 * doc/install.texi (Specific, *-*-solaris2*): Document it.
6612 2018-10-12 Jeff Law <law@redhat.com>
6614 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
6615 reg + sym +- const_int addressing modes.
6617 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6619 * common.opt (fdiagnostics-minimum-margin-width=): New option.
6620 * diagnostic-show-locus.c (layout::layout): Apply the minimum
6622 (layout::start_annotation_line): Only print up to 3 of the
6623 margin character, to avoid touching the left-hand side.
6624 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
6625 minimum margin width, as set by test_diagnostic_context's ctor.
6626 (selftest::test_fixit_insert_containing_newline): Likewise.
6627 (selftest::test_fixit_insert_containing_newline_2): Likewise.
6628 (selftest::test_line_numbers_multiline_range): Clear
6629 dc.min_margin_width.
6630 * diagnostic.c (diagnostic_initialize): Initialize
6632 * diagnostic.h (struct diagnostic_context): Add field
6634 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
6635 * opts.c (common_handle_option): Handle
6636 OPT_fdiagnostics_minimum_margin_width_.
6637 * selftest-diagnostic.c
6638 (selftest::test_diagnostic_context::test_diagnostic_context):
6639 Initialize min_margin_width to 6.
6640 * toplev.c (general_init): Initialize global_dc->min_margin_width.
6642 2018-10-15 David Malcolm <dmalcolm@redhat.com>
6644 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6645 Fix usage of "error_at_rich_loc" in the comment.
6647 2018-10-15 Renlin Li <renlin.li@arm.com>
6650 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
6651 if-conversioned loop when it contains ifn with types not
6652 supported by backend.
6653 * internal-fn.c (expand_direct_optab_fn): Add an assert.
6654 (direct_internal_fn_supported_p): New helper function.
6655 * internal-fn.h (direct_internal_fn_supported_p): Declare.
6657 2018-10-15 Jakub Jelinek <jakub@redhat.com>
6660 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6661 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6662 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6664 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
6666 PR tree-optimization/87022
6667 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
6668 bits in dist vector rather than the first one.
6670 2018-10-15 Richard Biener <rguenther@suse.de>
6673 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
6674 (visit_loadstore): When a used restrict tag escaped verify that
6675 the points-to solution of "other" pointers do not include
6677 (compute_dependence_clique): If a used restrict tag escaped
6678 communicated that down to visit_loadstore.
6680 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
6682 * config/s390/s390.c (s390_expand_vec_init): Force vector element
6683 into reg if it isn't a general operand.
6685 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6688 * config/i386/sse.md (*vec_dupv2di): Add register source to
6691 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
6694 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
6695 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
6696 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
6697 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
6699 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
6701 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
6702 (notice_args_size): Set it in the current trace if no insn that can
6703 throw internally has been seen yet.
6704 (connect_traces): When connecting args_size between traces, allow the
6705 incoming values not to match if there is an insn setting it before the
6706 first insn that can throw internally; in that case, force the creation
6707 of a CFI note on this latter insn.
6709 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
6711 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
6712 the base class, because using the injected-class-name was not clearly
6713 specified until DR 176.
6715 2018-10-12 Paul Koning <ni1d@arrl.net>
6717 * config/pdp11/pdp11.md (doloop_end): New expander.
6718 (doloop_end_insn): renamed from "doloop_end".
6719 (addqi3): New pattern.
6720 (subqi3): New pattern.
6721 * config/pdp11/predicates.md (incdec_operand): New predicate.
6723 2018-10-12 Yury Gribov <tetra2005@gmail.com>
6726 * real.c (format_helper::can_represent_integral_type_p): New function
6727 * real.h (format_helper::can_represent_integral_type_p): Ditto.
6728 * match.pd: New pattern.
6730 2018-10-12 Alexandre Oliva <oliva@adacore.com>
6732 * configure.ac: Introduce --enable-large-address-aware
6733 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
6734 * doc/install.texi: Document it.
6735 * configure, config.in: Rebuilt.
6736 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
6737 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
6738 (LINK_SPEC): Insert it.
6739 * config/i386/mingw-w64.h: Likewise.
6741 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
6743 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
6745 PR rtl-optimization/87600
6746 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
6748 2018-10-12 Paul Koning <ni1d@arrl.net>
6750 * doc/md.texi (doloop_end): Document that the pattern code may
6751 need to check operand mode.
6753 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
6755 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
6756 to zero-extend between int and floating-point registers.
6757 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
6758 ldp into floating-point registers. Add type and arch attributes.
6759 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
6760 Use f_loads for type attribute.
6762 2018-10-11 Martin Sebor <msebor@redhat.com>
6764 * doc/extend.texi (attribute packed): Correct typos.
6766 2018-10-11 Martin Sebor <msebor@redhat.com>
6768 * doc/extend.texi (attribute flatten): Mention interaction with
6771 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
6774 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
6775 Set new_decl virtual flag to zero.
6777 2018-10-11 Martin Sebor <msebor@redhat.com>
6780 * doc/extend.texi (attribute format_arg): Discuss using multiple
6781 attributes on a single function.
6783 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
6785 PR tree-optimization/86829
6786 * match.pd (sin (atan (x))): New simplification rules.
6787 (cos (atan (x))): Likewise.
6788 * real.c (build_sinatan_real): New function.
6789 * real.h (build_sinatan_real): Prototype.
6791 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
6793 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
6795 (fold_mergeeo_helper): New helper function.
6796 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
6797 intrinsics. Correct some whitespace indentation issues.
6799 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
6802 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
6803 Use HOST_WIDE_INT_1U for shift.
6805 2018-10-11 Doug Rupp <rupp@adacore.com>
6806 Olivier Hainque <hainque@adacore.com>
6808 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
6809 Pass --relax to the linker for RTPs.
6810 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
6812 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
6813 Jan Hubicka <jh@suse.cz>
6814 Martin Jambor <mjambor@suse.cz>
6816 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
6817 the same elements are repeated rather than printing all of them.
6818 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
6819 "repeated" elements.
6820 * read-rtl-function.c (test_loading_repeat): New function.
6821 (read_rtl_function_c_tests): Call test_loading_repeat.
6822 * rtl-tests.c (test_dumping_repeat): New function.
6823 (rtl_tests_c_tests): Call test_dumping_repeat.
6825 2018-10-11 Richard Biener <rguenther@suse.de>
6827 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
6828 bdver?_cost): Unify to ...
6829 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
6830 * config/i386/i386.c (processor_cost_table): Adjust.
6832 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
6835 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
6836 the thunk when expanding to GIMPLE.
6838 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
6841 * varasm.c (mergeable_string_section): Don't try to move zero-length
6842 strings to the merge section.
6844 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
6847 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
6849 2018-10-10 Jakub Jelinek <jakub@redhat.com>
6852 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
6853 to special_args set.
6855 2018-10-10 Richard Biener <rguenther@suse.de>
6857 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
6858 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
6859 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
6860 and recursing and pattern terminating the recursion on SSE
6861 vector width using ix86_expand_reduc.
6862 (reduc_sminmax_scal_<mode>): Split into part reducing to half
6863 width and recursing and SSE2 vector variant doing the final
6864 reduction with ix86_expand_reduc.
6865 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
6866 with terminating the recursion at AVX level, splitting that
6869 2018-10-09 David Malcolm <dmalcolm@redhat.com>
6871 * genmatch.c (error_cb): Rename to...
6872 (diagnostic_cb): ...this, converting int params to enums.
6873 (fatal_at): Update for renaming.
6874 (warning_at): Likewise.
6876 * input.c (selftest::ebcdic_execution_charset::apply):
6877 Update for renaming of...
6878 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
6880 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
6881 converting level and reason to enums.
6882 (class selftest::lexer_error_sink): Rename to...
6883 (class selftest::lexer_test_options): ...this, renaming field
6884 "m_errors" to "m_diagnostics".
6885 (selftest::lexer_test_options::apply): Update for renaming of...
6886 (selftest::lexer_test_options::on_error): ...this, renaming to...
6887 (selftest::lexer_test_options::on_diagnostic): ...this
6888 converting level and reason to enums.
6889 (selftest::test_lexer_string_locations_raw_string_unterminated):
6890 Update for renamings.
6891 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
6894 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
6896 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
6897 * config/rs6000/pmmintrin.h: New file.
6899 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
6901 PR tree-optimization/86659
6902 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
6904 2018-10-09 Richard Biener <rguenther@suse.de>
6906 PR tree-optimization/63155
6907 * tree-ssa-structalias.c: Include tree-ssa.h.
6908 (get_constraint_for_ssa_var): For undefs return nothing_id.
6909 (find_func_aliases): Cleanup PHI handling.
6911 2018-10-09 Richard Biener <rguenther@suse.de>
6913 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
6916 2018-10-09 Martin Liska <mliska@suse.cz>
6918 * asan.c (asan_emit_stack_protection): If a stack variable
6919 is located in a same file as current function, then emit
6920 line info into variable definition string.
6922 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6924 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
6927 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6929 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
6932 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6934 PR tree-optimization/86659
6935 * gimple-match.h (struct gimple_match_op): Add reverse field.
6936 (gimple_match_op::set_op): New overloaded method.
6937 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
6938 the REF_REVERSE_STORAGE_ORDER flag on the value.
6939 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
6940 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
6942 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
6945 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
6946 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
6948 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
6951 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
6952 Defined with __builtin_ia32_vfmaddsubpd512_mask.
6954 2018-10-08 Richard Biener <rguenther@suse.de>
6956 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
6957 cost the same as AVX128 ones.
6959 2018-10-08 Paul Koning <ni1d@arrl.net>
6961 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
6962 (expand_block_move): New function.
6963 * config/pdp11/pdp11.c (output_block_move): Remove.
6964 (expand_block_move): New function.
6965 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
6966 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
6967 (*movmemhi1): Remove.
6969 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
6971 * config/s390/2827.md: Increase latencies for some FP instructions.
6973 2018-10-08 Richard Biener <rguenther@suse.de>
6975 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
6977 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
6978 * tree-vectorizer.h (dump_stmt_cost): Adjust.
6979 (add_stmt_cost): Dump return value of the hook.
6981 2018-10-08 Richard Biener <rguenther@suse.de>
6983 PR tree-optimization/63155
6984 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
6985 (ssa_propagation_engine::ssa_propagate): Remove redundant
6986 bitmap bit clearing.
6988 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
6990 PR rtl-optimization/86939
6991 PR rtl-optimization/87479
6992 * ira.h (non_conflicting_reg_copy_p): New prototype.
6993 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
6994 (make_hard_regno_dead): Don't add conflicts for register
6995 ignore_reg_for_conflicts.
6996 (make_object_dead): Likewise.
6997 (non_conflicting_reg_copy_p): New function.
6998 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
6999 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
7000 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
7001 (make_hard_regno_dead): Don't add conflicts for register
7002 ignore_reg_for_conflicts. Remove special conflict handling of
7003 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
7004 check_pic_pseudo_p and update callers.
7005 (mark_pseudo_dead): Don't add conflicts for register
7006 ignore_reg_for_conflicts.
7007 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
7009 2018-10-05 Andrew Waterman <andrew@sifive.com>
7010 Jim Wilson <jimw@sifive.com>
7012 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
7013 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
7014 new pattern using HONOR_SNANS that emits one extra instruction.
7016 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7018 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
7019 patterns): Merge SI and DI patterns to a GPR pattern.
7020 (unnamed define_insn and define_split for record form of that): Merge
7021 to a single define_insn_and_split pattern.
7023 2018-10-05 David Malcolm <dmalcolm@redhat.com>
7026 * input.c (expand_location_to_spelling_point): Add param "aspect"
7027 and use rather than hardcoding LOCATION_ASPECT_CARET.
7028 (get_substring_ranges_for_loc): Handle the case of a single token
7029 within a macro expansion.
7030 * input.h (expand_location_to_spelling_point): Add "aspect" param,
7031 defaulting to LOCATION_ASPECT_CARET.
7033 2018-10-05 Paul Koning <ni1d@arrl.net>
7035 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
7036 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
7037 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
7038 (pdp11_guard_type): New function.
7040 2018-10-05 Paul Koning <ni1d@arrl.net>
7042 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
7043 * config/pdp11/pdp11.opt (mfloat32): Remove.
7045 * doc/invoke.texi (pdp11 -mfloat32): Remove:
7046 (pdp11 -mfloat64): Remove.
7048 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7050 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
7051 (*cmp<mode>_cc_i387): Ditto.
7052 (*cmpu<mode>_cc_i387): Ditto.
7053 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7054 * config/i386/i386.c (ix86_expand_fp_compare): Remove
7056 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
7057 Emit x86_sahf_1 pattern.
7058 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
7059 (ix86_expand_carry_flag_compare): Ditto.
7061 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7063 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
7064 to reg_or_0_operand. Add "C" constraint.
7065 (*cmpxf_cc_i387): Ditto.
7066 (*cmp<mode>_i387): Change operand 2 predicate
7067 to nonimm_or_0_operand. Add "C" constraint.
7068 (*cmp<mode>_cc_i387): Ditto.
7069 (*cmp<mode>_0_i387): Remove insn pattern.
7070 (*cmp<mode>_0_cc_i387): Ditto.
7072 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
7074 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
7075 * config/i386/predicates.md (nonimm_or_0_operand): Rename
7076 from vector_move_operand. Update all uses.
7078 2018-10-05 Martin Sebor <msebor@redhat.com>
7080 PR tree-optimization/87490
7081 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
7084 2018-10-05 Richard Biener <rguenther@suse.de>
7086 PR tree-optimization/63155
7087 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
7088 vertical space in dumpfiles.
7089 * tree-ssa-propagate.h
7090 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
7091 * tree-ssa-propagate.c (cfg_blocks_back): New global.
7092 (ssa_edge_worklist_back): Likewise.
7093 (curr_order): Likewise.
7094 (cfg_blocks_get): Remove abstraction.
7095 (cfg_blocks_add): Likewise.
7096 (cfg_blocks_empty_p): Likewise.
7097 (add_ssa_edge): Add to current or next worklist based on
7099 (add_control_edge): Likewise.
7100 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
7102 (ssa_propagation_engine::ssa_propagate): ... here. Unify
7103 iteration from CFG and SSA edge worklist so we process
7104 everything in RPO order, prioritizing forward progress
7106 (ssa_prop_init): Allocate new worklists, do not dump
7108 (ssa_prop_fini): Free new worklists.
7110 2018-10-05 Richard Biener <rguenther@suse.de>
7112 * tree-core.h (tree_block::abstract_flag): Remove.
7113 (tree_block::block_num): Make full 32bits.
7114 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
7115 * tree.h (BLOCK_ABSTRACT): Remove.
7116 * dwarf2out.c (gen_lexical_block_die): Remove dead code
7117 resulting from BLOCK_ABSTRACT being always false.
7118 (gen_inlined_subroutine_die): Likewise.
7119 (gen_block_die): Likewise.
7120 * tree.c (block_ultimate_origin): Likewise.
7121 * tree-pretty-print.c (dump_block_node): Remove code dealing
7122 with BLOCK_ABSTRACT.
7123 * tree-ssa-live.c (dump_scope_block): Likewise.
7124 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
7125 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
7127 2018-10-05 Richard Biener <rguenther@suse.de>
7129 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
7130 is asked for initialize mode to the component mode of the
7133 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
7136 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
7137 assembler for -mavx.
7138 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
7140 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
7143 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
7145 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
7146 to be DImode. When using mffscrn, force the operand to a register.
7148 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
7150 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
7151 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
7152 X87MODEF mode iterator.
7153 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
7154 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
7155 X87MODEF mode iterator.
7157 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
7159 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
7160 -Wno-prio-ctor-dtor.
7162 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7164 * Makefile.in (OBJS): Add opt-problem.o.
7165 * dump-context.h: Include "selftest.h.
7166 (selftest::temp_dump_context): New forward decl.
7167 (class dump_context): Make friend of class
7168 selftest::temp_dump_context.
7169 (dump_context::dump_loc_immediate): New decl.
7170 (class dump_pretty_printer): Move here from dumpfile.c.
7171 (class temp_dump_context): Move to namespace selftest.
7172 (temp_dump_context::temp_dump_context): Add param
7173 "forcibly_enable_dumping".
7174 (selftest::verify_dumped_text):
7175 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
7176 (selftest::verify_item):
7177 (ASSERT_IS_TEXT): Move here from dumpfile.c.
7178 (ASSERT_IS_TREE): Likewise.
7179 (ASSERT_IS_GIMPLE): Likewise.
7180 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
7182 (dump_context::dump_loc_immediate): ...this new function.
7183 (class dump_pretty_printer): Move to dump-context.h.
7184 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
7185 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
7186 (temp_dump_context::temp_dump_context): Move to "selftest"
7187 namespace. Add param "forcibly_enable_dumping", and use it to
7188 conditionalize the use of m_pp;
7189 (selftest::verify_dumped_text): Make non-static.
7190 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
7191 (selftest::verify_item): Make non-static.
7192 (ASSERT_IS_TEXT): Move to dump-context.h.
7193 (ASSERT_IS_TREE): Likewise.
7194 (ASSERT_IS_GIMPLE): Likewise.
7195 (selftest::test_capture_of_dump_calls): Pass "true" for new
7196 param of temp_dump_context.
7197 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
7198 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
7200 * opt-problem.cc: New file.
7201 * opt-problem.h: New file.
7202 * optinfo-emit-json.cc
7203 (selftest::test_building_json_from_dump_calls): Pass "true" for
7204 new param of temp_dump_context.
7205 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
7206 (optinfo::emit_for_opt_problem): New function.
7207 (optinfo::emit): Clarity which emit_item is used.
7208 * optinfo.h (optinfo::get_dump_location): New accessor.
7209 (optinfo::emit_for_opt_problem): New decl.
7210 (optinfo::emit): Make const.
7211 * selftest-run-tests.c (selftest::run_tests): Call
7212 selftest::opt_problem_cc_tests.
7213 * selftest.h (selftest::opt_problem_cc_tests): New decl.
7214 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
7215 bool to opt_result, converting fprintf messages to
7216 opt_result::failure_at calls. Add "stmt" param for use by the
7218 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
7219 (runtime_alias_check_p): Convert return type from bool to
7220 opt_result, converting dump_printf calls to
7221 opt_result::failure_at, using the statement DDR_A for their
7223 (find_data_references_in_stmt): Convert return type from bool to
7224 opt_result, converting "return false" to opt_result::failure_at
7226 * tree-data-ref.h: Include "opt-problem.h".
7227 (dr_analyze_innermost): Convert return type from bool to opt_result,
7228 and add a const gimple * param.
7229 (find_data_references_in_stmt): Convert return type from bool to
7231 (runtime_alias_check_p): Likewise.
7232 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
7233 dr_analyze_innermost.
7234 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
7235 Convert return type from bool to opt_result, adding a message for
7236 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
7237 (vect_analyze_data_ref_dependence): Convert return type from bool
7238 to opt_result. Change sense of return type from "false"
7239 effectively meaning "no problems" to "false" meaning a problem,
7240 so that "return false" becomes "return opt_result::success".
7241 Convert "return true" calls to opt_result::failure_at, using
7242 the location of statement A rather than vect_location.
7243 (vect_analyze_data_ref_dependences): Convert return type from bool
7245 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
7246 calls to opt_result::failure_at, using the stmt location rather
7248 (vect_verify_datarefs_alignment): Convert return type from bool
7250 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
7251 into multiple more-tightly-scoped copies.
7252 (vect_analyze_data_refs_alignment): Convert return type from bool
7254 (vect_analyze_data_ref_accesses): Likewise, converting a
7255 "return false" to a "return opt_result::failure_at", adding a
7257 (vect_prune_runtime_alias_test_list): Convert return type from
7258 bool to opt_result, converting dump_printf_loc to
7259 opt_result::failure_at. Add a %G to show the pertinent statement,
7260 and use the stmt's location rather than vect_location.
7261 (vect_find_stmt_data_reference): Convert return type from
7262 bool to opt_result, converting dump_printf_loc to
7263 opt_result::failure_at, using stmt's location.
7264 (vect_analyze_data_refs): Convert return type from bool to
7265 opt_result. Convert "return false" to "return
7266 opt_result::failure_at", adding messages as needed.
7267 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
7268 type from bool to opt_result.
7269 (vect_determine_vf_for_stmt): Likewise.
7270 (vect_determine_vectorization_factor): Likewise, converting
7271 dump_printf_loc to opt_result::failure_at, using location of phi
7272 rather than vect_location.
7273 (vect_analyze_loop_form_1): Convert return type from bool to
7274 opt_result, converting dump_printf_loc calls, retaining the use of
7276 (vect_analyze_loop_form): Convert return type from loop_vec_info
7277 to opt_loop_vec_info.
7278 (vect_analyze_loop_operations): Convert return type from bool to
7279 opt_result, converting dump_printf_loc calls, using the location
7280 of phi/stmt rather than vect_location where available. Convert
7281 various "return false" to "return opt_result::failure_at" with
7282 "unsupported phi" messages.
7283 (vect_get_datarefs_in_loop): Convert return type from bool to
7284 opt_result. Add a message for the
7285 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
7286 (vect_analyze_loop_2): Convert return type from bool to
7287 opt_result. Ensure "ok" is set to a opt_result::failure_at before
7288 each "goto again;", adding new messages where needed.
7289 Add "unsupported grouped {store|load}" messages.
7290 (vect_analyze_loop): Convert return type from loop_vec_info to
7292 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
7294 * tree-vect-stmts.c (process_use): Likewise, converting
7295 dump_printf_loc call and using stmt location, rather than
7297 (vect_mark_stmts_to_be_vectorized): Likeise.
7298 (vect_analyze_stmt): Likewise, adding a %G.
7299 (vect_get_vector_types_for_stmt): Convert return type from bool to
7300 opt_result, converting dump_printf_loc calls and using stmt
7301 location, rather than vect_location.
7302 (vect_get_mask_type_for_stmt): Convert return type from tree to
7303 opt_tree, converting dump_printf_loc calls and using stmt location.
7304 * tree-vectorizer.c: Include "opt-problem.h.
7305 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
7306 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
7307 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
7308 enabled, use it to report at the top level "couldn't vectorize
7309 loop" followed by the problem.
7310 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
7311 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
7313 (vect_analyze_stmt): Likewise.
7314 (vect_get_vector_types_for_stmt): Likewise.
7315 (tree vect_get_mask_type_for_stmt): Likewise.
7316 (vect_analyze_data_ref_dependences): Likewise.
7317 (vect_enhance_data_refs_alignment): Likewise.
7318 (vect_analyze_data_refs_alignment): Likewise.
7319 (vect_verify_datarefs_alignment): Likewise.
7320 (vect_analyze_data_ref_accesses): Likewise.
7321 (vect_prune_runtime_alias_test_list): Likewise.
7322 (vect_find_stmt_data_reference): Likewise.
7323 (vect_analyze_data_refs): Likewise.
7324 (vect_analyze_loop): Convert return type from loop_vec_info to
7326 (vect_analyze_loop_form): Likewise.
7327 (vect_analyze_slp): Convert return type from bool to opt_result.
7329 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7331 * doc/invoke.texi (-fopt-info): Document new "internals"
7333 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
7334 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
7336 (optinfo_verbosity_options): Add "internals".
7337 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
7338 (dump_context::apply_dump_filter_p): New member function.
7339 (dump_context::dump_loc): Use apply_dump_filter_p rather than
7340 explicitly masking the dump_kind.
7341 (dump_context::begin_scope): Increment the scope depth first. Use
7342 apply_dump_filter_p rather than explicitly masking the dump_kind.
7343 (dump_context::emit_item): Use apply_dump_filter_p rather than
7344 explicitly masking the dump_kind.
7345 (dump_dec): Likewise.
7346 (dump_hex): Likewise.
7347 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
7348 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
7349 (opt_info_switch_p): Update handling of default
7350 MSG_OPTIMIZED_LOCATIONS to cope with default of
7351 MSG_PRIORITY_USER_FACING.
7352 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
7353 masking the dump_kind.
7354 (selftest::test_capture_of_dump_calls): Update test_dump_context
7355 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
7356 than MSG_ALL. Generalize scope test to be run at all four
7357 combinations of with/without MSG_PRIORITY_USER_FACING and
7358 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
7359 for each of the two values.
7360 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
7361 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
7362 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
7363 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
7364 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
7365 with MSG_PRIORITY_*.
7366 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
7367 dump messages as MSG_PRIORITY_USER_FACING.
7368 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
7369 about the interaction with MSG_PRIORITY_*.
7371 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7373 * varasm.c (output_constant): Add new parameter merge_strings.
7374 Make strings properly zero terminated in merge string sections.
7375 (mergeable_string_section): Don't fail if the last char is non-zero.
7376 (assemble_variable_contents): Handle merge string sections.
7377 (assemble_variable): Likewise.
7378 (assemble_constant_contents): Likewise.
7379 (output_constant_def_contents): Likewise.
7380 (output_constructor_array_range,
7381 output_constructor_regular_field): Adjust call to output_constant.
7382 (output_object_block): Adjust call to assemble_constant_contents
7383 and assemble_variable_contents.
7385 2018-10-04 Martin Liska <mliska@suse.cz>
7388 * cgraphunit.c (process_function_and_variable_attributes):
7389 Warn about a function with alias attribute and a body.
7391 2018-10-04 Martin Liska <mliska@suse.cz>
7394 * multiple_target.c (redirect_to_specific_clone): New function.
7395 (ipa_target_clone): Use it.
7396 * tree-inline.c: Fix comment.
7398 2018-10-04 David Malcolm <dmalcolm@redhat.com>
7400 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
7402 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
7403 (gcc::dump_manager::register_pass): New member function, adapted
7404 from loop body in gcc::pass_manager::register_pass, adding a
7405 call to update_dfi_for_opt_info.
7406 (gcc::dump_manager::opt_info_enable_passes): Store the
7407 -fopt-info options into the new fields. Move the loop
7409 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
7411 * dumpfile.h (struct opt_pass): New forward decl.
7412 (gcc::dump_manager::register_pass): New decl.
7413 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
7414 (class gcc::dump_manager): Add fields "m_optgroup_flags",
7415 "m_optinfo_flags", and "m_optinfo_filename".
7416 * passes.c (gcc::pass_manager::register_pass): Move all of the
7417 dump-handling code to gcc::dump_manager::register_pass.
7419 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
7421 PR rtl-optimization/87466
7422 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
7423 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
7424 * doc/tm.texi: Regenerate.
7425 * ira-lives.c (process_bb_node_lives): Use the new target hook.
7426 * lra-lives.c (process_bb_lives): Likewise.
7427 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
7430 2018-10-04 Tamar Christina <tamar.christina@arm.com>
7432 * params.c (add_params): Fix initialization.
7434 2018-10-04 Martin Liska <mliska@suse.cz>
7436 PR gcov-profile/84107
7437 * tree-profile.c (init_ic_make_global_vars):
7438 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
7439 Come up with new ic_tuple* variables. Emit
7440 __gcov_indirect_call{,_topn} variables.
7441 (gimple_gen_ic_profiler): Access the variable
7443 (gimple_gen_ic_func_profiler): Access
7444 __gcov_indirect_call.callee field.
7445 (gimple_init_gcov_profiler): Use ptr_type_node.
7446 * value-prof.c (gimple_ic): Use ptr_type_node.
7448 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7450 PR tree-optimization/85787
7451 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
7452 malloc_candidate_p into this function and add support for
7453 detecting multiple phis.
7454 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
7456 2018-10-04 Martin Liska <mliska@suse.cz>
7459 * ipa-inline.c (inline_to_all_callers_1):
7460 Call ultimate_alias_target for node being inlined.
7462 2018-10-03 Jeff Law <law@redhat.com>
7464 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
7466 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
7467 * tree.h (get_typenode_from_name): Prototype.
7469 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
7471 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
7472 Change operand 2 predicate to nonimmediate_operand.
7473 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
7475 2018-10-03 Martin Sebor <msebor@redhat.com>
7476 Jeff Law <law@redhat.com>
7478 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
7480 (get_string_length): Detect unterminated arrays.
7481 (format_string): Same.
7482 (format_directive): Warn about unterminated arrays.
7483 (handle_gimple_call): Mark statements with no_warning as needed.
7485 2018-10-03 Jim Wilson <jimw@sifive.com>
7487 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
7488 also define __riscv_abi_rve. Delete trailing white space.
7490 2018-10-03 Paul Koning <ni1d@arrl.net>
7492 Enable LRA register allocator for PDP11.
7493 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
7496 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
7497 * config/pdp11/pdp11.opt (-mlra): New option.
7498 * doc/invoke.texi (PDP-11 Options): Document -mlra.
7500 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
7502 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
7503 (*<absneg:code>extend<mode>xf2): Ditto.
7505 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
7507 PR tree-optimization/87415
7508 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
7511 2018-10-02 Jeff Law <law@redhat.com>
7513 * gimple-fold.c (get_range_strlen): Only set *nonstr when
7514 an unterminated string is discovered. Bubble up range
7515 even for unterminated strings.
7516 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
7517 indicates the string was not terminated via NONSTR.
7519 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
7521 * tree-vrp.c (extract_range_from_unary_expr): Special case all
7522 pointer conversions.
7523 Do not do anything special for anti-ranges.
7525 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
7527 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
7528 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
7530 2018-10-03 Martin Liska <mliska@suse.cz>
7532 PR gcov-profile/86109
7533 * coverage.c (coverage_begin_function): Do not
7534 mark lambdas as artificial.
7535 * tree-core.h (struct GTY): Remove tm_clone_flag
7536 and introduce new lambda_function.
7537 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
7539 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7542 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
7543 P8_VECTOR and VSX are enabled.
7545 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
7547 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
7548 0x3907 as CPU model number.
7550 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
7552 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
7553 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
7554 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
7555 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
7556 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
7557 * config/s390/s390.md: Likewise. Rename also the cpu attribute
7558 value from arch12 to z14.
7560 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
7562 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
7564 (isinf<mode>2): Ditto.
7566 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
7568 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
7569 before emitting fxam. Perform calculations in XFmode.
7571 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
7573 * match.pd (((X /[ex] A) +- B) * A): New transformation.
7575 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
7578 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
7579 * tree.c (signed_or_unsigned_type_for): Handle complex.
7581 2018-10-02 Jeff Law <law@redhat.com>
7583 * gimple-fold.c (get_range_strlen): Remove dead code.
7585 2018-10-02 Martin Sebor <msebor@redhat.com>
7586 Jeff Law <law@redhat.com>
7588 * builtins.c (unterminated_array): Add new arguments.
7589 If argument is not terminated, bubble up size and exact
7591 (expand_builtin_strnlen): Detect, avoid expanding
7592 and diagnose unterminated arrays.
7593 (c_strlen): Fill in offset of start of unterminated strings.
7594 * builtins.h (unterminated_array): Update prototype.
7596 2018-10-02 Richard Biener <rguenther@suse.de>
7598 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
7599 of haddv4df, first reduce to SSE width and exploit the fact
7600 that we only need element zero with the reduction result.
7601 (reduc_plus_scal_v2df): Likewise.
7603 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7605 * dojump.h (do_jump): Delete.
7606 (do_jump_1): Likewise.
7607 (split_comparison): Move around.
7608 * dojump.c (do_jump): Make static.
7609 (do_jump_1): Likewise.
7610 (jumpifnot): Move around.
7611 (jumpifnot_1): Likewise.
7613 (jumpif_1): Likewise.
7614 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
7616 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
7618 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
7619 insns in the delay slot and add_insn_after for the jump insn.
7621 2018-10-02 Richard Biener <rguenther@suse.de>
7623 * tree-inline.c (expand_call_inline): Use the location of
7624 the callee declaration for the inline-entry marker.
7625 * final.c (notice_source_line): Remove special-casing of
7626 NOTE_INSN_INLINE_ENTRY.
7628 2018-10-01 Carl Love <cel@us.ibm.com>
7631 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
7632 (__builtin_mtfsb0): New.
7633 (__builtin_mtfsb1): New.
7634 ( __builtin_set_fpscr_rn): New.
7635 (__builtin_set_fpscr_drn): New.
7636 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
7637 (rs6000_expand_set_fpscr_rn_builtin): Add.
7638 (rs6000_expand_set_fpscr_drn_builtin): Add.
7639 (rs6000_expand_builtin): Add case statement entries for
7640 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
7641 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
7642 RS6000_BUILTIN_MFFSL.
7643 (rs6000_init_builtins): Add ftype initialization and def_builtin
7644 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
7645 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
7646 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
7647 rs6000_mffscdrn): Add define_insn.
7648 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
7649 * doc/extend.texi: Add documentation for the builtins.
7651 2018-10-01 Richard Biener <rguenther@suse.de>
7653 PR tree-optimization/87465
7654 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
7655 causing branch miscounts.
7657 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7659 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
7660 aarch64_option_default_param): New.
7661 (params.h): Include.
7662 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
7663 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
7664 stack-clash protection validation code.
7666 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7668 * params.c (validate_param): New.
7669 (add_params): Use it.
7670 (set_param_value): Refactor param validation into validate_param.
7671 (diagnostic.h): Include.
7672 * diagnostic.h (diagnostic_ready_p): New.
7674 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7676 * params.c (set_param_value):
7677 Add index of parameter being validated.
7678 * common/common-target.def (option_validate_param): New.
7679 * common/common-targhooks.h (default_option_validate_param): New.
7680 * common/common-targhooks.c (default_option_validate_param): New.
7681 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
7682 * doc/tm.texi: Regenerate.
7684 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7687 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7688 Add validation for stack-clash parameters and set defaults.
7690 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7693 * configure.ac: Add stack-clash-protection-guard-size.
7694 * doc/install.texi: Document it.
7695 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
7696 * params.def: Update comment for guard-size.
7697 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
7698 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
7699 * configure: Regenerate.
7701 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7704 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
7705 STACK_DYNAMIC_OFFSET): New.
7706 * config/aarch64/aarch64.c (aarch64_layout_frame):
7707 Update outgoing args size.
7708 (aarch64_stack_clash_protection_alloca_probe_range,
7709 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7711 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7714 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
7716 * target.def (stack_clash_protection_alloca_probe_range): New.
7717 (stack_clash_protection_final_dynamic_probe): Remove.
7718 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
7719 (default_stack_clash_protection_final_dynamic_probe): Remove.
7720 * targhooks.c: Likewise.
7721 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
7722 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
7723 * doc/tm.texi: Regenerate.
7725 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7728 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
7729 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
7730 aarch64_clamp_to_uimm12_shift): New.
7731 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
7732 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
7734 2018-10-01 Tamar Christina <tamar.christina@arm.com>
7737 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
7739 2018-10-01 Jeff Law <law@redhat.com>
7740 Richard Sandiford <richard.sandiford@linaro.org>
7741 Tamar Christina <tamar.christina@arm.com>
7744 * config/aarch64/aarch64.md
7745 (probe_stack_range): Add k (SP) constraint.
7746 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
7747 STACK_CLASH_MAX_UNROLL_PAGES): New.
7748 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
7749 stack probes for stack clash.
7750 (aarch64_allocate_and_probe_stack_space): New.
7751 (aarch64_expand_prologue): Use it.
7752 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
7753 (aarch64_sub_sp): Add emit_move_imm optional param.
7755 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
7757 PR tree-optimization/87261
7758 * match.pd: Remove trailing whitespace.
7759 Add (x & y) | ~(x | y) -> ~(x ^ y),
7760 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
7762 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7764 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
7766 (ashlsi3_insn): Update instruction constraints.
7767 (ashrsi3_insn): Likewise.
7768 (rotrsi3): Likewise.
7769 (add_shift): Likewise.
7770 * config/arc/constraints.md (Csz): New 32 bit constraint. It
7771 avoids placing in the limm field small constants which, otherwise,
7772 could end into a small instruction.
7774 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
7776 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
7777 destination register is not odd-even.
7778 (umaddsidi4_split): Likewise.
7780 2018-10-01 Richard Biener <rguenther@suse.de>
7782 * tree-inline.c (expand_call_inline): Store origin of fn
7783 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
7784 * tree.c (block_ultimate_origin): Simplify and do some
7787 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
7789 * config/i386/mmx.md (EMMS): New int iterator.
7790 (emms): New int attribute.
7791 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
7792 EMMS int iterator. Explicitly declare clobbers.
7793 (mmx_emms): Remove expander.
7795 * config/i386/predicates.md (emms_operation): Remove predicate.
7796 (vzeroall_pattern): New predicate.
7797 (vzeroupper_pattern): Rename from vzeroupper_operation.
7798 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
7799 vzeroupper_pattern and vzeroall_pattern predicates.
7801 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
7803 PR rtl-optimization/86939
7804 * ira-lives.c (make_hard_regno_born): Rename from this...
7805 (make_hard_regno_live): ... to this. Remove update to conflict
7806 information. Update function comment.
7807 (make_hard_regno_dead): Add conflict information update. Update
7809 (make_object_born): Rename from this...
7810 (make_object_live): ... to this. Remove update to conflict information.
7811 Update function comment.
7812 (make_object_dead): Add conflict information update. Update function
7814 (mark_pseudo_regno_live): Call make_object_live.
7815 (mark_pseudo_regno_subword_live): Likewise.
7816 (mark_hard_reg_dead): Update function comment.
7817 (mark_hard_reg_live): Call make_hard_regno_live.
7818 (process_bb_node_lives): Likewise.
7819 * lra-lives.c (make_hard_regno_born): Rename from this...
7820 (make_hard_regno_live): ... to this. Remove update to conflict
7821 information. Remove now uneeded check_pic_pseudo_p argument.
7822 Update function comment.
7823 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
7824 to conflict information. Update function comment.
7825 (mark_pseudo_live): Remove update to conflict information. Update
7827 (mark_pseudo_dead): Add conflict information update.
7828 (mark_regno_live): Call make_hard_regno_live.
7829 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
7830 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
7832 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
7835 * config/i386/i386.c (construct_container): Use TImode for
7836 BLKmode values in 2 integer registers.
7838 2018-09-29 Jeff Law <law@redhat.com>
7840 * builtins.c (unterminated_array): Pass in c_strlen_data * to
7841 c_strlen rather than just a tree *.
7842 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
7843 Update recursive calls appropriately. If caller did not provide a
7844 suitable data pointer, create a local one. When a non-terminated
7845 string is discovered, bubble up information about the string via the
7846 c_strlen_data object.
7847 * builtins.h (c_strlen): Update prototype.
7848 (c_strlen_data): New structure.
7849 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
7850 For a type 2 call, if c_strlen indicates a non-terminated string
7851 use the length of the non-terminated string.
7852 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
7854 2018-09-29 Jakub Jelinek <jakub@redhat.com>
7857 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
7858 __m512d type for __A argument rather than __m512.
7860 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
7862 * match.pd (simple_comparison): Don't optimize if either operand is
7863 a function pointer when target needs function pointer canonicalization.
7865 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7867 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
7868 power5 .. power9 to remove indirection.
7869 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
7870 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
7871 ASM_CPU_476_SPEC): Delete.
7872 (ASM_CPU_SPEC): Adjust.
7873 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
7874 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
7876 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7878 * config.in: Delete HAVE_AS_DCI.
7879 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
7880 * config/rs6000/rs6000.h: Ditto.
7881 * configure.ac: Delete HAVE_AS_DCI.
7882 * configure: Regenerate.
7884 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7886 * config.in (HAVE_AS_LWSYNC): Delete.
7887 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7888 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
7890 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7891 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
7893 * configure.ac: Delete HAVE_AS_LWSYNC.
7894 * configure: Regenerate.
7896 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7897 Pierre-Marie de Rodat <derodat@adacore.com>
7899 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
7900 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
7901 (cgraph_node::create_thunk): Add indirect_offset parameter.
7902 (thunk_adjust): Likewise.
7903 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
7904 and initialize the corresponding field with it.
7905 (cgraph_node::dump): Dump indirect_offset field.
7906 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
7907 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
7908 (thunk_adjust): Add indirect_offset parameter and deal with it.
7909 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
7910 pass it to thunk_adjust. Do not call the target hook if it's non-zero
7911 or if the thunk is external or local. Fix formatting. Do not chain
7912 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
7913 if any, in the GIMPLE representation.
7914 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
7915 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
7916 (input_node): Read indirect_offset field.
7917 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
7918 call to thunk_adjust.
7919 * tree-nested.c (struct nesting_info): Add thunk_p field.
7920 (create_nesting_tree): Set it.
7921 (convert_all_function_calls): Copy static chain from targets to thunks.
7922 (finalize_nesting_tree_1): Return early for thunks.
7923 (unnest_nesting_tree_1): Do not finalize thunks.
7924 (gimplify_all_functions): Do not gimplify thunks.
7926 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7928 * opt-suggestions.c (option_proposer::build_option_suggestions):
7929 Release "option_values".
7931 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7933 * coverage.c (get_coverage_counts): Convert problem-reporting dump
7934 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
7935 * dumpfile.c (kind_as_string): New function.
7936 (dump_loc): Rather than a hardcoded prefix of "note: ", use
7937 kind_as_string to vary the prefix based on dump_kind.
7938 (selftest::test_capture_of_dump_calls): Update for above.
7940 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7942 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
7943 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
7945 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7947 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
7948 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
7949 INVALID_REGNUM instead of FPSR_REG.
7950 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
7951 * config/i386/i386.md: Update comment of FP compares.
7952 (fldenv): Do not clobber FPSR_REG.
7954 2018-09-28 Richard Biener <rguenther@suse.de>
7956 * tree.h (BLOCK_ORIGIN): New.
7957 * omp-expand.c (grid_expand_target_grid_body): Assign
7958 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
7959 * tree-inline.c (remap_block): Likewise.
7960 * auto-profile.c (get_function_decl_from_block): Simplify
7961 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
7962 * langhooks.c (lhd_print_error_function): Likewise.
7963 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
7965 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
7966 * tree.c (block_nonartificial_location): Likewise.
7967 (block_ultimate_origin): Likewise.
7968 * tree-pretty-print.c (percent_K_format): Likewise. Remove
7969 no longer needed LTO case.
7971 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
7972 Jan Hubicka <jh@suse.cz>
7973 Martin Jambor <mjambor@suse.cz>
7975 * simplify-rtx.c (simplify_merge_mask): New function.
7976 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
7977 same masks are used in op1 or op2.
7978 (test_vec_merge): New function.
7979 (test_vector_ops): Call test_vec_merge.
7981 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7983 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
7984 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
7985 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
7986 (cypress_costs): Set it.
7987 (supersparc_costs): Likewise.
7988 (hypersparc_costs): Likewise.
7989 (leon_cost): Likewise.
7990 (leon3_costs): Likewise.
7991 (sparclet_costs): Likewise.
7992 (ultrasparc_costs): Likewise.
7993 (ultrasparc_costs): Likewise.
7994 (niagara_costs): Likewise.
7995 (niagara2_costs): Likewise.
7996 (niagara3_costs): Likewise.
7997 (niagara4_costs): Likewise.
7998 (niagara7_costs): Likewise.
7999 (m8_costs): Likewise.
8000 (TARGET_CAN_FOLLOW_JUMP): Define.
8001 (pass_work_around_errata::gate): Minor tweak.
8002 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
8003 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
8004 Automaitcally clear MASK_FSMULD mask for V7 processors.
8005 (sparc_can_follow_jump): New static function.
8006 (output_ubranch): Deal with CROSSING_JUMP_P.
8007 (sparc_use_sched_lookahead): Rewrite using switch statement.
8008 (sparc_issue_rate): Reorder.
8009 (sparc_branch_cost): New function.
8011 2018-09-27 Martin Sebor <msebor@redhat.com>
8013 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
8014 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
8015 (int_fits_type_p): Same.
8017 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
8019 * config/i386/i386.md (FPCR_REG): Remove.
8020 (UNSPEC_FLDCW): Remove.
8021 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
8022 (x86_fldcw_1): Remove insn pattern.
8023 (fnstenv): Do not clobber FPCR_REG.
8025 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
8026 (CALL_USED_REGISTERS): Ditto.
8027 (REG_ALLOC_ORDER): Ditto.
8028 (REG_CLASS_CONTENTS): Ditto.
8029 (HI_REGISTER_NAMES): Ditto.
8030 (ADDITIONAL_REGISTER_NAMES): Use defines instead
8031 of numerical constants.
8032 * config/i386/i386.c (regclass_map): Remove fpsr register.
8033 (dbx_register_map): Ditto.
8034 (dbx64_register_map): Ditto.
8035 (svr4_dbx_register_map): Ditto.
8036 (print_reg): Do not handle FPCR_REG.
8038 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
8041 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
8042 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
8043 Delete, always treat as true.
8044 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
8045 Ditto. Simplify remaining code.
8046 * config/powerpcspe/powerpcspe.h: Ditto.
8047 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
8048 Simplify remaining code.
8049 (rs6000_expand_builtin): Ditto.
8050 * config/rs6000/rs6000.h: Ditto.
8051 * configure.ac: Ditto.
8052 * configure: Regenerate.
8054 2018-09-27 Martin Liska <mliska@suse.cz>
8056 * coverage.c (get_coverage_counts): Revert the formatting
8057 of missing profile opt info.
8059 2018-09-27 Richard Biener <rguenther@suse.de>
8063 * dwarf2out.c (set_block_origin_self): Do not mark outermost
8064 block as we do not output that.
8065 (gen_inlined_subroutine_die): Elide the originally outermost
8066 block, matching what we do for concrete instances.
8067 (decls_for_scope): Add parameter specifying whether to recurse
8070 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
8071 Tom de Vries <tom@codesourcery.com>
8075 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
8076 STORE_FLAG_VALUE == 1.
8078 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
8080 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
8081 constant definitions.
8082 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
8083 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
8084 ("speculation_barrier"): New expander definition.
8086 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
8088 PR gcov-profile/86957
8089 * common.opt: New warning option -Wmissing-profile.
8090 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
8091 * doc/invoke.texi: Document -Wmissing-profile.
8093 2018-09-26 Jim Wilson <jimw@sifive.com>
8095 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
8096 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
8098 2018-09-26 Martin Sebor <msebor@redhat.com>
8100 * tree.c (zerop): Change return type to bool.
8101 (integer_zerop, integer_onep, integer_each_onep): Same.
8102 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
8103 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
8104 (real_onep, real_minus_onep, chain_index): Same.
8105 (print_type_hash_statistics, type_list_equal): Same.
8106 * tree.h (zerop): Same.
8107 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
8108 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
8109 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
8110 (real_onep, real_minus_onep, chain_index): Same.
8111 (print_type_hash_statistics, type_list_equal): Same.
8113 2018-09-26 Jim Wilson <jimw@sifive.com>
8115 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
8117 2018-09-26 Jakub Jelinek <jakub@redhat.com>
8120 * config/i386/i386.c: Include debug.h and dwarf2out.h.
8121 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
8124 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
8126 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
8128 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
8130 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
8131 and FP_SECOND_SSE_REGS.
8132 (REG_CLASS_NAMES): Ditto.
8133 (REG_CLASS_CONTENTS): Ditto.
8134 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
8135 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
8136 (ix86_preferred_output_reload_class): Ditto.
8137 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
8138 clobber constraint to "=&f".
8139 (fix_truncdi_i387): Ditto.
8140 (lrintxfdi2): Ditto.
8141 (fistdi2_<rounding>): Ditto.
8142 (fpremxf4_i387): Change "=u" constraint to "=f".
8143 (fprem1xf4_i387): Ditto.
8145 (fptanxf4_i387): Ditto.
8146 (fxtractxf3_i387): Ditto.
8147 (fscalexf4_i387): Ditto.
8148 (atan2xf3): Change "u" constraint to "f".
8149 (fyl2xxf3_i387): Ditto.
8150 (fyl2xp1xf3_i387): Ditto.
8152 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
8155 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
8156 for removed I387_MASK_PM entity.
8159 2018-09-26 Jeff Law <law@redhat.com>
8161 2018-09-26 Alexey Neyman <stilor@att.net>
8163 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
8164 headers are no longer pulled in by <isl/val.h>.
8166 2018-09-26 Richard Biener <rguenther@suse.de>
8169 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
8170 or concrete instance DIE to the tree. Create abstract origin
8171 attributes also for concrete instances.
8173 2018-09-26 Alexey Neyman <stilor@att.net>
8175 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
8176 headers are no longer pulled in by <isl/val.h>.
8178 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
8180 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
8181 Use new helper functions.
8182 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
8183 Use new helper functions.
8184 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
8185 aarch_mm_needs_release): New declarations.
8186 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
8187 aarch_mm_needs_release): New.
8189 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
8191 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
8192 (arm32_output_mi_thunk): Deal with long calls.
8194 2018-09-26 Richard Biener <rguenther@suse.de>
8198 * tree-inline.c (expand_call_inline): When the location
8199 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
8200 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
8201 the inserted BLOCK to make inlined_function_outer_scope_p
8203 * dwarf2out.c (add_call_src_coords_attributes): Do not add
8204 coords for reserved locations.
8206 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8208 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
8209 (*call_indirect_nonlocal_sysv<mode>): Ditto.
8210 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
8211 (*sibcall_nonlocal_sysv<mode>): Ditto.
8212 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
8213 (<bd>_<mode>): Ditto.
8214 (<bd>tf_<mode>): Ditto.
8216 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8218 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
8219 control string as a list of templates instead of as C code.
8220 (*altivec_movti): Ditto.
8221 * config/rs6000/darwin.md (movdf_low_di): Ditto.
8223 2018-09-25 Jim Wilson <jimw@sifive.com>
8225 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
8226 when target symbol is weak.
8228 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
8231 * builtins.c (unterminated_array): Simplify.
8232 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
8233 where pointer arithmetic is safe.
8235 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
8238 * config/rs6000/altivec.md (altivec_vspltb): Use
8239 const_0_to_15_operand instead of u5bit_cint_operand.
8240 (*altivec_vspltb_internal): Ditto.
8241 (altivec_vspltb_direct): Ditto.
8242 (altivec_vsplth): Use const_0_to_7_operand instead of
8244 (*altivec_vsplth_internal): Ditto.
8245 (altivec_vsplth_direct): Ditto.
8246 (altivec_vspltw): Use const_0_to_3_operand instead of
8248 (*altivec_vspltw_internal): Ditto.
8249 (altivec_vspltw_direct): Ditto.
8250 (altivec_vspltsf): Ditto.
8251 (*altivec_vspltsf_internal): Ditto.
8252 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
8253 various splats with the proper size immediate. Reorder the various
8254 cases by ascending size of immediate, and put all such together.
8256 2018-09-25 Richard Biener <rguenther@suse.de>
8259 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
8261 (maybe_create_die_with_external_ref): Do not create
8262 DW_TAG_imported_unit here.
8263 (add_abstract_origin_attribute): Handle external BLOCK refs.
8264 (dwarf2out_abstract_function): Simplify LTO case.
8265 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
8266 rather than using maybe_create_die_with_external_ref.
8268 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
8271 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
8272 (frndintxf2_mask_pm_i387): Ditto.
8273 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
8274 Enable for !flag_trapping_math.
8275 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
8276 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
8277 Change operand 1 predicate to nonimmediate_operand.
8278 (attr "i387_cw"): Remove mask_pm.
8279 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
8280 (enum ix86_entity): Remove I387_MASK_PM.
8281 * config/i386/i386.c (ix86_i387_mode_needed): Do not
8282 handle I387_MASK_PM.
8283 (ix86_mode_needed): Ditto.
8284 (ix86_mode_after): Ditto.
8285 (ix86_mode_entry): Ditto.
8286 (ix86_mode_exit): Ditto.
8287 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
8289 2018-09-25 Jakub Jelinek <jakub@redhat.com>
8291 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
8292 to_update_switch_stmts to vNULL instead of calling create on them
8295 2018-09-25 Richard Biener <rguenther@suse.de>
8297 PR tree-optimization/87402
8298 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
8299 (visit_phi): Re-instantiate handling of supposed to be VARYING
8300 but non-VARYING backedge value.
8302 2018-09-25 Richard Biener <rguenther@suse.de>
8305 * dwarf2out.c (struct sym_off_pair): New.
8306 (external_die_map): New global.
8307 (lookup_decl_die): When in LTO create DIEs lazily from the
8309 (lookup_block_die): New function, create DIEs lazily in LTO.
8310 (equate_block_to_die): New function.
8311 (dwarf2out_die_ref_for_decl): During WPA get the association
8312 from the external DIE map.
8313 (dwarf2out_register_external_die): Record mapping into the
8315 (maybe_create_die_with_external_ref): New function split out from
8316 DIE generation part of old dwarf2out_register_external_die.
8317 (add_abstract_origin_attribute): Do not return the DIE. When
8318 in LTO reference externals directly.
8319 (dwarf2out_abstract_function): When in LTO ignore calls for
8320 decls with external DIEs (already present abstract instances).
8321 (gen_call_site_die): Adjust.
8322 (add_high_low_attributes): Likewise.
8323 (gen_lexical_block_die): Likewise.
8324 (gen_inlined_subroutine_die): Likewie.
8325 (gen_block_die): Likewise.
8326 (dwarf2out_inline_entry): Likewise.
8327 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
8330 2018-09-25 Martin Liska <mliska@suse.cz>
8332 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
8333 integers and not by a float value.
8335 2018-09-25 Martin Liska <mliska@suse.cz>
8338 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
8340 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
8341 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
8342 * godump.c (go_format_type): Remove extra parenthesis.
8344 2018-09-25 Martin Liska <mliska@suse.cz>
8346 * alias.c (set_dest_equal_p): Remove unused function.
8347 * config/i386/i386.c (def_builtin_pure2): Likewise.
8348 * diagnostic-show-locus.c (class layout): Remove
8350 (layout::layout): Likewise here.
8351 * dump-context.h (class temp_dump_context): Likewise.
8352 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
8353 (add_AT_loclistsptr): Likewise.
8354 (add_AT_offset): Likewise.
8355 (get_AT_hi_pc): Likewise.
8356 (is_comdat_die): Likewise.
8357 (type_is_enum): Likewise.
8358 (ceiling): Likewise.
8359 (add_AT_vms_delta): Likewise.
8360 (is_class_die): Likewise.
8361 * edit-context.c (class line_event): Remove unused field.
8362 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
8364 * ipa-cp.c (ipa_get_vr_lat): Likewise.
8365 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
8366 (ok_for_base_p_nonstrict): Likewise.
8367 * tree-chrec.c (is_not_constant_evolution): Likewise.
8368 (chrec_fold_poly_cst): Likewise.
8369 * tree-if-conv.c (has_pred_critical_p): Likewise.
8370 * tree-ssa-coalesce.c (print_exprs): Likewise.
8371 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
8372 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
8373 * tree-vrp.c (value_ranges_intersect_p): Likewise.
8374 (value_range_nonnegative_p): Likewise.
8376 2018-09-25 Martin Liska <mliska@suse.cz>
8378 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
8379 Do not handle "GNU Pascal".
8380 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
8382 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
8383 from documentation. Likewise.
8384 * dbxout.c (dbxout_range_type): Likewise.
8385 * doc/cpp.texi: Likewise.
8386 * doc/extend.texi: Likewise.
8387 * doc/frontends.texi: Likewise.
8388 * doc/invoke.texi: Remove Pascal entry.
8389 * tree.def (CLEANUP_POINT_EXPR): Likewise.
8390 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
8392 2018-09-25 Martin Liska <mliska@suse.cz>
8395 * doc/invoke.texi: Document all parameters and remove default
8398 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
8401 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
8402 contains HOST_WIDE_INTs when computing its size.
8404 2018-09-24 Jim Wilson <jimw@sifive.com>
8407 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
8409 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
8411 2018-09-24 Andrew Pinski <apinski@marvell.com>
8413 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
8414 access prev before checking it for NULLness in the
8415 AARCH64_FUSE_CMP_BRANCH case.
8417 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
8420 * config/i386/i386.c (rest_of_insert_endbranch): Set
8421 endbr_queued_at_entrance to true and don't insert ENDBR if
8422 x86_function_profiler will be called.
8423 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
8425 * config/i386/i386.h (machine_function): Add
8426 endbr_queued_at_entrance.
8428 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
8430 * genattrtab.c (mk_attr_alt): Use alternative_mask.
8431 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
8433 (check_attr_test): Use alternative_mask.
8434 (get_attr_value): Likewise.
8435 (compute_alternative_mask): Use alternative_mask and XWINT.
8436 (make_alternative_compare): Use alternative_mask.
8437 (attr_alt_subset_p): Use XWINT.
8438 (attr_alt_subset_of_compl_p): Likewise.
8439 (attr_alt_intersection): Use alternative_mask and XWINT.
8440 (attr_alt_union): Likewise.
8441 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
8442 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
8443 (simplify_test_exp): Use alternative_mask and XWINT.
8444 (write_test_expr): Use alternative_mask and XWINT, adjust bit
8445 number calculation to support 64 bits. Generate code that
8446 checks 64-bit masks.
8447 (main): Use alternative_mask.
8448 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
8450 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
8453 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
8454 RETURN+USE when returning via %r14.
8456 2018-09-24 Martin Liska <mliska@suse.cz>
8458 * gcov.c (output_lines): Print colorization legend
8459 for both flag_use_colors and flag_use_hotness_colors.
8462 2018-09-24 Martin Liska <mliska@suse.cz>
8464 * coverage.c (get_coverage_counts): Use warning_at
8465 with current_function_decl location. Use %qD in warning
8468 2018-09-24 Martin Liska <mliska@suse.cz>
8470 * memory-block.h (memory_block_pool::release): Annotate with
8471 valgrind that the memory is not accessible.
8473 2018-09-24 Martin Liska <mliska@suse.cz>
8476 * asan.c: Make asan_handled_variables extern.
8478 * cfgexpand.c (expand_stack_vars): Make sure
8479 a representative is unpoison if another
8480 variable in the partition is handled by
8481 use-after-scope sanitization.
8483 2018-09-24 Richard Biener <rguenther@suse.de>
8485 PR tree-optimization/63155
8486 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
8487 the worklist when the edge of the respective argument isn't
8490 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8492 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
8493 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
8494 (MASK_CLASS_P): Update for rename.
8495 (MAYBE_MASK_CLASS_P): Ditto.
8496 (REG_CLASS_NAMES): Update.
8497 (REG_CLASS_CONTENT): Update.
8498 * config/i386/i386.c (regclass_map): Update for MASK_REG
8499 and ALL_MASK_REGS rename.
8500 * config/i386/constraints.md (Yk): Update for rename.
8503 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8505 * config/i386/i386.h (enum reg_class): Remove
8506 EVEX_SSE_REGS and MOD4_SSE_REGS.
8507 (REG_CLASS_NAMES): Update.
8508 (REG_CLASS_CONTENT): Update.
8509 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
8510 registers as ALL_SSE_REGS.
8511 (ix86_additional_allocno_class_p): Remove.
8512 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
8513 (ix86_register_priority): Lower priority of EVEX SSE registers.
8514 Use IN_RANGE macro where appropriate.
8515 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
8516 AVX-5124VNNIW checks.
8517 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
8518 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
8519 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
8520 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
8521 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
8522 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
8523 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
8524 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
8525 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
8526 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
8527 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
8528 * config/i386/constraints.md (Yh): Remove.
8530 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
8532 * config/i386/i386.c (regclass_map): Declare integer REX registers
8535 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
8537 * doc/service.texi (Service): Switch the fsf.org link to https.
8539 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
8542 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8543 Define to speculation_safe_value_not_needed.
8545 2018-09-21 Florian Weimer <fweimer@redhat.com>
8548 * doc/extend.texi (Common Function Attributes): Mention that
8549 noreturn suppresses tail call optimization.
8551 2018-09-21 Jeff Law <law@redhat.com>
8553 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
8554 vr_values::cleanup_edges_and_switches.
8555 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
8557 (identify_jump_threads): Remove EDGE_IGNORE handling.
8558 (execute_vrp): Move handling of to_remove_edges and
8559 to_update_switch_stmts into vr_values class member functions.
8560 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
8561 (to_update_switch_stmts): Likewise.
8562 * vr-values.c: Include cfghooks.h.
8563 (vr_values::vr_values): Initialize to_remove_edges and
8564 to_update_switch_stmts.
8565 (vr_values::~vr_values): Verify to_remove_edges and
8566 to_update_switch_stmts are empty.
8567 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
8568 (vr_values::cleanup_edges_and_switches): New member function.
8569 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
8570 function. Add new data members.
8572 2018-09-21 David Malcolm <dmalcolm@redhat.com>
8574 PR tree-optimization/87309
8575 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
8576 calls with pflags and alt_flags.
8577 (selftest::test_capture_of_dump_calls): Add test of interaction of
8578 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
8580 2018-09-21 Olivier Hainque <hainque@adacore.com>
8582 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
8584 2018-09-21 Olivier Hainque <hainque@adacore.com>
8586 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
8588 2018-09-21 Olivier Hainque <hainque@adacore.com>
8590 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
8591 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
8592 (VXWORKS_LIBS_RTP): Minor reordering.
8594 2018-09-21 Olivier Hainque <hainque@adacore.com>
8596 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
8597 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
8599 2018-09-21 Olivier Hainque <hainque@adacore.com>
8601 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
8602 (PTRDIFF_TYPE): Likewise.
8604 2018-09-21 Olivier Hainque <hainque@adacore.com>
8606 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
8607 triplet, similar to support for VxWorks7.
8608 * config/vxworks-dummy.h: Provide a default definition
8609 of TARGET_VXWORKS64 to 0.
8611 2018-09-21 Olivier Hainque <hainque@adacore.com>
8613 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
8614 * config/vxworks-dummy.h: here.
8616 2018-09-21 Olivier Hainque <hainque@adacore.com>
8618 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
8620 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
8621 Bo Zhou <zbo.zhou@hisilicon.com>
8623 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
8624 * config/aarch64/aarch64-tune.md: Regenerated.
8625 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
8626 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
8627 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
8629 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
8630 Julian Brown <julian@codesourcery.com>
8632 * builtins.c (get_builtin_sync_mem): Handle address spaces.
8634 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8636 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
8637 if the call takes a static chain.
8639 2018-09-21 Martin Liska <mliska@suse.cz>
8641 * auto-profile.c (autofdo_source_profile::read): Do not
8643 (read_profile): Do not add working sets.
8644 (read_autofdo_file): Remove sum_all.
8645 (afdo_callsite_hot_enough_for_early_inline): Remove const
8647 * coverage.c (struct counts_entry): Remove gcov_summary.
8648 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
8649 do not support GCOV_TAG_PROGRAM_SUMMARY.
8650 (get_coverage_counts): Remove summary and expected
8652 * coverage.h (get_coverage_counts): Likewise.
8653 * doc/gcov-dump.texi: Remove -w option.
8654 * gcov-dump.c (dump_working_sets): Remove.
8655 (main): Do not support '-w' option.
8656 (print_usage): Likewise.
8657 (tag_summary): Likewise.
8658 * gcov-io.c (gcov_write_summary): Do not dump
8660 (gcov_read_summary): Likewise.
8661 (gcov_histo_index): Remove.
8662 (gcov_histogram_merge): Likewise.
8663 (compute_working_sets): Likewise.
8664 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
8666 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
8667 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
8668 (GCOV_HISTOGRAM_SIZE): Remove.
8669 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
8670 (struct gcov_summary): Simplify rapidly just
8671 to runs and sum_max fields.
8672 (gcov_histo_index): Remove.
8673 (NUM_GCOV_WORKING_SETS): Likewise.
8674 (compute_working_sets): Likewise.
8675 * gcov-tool.c (print_overlap_usage_message): Remove
8676 trailing empty line.
8677 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
8678 (output_lines): Remove program related line.
8679 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
8680 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
8682 (input_profile_summary): Do not read it.
8683 (merge_profile_summaries): And do not merge it.
8684 (input_symtab): Do not call removed function.
8685 * modulo-sched.c (sms_schedule): Do not print sum_max.
8686 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
8687 removed when histogram method was invented.
8688 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
8690 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
8692 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
8693 and dump selected value.
8694 * profile.c (add_working_set): Remove.
8695 (get_working_sets): Likewise.
8696 (find_working_set): Likewise.
8697 (get_exec_counts): Do not work with working sets.
8698 (read_profile_edge_counts): Do not inform as sum_max is removed.
8699 (compute_branch_probabilities): Likewise.
8700 (compute_value_histograms): Remove argument for call of
8701 get_coverage_counts.
8702 * profile.h: Do not make gcov_summary const.
8704 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
8706 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
8708 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
8710 PR tree-optimization/86990
8711 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
8712 Check that the entire merged store group is made of constants only for
8715 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
8717 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
8718 (VTABLE_VERIFICATION_SPEC): Likewise.
8719 (SANITIZER_EARLY_SPEC): Likewise.
8720 (SANITIZER_SPEC): Likewise.
8721 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
8722 * doc/invoke.texi (Link Options): Document -r.
8724 2018-09-20 Richard Biener <rguenther@suse.de>
8727 * gimplify.c (gimplify_expr): Retain alignment of
8728 addressable lvalue in dereference.
8730 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
8733 * configure.ac: Check for .loc is_stmt support.
8734 * configure, config.in: Rebuilt.
8735 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
8738 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8740 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
8741 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
8744 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
8746 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
8748 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
8751 2018-09-20 Marek Polacek <polacek@redhat.com>
8753 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
8755 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8757 PR tree-optimization/87288
8758 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
8759 into account when determining PEELING_FOR_NITERS.
8761 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
8763 PR tree-optimization/86877
8764 * tree-vect-loop.c (vect_analyze_loop_2): Call
8765 vect_verify_datarefs_alignment.
8767 2018-09-19 Marek Polacek <polacek@redhat.com>
8769 * doc/invoke.texi: Document -Wclass-conversion.
8771 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
8773 * config/pa/pa.c (pa_adjust_priority): Delete.
8774 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
8776 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
8777 (atomic_storehi): Likewise.
8778 (atomic_storesi): Likewise.
8779 (atomic_loaddi): Restore compare and swap exchange loop code.
8781 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8783 PR rtl-optimization/86902
8784 * combine.c (try_combine): When changing the CC mode used, don't change
8785 an unrelated mode in other_insn to that new CC mode.
8787 2018-09-19 David Malcolm <dmalcolm@redhat.com>
8789 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
8790 with %T in place of calls to dump_generic_expr.
8791 (prune_runtime_alias_test_list): Likewise.
8792 (create_runtime_alias_checks): Likewise.
8793 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
8794 (vect_analyze_data_ref_dependence): Likewise.
8795 (vect_slp_analyze_data_ref_dependence): Likewise.
8796 (vect_record_base_alignment): Likewise. Use %G in place of call
8797 to dump_gimple_stmt.
8798 (vect_compute_data_ref_alignment): Likewise.
8799 (verify_data_ref_alignment): Likewise.
8800 (vect_find_same_alignment_drs): Likewise.
8801 (vect_analyze_group_access_1): Likewise.
8802 (vect_analyze_data_ref_accesses): Likewise.
8803 (dependence_distance_ge_vf): Likewise.
8804 (dump_lower_bound): Likewise.
8805 (vect_prune_runtime_alias_test_list): Likewise.
8806 (vect_find_stmt_data_reference): Likewise.
8807 (vect_analyze_data_refs): Likewise.
8808 (vect_create_addr_base_for_vector_ref): Likewise.
8809 (vect_create_data_ref_ptr): Likewise.
8810 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
8811 (vect_can_advance_ivs_p): Likewise.
8812 (vect_update_ivs_after_vectorizer): Likewise.
8813 (vect_gen_prolog_loop_niters): Likewise.
8814 (vect_prepare_for_masked_peels): Likewise.
8815 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8816 (vect_determine_vectorization_factor): Likewise.
8817 (vect_is_simple_iv_evolution): Likewise.
8818 (vect_analyze_scalar_cycles_1): Likewise.
8819 (vect_analyze_loop_operations): Likewise.
8820 (report_vect_op): Likewise.
8821 (vect_is_slp_reduction): Likewise.
8822 (check_reduction_path): Likewise.
8823 (vect_is_simple_reduction): Likewise.
8824 (vect_create_epilog_for_reduction): Likewise.
8825 (vect_finalize_reduction:): Likewise.
8826 (vectorizable_induction): Likewise.
8827 (vect_transform_loop_stmt): Likewise.
8828 (vect_transform_loop): Likewise.
8829 (optimize_mask_stores): Likewise.
8830 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
8831 (vect_split_statement): Likewise.
8832 (vect_recog_over_widening_pattern): Likewise.
8833 (vect_recog_average_pattern): Likewise.
8834 (vect_determine_min_output_precision_1): Likewise.
8835 (vect_determine_precisions_from_range): Likewise.
8836 (vect_determine_precisions_from_users): Likewise.
8837 (vect_mark_pattern_stmts): Likewise.
8838 (vect_pattern_recog_1): Likewise.
8839 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8840 (vect_record_max_nunits): Likewise.
8841 (vect_build_slp_tree_1): Likewise.
8842 (vect_build_slp_tree_2): Likewise.
8843 (vect_print_slp_tree): Likewise.
8844 (vect_analyze_slp_instance): Likewise.
8845 (vect_detect_hybrid_slp_stmts): Likewise.
8846 (vect_detect_hybrid_slp_1): Likewise.
8847 (vect_slp_analyze_operations): Likewise.
8848 (vect_slp_analyze_bb_1): Likewise.
8849 (vect_transform_slp_perm_load): Likewise.
8850 (vect_schedule_slp_instance): Likewise.
8851 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8852 (vect_mark_stmts_to_be_vectorized): Likewise.
8853 (vect_init_vector_1): Likewise.
8854 (vect_get_vec_def_for_operand): Likewise.
8855 (vect_finish_stmt_generation_1): Likewise.
8856 (vect_check_load_store_mask): Likewise.
8857 (vectorizable_call): Likewise.
8858 (vectorizable_conversion): Likewise.
8859 (vectorizable_operation): Likewise.
8860 (vectorizable_load): Likewise.
8861 (vect_analyze_stmt): Likewise.
8862 (vect_is_simple_use): Likewise.
8863 (vect_get_vector_types_for_stmt): Likewise.
8864 (vect_get_mask_type_for_stmt): Likewise.
8865 * tree-vectorizer.c (increase_alignment): Likewise.
8867 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
8869 * doc/rtl.texi: Adjust vec_select description.
8870 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
8871 non-constant selectors.
8873 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
8875 * config/aarch64/aarch64-protos.h
8876 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
8877 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
8878 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
8879 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
8880 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
8881 (AARCH64_FL_PROFILE): Move index so flags are ordered.
8882 (AARCH64_ISA_RCPC8_4): New flag.
8883 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
8884 to aarch64_offset_9bit_signed_unscaled_p.
8885 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
8887 * config/aarch64/constraints.md (Ust): New constraint.
8888 * config/aarch64/predicates.md.
8889 (aarch64_9bit_offset_memory_operand): New predicate.
8890 (aarch64_rcpc_memory_operand): New predicate.
8892 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
8894 PR rtl-optimization/87361
8895 * rtlanal.c (nonzero_bits1): Revert accidental change.
8897 2018-09-19 Richard Biener <rguenther@suse.de>
8899 PR tree-optimization/87349
8900 PR tree-optimization/87342
8901 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
8903 2018-09-18 Marek Polacek <polacek@redhat.com>
8905 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
8906 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
8908 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8910 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
8912 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8914 PR rtl-optimization/86882
8915 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
8917 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
8919 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
8920 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
8922 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
8925 * doc/invoke.texi (Link Options): Fix formatting and grammar.
8927 2018-09-18 Richard Biener <rguenther@suse.de>
8930 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
8931 (compute_samebase_partition_bases): Likewise.
8932 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
8933 (gimple_can_coalesce_p): Simplify.
8935 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
8937 Handle a library implementation of ffs calling __builtin_ffs.
8938 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
8939 (mmix_init_libfuncs): New function: make __builtin_ffs expand
8942 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8944 * diagnostic-show-locus.c (class layout_range): Add field
8946 (layout_range::layout_range): Add "original_idx" param and use it
8947 to initialize new field.
8948 (make_range): Use 0 for original_idx.
8949 (layout::layout): Pass in index to calls to
8950 maybe_add_location_range.
8951 (layout::maybe_add_location_range): Add param "original_idx" and
8952 pass it on to layout_range.
8953 (layout::print_any_labels): Pass on range->m_original_idx to
8955 (gcc_rich_location::add_location_if_nearby): Use 0 for
8957 * gcc-rich-location.h (text_range_label::get_text): Update for new
8959 (range_label_for_type_mismatch::get_text): Likewise.
8961 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
8963 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
8965 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8967 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
8968 format_string_diagnostic_t.
8969 (fmtwarn_n): Likewise.
8970 * substring-locations.c
8971 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
8972 (format_warning_n_va): Convert to...
8973 (format_string_diagnostic_t::emit_warning_n_va): ...this.
8974 (format_warning_va): Convert to...
8975 (format_string_diagnostic_t::emit_warning_va): ...this.
8976 (format_warning_at_substring): Convert to...
8977 (format_string_diagnostic_t::emit_warning): ...this.
8978 (format_warning_at_substring_n): Convert to...
8979 (format_string_diagnostic_t::emit_warning_n): ...this.
8980 * substring-locations.h (class format_string_diagnostic_t): New
8982 (format_warning_va): Convert to
8983 format_string_diagnostic_t::emit_warning_va.
8984 (format_warning_n_va): Convert to
8985 format_string_diagnostic_t::emit_warning_n_va.
8986 (format_warning_at_substring): Convert to
8987 format_string_diagnostic_t::emit_warning.
8988 (format_warning_at_substring_n): Convert to
8989 format_string_diagnostic_t::emit_warning_n.
8991 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
8992 Bernd Schmidt <bernds_cb1@t-online.de>
8994 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
8997 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
8999 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
9000 operand 0 predicate to nonimmediate operand.
9001 (rint<mode>2_frndint): Remove insn pattern.
9002 (rint<mode>2): Change operand 1 predicate to general_operand.
9003 Extend operand 1 to XFmode and generate rintxf2 insn.
9004 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
9005 Do not use X87MODEF mode macro.
9006 (frndintxf2_<rounding>_i387): Rename from
9007 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
9008 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
9009 to XFmode and generate significandxf3 insn.
9011 2018-09-17 Richard Biener <rguenther@suse.de>
9013 PR tree-optimization/87328
9014 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
9015 visiting unexecutable backedges when not iterating.
9016 (do_rpo_vn): Mark all edges not executable even when not
9019 2018-09-17 Martin Jambor <mjambor@suse.cz>
9022 * doc/invoke.texi (Warning Options): Likewise.
9024 2018-09-17 Richard Biener <rguenther@suse.de>
9026 PR tree-optimization/87301
9027 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
9028 clean EH info from leftover copy assignments.
9030 2018-09-17 Martin Liska <mliska@suse.cz>
9032 PR gcov-profile/85871
9033 * gcov.c (output_intermediate_file): Fix out of bounds
9036 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
9038 * config/arc/arc.c: Object attributes for core4 not reflected
9040 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
9043 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
9045 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
9047 2018-09-17 Martin Liska <mliska@suse.cz>
9049 * doc/gcov.texi: Document new option --use-hotness-colors.
9050 * gcov.c (struct source_info): Declare new field.
9051 (source_info::source_info): Set default for maximum_count.
9052 (print_usage): Add new -q option.
9053 (process_args): Process it.
9054 (accumulate_line_info): Save src->maximum_count.
9055 (output_line_beginning): Make color line number if
9056 flag_use_hotness_colors is set.
9057 (output_line_details): Pass default argument value.
9058 (output_lines): Pass src->maximum_count.
9060 2018-09-17 Martin Liska <mliska@suse.cz>
9062 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
9063 Use processor_names table.
9064 * config/i386/i386.c (ix86_default_align): Use
9065 processor_cost_table for alignment values.
9066 (ix86_option_override_internal): Use processor_names.
9067 (ix86_function_specific_print): Likewise.
9068 * config/i386/i386.h (struct processor_costs):
9069 Add alignment values.
9070 (struct ptt): Remove and replace with const char *.
9071 * config/i386/x86-tune-costs.h (struct processor_costs):
9072 Declare default alignments for all costs.
9074 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
9076 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
9077 symbolics or VR_VARYING ranges for ABS_EXPR.
9078 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
9079 when range will wrap.
9081 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
9084 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
9085 before and after a JUMP_TABLE_DATA.
9087 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
9090 * dojump.c (do_compare_and_jump): Canonicalize function pointers
9091 when one operand is a function pointer. Use POINTER_TYPE_P and
9092 FUNC_OR_METHOD_TYPE_P.
9093 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
9094 * fold-const.c (build_range_check): Likewise.
9095 * match.pd (simple_comparison): Likewise.
9097 2018-09-14 David Malcolm <dmalcolm@redhat.com>
9100 * spellcheck.c (get_edit_distance_cutoff): New function.
9101 (selftest::test_edit_distance_unit_test_oneway): Rename to...
9102 (selftest::test_get_edit_distance_one_way): ...this.
9103 (selftest::test_get_edit_distance_unit): Rename to...
9104 (selftest::test_get_edit_distance_both_ways): ...this.
9105 (selftest::test_edit_distances): Move tests to this new function,
9106 and test some more pairs of strings. Update for above renaming.
9107 (selftest::get_old_cutoff): New function.
9108 (selftest::test_get_edit_distance_cutoff): New function.
9109 (selftest::assert_suggested_for): New function.
9110 (ASSERT_SUGGESTED_FOR): New macro.
9111 (selftest::assert_not_suggested_for): New function.
9112 (ASSERT_NOT_SUGGESTED_FOR): New macro.
9113 (selftest::test_suggestions): New function.
9114 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
9115 tests to selftest::test_edit_distances and call it. Add calls to
9116 selftest::test_get_edit_distance_cutoff and
9117 selftest::test_suggestions.
9118 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
9119 (best_match::consider): Replace hard-coded cutoff calculation with
9121 (best_match::get_cutoff): New declaration.
9122 (best_match::get_best_meaningful_candidate): Likewise.
9124 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
9126 * builtins.c (fold_builtin_strlen): Remove TODO comment.
9128 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
9131 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9133 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9134 terminated string literal.
9136 2018-09-14 Martin Sebor <msebor@redhat.com>
9138 * builtins.c (unterminated_array): Handle ARRAY_REF.
9139 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
9140 * builtins.h (unterminated_array): Declare extern.
9141 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
9143 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
9146 2018-09-14 Martin Sebor <msebor@redhat.com>
9147 Jeff Law <law@redhat.com>
9149 * builtins.c (unterminated_array): New.
9150 (expand_builtin_strcpy): Adjust.
9151 (expand_builtin_strcpy_args): Detect unterminated arrays.
9152 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
9153 unterminated arrays.
9154 * gimple-fold.h (get_maxval_strlen): Add argument.
9155 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
9157 * gimple-fold.c (get_range_strlen): Add argument.
9158 (get_maxval_strlen): Adjust.
9159 * gimple-fold.h (get_range_strlen): Add argument.
9161 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
9163 * config/i386/movdirintrin.h: Fix copyright year.
9165 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
9167 * reg-stack.c: Include regs.h.
9168 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
9169 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
9170 FIRST_STACK_REG, not DFmode.
9171 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
9172 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
9173 (change stack): Default register mode to the reg_raw_mode of
9174 FIRST_STACK_REG, not DFmode.
9175 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
9176 (*swapxf): Rename from swapxf.
9178 2018-09-14 Carl Love <cel@us.ibm.com>
9180 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
9181 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
9183 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
9186 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
9189 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
9192 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
9194 2018-09-14 Jason Merrill <jason@redhat.com>
9196 Fix --enable-gather-detailed-mem-stats.
9197 * hash-table.c (hash_table_usage): Change from variable to function.
9198 * hash-table.h: Adjust.
9199 * Makefile.in: Add missing dependencies on hash-table.h.
9201 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9203 PR tree-optimization/87259
9205 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
9206 execute_cse_reciprocals_1 has tried transforming.
9208 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
9210 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
9211 VR_VARYING for PLUS/MINUS_EXPR.
9213 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
9215 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
9218 2018-09-14 Richard Biener <rguenther@suse.de>
9221 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
9222 bits for the merged partition.
9224 2018-09-13 Martin Sebor <msebor@redhat.com>
9225 Bernd Edlinger <bernd.edlinger@hotmail.de>
9227 * builtins.h (c_srlen): Add argument.
9228 * builtins.c (warn_string_no_nul): New function.
9229 (c_strlen): Add argument and use it. Update recursive calls.
9230 Pass DECL argument to string_constant to get info on non
9231 terminated strings. Update *NONSTR as needed.
9232 (fold_builtin_strlen): Add argument to calls to c_strlen.
9233 Warn for unterminated arrays.
9234 (warn_string_no_null): Add prototype.
9235 * expr.c (string_constant): Update arguments. Update recursive
9236 calls appropriately. Detect missing NUL terminator and outermost
9237 declaration its missing in.
9238 Improve checks for arrays with nonzero lower bound or elements
9239 that are not a single byte. Simplify offset computation.
9240 Simplify checks for non-NUL terminated strings.
9241 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
9242 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
9244 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
9246 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
9248 * fold-const.c (c_getstr): Fix function comment. Remove unused third
9249 argument. Fix range checks.
9250 * fold-const.h (c_getstr): Adjust protoype.
9251 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
9252 string is constant but contains no NUL byte.
9254 * expr.c (string_constant): Adjust function comment.
9255 Remove bogus check for zero termination.
9257 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
9259 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
9260 (get_constant_size): Don't make STRING_CSTs larger than they are.
9261 (check_string_literal): New check function for STRING_CSTs.
9262 (output_constant): Use it.
9264 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
9267 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
9269 2018-09-13 Richard Biener <rguenther@suse.de>
9271 PR tree-optimization/87263
9272 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
9273 (struct unwind_state): Add max_rpo field.
9274 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
9275 Compute max_rpo, the max RPO number a block can be backwards reached
9276 from. Re-write non-iterating mode to a RPO ordered worklist approach,
9277 separating it from the iterating mode.
9279 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
9281 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
9282 (rfs_decision): New scheduling decision.
9284 2018-09-13 Richard Biener <rguenther@suse.de>
9287 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
9288 (vn_nary_op_insert_pieces_predicated): Do not write useless
9289 valid_dominated_by_p entry outside of the allocated storage.
9291 2018-09-13 Omar Sandoval <osandov@osandov.com>
9292 Tom de Vries <tdevries@suse.de>
9295 * dwarf2out.c (is_c): New function.
9296 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
9298 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
9301 * config/aarch64/aarch64.md (*aarch64_bfxil):
9303 * config/aarch64/constraints.md (Ulc): Define.
9304 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
9306 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
9309 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
9311 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
9312 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
9313 aarch64_layout_frame call.
9314 (aarch64_expand_epilogue): Likewise.
9315 (aarch64_initial_elimination_offset): Likewise.
9316 (aarch64_get_separate_components): Likewise.
9317 (aarch64_use_return_insn_p): Likewise.
9318 (aarch64_layout_frame): Remove unneeded check.
9320 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9322 * configure.ac: Only append
9323 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
9324 gcc_config_arguments if it was never reconfigured or last reconfigure
9325 was with different arguments.
9326 * configure: Regenerated.
9328 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9329 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9332 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
9333 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
9335 2018-09-13 Jakub Jelinek <jakub@redhat.com>
9337 PR tree-optimization/87287
9338 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
9339 X % C == 0 to X % (unsigned) C == 0 optimization to ...
9340 * match.pd (X % C == 0): ... here. New optimization.
9342 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9345 * expr.h (maybe_optimize_mod_cmp): Declare.
9346 * expr.c (mod_inv): New function.
9347 (maybe_optimize_mod_cmp): New function.
9348 (do_store_flag): Use it.
9349 * cfgexpand.c (expand_gimple_cond): Likewise.
9351 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
9352 Julian Brown <julian@codesourcery.com>
9355 * gimplify.c (gimplify_scan_omp_clauses): Set
9356 target_firstprivatize_array_bases in OpenACC parallel and kernels
9357 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
9358 OpenACC data regions.
9360 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
9362 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
9363 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
9365 2018-09-12 Richard Biener <rguenther@suse.de>
9367 PR tree-optimization/87280
9368 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
9369 edge but unreachable target.
9370 (do_rpo_vn): For conservatively handling a PHI only mark
9371 the backedge executable but not the block reachable.
9373 2018-09-12 Richard Biener <rguenther@suse.de>
9375 PR tree-optimization/87266
9376 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
9379 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
9381 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
9383 ("trunc<BFP:mode><DFP_ALL:mode>2")
9384 ("trunc<DFP_ALL:mode><BFP:mode>2")
9385 ("extend<BFP:mode><DFP_ALL:mode>2")
9386 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
9387 according to the target operand type.
9389 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9390 Andreas Krebbel <krebbel@linux.ibm.com>
9392 PR tree-optimization/86844
9393 * gimple-ssa-store-merging.c
9394 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
9395 there are any overlapping stores in between them, make sure they are
9396 also coalesced or we give up completely.
9398 2018-09-12 Jakub Jelinek <jakub@redhat.com>
9401 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
9402 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
9404 2018-09-12 Tom de Vries <tdevries@suse.de>
9406 * common.opt (gdescribe-dies): Add option.
9407 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
9408 attribute for artifical and nameless decls.
9409 (dwarf2out_register_external_die): Add description attribute to
9410 external reference die.
9411 (add_desc_attribute): New functions.
9412 (gen_subprogram_die): Add description attribute to
9413 DW_TAG_call_site_parameter.
9414 * tree-pretty-print.c (print_generic_expr_to_str): New function.
9415 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
9416 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
9418 (@item -gdescribe-dies): Add.
9420 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
9422 * tree-vrp.c (vrp_shift_undefined_p): Remove.
9423 (extract_range_from_binary_expr_1: Call
9424 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
9425 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
9428 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
9430 * gimple-ssa-warn-alloca.c
9431 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
9432 field for ALLOCA_BOUND_*_LARGE.
9434 2018-09-11 Nathan Sidwell <nathan@acm.org>
9436 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
9438 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9440 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
9441 for clobbers. Remove obsolete comment.
9443 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9445 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
9446 mpxchk, mpxld and mpxst types.
9447 (define_attr length_immediate): Remove all processing of mpx types.
9448 (define_attr prefix_0f): Ditto.
9449 (define_attr memory): Ditto.
9451 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
9453 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
9454 (log<mode>2): Change operand 1 predicate to general_operand.
9455 Extend operand 1 to XFmode and generate logxf3 insn.
9456 (log10<mode>2): Change operand 1 predicate to general_operand.
9457 Extend operand 1 to XFmode and generate log10xf3 insn.
9458 (log2<mode>2): Change operand 1 predicate to general_operand.
9459 Extend operand 1 to XFmode and generate log2xf3 insn.
9460 (fyl2xp1_extend<mode>xf3_i387): Remove.
9461 (log1p<mode>2): Change operand 1 predicate to general_operand.
9462 Extend operand 1 to XFmode and generate log1pxf3 insn.
9463 (fxtract_extend<mode>xf3_i387): Remove.
9464 (logb<mode>2): Change operand 1 predicate to general_operand.
9465 Extend operand 1 to XFmode and generate logbxf3 insn.
9466 (ilogb<mode>2): Change operand 1 predicate to general_operand.
9467 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
9468 (significand<mode>2): Change operand 1 predicate to general_operand.
9469 Extend operand 1 to XFmode and generate significandxf3 insn.
9471 2018-09-11 Nathan Sidwell <nathan@acm.org>
9473 * gcc.c (perror_with_name, pfatal_with_name): Delete.
9474 (load_specs): Use fatal_error.
9475 (DELETE_IF_ORDINARY, process_command): Use error.
9476 (execute, run_attempt): Use fatal_error.
9478 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
9480 * diagnostic-core.h (sorry_at): New prototype.
9481 * diagnostic.c (sorry_at): New function.
9483 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
9485 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
9486 by zero as VR_UNDEFINED.
9488 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
9490 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
9491 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
9492 (<sincos>mode2): New expander.
9493 (sincos_extend<mode>xf3_i387): Remove insn pattern.
9494 (sincos -> sin, cos splitters): Remove splitter patterns.
9495 (sincos<mode>3): Change operand 2 predicate to general_operand.
9496 Extend operand 2 to XFmode and generate sincosxf3 insn.
9497 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
9498 Change operand 3 predicate to const1_operand.
9499 (fptan_extend<mode>xf4_i387): Remove insn pattern.
9500 (tanxf2): Update operands in the call to fptanxf4_i387.
9501 (tan<mode>2): Change operand 1 predicate to general_operand.
9502 Extend operand 1 to XFmode and generate tanxf3 insn.
9503 (atan2xf3): Rename from *fpatanxf3_i387.
9504 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
9505 (atan2xf3): Remove expander.
9506 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
9507 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
9508 (atan<mode>2): Change operand 1 predicate to general_operand.
9509 Extend operand 1 to XFmode and generate atanxf3 insn.
9511 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
9513 * config/i386/i386.md (x87/SSE constant load splitter): Use
9514 memory_operand instead of nonimmediate_operand for input operand
9517 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
9519 * config/i386/i386.md (float partial SSE register stall splitter): Move
9520 splitter near its instruction pattern.
9521 (float_extend partial SSE register stall splitter): Ditto.
9522 (float_truncate partial SSE register stall splitter): Ditto.
9524 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
9527 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9528 to speculation_safe_value_not_needed.
9531 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
9532 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
9533 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
9534 leaf_function_p, instead use has_hard_reg_initial_val.
9536 2018-09-09 Nathan Sidwell <nathan@acm.org>
9538 * gcc.h (pfatal_with_name): Don't declare here.
9539 * gcc.c (pfatal_with_name): Make static.
9541 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
9543 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
9546 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
9548 PR rtl-optimization/85458
9549 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
9550 priority hook to reduce the priority of EXPR.
9552 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
9554 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
9555 DImode for x87 on 32bit targets. Conditionally disable x87 modes
9556 with X87_ENABLE_FLOAT. Remove preparation code.
9557 (*float<SWI48:mode><MODEF:mode>2): Rename from
9558 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
9559 math using "enabled" attribute.
9560 (*floatdi<MODEF:mode>2_i387): Rename from
9561 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
9562 enable for 32bit targets only.
9563 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
9565 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
9566 as operand 1 predicate. Rewrite as define_insn_and_split.
9567 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
9569 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
9571 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
9572 to fallthru to FLOAT case.
9574 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9577 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
9578 around folding of vec_sl to handle out of range shift values.
9580 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
9582 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
9583 Update callers to gen_fix_trunc<mode>_i387_fisttp
9584 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
9585 nonimmediate_operand.
9586 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
9587 and corresponding splitters.
9588 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
9589 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
9590 (fix_truncdi_i387_with_temp): Remove insn pattern
9591 and corresponding splitters.
9592 (fix_trunc<mode>_i387): Change operand 0 predicate to
9593 nonimmediate_operand.
9594 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
9595 and corresponding splitters.
9596 (*fistdi2_1): Remove.
9598 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
9599 (lrintxfdi2): Remove expander. Reimplement as define_insn.
9600 (*fist<mode>2_1): Remove.
9601 (fist<mode>2): Ditto.
9602 (fist<mode>2_with_temp): Remove insn pattern and corresponding
9604 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
9605 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
9606 (fistdi2_<rounding>): Change operand 0 predicate to
9607 nonimmediate_operand.
9608 (fistdi2_<rounding>_with_temp): Remove insn pattern
9609 and corresponding splitters.
9610 (fist<mode>2_<rounding>): Change operand 0 predicate to
9611 nonimmediate_operand.
9612 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
9613 and corresponding splitters.
9615 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
9617 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9619 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
9622 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9624 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
9625 early gimple folding of vec_splat().
9626 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
9627 * gimple-fold.h: Add an extern define for tree_vec_extract().
9629 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
9631 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
9632 wrappers around TREE_TYPE comparisons.
9634 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9637 * config/s390/predicates.md: Add nonsym_memory_operand.
9638 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
9639 contains a SYMBOL_REF, load it into an intermediate pseudo.
9640 (s390_emit_compare_and_swap): Legitimize operand.
9641 * config/s390/s390.md: Use the new nonsym_memory_operand
9642 with UNSPECV_CAS patterns.
9644 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9647 * config/s390/s390-passes.def: New file.
9648 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
9650 (make_pass_s390_early_mach): Add declaration.
9651 * config/s390/s390.c (make_pass_s390_early_mach):
9652 (s390_option_override): Remove dynamic registration.
9653 * config/s390/t-s390: Add s390-passes.def.
9655 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
9657 * config/s390/s390.c (s390_decompose_constant_pool_ref):
9658 Remove UNSPEC_LTREL_BASE check.
9659 (annotate_constant_pool_refs): Likewise.
9660 (find_constant_pool_ref): Likewise.
9661 (find_ltrel_base): Removed.
9662 (replace_ltrel_base): Removed.
9663 (s390_mainpool_finish): Remove replace_ltrel_base call.
9664 (s390_chunkify_start): Remove pending LTREL_BASE logic.
9665 (s390_chunkify_finish): Remove replace_ltrel_base call.
9666 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
9668 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
9671 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9672 to speculation_safe_value_not_needed.
9674 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
9675 Bernd Schmidt <bernds_cb1@t-online.de>
9677 * config/nvptx/nvptx-opts.h: New file.
9678 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
9679 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
9681 (TARGET_SM35): New macro.
9682 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
9684 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
9686 (misa=): New option.
9687 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
9689 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
9691 * config/i386/i386.md (truncdfsf2): Remove expander.
9692 (truncdfsf2_with_temp): Ditto.
9693 (truncxf<mode>2): Ditto.
9694 (*truncdfsf_fast_mixed): Remove insn pattern.
9695 (*truncdfsf_fast_i387): Ditto.
9696 (*truncdfsf_mixed): Ditto.
9697 (*truncdfsf_i387): Ditto.
9698 (*truncdfsf2_i387_1): Ditto.
9699 (*truncxfsf2_mixed): Ditto.
9700 (*truncxfdf2_mixed): Ditto.
9701 (*truncxf<mode>2_i387_noop): Ditto. Update callers
9702 to call gen_truncxf<mode>2 instead.
9703 (*truncxf<mode>2_i387): Remove.
9704 (reg->reg splitters): Remove splitter pattern.
9705 (reg->mem splitters): Ditto.
9707 (truncdfsf2): New insn pattern.
9708 (truncxf<mode>2): Ditto.
9710 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9712 * tree-ssa-math-opts.c (is_mult_by): New function.
9713 (is_square_of): Use the above.
9714 (optimize_recip_sqrt): New function.
9715 (pass_cse_reciprocals::execute): Use the above.
9717 2018-09-05 Richard Biener <rguenther@suse.de>
9720 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
9721 to zero-init the emplaced vec.
9723 2018-09-05 Martin Liska <mliska@suse.cz>
9725 PR tree-optimization/87205
9726 * tree-switch-conversion.c (pass_lower_switch::execute):
9727 Group cases for switch statements.
9729 2018-09-05 Richard Biener <rguenther@suse.de>
9731 PR tree-optimization/87217
9732 * tree-ssa-sccvn.c (vuse_valueize): New.
9733 (vn_reference_lookup_pieces): Use it.
9734 (vn_reference_lookup): Likewise.
9736 2018-09-05 Nathan Sidwell <nathan@acm.org>
9739 * stor-layout.c (place_field): Scan forwards to check last
9740 bitfield when ms_bitfield_placement is in effect.
9742 2018-09-05 Richard Biener <rguenther@suse.de>
9745 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
9748 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
9749 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
9751 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
9752 * config.gcc (extra_objs): Build it.
9753 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
9755 * config/aarch64/aarch64-passes.def
9756 (pass_tag_collision_avoidance): New pass.
9757 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
9758 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
9759 (aarch64_classify_address): Remove static qualifier.
9760 (aarch64_address_info, aarch64_address_type): Move to...
9761 * config/aarch64/aarch64-protos.h: ... here.
9762 (make_pass_tag_collision_avoidance): New function.
9763 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
9766 2018-09-05 Martin Liska <mliska@suse.cz>
9768 * doc/gcov.texi: Update documentation of humar
9770 * gcov.c (format_count): Print one decimal place, it provides
9771 more fine number of situations like '1G' vs. '1.4G'.
9773 2018-09-05 Martin Liska <mliska@suse.cz>
9776 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
9777 * optc-gen.awk: Allow 'Var' for Deprecated options in order
9778 to generate a MASK value.
9780 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
9783 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
9784 if frame pointer isn't used.
9785 (compute_frame_pointer_to_fb_displacement): Likewise.
9787 2018-09-04 Jakub Jelinek <jakub@redhat.com>
9790 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
9791 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
9792 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
9793 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
9794 and OPTION_MASK_ISA_XSAVEC_UNSET.
9796 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
9798 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
9799 XOR operations in NAND case.
9801 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9803 * wide-int-range.cc (wide_int_range_convert): New.
9804 * wide-int-range.h (wide_int_range_convert): New.
9805 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
9806 code into wide_int_range_convert.
9807 (extract_range_into_wide_ints): Do not munge anti range constants
9808 into the entire domain. Just return the range back.
9810 2018-09-04 Martin Liska <mliska@suse.cz>
9812 * genmatch.c (output_line_directive): Add new argument
9814 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
9816 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9818 * doc/invoke.texi (Option Summary): Add whitespace.
9820 * doc/invoke.texi (Option Summary): Add -Waligned-new.
9822 2018-09-04 Richard Biener <rguenther@suse.de>
9824 PR tree-optimization/87211
9825 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
9826 backedge value we're supposed to treat as VARYING also number
9827 the PHI to VARYING in case it got a different value-number already.
9829 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9831 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
9832 (extract_range_from_binary_expr_1): Do not call
9833 vrp_can_optimize_bit_op.
9834 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
9836 (wide_int_range_get_mask_and_bounds): New.
9837 (wide_int_range_optimize_bit_op): New.
9838 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
9839 (wide_int_range_bit_and): Same.
9840 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
9841 (wide_int_range_optimize_bit_op): New.
9842 (wide_int_range_get_mask_and_bounds): New.
9844 2018-09-04 Richard Biener <rguenther@suse.de>
9846 PR tree-optimization/87176
9847 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
9848 variable. When value-numbering a virtual PHI node make sure
9849 to not value-number to the backedge value.
9851 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9853 * doc/extend.texi (Long Long, Hex Floats): Document support for
9854 long long and hex floats in more recent versions of ISO C++.
9856 2018-09-03 Richard Biener <rguenther@suse.de>
9858 PR tree-optimization/87177
9859 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
9862 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9864 * bb-reorder.c (edge_order): Convert to C-qsort-style
9865 tri-state comparator.
9866 (reorder_basic_blocks_simple): Change std::stable_sort to
9869 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9871 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
9872 tri-state comparator.
9873 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
9875 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9877 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
9878 (mergesort): ... here as maximum count for using netsort.
9879 (gcc_qsort): Set nlim to 3 if stable sort is requested.
9880 (gcc_stablesort): New.
9881 * system.h (gcc_stablesort): Declare.
9883 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9885 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
9886 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
9887 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
9889 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
9891 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
9892 lxsdx and stxsdx alternatives.
9893 (*mov<mode>_hardfloat64): Ditto.
9894 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
9896 2018-09-03 Richard Biener <rguenther@suse.de>
9898 PR tree-optimization/87200
9899 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
9902 2018-09-03 Martin Liska <mliska@suse.cz>
9904 PR tree-optimization/87201
9905 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9906 Fix parenthesis in an expression.
9908 2018-09-03 Richard Biener <rguenther@suse.de>
9910 PR tree-optimization/87197
9911 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
9912 visited. CSE the VN_INFO hashtable lookup.
9914 PR tree-optimization/87169
9915 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
9916 iterating make sure there's no extra backedges from irreducible
9917 regions feeding the header. Mark the destination block
9920 2018-09-03 Martin Liska <mliska@suse.cz>
9923 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
9924 * common/common-targhooks.c (default_get_valid_option_values):
9926 * common/common-targhooks.h (default_get_valid_option_values):
9928 * common/config/i386/i386-common.c: Move processor_target_table
9930 (ix86_get_valid_option_values): New function.
9931 (TARGET_GET_VALID_OPTION_VALUES): New macro.
9932 * config/i386/i386.c (struct ptt): Move to i386-common.c.
9933 (PTA_*): Move all defined masks into i386-common.c.
9934 (ix86_function_specific_restore): Use new processor_cost_table.
9935 * config/i386/i386.h (struct ptt): Moved from i386.c.
9936 (struct pta): Likewise.
9937 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
9938 * doc/tm.texi.in: Likewise.
9939 * opt-suggestions.c (option_proposer::suggest_option):
9940 Pass prefix to build_option_suggestions.
9941 (option_proposer::get_completions): Likewise.
9942 (option_proposer::build_option_suggestions): Use the new target
9944 * opts.c (struct option_help_tuple): New struct.
9945 (print_filtered_help): Use the new target hook.
9947 2018-09-03 Martin Liska <mliska@suse.cz>
9950 * predict.c (set_even_probabilities): Add likely_edges
9951 argument and handle cases where we have precisely one
9953 (combine_predictions_for_bb): Catch also likely_edges.
9954 (tree_predict_by_opcode): Handle gswitch statements.
9955 * tree-cfg.h (find_case_label_for_value): New declaration.
9956 (find_taken_edge_switch_expr): Likewise.
9957 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9958 Find pivot in decision tree based on probabily, not by number of
9961 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
9963 * doc/standards.texi (Standards): Update Objective-C reference.
9965 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9967 * doc/install.texi (Prerequisites): Update link for MPC.
9969 2018-09-01 Michael Matz <matz@suse.de>
9971 PR tree-optimization/87074
9972 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
9973 PHIs for outer-loop uses.
9975 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9977 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
9978 * doc/invoke.texi (C Dialect Options): Ditto.
9980 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9982 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
9984 2018-08-31 Richard Biener <rguenther@suse.de>
9986 PR tree-optimization/87168
9987 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
9988 (rpo_elim::eliminate_avail): When OP was not visited it must
9991 2018-08-31 David Malcolm <dmalcolm@redhat.com>
9993 * tree-vrp.c (copy_value_range): Convert param "from" from
9994 "value_range *" to "const value_range *".
9995 (range_is_null): Likewise for param "vr".
9996 (range_int_cst_p): Likewise.
9997 (range_int_cst_singleton_p): Likewise.
9998 (symbolic_range_p): Likewise.
9999 (value_ranges_intersect_p): Likewise for both params.
10000 (value_range_nonnegative_p): Likewise for param "vr".
10001 (value_range_constant_singleton): Likewise.
10002 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
10003 (extract_range_into_wide_ints): Likewise for param "vr".
10004 (extract_range_from_multiplicative_op): Likewise for params "vr0"
10006 (vrp_can_optimize_bit_op): Likewise.
10007 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
10009 (extract_range_from_unary_expr): Likewise.
10010 (debug_value_range): Likewise for param "vr".
10011 (value_range::dump): Add "const" qualifier.
10012 (vrp_prop::check_array_ref): Convert local "vr" from
10013 "value_range *" to "const value_range *".
10014 (vrp_prop::check_mem_ref): Likewise.
10015 (vrp_prop::visit_stmt): Likewise for local "old_vr".
10016 (vrp_intersect_ranges_1): Likewise for param "vr_1".
10017 (vrp_intersect_ranges): Likewise.
10018 (simplify_stmt_for_jump_threading): Likewise for local "vr".
10019 (vrp_prop::vrp_finalize): Likewise.
10020 * tree-vrp.h (value_range::dump): Add "const" qualifier.
10021 (vrp_intersect_ranges): Add "const" qualifier to params as above.
10022 (extract_range_from_unary_expr): Likewise.
10023 (value_range_constant_singleton): Likewise.
10024 (symbolic_range_p): Likewise.
10025 (copy_value_range): Likewise.
10026 (extract_range_from_binary_expr_1): Likewise.
10027 (range_int_cst_p): Likewise.
10028 (vrp_set_zero_nonzero_bits): Likewise.
10029 (range_int_cst_singleton_p): Likewise.
10031 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
10033 * config/aarch64/arm_neon.h (vabsd_s64): New.
10034 (vnegd_s64): Likewise.
10036 2018-08-31 Martin Jambor <mjambor@suse.cz>
10038 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
10040 2018-08-31 Martin Liska <mliska@suse.cz>
10042 * ipa-icf.c (sem_item::add_type): Use
10043 sem_item::m_type_hash_cache.
10044 * ipa-icf.h: Move the cache from sem_item_optimizer
10047 2018-08-31 Nathan Sidwell <nathan@acm.org>
10049 * doc/extend.texi (Backwards Compatibility): Remove implicit
10050 extern C leeway of () being (...).
10052 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10054 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
10056 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
10060 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
10062 2018-08-31 Jakub Jelinek <jakub@redhat.com>
10064 PR middle-end/87138
10065 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
10066 gen_int_mode. Formatting fixes.
10068 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
10070 * target.def (custom_function_descriptors): Improve documentation.
10071 * doc/tm.texi.in (Trampolines): Expand discussion of function
10072 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
10073 beginning of the section.
10074 * doc/tm.texi: Regenerated.
10076 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
10078 * cfg.h (class auto_edge_flag): Spell out the template-id of the
10079 base class in the initializer list. This is a workaround for
10080 building with older GCC.
10081 (class auto_bb_flag): Likewise.
10083 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
10085 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
10086 (altivec_vcmpequ<VI_char>_p): Remove star.
10087 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
10089 (expand_strncmp_vec_sequence): New function.
10090 (emit_final_str_compare_vec): New function.
10091 (expand_strn_compare): Add support for vector strncmp.
10092 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
10093 length specification to bytes.
10094 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
10095 (vcmpnezb_p): New pattern.
10096 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
10097 for option -mstring-compare-inline-limit.
10099 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
10101 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
10102 (PTA_SKYLAKE): Add PTA_AES.
10103 (PTA_GOLDMONT): Likewise.
10105 2018-08-29 Jan Hubicka <jh@suse.cz>
10108 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
10109 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
10111 2018-08-29 Jan Hubicka <jh@suse.cz>
10113 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
10115 (hash_tree): Do not visit TYPE_STUB_DECL.
10116 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
10117 stream TYPE_STUB_DECL.
10118 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
10119 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
10120 after free_lang_data.
10121 (type_in_anonymous_namespace_p): Likewise.
10123 2018-08-29 Jan Hubicka <jh@suse.cz>
10125 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
10126 comment that it has to be even number.
10127 (class sreal): Change m_sig type to int32_t.
10128 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
10129 int64_t for temporary calculations.
10130 (sreal_verify_basics): Drop one bit from minimum and maximum.
10132 2018-08-30 Richard Biener <rguenther@suse.de>
10134 PR tree-optimization/87147
10135 * tree-ssa-sccvn.c (SSA_VISITED): New function.
10136 (visit_phi): When the degenerate result is from the backedge and
10137 we didn't visit its definition yet drop to VARYING.
10138 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
10140 2018-08-29 Jan Hubicka <jh@suse.cz>
10142 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
10144 (hash_tree): Likewise.
10146 2018-08-29 Jan Hubicka <jh@suse.cz>
10148 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
10149 and TYPE_NEXT_REF_TO.
10151 2018-08-29 Jan Hubicka <jh@suse.cz>
10153 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
10154 comment that it has to be even number.
10155 (class sreal): Change m_sig type to int32_t.
10156 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
10157 int64_t for temporary calculations.
10158 (sreal_verify_basics): Drop one bit from minimum and maximum.
10160 2018-08-30 Tamar Christina <tamar.christina@arm.com>
10162 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
10164 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
10166 PR middle-end/86995
10167 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
10168 if to_add is negative.
10170 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
10172 PR middle-end/87053
10173 * builtins.c (c_strlen): Improve range checks.
10175 2018-08-29 Martin Sebor <msebor@redhat.com>
10176 Jeff Law <law@redhat.com>
10178 PR tree-optimization/86714
10179 PR tree-optimization/86711
10180 * builtins.c (c_strlen): Add arguments to call to string_constant.
10181 * expr.c (string_constant): Add argument. Detect missing nul
10182 terminator and outermost declaration it's missing in.
10183 * expr.h (string_constant): Add argument.
10184 * fold-const.c (read_from_constant_string): Add arguments to call to
10186 (c_getstr): Likewise.
10187 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
10188 to string_constant.
10189 * tree-ssa-strlen.c (get_stridx): Likewise.
10191 2018-08-29 Jan Hubicka <jh@suse.cz>
10193 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
10194 Do not stream DECL_VINDEX.
10195 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
10196 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
10197 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
10199 2018-08-29 Richard Biener <rguenther@suse.de>
10201 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
10202 virtual operands that are not default defs to honor region
10204 (rpo_vn_valueize): Remove ineffective code here.
10206 2018-08-29 Richard Biener <rguenther@suse.de>
10208 PR tree-optimization/87132
10209 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
10210 when skipping defs reachable over backedges.
10212 2018-08-29 Richard Biener <rguenther@suse.de>
10214 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
10215 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
10216 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
10217 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
10218 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
10219 (vn_reference_lookup_3): Remove use of const_parms.
10220 (free_rpo_vn): Do not free const_parms.
10221 (do_rpo_vn): Do not call init_const_parms.
10222 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
10223 SSA_NAME_POINTS_TO_READONLY_MEMORY.
10224 (call_may_clobber_ref_p_1): Likewise.
10226 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
10229 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
10231 (-ftree-scev-cprop): Document.
10233 2018-08-29 Jan Hubicka <jh@suse.cz>
10235 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
10237 (sreal constructor): Update.
10238 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
10239 sreal:operator/): Update.
10241 2018-08-29 Martin Liska <mliska@suse.cz>
10243 * tree-switch-conversion.c (switch_conversion::expand):
10244 Strenghten assumption about gswitch statements.
10246 2018-08-29 Richard Biener <rguenther@suse.de>
10248 PR tree-optimization/87117
10249 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
10250 re-value-number released SSA VDEFs.
10252 2018-08-29 Richard Biener <rguenther@suse.de>
10254 PR tree-optimization/87126
10255 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
10257 2018-08-28 Jim Wilson <jimw@sifive.com>
10259 * config/riscv/pic.md: Rewrite.
10260 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
10262 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
10263 (SOFTF, default_load, softload, softstore): New.
10265 2018-08-28 Jeff Law <law@redhat.com>
10267 * fold-const.c (fold_binary_loc): Remove recently added assert.
10269 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
10271 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
10272 to OP parmeter of generated function.
10274 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
10276 PR tree-optimization/87009
10277 * match.pd: Add boolean optimizations.
10279 2018-08-28 Martin Sebor <msebor@redhat.com>
10281 PR middle-end/86631
10282 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
10283 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
10284 (pass_walloca::gate): Use it.
10285 (alloca_call_type): Same.
10286 (pass_walloca::execute): Same.
10287 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
10289 2018-08-28 David Malcolm <dmalcolm@redhat.com>
10291 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
10292 GCC_VERSION for usage of "__gcc_dump_printf__" format from
10293 >= 3005 to >= 9000.
10295 2018-08-28 Richard Biener <rguenther@suse.de>
10297 PR tree-optimization/87124
10298 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
10299 constants before looking up avail.
10301 2018-08-28 Jakub Jelinek <jakub@redhat.com>
10303 PR middle-end/87099
10304 * calls.c (maybe_warn_nonstring_arg): Punt early if
10305 warn_stringop_overflow is zero. Don't call get_range_strlen
10306 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
10307 Swap comparison operands to have constants on rhs. Only use
10308 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
10309 increment lenrng[0].
10311 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
10313 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
10314 use of tree_to_shwi. Remove duplicated test for the size being
10315 a whole number of bytes.
10317 2018-08-28 Richard Biener <rguenther@suse.de>
10319 PR tree-optimization/87117
10320 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
10321 Handle removed stmt without LHS (GIMPLE_NOP).
10323 2018-08-28 Richard Biener <rguenther@suse.de>
10325 PR tree-optimization/87117
10326 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
10327 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
10329 2018-08-28 Richard Biener <rguenther@suse.de>
10331 PR tree-optimization/87117
10332 * tree-ssa-pre.c (compute_avail): Do not make expressions
10333 with predicated values available.
10334 (get_expr_value_id): Assert we do not run into predicated value
10337 2018-08-28 Richard Biener <rguenther@suse.de>
10339 PR tree-optimization/87117
10340 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
10341 get virtual operands.
10342 (get_expr_operands): Handle STRING_CST like other decls.
10344 2018-08-28 Martin Liska <mliska@suse.cz>
10346 * tree.h: Update documentation of fndecl_built_in_p
10350 2018-08-27 Jeff Law <law@redhat.com>
10351 PR tree-optimization/87110
10352 * tree-ssa-dse.c (compute_trims): Handle non-constant
10355 2018-08-27 Martin Sebor <msebor@redhat.com>
10357 PR tree-optimization/86914
10358 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
10360 2018-08-27 Martin Sebor <msebor@redhat.com>
10362 PR tree-optimization/87112
10363 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
10364 the type of the bound argument.
10366 2018-08-27 Jeff Law <law@redhat.com>
10368 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
10369 type does not have a TYPE_SIZE_UNIT.
10371 2018-08-27 Steve Ellcey <sellcey@cavium.com>
10373 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
10374 with include of backend.h.
10376 2018-08-27 Richard Biener <rguenther@suse.de>
10378 PR tree-optimization/86927
10379 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
10380 use const cond reduction code.
10382 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
10384 PR tree-optimization/85758
10385 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
10387 2018-08-27 David Malcolm <dmalcolm@redhat.com>
10390 * diagnostic-show-locus.c (class layout_range): Update for
10391 conversion of show_caret_p to a tri-state.
10392 (layout_range::layout_range): Likewise.
10393 (make_range): Likewise.
10394 (layout::maybe_add_location_range): Likewise.
10395 (layout::should_print_annotation_line_p): Don't show annotation
10396 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
10397 (layout::get_state_at_point): Update for conversion of
10398 show_caret_p to a tri-state. Bail out early for
10399 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
10400 underlining or source colorization.
10401 (gcc_rich_location::add_location_if_nearby): Update for conversion
10402 of show_caret_p to a tri-state.
10403 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
10404 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
10406 (selftest::test_one_liner_labels): Likewise.
10407 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
10408 conversion of show_caret_p to a tri-state.
10409 * pretty-print.c (text_info::set_location): Likewise.
10410 * pretty-print.h (text_info::set_location): Likewise.
10411 * substring-locations.c (format_warning_n_va): Likewise.
10412 * tree-diagnostic.c (default_tree_printer): Likewise.
10413 * tree-pretty-print.c (newline_and_indent): Likewise.
10415 2018-08-27 David Malcolm <dmalcolm@redhat.com>
10418 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
10419 line above for line-insertion fix-it hints.
10420 (selftest::test_fixit_insert_containing_newline): Update the
10421 expected results, and add a test with line-numbering enabled.
10423 2018-08-27 Martin Liska <mliska@suse.cz>
10426 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
10427 params with DECL_HAS_VALUE_EXPR_P.
10429 2018-08-27 Martin Liska <mliska@suse.cz>
10431 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
10432 selected expansion strategy.
10434 2018-08-27 Martin Liska <mliska@suse.cz>
10436 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
10437 * builtins.c (is_builtin_fn): Likewise.
10438 * attribs.c (diag_attr_exclusions): Use new function
10439 fndecl_built_in_p and remove check for FUNCTION_DECL if
10441 (builtin_mathfn_code): Likewise.
10442 (fold_builtin_expect): Likewise.
10443 (fold_call_expr): Likewise.
10444 (fold_builtin_call_array): Likewise.
10445 (fold_call_stmt): Likewise.
10446 (set_builtin_user_assembler_name): Likewise.
10447 (is_simple_builtin): Likewise.
10448 * calls.c (gimple_alloca_call_p): Likewise.
10449 (maybe_warn_nonstring_arg): Likewise.
10450 * cfgexpand.c (expand_call_stmt): Likewise.
10451 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
10452 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
10453 (cgraph_node::verify_node): Likewise.
10454 * cgraphclones.c (build_function_decl_skip_args): Likewise.
10455 (cgraph_node::create_clone): Likewise.
10456 * config/arm/arm.c (arm_insert_attributes): Likewise.
10457 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
10458 * dse.c (scan_insn): Likewise.
10459 * expr.c (expand_expr_real_1): Likewise.
10460 * fold-const.c (operand_equal_p): Likewise.
10461 (fold_binary_loc): Likewise.
10462 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
10463 * gimple-low.c (lower_stmt): Likewise.
10464 * gimple-pretty-print.c (dump_gimple_call): Likewise.
10465 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
10467 * gimple.c (gimple_build_call_from_tree): Likewise.
10468 (gimple_call_builtin_p): Likewise.
10469 (gimple_call_combined_fn): Likewise.
10470 * gimplify.c (gimplify_call_expr): Likewise.
10471 (gimple_boolify): Likewise.
10472 (gimplify_modify_expr): Likewise.
10473 (gimplify_addr_expr): Likewise.
10474 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
10475 * ipa-cp.c (determine_versionability): Likewise.
10476 * ipa-fnsummary.c (compute_fn_summary): Likewise.
10477 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
10478 * ipa-split.c (visit_bb): Likewise.
10479 (split_function): Likewise.
10480 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10481 * lto-cgraph.c (input_node): Likewise.
10482 * lto-streamer-out.c (write_symbol): Likewise.
10483 * omp-low.c (setjmp_or_longjmp_p): Likewise.
10484 (lower_omp_1): Likewise.
10485 * predict.c (strip_predict_hints): Likewise.
10486 * print-tree.c (print_node): Likewise.
10487 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
10488 * trans-mem.c (is_tm_irrevocable): Likewise.
10489 (is_tm_load): Likewise.
10490 (is_tm_simple_load): Likewise.
10491 (is_tm_store): Likewise.
10492 (is_tm_simple_store): Likewise.
10493 (is_tm_abort): Likewise.
10494 (tm_region_init_1): Likewise.
10495 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
10496 * tree-cfg.c (verify_gimple_call): Likewise.
10497 (move_stmt_r): Likewise.
10498 (stmt_can_terminate_bb_p): Likewise.
10499 * tree-eh.c (lower_eh_constructs_2): Likewise.
10500 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
10501 * tree-inline.c (remap_gimple_stmt): Likewise.
10502 (copy_bb): Likewise.
10503 (estimate_num_insns): Likewise.
10504 (fold_marked_statements): Likewise.
10505 * tree-sra.c (scan_function): Likewise.
10506 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
10507 (optimize_stack_restore): Likewise.
10508 (pass_fold_builtins::execute): Likewise.
10509 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
10510 (mark_all_reaching_defs_necessary_1): Likewise.
10511 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
10512 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
10513 (pass_forwprop::execute): Likewise.
10514 * tree-ssa-loop-im.c (stmt_cost): Likewise.
10515 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10516 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
10517 * tree-ssa-strlen.c (get_string_length): Likewise.
10518 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
10519 (find_func_aliases_for_call): Likewise.
10520 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
10521 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
10522 * tree-tailcall.c (find_tail_calls): Likewise.
10523 * tree.c (need_assembler_name_p): Likewise.
10524 (free_lang_data_in_decl): Likewise.
10525 (get_call_combined_fn): Likewise.
10526 * ubsan.c (is_ubsan_builtin_p): Likewise.
10527 * varasm.c (incorporeal_function_p): Likewise.
10528 * tree.h (DECL_BUILT_IN): Remove and replace with
10530 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
10531 (fndecl_built_in_p): New.
10533 2018-08-27 Martin Liska <mliska@suse.cz>
10535 PR tree-optimization/86847
10536 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
10537 Dump also subtree probability.
10538 (switch_decision_tree::do_jump_if_equal): New function.
10539 (switch_decision_tree::emit_case_nodes): Handle special
10540 situations in balanced tree that can be emitted much simpler.
10541 Fix calculation of probabilities that happen in tree expansion.
10542 * tree-switch-conversion.h (struct cluster): Add
10544 (struct simple_cluster): Likewise.
10545 (struct case_tree_node): Add new function has_child.
10546 (do_jump_if_equal): New.
10548 2018-08-27 Martin Liska <mliska@suse.cz>
10550 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
10551 Add new argument to bit_test_cluster constructor.
10552 (bit_test_cluster::emit): Set bits really number of values
10554 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
10555 probability argument.
10556 * tree-switch-conversion.h (struct bit_test_cluster):
10557 Add m_handles_entire_switch.
10559 2018-08-27 Martin Liska <mliska@suse.cz>
10561 PR tree-optimization/86702
10562 * tree-switch-conversion.c (jump_table_cluster::emit):
10563 Make probabilities even for values in jump table
10564 according to number of cases handled.
10565 (switch_decision_tree::compute_cases_per_edge): Pass
10566 argument to reset_out_edges_aux function.
10567 (switch_decision_tree::analyze_switch_statement): Likewise.
10568 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
10571 2018-08-27 Martin Liska <mliska@suse.cz>
10573 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
10574 cfun argument explicitly.
10575 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
10576 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
10577 function gimple_switch_default_bb.
10578 (convert_switch_statements):
10580 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
10581 * stmt.c (label_to_block_fn): Use label_to_block and pass
10582 cfun argument explicitly and use gimple_switch_label_bb.
10583 (expand_case): Likewise.
10584 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
10585 cfun argument explicitly. Likewise.
10586 (make_edges_bb): Likewise.
10587 (make_cond_expr_edges): Likewise.
10588 (get_cases_for_edge): Likewise.
10589 (make_gimple_switch_edges): Likewise.
10590 (label_to_block_fn): Likewise.
10591 (label_to_block): Likewise.
10592 (make_goto_expr_edges): Likewise.
10593 (make_gimple_asm_edges): Likewise.
10594 (main_block_label): Likewise.
10595 (group_case_labels_stmt): Likewise.
10596 (find_taken_edge_computed_goto): Likewise.
10597 (find_taken_edge_switch_expr): Likewise.
10598 (gimple_verify_flow_info): Likewise.
10599 (gimple_redirect_edge_and_branch): Likewise.
10600 (gimple_switch_label_bb): New function.
10601 (gimple_switch_default_bb): Likewise.
10602 (gimple_switch_edge): Likewise.
10603 (gimple_switch_default_edge): Likewise.
10604 * tree-cfg.h (label_to_block_fn): Remove and replace ...
10605 (label_to_block): ... with this.
10606 (gimple_switch_label_bb): New.
10607 (gimple_switch_default_bb): Likewise.
10608 (gimple_switch_edge): Likewise.
10609 (gimple_switch_default_edge): Likewise.
10610 * tree-cfgcleanup.c (convert_single_case_switch): Use
10611 new gimple functions and pass new argument to label_to_block.
10612 (cleanup_control_flow_bb):
10613 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
10614 cfun argument explicitly.
10615 (make_eh_edges): Likewise.
10616 (redirect_eh_dispatch_edge): Likewise.
10617 (lower_resx): Likewise.
10618 (lower_eh_dispatch): Likewise.
10619 (maybe_remove_unreachable_handlers): Likewise.
10620 (unsplit_eh): Likewise.
10621 (cleanup_empty_eh): Likewise.
10622 (verify_eh_edges): Likewise.
10623 (verify_eh_dispatch_edge): Likewise.
10624 * tree-ssa-dom.c (record_edge_info): Likewise.
10625 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
10626 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
10627 (thread_through_normal_block): Likewise.
10628 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
10629 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
10630 * tree-switch-conversion.c (switch_conversion::collect): Use new
10632 (switch_conversion::check_final_bb): Likewise.
10633 (switch_conversion::gather_default_values): Pass new argument
10635 (switch_conversion::build_constructors): Likewise.
10636 (switch_decision_tree::compute_cases_per_edge): Use new
10637 gimple_switch_edge function.
10638 (switch_decision_tree::analyze_switch_statement): Pass new argument
10640 (switch_decision_tree::try_switch_expansion): Use
10641 gimple_switch_default_edge.
10642 * tree-vrp.c (find_switch_asserts): Pass new argument
10644 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
10645 (vr_values::simplify_switch_using_ranges): Likewise.
10647 2018-08-27 Richard Biener <rguenther@suse.de>
10649 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
10650 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
10652 * tree-ssa-sccvn.h (struct vn_pval): New structure.
10653 (struct vn_nary_op_s): Add unwind_to member. Add
10654 predicated_values flag and put result into a union together
10655 with a linked list of vn_pval.
10656 (struct vn_ssa_aux): Add name member to make maintaining
10657 a map of SSA name to vn_ssa_aux possible. Remove no longer
10658 needed info, dfsnum, low, visited, on_sccstack, use_processed
10659 and range_info_anti_range_p members.
10660 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
10661 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
10663 (vn_valueize): New global.
10664 (vn_context_bb): Likewise.
10665 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
10666 VN_INFO_PTR_INFO): Remove.
10667 * tree-ssa-sccvn.c: ... (rewrite)
10668 (pass_fre::execute): For -O2+ initialize loops and run
10669 RPO VN in optimistic mode (iterating). For -O1 and -Og
10670 run RPO VN in non-optimistic mode.
10671 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
10672 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
10673 * doc/invoke.texi (sccvn-max-scc-size): Remove.
10674 (rpo-vn-max-loop-depth): Document.
10675 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
10676 when valuezing the VUSE signals we walked out of the region.
10677 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
10678 (phi_translate): Set VN context block to use for availability
10680 (compute_avail): Likewise.
10681 (pre_valueize): New function.
10682 (pass_pre::execute): Adjust to the RPO VN API.
10684 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
10685 (propagate_constants_for_unrolling): Remove.
10686 (tree_unroll_loops_completely): Perform value-numbering
10687 on the unrolled bodies loop parent.
10689 2018-08-27 Richard Biener <rguenther@suse.de>
10691 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
10692 for partial antic compute.
10694 2018-08-27 Jakub Jelinek <jakub@redhat.com>
10696 PR rtl-optimization/87065
10697 * combine.c (simplify_if_then_else): Formatting fix.
10698 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
10700 (known_cond): Don't return const_true_rtx for vector modes. Use
10701 CONST0_RTX instead of const0_rtx. Formatting fixes.
10703 2018-08-27 Martin Liska <mliska@suse.cz>
10705 PR gcov-profile/87069
10706 * gcov.c (process_file): Record files already processed
10707 and warn about a file being processed multiple times.
10709 2018-08-27 Martin Liska <mliska@suse.cz>
10712 * config/aarch64/aarch64.c (aarch64_override_options_internal):
10713 Set default values for x_aarch64_*_string strings.
10714 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
10715 prefix. For -mabi do not print '=ABI' in help and use
10716 <option_value> format for -msve-vector-bits and -moverride
10719 2018-08-26 Jeff Law <law@redhat.com>
10721 * config/mips/frame-header-opt.c: Include "backend.h" rather than
10724 2018-08-26 Marek Polacek <polacek@redhat.com>
10726 PR c++/87029, Implement -Wredundant-move.
10727 * doc/invoke.texi: Document -Wredundant-move.
10729 2018-08-25 Martin Sebor <msebor@redhat.com>
10731 PR tree-optimization/87059
10732 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
10733 to the same type as the other.
10734 * fold-const.c (fold_binary_loc): Assert expectation.
10736 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
10738 * config/darwin.c (machopic_legitimize_pic_address): Clean up
10739 extraneous parentheses, dead code section and formatting.
10741 2018-08-24 David Malcolm <dmalcolm@redhat.com>
10744 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
10745 wide enough for jumps in the line-numbering to be visible.
10746 (layout::print_gap_in_line_numbering): New member function.
10747 (layout::calculate_line_spans): When using line numbering, merge
10748 line spans that are only 1 line apart.
10749 (diagnostic_show_locus): When printing line numbers, show gaps in
10750 line numbering directly, rather than printing headers.
10751 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
10752 line-numbering with multiple line spans.
10753 (selftest::test_fixit_insert_containing_newline_2): Add test of
10754 line-numbering, in which the spans are close enough to be merged.
10756 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10758 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
10759 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
10760 * tree-vrp.c (range_is_nonnull): Remove.
10761 (range_includes_zero_p): Accept value_range instead of min/max.
10762 (extract_range_from_binary_expr_1): Do not early bail on
10764 Use range_includes_zero_p instead of range_is_nonnull.
10765 (extract_range_from_unary_expr): Use range_includes_zero_p instead
10766 of range_is_nonnull.
10767 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
10768 special case VR_ANTI_RANGE.
10769 (vrp_finalize): Same.
10770 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
10771 instead of min/max.
10772 (range_is_nonnull): Remove.
10773 * vr-values.c (vrp_stmt_computes_nonzero): Use
10774 range_includes_zero_p instead of range_is_nonnull.
10775 (extract_range_basic): Pass value_range to range_includes_zero_p
10776 instead of range_is_nonnull.
10778 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
10780 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
10781 * emit-rtl.h (rtl_data): Remove return_bnd.
10782 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
10783 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
10784 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
10785 (POINTER_BOUNDS_MODE): Remove definition.
10786 (make_pointer_bounds_mode): Remove.
10787 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
10788 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
10789 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
10790 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
10791 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
10792 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
10793 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
10795 * config/i386/i386-modes.def (BND32, BND64): Remove.
10796 * config/i386/i386.c (dbx_register_map): Remove bound registers.
10797 (dbx64_register_map): Ditto.
10798 (svr4_dbx_register_map): Ditto.
10799 (indirect_thunk_bnd_needed): Remove.
10800 (indirect_thunks_bnd_used): Ditto.
10801 (indirect_return_bnd_needed): Ditto.
10802 (indirect_return_via_cx_bnd): Ditto.
10803 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
10804 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
10805 (output_indirect_thunk): Ditto. Remove need_prefix argument.
10806 (output_indirect_thunk_function): Remove handling of
10807 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
10808 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
10809 (ix86_save_reg): Remove handling of crtl->return_bnd.
10810 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
10811 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
10812 and UNSPEC_BNDLX_ADDR.
10813 (ix86_output_indirect_branch_via_reg): Remove handling of
10814 indirect_thunk_prefix_bnd.
10815 (ix86_output_indirect_branch_via_push): Ditto.
10816 (ix86_output_function_return): Ditto.
10817 (ix86_output_indirect_function_return): Ditto.
10818 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
10819 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
10820 (CALL_USED_REGISTERS): Ditto.
10821 (REG_ALLOC_ORDER): Update for removal of bound registers.
10822 (HI_REGISTER_NAMES): Ditto.
10823 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
10824 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
10825 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
10826 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
10827 (FIRST_PSEUDO_REG): Update.
10828 (BND): Remove mode iterator.
10829 * config/i386/predicates.md (bnd_mem_operator): Remove.
10831 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10833 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
10836 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10838 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
10839 the case in which the permute needs only a single element and
10840 repeats for every vector of the result. Extend that case to
10841 handle variable-length vectors.
10842 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
10844 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
10847 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
10848 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
10850 2018-08-24 Richard Biener <rguenther@suse.de>
10852 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
10853 bb_flags_allocated members.
10854 (auto_flag): New RAII class for allocating flags.
10855 (auto_edge_flag): New RAII class for allocating edge flags.
10856 (auto_bb_flag): New RAII class for allocating bb flags.
10857 * cfgloop.c (verify_loop_structure): Allocate temporary edge
10859 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
10860 in favor of temporarily allocated BB flag.
10861 * hsa-brig.c: Re-order includes.
10862 * hsa-dump.c: Likewise.
10863 * hsa-regalloc.c: Likewise.
10864 * print-rtl.c: Likewise.
10865 * profile-count.c: Likewise.
10867 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
10870 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
10873 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10876 * wide-int-range.cc (wide_int_range_div): Do not ignore result
10877 from wide_int_range_multiplicative_op.
10879 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10881 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
10884 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
10886 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
10889 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
10891 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
10892 full GENERIC expression used for replacement.
10894 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
10896 * tree-vrp.c (abs_extent_range): Remove.
10897 (extract_range_into_wide_ints): Pass wide ints by reference.
10898 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
10899 Pass wide ints by reference in all calls to
10900 extract_range_into_wide_ints.
10901 * wide-int-range.cc (wide_int_range_div): New.
10902 * wide-int-range.h (wide_int_range_div): New.
10903 (wide_int_range_includes_zero_p): New.
10904 (wide_int_range_zero_p): New.
10906 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
10908 * config/aarch64/aarch64.md (arches): New enum.
10909 (arch): New enum attr.
10910 (arch_enabled): New attr.
10911 (enabled): Now uses arch_enabled only.
10912 (simd, sve, fp16): Removed attribute.
10913 (fp): Attr now defined in terms of 'arch'.
10914 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
10915 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
10916 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
10917 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
10918 attributes into 'arch'.
10919 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
10920 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
10921 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
10922 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
10923 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
10924 'simd' attribute into 'arch'.
10925 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
10926 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
10927 Convert use of 'fp' attribute to 'arch'.
10928 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
10929 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
10931 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
10932 (different modes) Merge 'fp' and 'simd' into 'arch'.
10933 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
10934 'simd' into 'arch'.
10936 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
10938 PR rtl-optimization/87026
10939 * expmed.c (canonicalize_comparison): If we can no longer create
10940 pseudoregisters, don't.
10942 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
10945 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
10947 * config/arm/arm.c (speculation_barrier_libfunc): New static
10949 (arm_init_libfuncs): Initialize it.
10950 (arm_emit_speculation_barrier): New function.
10951 * config/arm/arm.md (speculation_barrier): Call
10952 arm_emit_speculation_barrier for architectures that do not have
10954 (speculation_barrier_insn): Only match on Armv7 or later.
10956 2018-08-23 Richard Biener <rguenther@suse.de>
10958 PR middle-end/87024
10959 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
10962 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
10964 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
10965 of single-vector TBLs.
10966 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
10967 one input is given.
10969 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
10972 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
10973 aarch64_evpc_tbl guard.
10975 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
10977 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
10980 2018-08-22 Martin Sebor <msebor@redhat.com>
10982 PR middle-end/87052
10983 * tree-pretty-print.c (pretty_print_string): Add argument.
10984 (dump_generic_node): Call to pretty_print_string with string size.
10986 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
10988 PR rtl-optimization/86771
10989 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
10990 of two SETs into those two SETs, one to be placed at i2, if that SETs
10991 destination is modified between i2 and i3.
10993 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10995 PR tree-optimization/86725
10996 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
10998 (vect_analyze_scalar_cycles_1): Check it.
11000 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11002 PR tree-optimization/86725
11003 * tree-vect-loop.c (vect_is_simple_reduction): When treating
11004 an outer loop phi as a double reduction, make sure that the
11005 single user of the phi result is an inner loop phi.
11007 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11009 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
11010 grouped stores with gaps to a strided group.
11012 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
11014 * tree-vect-stmts.c (get_group_load_store_type)
11015 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
11016 first statement in a group.
11018 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11020 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
11021 the sequence used in gcc/gcc.c.
11023 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11026 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
11029 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11031 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
11032 Darwin10-specific unwinder-shim.
11033 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
11034 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
11035 New to cater for Darwin10 Rosetta.
11037 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11039 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
11040 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
11042 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
11047 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
11048 (dwarf2out_switch_text_section): Generate a local label for the
11049 second function sub-section and apply it as the second FDE start
11051 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
11052 second sub-section start.
11054 2018-08-22 Richard Biener <rguenther@suse.de>
11056 PR tree-optimization/86988
11057 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
11059 2018-08-22 Richard Biener <rguenther@suse.de>
11061 PR tree-optimization/86945
11062 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
11064 2018-08-22 Alexandre Oliva <oliva@adacore.com>
11066 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
11067 a comment about how uses of r2 for .sdata2 come about.
11069 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
11071 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
11073 2018-08-21 Marek Polacek <polacek@redhat.com>
11075 PR c++/86981, Implement -Wpessimizing-move.
11076 * doc/invoke.texi: Document -Wpessimizing-move.
11078 2018-08-21 Jan Hubicka <jh@suse.cz>
11080 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
11081 * tree.h (is_redundant_typedef): Remove.
11082 * dwarf2out.c (is_redundant_typedef): Turn into static function.
11084 2018-08-21 Jan Hubicka <jh@suse.cz>
11086 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
11089 2018-08-21 Tamar Christina <tamar.christina@arm.com>
11091 * expmed.c (extract_low_bits): Reject invalid subregs early.
11093 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
11095 PR middle-end/86121
11096 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
11099 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11101 * config/vxworks.h: Guard vxworks_asm_out_constructor and
11102 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
11103 * config/vxworks.c: Likewise.
11105 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11107 * config/vxworks.c: Set targetm.have_ctors_dtors
11108 if HAVE_INITFINI_ARRAY_SUPPORT.
11109 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
11110 if HAVE_INITFINI_ARRAY_SUPPORT.
11112 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
11114 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
11115 default search path for VxWorks < 7.
11117 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
11119 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
11120 (get_format_string): Refer to c_getstr.
11122 2018-08-21 Tom de Vries <tdevries@suse.de>
11124 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
11125 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
11126 (debuginfo_early_stop): Declare.
11127 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
11128 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
11129 (debuginfo_early_stop): New function.
11130 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
11131 and debuginfo_early_stop.
11132 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
11133 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
11134 (general_init): Call debuginfo_early_init.
11135 (finalize): Call debuginfo_fini.
11136 (do_compile): Call debuginfo_init.
11137 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
11138 -fdump-early-debug.
11139 (@item -fdump-debug, @item -fdump-earlydebug): Add.
11141 2018-08-21 Tom de Vries <tdevries@suse.de>
11143 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
11144 flag_dump_noaddr and flag_dump_unnumbered.
11146 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
11148 * wide-int-range.cc (wide_int_range_abs): New.
11149 (wide_int_range_order_set): Rename from wide_int_range_min_max.
11150 * wide-int-range.h (wide_int_range_abs): New.
11151 (wide_int_range_min_max): New.
11152 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
11153 case to call wide_int_range_abs.
11154 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
11155 (extract_range_from_abs_expr): Delete.
11157 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
11160 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
11161 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
11164 2018-08-20 Nathan Sidwell <nathan@acm.org>
11165 Jeff Law <law@redhat.com>
11167 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
11168 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
11170 2018-08-20 David Malcolm <dmalcolm@redhat.com>
11173 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
11174 (decl_attributes): Likewise.
11175 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
11177 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11178 * diagnostic-core.h (class auto_diagnostic_group): New class.
11179 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
11180 (diagnostic_report_diagnostic): Handle the first diagnostics within
11182 (emit_diagnostic): Add auto_diagnostic_group instance.
11183 (inform): Likewise.
11184 (inform_n): Likewise.
11185 (warning): Likewise.
11186 (warning_at): Likewise.
11187 (warning_n): Likewise.
11188 (pedwarn): Likewise.
11189 (permerror): Likewise.
11191 (error_n): Likewise.
11192 (error_at): Likewise.
11194 (fatal_error): Likewise.
11195 (internal_error): Likewise.
11196 (internal_error_no_backtrace): Likewise.
11197 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
11198 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
11199 * diagnostic.h (struct diagnostic_context): Add fields
11200 "diagnostic_group_nesting_depth",
11201 "diagnostic_group_emission_count", "begin_group_cb",
11203 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
11204 Add auto_diagnostic_group instance(s).
11205 (find_explicit_erroneous_behavior): Likewise.
11206 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
11207 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
11208 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
11209 (gimplify_va_arg_expr): Likewise.
11210 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
11211 (HSA_SORRY_AT): Likewise.
11212 * ipa-devirt.c (compare_virtual_tables): Likewise.
11213 (warn_odr): Likewise.
11214 * multiple_target.c (expand_target_clones): Likewise.
11215 * opts-common.c (cmdline_handle_error): Likewise.
11216 * reginfo.c (globalize_reg): Likewise.
11217 * substring-locations.c (format_warning_n_va): Likewise.
11218 * tree-inline.c (expand_call_inline): Likewise.
11219 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
11220 * tree-ssa-loop-niter.c
11221 (do_warn_aggressive_loop_optimizations): Likewise.
11222 * tree-ssa-uninit.c (warn_uninit): Likewise.
11223 * tree.c (warn_deprecated_use): Likewise.
11225 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
11228 * config/i386/i386.md (eh_return): Always update EH return
11229 address in word_mode.
11231 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
11233 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
11234 TARGET_SPLIT_COMPLEX_ARG is defined.
11236 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11238 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
11240 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
11243 * expr.c (expand_assignment): Assert that bitpos is positive.
11244 (store_field): Likewise
11245 (expand_expr_real_1): Make sure that bitpos is positive.
11246 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
11249 2018-08-20 Nathan Sidwell <nathan@acm.org>
11251 * Makefile.in (CPP_ID_DATA_H): Delete.
11252 (CPP_INTERNAL_H): Don't add it.
11253 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
11254 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
11256 2018-08-20 Richard Biener <rguenther@suse.de>
11258 PR tree-optimization/78655
11259 * tree-vrp.c (extract_range_from_binary_expr_1): Make
11260 pointer + offset nonnull if either operand is nonnull work.
11262 2018-08-20 Tom de Vries <tdevries@suse.de>
11264 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
11265 unless the referenced die describes the added property using
11266 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
11267 Otherwise, add a DW_AT_location to the referenced die.
11269 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
11272 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
11273 register_operand when calling ix86_set_reg_reg_cost.
11274 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
11275 Set *total to 0 for operands that satisfy x86_64_immediate_operand
11276 predicate and to 1 otherwise.
11278 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
11280 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
11281 emit a diagnostic that it is not supported and reset the option.
11282 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
11283 supported and consume the option. (ASM_FINAL_SPEC): New.
11285 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
11287 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
11290 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
11292 C-SKY port: Documentation
11294 * doc/extend.texi (C-SKY Function Attributes): New section.
11295 * doc/invoke.texi (Option Summary): Add C-SKY options.
11296 (C-SKY Options): New section.
11297 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
11299 2018-08-17 Jojo <jijie_rong@c-sky.com>
11300 Huibin Wang <huibin_wang@c-sky.com>
11301 Sandra Loosemore <sandra@codesourcery.com>
11302 Chung-Lin Tang <cltang@codesourcery.com>
11304 C-SKY port: Backend implementation
11306 * config/csky/*: New.
11307 * common/config/csky/*: New.
11309 2018-08-17 Jojo <jijie_rong@c-sky.com>
11310 Huibin Wang <huibin_wang@c-sky.com>
11311 Sandra Loosemore <sandra@codesourcery.com>
11312 Chung-Lin Tang <cltang@codesourcery.com>
11313 Andrew Jenner <andrew@codesourcery.com>
11315 C-SKY port: Configury
11317 * config.gcc (csky-*-*): New.
11318 * configure.ac: Add csky to targets for dwarf2 debug_line support.
11319 * configure: Regenerated.
11321 2018-08-17 David Malcolm <dmalcolm@redhat.com>
11323 * dump-context.h: Include "dumpfile.h".
11324 (dump_context::dump_printf_va): Convert final param from va_list
11325 to va_list *. Convert from ATTRIBUTE_PRINTF to
11326 ATTRIBUTE_GCC_DUMP_PRINTF.
11327 (dump_context::dump_printf_loc_va): Likewise.
11328 * dumpfile.c: Include "stringpool.h".
11329 (make_item_for_dump_printf_va): Delete.
11330 (make_item_for_dump_printf): Delete.
11331 (class dump_pretty_printer): New class.
11332 (dump_pretty_printer::dump_pretty_printer): New ctor.
11333 (dump_pretty_printer::emit_items): New member function.
11334 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
11336 (dump_pretty_printer::emit_item): New member function.
11337 (dump_pretty_printer::stash_item): New member function.
11338 (dump_pretty_printer::format_decoder_cb): New member function.
11339 (dump_pretty_printer::decode_format): New member function.
11340 (dump_context::dump_printf_va): Reimplement in terms of
11341 dump_pretty_printer.
11342 (dump_context::dump_printf_loc_va): Convert final param from va_list
11344 (dump_context::begin_scope): Reimplement call to
11345 make_item_for_dump_printf.
11346 (dump_printf): Update for change to dump_printf_va.
11347 (dump_printf_loc): Likewise.
11348 (selftest::test_capture_of_dump_calls): Convert "stmt" from
11349 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
11350 with %T, %E, and %G.
11351 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
11352 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
11353 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
11354 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
11355 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
11356 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
11357 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
11358 within a dump_printf_loc call to "%wu".
11359 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
11360 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
11361 missing space after "=".
11362 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
11363 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
11364 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
11365 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
11366 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
11367 duplicate "vectorized" from message.
11369 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
11371 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
11372 polyNxK_t element's TYPE_STRING_FLAG.
11374 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
11376 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
11377 (they were unnamed before). Fix comments.
11379 2018-08-17 Nathan Sidwell <nathan@acm.org>
11381 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
11383 2018-08-17 Richard Biener <rguenther@suse.de>
11385 PR tree-optimization/86841
11386 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
11388 2018-08-17 Martin Liska <mliska@suse.cz>
11390 * common.opt: Remove Warn, Init and Report for options with
11391 Ignore/Deprecated flag. Warning is done automatically for
11393 * config/i386/i386.opt: Likewise.
11394 * config/ia64/ia64.opt: Likewise.
11395 * config/rs6000/rs6000.opt: Likewise.
11396 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
11397 Remove usage of flag_check_pointer_bounds.
11398 * lto-wrapper.c (merge_and_complain): Do not handle
11399 OPT_fcheck_pointer_bounds.
11400 (append_compiler_options): Likewise.
11401 * opt-functions.awk: Do not handle Deprecated.
11402 * optc-gen.awk: Check that Var, Report and Init are not
11403 used for an option with Ignore/Deprecated flag.
11404 * opts-common.c (decode_cmdline_option): Do not report
11406 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
11408 * opts.h (struct cl_option): Remove cl_deprecated flag.
11409 (CL_ERR_DEPRECATED): Remove error enum value.
11411 2018-08-17 Richard Biener <rguenther@suse.de>
11413 PR middle-end/86505
11414 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
11415 across a va-arg-pack using call adjust its return value accordingly.
11417 2018-08-16 Martin Sebor <msebor@redhat.com>
11419 PR tree-optimization/86853
11420 * gimple-ssa-sprintf.c (struct format_result): Rename member.
11421 (struct fmtresult): Add member and initialize it in ctors.
11422 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
11423 (format_string): Handle %S the same as %ls. Set MAYFAIL.
11424 (format_directive): Set POSUNDER4K when MAYFAIL is set.
11425 (parse_directive): Handle %C same as %c.
11426 (sprintf_dom_walker::compute_format_length): Adjust.
11427 (is_call_safe): Adjust.
11429 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
11431 * builtins.c (c_strlen): Add new parameter eltsize. Use it
11432 for determining how to count the elements.
11433 * builtins.h (c_strlen): Adjust prototype.
11434 * expr.c (string_constant): Add new parameter mem_size.
11435 Set *mem_size appropriately.
11436 * expr.h (string_constant): Adjust protoype.
11437 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
11438 * gimple-fold.h (get_range_strlen): Adjust prototype.
11439 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
11440 (format_string): Call get_string_length with eltsize.
11442 2018-08-16 David Malcolm <dmalcolm@redhat.com>
11444 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
11445 to emit the span, rather than setting it as the prefix.
11447 2018-08-16 David Malcolm <dmalcolm@redhat.com>
11449 * diagnostic-show-locus.c (layout::start_annotation_line): Add
11450 "margin_char" parameter, defaulting to space. Use it in place
11451 of pp_space for the initial part of the margin.
11452 (layout::print_leading_fixits): Use '+' when filling the margin
11453 of line-insertion fix-it hints.
11455 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11457 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
11460 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11462 * config/rs6000/altivec.md: Don't set length attribute to the default
11464 * config/rs6000/darwin.md: Ditto.
11465 * config/rs6000/dfp.md: Ditto.
11466 * config/rs6000/htm.md: Ditto.
11467 * config/rs6000/rs6000.md: Ditto.
11468 * config/rs6000/sync.md: Ditto.
11469 * config/rs6000/vsx.md: Ditto.
11471 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11473 * config/rs6000/altivec.md: Don't set length attribute to the default
11474 value, for branch instructions.
11475 * config/rs6000/darwin.md: Ditto.
11476 * config/rs6000/rs6000.md: Ditto.
11478 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
11480 * config/rs6000/rs6000.md (length): Always define as const_int 4.
11481 (unnamed conditional branch define_insn): Set length to 4 or 8
11482 depending on offset.
11483 (<bd>_<mode>): Similar, for alternative 0.
11484 (<bd>tf_<mode>): Ditto.
11486 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11488 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
11490 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
11492 * doc/rtl.texi: Replace old RTX class names with new names.
11495 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
11497 * expmed.h (canonicalize_comparison): New declaration.
11498 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
11499 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
11500 * optabs.c (prepare_cmp_insn): Likewise.
11501 * rtl.h (unsigned_condition_p): New function which checks if a
11502 comparison operator is unsigned.
11504 2018-08-16 Nathan Sidwell <nathan@acm.org>
11506 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
11507 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
11509 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11512 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
11513 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
11514 (mov<mov>): ..this and enable unconditionally.
11516 2018-08-16 Tamar Christina <tamar.christina@arm.com>
11518 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
11520 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
11522 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
11523 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
11524 "Common" with "Target".
11526 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
11528 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
11529 * doc/invoke.texi (mmitigate-rop): Remove.
11530 * config/i386/i386.c: Do not include "regrename.h".
11531 (ix86_rop_should_change_byte_p, reg_encoded_number)
11532 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
11534 (ix86_reorg): Remove call to ix86_mitigate_rop.
11535 * config/i386/i386.md (attr "modrm_class"): Remove.
11536 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
11537 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
11538 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
11540 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
11542 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
11543 allow folding of mergeh() and mergel() for the float and double types.
11544 (fold_mergehl_helper): Rework to handle building a permute tree
11547 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
11549 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
11552 2018-08-15 David Malcolm <dmalcolm@redhat.com>
11554 * common.opt (fdiagnostics-show-labels): New option.
11555 * diagnostic-show-locus.c (class layout_range): Add field
11557 (class layout): Add field "m_show_labels_p".
11558 (layout_range::layout_range): Add param "label" and use it to
11559 initialize m_label.
11560 (make_range): Pass in NULL for new "label" param of layout_range's
11562 (layout::layout): Initialize m_show_labels_p.
11563 (layout::maybe_add_location_range): Pass in loc_range->m_label
11564 when constructing layout_range instances.
11565 (struct line_label): New struct.
11566 (layout::print_any_labels): New member function.
11567 (layout::print_line): Call it if label-printing is enabled.
11568 (selftest::test_one_liner_labels): New test.
11569 (selftest::test_diagnostic_show_locus_one_liner): Call it.
11570 * diagnostic.c (diagnostic_initialize): Initialize
11571 context->show_labels_p.
11572 * diagnostic.h (struct diagnostic_context): Add field
11574 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11575 -fno-diagnostics-show-labels.
11576 * dwarf2out.c (gen_producer_string): Add
11577 OPT_fdiagnostics_show_labels to the ignored options.
11578 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
11580 (gcc_rich_location::maybe_add_expr): Likewise.
11581 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
11582 label" param, defaulting to NULL.
11583 (gcc_rich_location::add_expr): Add "label" param.
11584 (gcc_rich_location::maybe_add_expr): Likewise.
11585 (class text_range_label): New class.
11586 (class range_label_for_type_mismatch): New class.
11587 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
11588 of format_warning_va.
11589 (fmtwarn_n): Likewise for new params of format_warning_n_va.
11590 * lto-wrapper.c (merge_and_complain): Add
11591 OPT_fdiagnostics_show_labels to the "pick one setting" options.
11592 (append_compiler_options): Likewise to the dropped options.
11593 (append_diag_options): Likewise to the passed-on options.
11594 * opts.c (common_handle_option): Handle the new option.
11595 * selftest-diagnostic.c
11596 (test_diagnostic_context::test_diagnostic_context): Enable
11598 * substring-locations.c: Include "gcc-rich-location.h".
11599 (format_warning_n_va): Add "fmt_label" and "param_label" params
11600 and use them as appropriate.
11601 (format_warning_va): Add "fmt_label" and "param_label" params,
11602 passing them on to format_warning_n_va.
11603 (format_warning_at_substring): Likewise.
11604 (format_warning_at_substring_n): Likewise.
11605 * substring-locations.h (format_warning_va): Add "fmt_label" and
11606 "param_label" params.
11607 (format_warning_n_va): Likewise.
11608 (format_warning_at_substring): Likewise.
11609 (format_warning_at_substring_n): Likewise.
11610 * toplev.c (general_init): Initialize global_dc->show_labels_p.
11612 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
11615 * builtins.c (expand_builtin_memcmp): Do not expand the call
11616 when overflow is detected.
11618 2018-08-15 Martin Sebor <msebor@redhat.com>
11620 PR tree-optimization/71625
11621 * config/aarch64/aarch64-builtins.c
11622 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
11624 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
11626 * config/s390/s390.c (s390_reorg): Remove loop.
11628 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11631 (darwin_function_switched_text_sections): Delete.
11632 * gcc/config/darwin.h
11633 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
11635 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
11638 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
11639 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
11640 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
11642 2018-08-15 Martin Liska <mliska@suse.cz>
11644 PR tree-optimization/86925
11645 * predict.c (expr_expected_value_1): When taking
11646 later predictor, assign also probability.
11647 Use fold_build2_initializer_loc in order to fold
11648 the expression in -frounding-math.
11650 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
11652 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
11654 (expand_vec_perm_1): Try the new method.
11656 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
11659 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
11660 Check whether lra_live_max_point is 0 before dividing.
11662 2018-08-14 Martin Sebor <msebor@redhat.com>
11664 PR tree-optimization/86650
11665 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
11666 (vrp_prop::check_mem_ref): Same.
11668 2018-08-13 Liu Hao <lh_mouse@126.com>
11670 * pretty-print.c (eat_esc_sequence): Swap the foreground and
11671 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
11672 and clear it thereafter, as it only works for DBCS.
11674 2018-08-13 Liu Hao <lh_mouse@126.com>
11676 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
11677 handle returned by _get_osf_handle().
11679 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11681 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
11682 for folding vec_perm.
11684 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11686 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
11687 Add support for gimple-folding of vec_pack() and vec_unpack()
11690 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11692 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
11693 vec_xst variants to the list.
11694 (rs6000_gimple_fold_builtin): Add support for folding unaligned
11695 vector loads and stores.
11697 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
11699 * config.gcc (rs6000-ibm-aix4.x): Delete.
11700 (rs6000-ibm-aix5.1): Delete.
11701 (rs6000-ibm-aix5.2): Delete.
11702 (rs6000-ibm-aix5.3): Delete.
11703 * config/rs6000/aix43.h: Delete.
11704 * config/rs6000/aix51.h: Delete.
11705 * config/rs6000/aix52.h: Delete.
11706 * config/rs6000/t-aix43: Delete.
11708 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
11710 * config/s390/s390.c (s390_decompose_constant_pool_ref):
11712 (s390_decompose_address): Factor out constant pool ref
11715 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11717 * config/nds32/nds32-predicates.c
11718 (nds32_can_use_bclr_p): Change return type as bool.
11719 (nds32_can_use_bset_p): Ditto.
11720 (nds32_can_use_btgl_p): Ditto.
11721 (nds32_can_use_bitci_p): Ditto.
11722 * config/nds32/nds32-protos.h
11723 (nds32_can_use_bclr_p): Change declaration.
11724 (nds32_can_use_bset_p): Ditto.
11725 (nds32_can_use_btgl_p): Ditto.
11726 (nds32_can_use_bitci_p): Ditto.
11728 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11730 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
11731 Support -msched-prolog-epilog option.
11732 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
11734 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11736 * common/config/nds32/nds32-common.c
11737 (nds32_option_optimization_table): Enalbe -malways-align.
11739 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
11741 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
11743 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
11744 OPT_misr_secure_ case.
11745 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
11746 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
11747 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
11749 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
11750 (nds32_isr_info): New field security_level.
11751 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
11752 * config/nds32/nds32.md (return_internal): Consider critical attribute.
11753 * config/nds32/nds32.opt (misr-secure): New option.
11754 * config/nds32/nds32_init.inc: New file.
11755 * config/nds32/nds32_isr.h: New file.
11757 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
11759 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
11760 Update comment for atomic instructions.
11761 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
11762 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
11764 (atomic_loaddi): Revise fence expansion to only emit fence prior to
11765 load for __ATOMIC_SEQ_CST model.
11766 (atomic_loaddi_1): Remove float register target.
11767 (atomic_storedi): Handle CONST_INT values.
11768 (atomic_storedi_1): Remove float register source. Add special case
11770 (memory_barrier): New expander and insn.
11772 2018-08-11 Jakub Jelinek <jakub@redhat.com>
11774 PR tree-optimization/86835
11775 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
11776 new_stmt after def_gsi, make sure to insert new_square_stmt after
11777 that stmt, not 2 stmts before it.
11779 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
11782 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
11785 2018-08-10 Martin Liska <mliska@suse.cz>
11788 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
11790 * builtins.c (expand_builtin_expect_with_probability):
11792 (expand_builtin_expect_with_probability): New function.
11793 (build_builtin_expect_predicate): Add new argumnet probability
11794 for BUILT_IN_EXPECT_WITH_PROBABILITY.
11795 (fold_builtin_expect):
11798 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
11799 * builtins.h (fold_builtin_expect): Set new argument.
11800 * doc/extend.texi: Document __builtin_expect_with_probability.
11801 * doc/invoke.texi: Likewise.
11802 * gimple-fold.c (gimple_fold_call): Pass new argument.
11803 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
11804 also BUILT_IN_EXPECT_WITH_PROBABILITY.
11805 * predict.c (get_predictor_value): New function.
11806 (expr_expected_value): Add new argument probability. Assume
11807 that predictor and probability are always non-null.
11808 (expr_expected_value_1): Likewise. For __builtin_expect and
11809 __builtin_expect_with_probability set probability. Handle
11810 combination in binary expressions.
11811 (tree_predict_by_opcode): Simplify code by simply calling
11812 get_predictor_value.
11813 (pass_strip_predict_hints::execute): Add handling of
11814 BUILT_IN_EXPECT_WITH_PROBABILITY.
11815 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
11817 * tree.h (DECL_BUILT_IN_P): New function.
11819 2018-08-10 Martin Liska <mliska@suse.cz>
11821 PR tree-optimization/85799
11822 * passes.def: Add argument for pass_strip_predict_hints.
11823 * predict.c (class pass_strip_predict_hints): Add new argument
11825 (strip_predictor_early): New function.
11826 (pass_strip_predict_hints::execute): Call the function to
11828 (strip_predict_hints): New function.
11829 * predict.def: Fix comment.
11831 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
11833 * Makefile.in: Clarify which tm.texi to copy over to assert the
11834 right to grant a GFDL license for all.
11836 2018-08-09 Jeff Law <law@redhat.com>
11838 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
11841 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
11843 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
11846 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11848 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
11849 reductions for variable-length vectors.
11851 2018-08-09 David Malcolm <dmalcolm@redhat.com>
11854 * common.opt (fdiagnostics-show-line-numbers): New option.
11855 * diagnostic-show-locus.c (class layout): Add fields
11856 "m_show_line_numbers_p" and "m_linenum_width";
11857 (num_digits): New function.
11858 (test_num_digits): New function.
11859 (layout::layout): Initialize new fields. Update m_x_offset
11860 logic to handle any left margin.
11861 (layout::print_source_line): Print line number when requested.
11862 (layout::start_annotation_line): New member function.
11863 (layout::print_annotation_line): Call it.
11864 (layout::print_leading_fixits): Likewise.
11865 (layout::print_trailing_fixits): Likewise. Update calls to
11866 move_to_column for new parameter.
11867 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
11868 it to potentially call start_annotation_line.
11869 (layout::show_ruler): Call start_annotation_line.
11870 (selftest::test_line_numbers_multiline_range): New selftest.
11871 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
11872 and selftest::test_line_numbers_multiline_range.
11873 * diagnostic.c (diagnostic_initialize): Initialize
11874 show_line_numbers_p.
11875 * diagnostic.h (struct diagnostic_context): Add field
11876 "show_line_numbers_p".
11877 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11878 -fno-diagnostics-show-line-numbers.
11879 * dwarf2out.c (gen_producer_string): Add
11880 OPT_fdiagnostics_show_line_numbers to the ignored options.
11881 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
11882 one setting" options.
11883 (append_compiler_options): Likewise to the dropped options.
11884 (append_diag_options): Likewise to the passed-on options.
11885 * opts.c (common_handle_option): Handle the new option.
11886 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
11888 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
11890 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
11891 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
11892 third argument of type "const signed char" to descriptions of
11893 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
11894 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
11895 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
11896 __builtin_bcdsub_ov functions.
11898 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11900 PR tree-optimization/86858
11901 * tree-vect-loop.c (vect_is_simple_reduction): Restore
11902 flow_bb_inside_loop_p calls.
11904 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11906 PR tree-optimization/86871
11907 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
11908 instead of gimple_assign_lhs.
11910 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
11913 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
11914 register constraint to operand 0.
11915 (add<mode>3_carryinC): Likewise.
11916 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
11918 2018-08-09 Martin Liska <mliska@suse.cz>
11921 * common.opt: Remove extra line.
11923 2018-08-09 Martin Liska <mliska@suse.cz>
11925 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
11926 at the end of a line, make first letter capital and end up
11927 a sentence with a dot.
11928 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
11929 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
11930 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
11931 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
11932 (PARAM_MAX_ISL_OPERATIONS): Likewise.
11933 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
11934 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
11935 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
11936 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
11937 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
11938 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
11939 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
11940 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
11941 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
11942 (PARAM_TREE_REASSOC_WIDTH): Likewise.
11943 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
11944 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
11945 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
11947 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
11950 * config/s390/s390.c (s390_option_override_internal): Reduce the
11951 stack-clash-protection-probe-interval param if it would be too big
11954 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
11957 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
11958 * config/m68k/m68k.c (handle_move_double): Don't call
11959 m68k_final_prescan_insn.
11960 (m68k_adjust_decorated_operand): Renamed from
11961 m68k_final_prescan_insn, remove first and third operand and
11963 (print_operand): Call it.
11964 (print_operand_address): Call it.
11966 2018-08-08 Nathan Sidwell <nathan@acm.org>
11968 * diagnostic.c (diagnostic_report_current_module): Use
11969 linemap_included_from & linemap_included_from_linemap.
11971 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
11973 * config/aarch64/aarch64-cores.def: Add phecda core.
11974 * config/aarch64/aarch64-tune.md: Regenerate.
11975 * doc/invoke.texi: Add phecda core.
11977 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
11980 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
11982 * config/s390/s390.md ("movti"): Add more alternatives for
11983 constant to GPR copies.
11985 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
11987 * config/s390/s390.c: Fix whitespace damage throughout the file.
11988 * config/s390/s390.h: Likewise.
11989 * config/s390/tpf.h: Likewise.
11991 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
11993 * config/s390/s390.c (s390_loadrelative_operand_p):
11994 Remove TARGET_CPU_ZARCH usages.
11995 (s390_rtx_costs): Likewise.
11996 (s390_legitimate_constant_p): Likewise.
11997 (s390_cannot_force_const_mem): Likewise.
11998 (legitimate_reload_constant_p): Likewise.
11999 (s390_preferred_reload_class): Likewise.
12000 (legitimize_pic_address): Likewise.
12001 (legitimize_tls_address): Likewise.
12002 (s390_split_branches): Removed.
12003 (s390_add_execute): Removed.
12004 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
12005 (s390_mainpool_start): Likewise.
12006 (s390_mainpool_finish): Likewise.
12007 (s390_mainpool_cancel): Removed.
12008 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
12009 (s390_chunkify_cancel): Likewise.
12010 (s390_return_addr_rtx): Likewise.
12011 (s390_register_info): Remove split_branches_pending_p uages.
12012 (s390_optimize_register_info): Likewise.
12013 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
12014 split_branches_pending_p usages.
12015 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
12016 (s390_load_got): Likewise.
12017 (s390_expand_split_stack_prologue): Likewise.
12018 (output_asm_nops): Likewise.
12019 (s390_function_profiler): Likewise.
12020 (s390_emit_call): Likewise.
12021 (s390_conditional_register_usage): Likewise.
12022 (s390_optimize_prologue): Likewise.
12023 (s390_reorg): Remove TARGET_CPU_ZARCH and
12024 split_branches_pending_p usages.
12025 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
12027 (s390_output_indirect_thunk_function): Likewise.
12028 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
12029 (TARGET_CPU_ZARCH_P): Removed.
12030 (struct machine_function): Remove split_branches_pending_p.
12031 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
12033 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
12035 * common/config/s390/s390-common.c (processor_flags_table):
12037 * config.gcc: Remove with_arch/with_tune support.
12038 * config/s390/2064.md: Remove cpu attribute comparisons.
12039 * config/s390/driver-native.c (s390_host_detect_local_cpu):
12041 * config/s390/linux.h (ASM_SPEC):
12042 Remove -march support.
12043 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
12044 Use a table to get an arch level.
12045 * config/s390/s390-opts.h (enum processor_type):
12046 Remove enum values.
12047 * config/s390/s390.c
12048 (processor_table): Remove entries, add arch_level values.
12049 (s390_issue_rate): Remove cases.
12050 (s390_option_override): Adjust
12051 s390_option_override_internal() call.
12052 (s390_option_override_internal): Remove deprecation warning.
12053 (s390_valid_target_attribute_tree): Adjust
12054 s390_option_override_internal() call.
12055 * config/s390/s390.h (struct s390_processor):
12056 Share with s390-c.c, add arch_level field.
12057 * config/s390/s390.md:
12058 Remove occurrences in cpu attribute.
12059 * config/s390/s390.opt: Remove -march/-mtune support.
12060 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
12061 * doc/invoke.texi: Remove deprecation warning.
12063 2018-08-08 Luis Machado <luis.machado@linaro.org>
12065 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
12067 (qdf24xx_tunings): Set vector cost structure to
12068 qdf24xx_vector_cost.
12070 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
12071 <register_sextend>: Set to 3.
12073 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
12076 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
12077 * config/aarch64/aarch64-simd.md
12078 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
12079 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
12080 (aarch64_frecpx<mode>): ...this new pattern.
12081 * config/aarch64/aarch64-simd-builtins.def: Remove comment
12082 about aarch64_frecp<FRECP:frecp_suffix><mode>.
12084 2018-08-07 Martin Liska <mliska@suse.cz>
12086 PR middle-end/83023
12087 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
12088 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
12089 * predict.def (PRED_MALLOC_NONNULL): New predictor.
12090 * doc/extend.texi: Document that malloc attribute adds
12093 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
12096 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12097 Define to speculation_safe_value_not_needed.
12099 2018-08-06 Jeff Law <law@redhat.com>
12101 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
12102 the vr_values instance to cprop_into_stmt.
12103 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
12104 (cprop_operand): Also query EVRP to determine if OP is a constant.
12106 2018-08-06 Nathan Sidwell <nathan@acm.org>
12108 * diagnostic.c (diagnostic_report_current_module): Reroll
12109 included-at loop. Translate text.
12111 2018-08-06 David Malcolm <dmalcolm@redhat.com>
12113 * function-tests.c (selftest::test_expansion_to_rtl): Call
12114 free_after_compilation.
12116 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12118 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
12120 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
12122 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
12123 loops with memory block operations from getting unrolled.
12125 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
12128 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12129 Define to speculation_safe_value_not_needed.
12131 2018-08-06 Jeff Law <law@redhat.com>
12133 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
12136 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12139 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
12140 with all enabled __intN types.
12142 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
12144 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12146 * alias.c (record_set): Check for clobber high.
12147 * cfgexpand.c (expand_gimple_stmt): Likewise.
12148 * combine-stack-adj.c (single_set_for_csa): Likewise.
12149 * combine.c (find_single_use_1): Likewise.
12150 (set_nonzero_bits_and_sign_copies): Likewise.
12151 (get_combine_src_dest): Likewise.
12152 (is_parallel_of_n_reg_sets): Likewise.
12153 (try_combine): Likewise.
12154 (record_dead_and_set_regs_1): Likewise.
12155 (reg_dead_at_p_1): Likewise.
12156 (reg_dead_at_p): Likewise.
12157 * dce.c (deletable_insn_p): Likewise.
12158 (mark_nonreg_stores_1): Likewise.
12159 (mark_nonreg_stores_2): Likewise.
12160 * df-scan.c (df_find_hard_reg_defs): Likewise.
12161 (df_uses_record): Likewise.
12162 (df_get_call_refs): Likewise.
12163 * dwarf2out.c (mem_loc_descriptor): Likewise.
12164 * haifa-sched.c (haifa_classify_rtx): Likewise.
12165 * ira-build.c (create_insn_allocnos): Likewise.
12166 * ira-costs.c (scan_one_insn): Likewise.
12167 * ira.c (equiv_init_movable_p): Likewise.
12168 (rtx_moveable_p): Likewise.
12169 (interesting_dest_for_shprep): Likewise.
12170 * jump.c (mark_jump_label_1): Likewise.
12171 * postreload-gcse.c (record_opr_changes): Likewise.
12172 * postreload.c (reload_cse_simplify): Likewise.
12173 (struct reg_use): Add source expr.
12174 (reload_combine): Check for clobber high.
12175 (reload_combine_note_use): Likewise.
12176 (reload_cse_move2add): Likewise.
12177 (move2add_note_store): Likewise.
12178 * print-rtl.c (print_pattern): Likewise.
12179 * recog.c (decode_asm_operands): Likewise.
12180 (store_data_bypass_p): Likewise.
12181 (if_test_bypass_p): Likewise.
12182 * regcprop.c (kill_clobbered_value): Likewise.
12183 (kill_set_value): Likewise.
12184 * reginfo.c (reg_scan_mark_refs): Likewise.
12185 * reload1.c (maybe_fix_stack_asms): Likewise.
12186 (eliminate_regs_1): Likewise.
12187 (elimination_effects): Likewise.
12188 (mark_not_eliminable): Likewise.
12189 (scan_paradoxical_subregs): Likewise.
12190 (forget_old_reloads_1): Likewise.
12191 * reorg.c (find_end_label): Likewise.
12192 (try_merge_delay_insns): Likewise.
12193 (redundant_insn): Likewise.
12194 (own_thread_p): Likewise.
12195 (fill_simple_delay_slots): Likewise.
12196 (fill_slots_from_thread): Likewise.
12197 (dbr_schedule): Likewise.
12198 * resource.c (update_live_status): Likewise.
12199 (mark_referenced_resources): Likewise.
12200 (mark_set_resources): Likewise.
12201 * rtl.c (copy_rtx): Likewise.
12202 * rtlanal.c (reg_referenced_p): Likewise.
12203 (single_set_2): Likewise.
12204 (noop_move_p): Likewise.
12205 (note_stores): Likewise.
12206 * sched-deps.c (sched_analyze_reg): Likewise.
12207 (sched_analyze_insn): Likewise.
12209 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12211 * cse.c (invalidate_reg): New function extracted from...
12212 (invalidate): ...here.
12213 (canonicalize_insn): Check for clobber high.
12214 (invalidate_from_clobbers): invalidate clobber highs.
12215 (invalidate_from_sets_and_clobbers): Likewise.
12216 (count_reg_usage): Check for clobber high.
12217 (insn_live_p): Likewise.
12218 * cselib.c (cselib_expand_value_rtx_1):Likewise.
12219 (cselib_invalidate_regno): Check for clobber in setter.
12220 (cselib_invalidate_rtx): Pass through setter.
12221 (cselib_invalidate_rtx_note_stores):
12222 (cselib_process_insn): Check for clobber high.
12223 * cselib.h (cselib_invalidate_rtx): Add operand.
12225 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12227 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
12228 (mark_not_eliminable): Likewise.
12229 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
12230 * lra-lives.c (process_bb_lives): Check for clobber high.
12231 * lra.c (new_insn_reg): Remember clobber highs.
12232 (collect_non_operand_hard_regs): Check for clobber high.
12233 (lra_set_insn_recog_data): Likewise.
12234 (add_regs_to_insn_regno_info): Likewise.
12235 (lra_update_insn_regno_info): Likewise.
12237 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12239 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
12240 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
12242 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12244 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
12245 (copy_insn_1): Likewise.
12246 (gen_hard_reg_clobber_high): New gen function.
12247 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
12248 * genemit.c (gen_exp): Likewise.
12249 (gen_emit_seq): Pass through info.
12250 (gen_insn): Check for CLOBBER_HIGH.
12251 (gen_expand): Pass through info.
12252 (gen_split): Likewise.
12253 (output_add_clobbers): Likewise.
12254 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
12255 (remove_clobbers): Likewise.
12256 * rtl.h (gen_hard_reg_clobber_high): New declaration.
12258 2018-08-06 Alan Hayward <alan.hayward@arm.com>
12260 * doc/rtl.texi (clobber_high): Add.
12261 (parallel): Add in clobber high
12262 * rtl.c (rtl_check_failed_code3): Add function.
12263 * rtl.def (CLOBBER_HIGH): Add expression.
12264 * rtl.h (RTL_CHECKC3): Add macro.
12265 (rtl_check_failed_code3): Add declaration.
12266 (XC3EXP): Add macro.
12268 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
12271 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
12272 cfun->machine->max_used_stack_alignment if needed.
12274 2018-08-04 Martin Sebor <msebor@redhat.com>
12276 PR tree-optimization/86571
12277 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
12280 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
12283 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12286 2018-08-03 Jeff Law <law@redhat.com>
12289 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12290 Define to speculation_safe_value_not_needed.
12292 2018-08-03 David Malcolm <dmalcolm@redhat.com>
12294 * doc/gcov.texi (-x): Remove duplicate "to".
12295 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
12296 (-Wif-not-aligned): Remove duplicate "is".
12297 (-flto): Remove duplicate "the".
12298 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
12299 duplicate "v5.00.b".
12300 (MSP430 Options): Remove duplicate "and" from the description
12301 of "-mgprel-sec=regexp".
12302 (x86 Options): Remove duplicate copies of "vmldLog102" and
12303 vmlsLog104 from description of "-mveclibabi=type".
12305 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
12307 * internal-fn.h (first_commutative_argument): Declare.
12308 * internal-fn.c (first_commutative_argument): New function.
12309 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
12310 restrictions for pattern statements. Use first_commutative_argument
12311 to look for commutative operands in calls to internal functions.
12313 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
12315 * Makefile.in (wide-int-range.o): New.
12316 * tree-vrp.c: Move all the wide_int_* functions to...
12317 * wide-int-range.cc: ...here.
12318 * tree-vrp.h: Move all the wide_int_* prototypes to...
12319 * wide-int-range.h: ...here.
12321 2018-08-03 Tom de Vries <tdevries@suse.de>
12323 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
12325 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
12326 * except.c (output_function_exception_table): Do early exit if
12327 targetm_common.except_unwind_info (&global_options) == UI_NONE.
12329 2018-08-03 Martin Liska <mliska@suse.cz>
12331 * predict.c (dump_prediction): Change to 2 digits
12334 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
12336 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
12338 (falkor_am_1_gtov_gtov): ... a new insn reservation.
12340 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
12342 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
12343 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
12344 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
12345 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
12346 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
12348 2018-08-02 David Malcolm <dmalcolm@redhat.com>
12350 * diagnostic-show-locus.c (diagnostic_show_locus): Use
12351 pp_take_prefix when saving the existing prefix.
12352 * diagnostic.c (diagnostic_append_note): Likewise.
12353 * langhooks.c (lhd_print_error_function): Likewise.
12354 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
12355 param's type. Free the existing prefix.
12356 (pp_take_prefix): New function.
12357 (pretty_printer::pretty_printer): Drop the prefix parameter.
12358 Rename the length parameter to match the comment.
12359 (pretty_printer::~pretty_printer): Free the prefix.
12360 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
12362 (struct pretty_printer): Drop the "const" from "prefix" field's
12363 type and clarify memory management.
12364 (pp_set_prefix): Drop the "const" from the 2nd param.
12365 (pp_take_prefix): New decl.
12367 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
12369 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
12370 for word_mode_ok here instead of passing as argument.
12371 (expand_block_compare): Change select_block_compare_mode() call.
12372 (expand_strncmp_gpr_sequence): New function.
12373 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
12375 2018-08-02 Jeff Law <law@redhat.com>
12378 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12379 Define to speculation_safe_value_not_needed.
12382 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12383 Define to speculation_safe_value_not_needed.
12385 2018-08-02 Tom de Vries <tdevries@suse.de>
12388 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
12389 function. Return UI_TARGET unconditionally.
12390 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
12391 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
12393 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12395 * genemit.c (print_overload_test): Fix typo.
12397 2018-08-02 Richard Biener <rguenther@suse.de>
12399 PR tree-optimization/86816
12400 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
12401 which checks for value availability before querying it.
12402 (gvn_uses_equal): Use it.
12403 (same_succ_hash): Likewise.
12404 (gimple_equal_p): Likewise.
12406 2018-08-02 Nick Clifton <nickc@redhat.com>
12409 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12410 Define to speculation_safe_value_not_needed.
12413 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12414 Define to speculation_safe_value_not_needed.
12417 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12418 Define to speculation_safe_value_not_needed.
12421 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12422 Define to speculation_safe_value_not_needed.
12425 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12426 Define to speculation_safe_value_not_needed.
12429 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12430 Define to speculation_safe_value_not_needed.
12433 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12434 Define to speculation_safe_value_not_needed.
12437 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
12438 Define to speculation_safe_value_not_needed.
12441 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12442 speculation_safe_value_not_needed.
12445 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
12446 to speculation_safe_value_not_needed.
12448 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12450 * doc/md.texi: Expand the documentation of instruction names
12451 to mention port-local uses. Document '@' in pattern names.
12452 * read-md.h (overloaded_instance, overloaded_name): New structs.
12453 (mapping): Declare.
12454 (md_reader::handle_overloaded_name): New member function.
12455 (md_reader::get_overloads): Likewise.
12456 (md_reader::m_first_overload): New member variable.
12457 (md_reader::m_next_overload_ptr): Likewise.
12458 (md_reader::m_overloads_htab): Likewise.
12459 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
12460 m_next_overload_ptr and m_overloads_htab.
12461 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
12462 (get_mode_token, get_code_token, get_int_token): New functions.
12463 (map_attr_string): Add an optional argument that passes back
12464 the associated iterator.
12465 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
12466 (md_reader::handle_overloaded_name, add_overload_instance): New
12468 (apply_iterators): Handle '@' names. Report an error if '@'
12469 is used without iterators.
12470 (initialize_iterators): Initialize the new iterator_group fields.
12471 * genopinit.c (handle_overloaded_code_for)
12472 (handle_overloaded_gen): New functions.
12473 (main): Use them to print declarations of maybe_code_for_* and
12474 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
12475 * genemit.c (print_overload_arguments, print_overload_test)
12476 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
12477 (main): Use it to print definitions of maybe_code_for_* and
12478 maybe_gen_* functions.
12479 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
12480 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
12481 instead of explicit mode checks.
12482 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
12483 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
12484 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
12485 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
12486 (aarch64_expand_compare_and_swap): Likewise
12487 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
12488 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
12489 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
12490 (aarch64_constant_pool_reload_icode): Delete.
12491 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
12492 instead of aarch64_constant_pool_reload_icode. Use
12493 code_for_aarch64_reload_mov instead of explicit mode checks.
12494 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
12495 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
12496 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
12497 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
12498 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
12499 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
12500 (aarch64_atomic_load_op_code): Delete.
12501 (aarch64_emit_atomic_load_op): Likewise.
12502 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
12503 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
12504 instead of aarch64_emit_atomic_load_op.
12505 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
12506 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
12507 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
12508 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
12509 character before the pattern name.
12510 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
12511 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
12512 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
12513 (aarch64_frecps<mode>): Likewise.
12514 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
12515 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
12516 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
12517 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
12518 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
12520 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
12522 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
12523 Allow HFmode constants if TARGET_FP_F16INST.
12525 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
12528 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
12529 No longer check last store for clobber of address register.
12531 2018-08-02 Martin Liska <mliska@suse.cz>
12533 PR gcov-profile/86817
12534 * gcov.c (process_all_functions): New function.
12536 (process_file): Move functions processing to
12537 process_all_functions.
12539 2018-08-02 David Malcolm <dmalcolm@redhat.com>
12541 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
12542 "const" to the "gimple *" and "rtx_insn *" parameters.
12543 * dumpfile.h (dump_user_location_t::dump_user_location_t):
12545 (dump_location_t::dump_location_t): Likewise.
12547 2018-08-01 Martin Sebor <msebor@redhat.com>
12549 PR tree-optimization/86650
12550 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
12551 rather than a "gcall *". Directly pass the data of interest
12552 to percent_K_format, rather than building a temporary CALL_EXPR
12554 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
12555 (gimple_fold_builtin_strncat): Adjust.
12556 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
12557 gcall* argument with gimple*.
12558 * gimple-ssa-warn-restrict.c (check_call): Same.
12559 (wrestrict_dom_walker::before_dom_children): Same.
12560 (builtin_access::builtin_access): Same.
12561 (check_bounds_or_overlap): Same
12562 (maybe_diag_overlap): Same.
12563 (maybe_diag_offset_bounds): Same.
12564 * tree-diagnostic.c (default_tree_printer): Move usage of
12565 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
12567 * tree-pretty-print.c (percent_K_format): Add argument.
12568 * tree-pretty-print.h: Add argument.
12569 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
12570 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
12571 (maybe_diag_stxncpy_trunc): Same.
12572 (handle_builtin_stxncpy): Same.
12573 (handle_builtin_strcat): Same.
12575 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12577 * match.pd: Optimise pointer range checks.
12579 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12581 PR tree-optimization/86758
12582 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
12583 to remove pattern statements.
12585 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12587 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
12588 result of dfs_enumerate_from when constructing stmt_vec_infos,
12589 instead of additionally calling get_loop_body.
12591 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12593 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
12595 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
12596 When creating an iv, assert that the step is not known to be zero.
12597 (vect_setup_realignment): Update call accordingly.
12598 * tree-vect-stmts.c (vectorizable_store): Likewise.
12599 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
12601 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12603 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
12604 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
12605 (vectorizable_reduction): Likewise.
12606 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12607 (vect_detect_hybrid_slp_stmts): Likewise.
12608 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12610 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
12612 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
12613 (wide_int_set_zero_nonzero_bits): ...this.
12614 (zero_nonzero_bits_from_vr): Rename to...
12615 (vrp_set_zero_nonzero_bits): ...this.
12616 (extract_range_from_multiplicative_op_1): Abstract wide int
12618 (wide_int_range_multiplicative_op): ...here.
12619 (extract_range_from_binary_expr_1): Extract wide int binary
12620 operations into their own functions.
12621 (wide_int_range_lshift): New.
12622 (wide_int_range_can_optimize_bit_op): New.
12623 (wide_int_range_shift_undefined_p): New.
12624 (wide_int_range_bit_xor): New.
12625 (wide_int_range_bit_ior): New.
12626 (wide_int_range_bit_and): New.
12627 (wide_int_range_trunc_mod): New.
12628 (extract_range_into_wide_ints): New.
12629 (vrp_shift_undefined_p): New.
12630 (extract_range_from_multiplicative_op): New.
12631 (vrp_can_optimize_bit_op): New.
12632 * tree-vrp.h (value_range::dump): New.
12633 (wide_int_range_multiplicative_op): New.
12634 (wide_int_range_lshift):New.
12635 (wide_int_range_shift_undefined_p): New.
12636 (wide_int_range_bit_xor): New.
12637 (wide_int_range_bit_ior): New.
12638 (wide_int_range_bit_and): New.
12639 (wide_int_range_trunc_mod): New.
12640 (zero_nonzero_bits_from_bounds): Rename to...
12641 (wide_int_set_zero_nonzero_bits): ...this.
12642 (zero_nonzero_bits_from_vr): Rename to...
12643 (vrp_set_zero_nonzero_bits): ...this.
12644 (range_easy_mask_min_max): Rename to...
12645 (wide_int_range_can_optimize_bit_op): this.
12646 * vr-values.c (simplify_bit_ops_using_ranges): Rename
12647 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
12649 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12651 * tree-vectorizer.h (vect_orig_stmt): New function.
12652 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
12653 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
12654 (vect_create_epilog_for_reduction): Likewise.
12655 (vectorizable_live_operation): Likewise.
12656 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12657 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
12658 * tree-vect-stmts.c (vectorizable_call): Likewise.
12659 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
12661 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12663 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
12665 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
12666 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
12667 (vect_transform_loop): Likewise.
12668 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12670 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12672 * tree-vectorizer.h (vect_schedule_slp): Return void.
12673 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
12674 (vect_schedule_slp): Likewise.
12676 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12678 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
12680 (vect_transform_loop): Update calls accordingly. Schedule SLP
12681 instances before the main loop, if any exist.
12683 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
12685 PR tree-optimization/86749
12686 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
12687 If the lhs is used in a COND_EXPR, check that it is being used
12688 as the "then" or "else" value.
12690 2018-08-01 Tom de Vries <tdevries@suse.de>
12693 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
12694 speculation_safe_value_not_needed.
12696 2018-08-01 Richard Biener <rguenther@suse.de>
12698 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
12699 as base and offset.
12701 2018-08-01 Martin Liska <mliska@suse.cz>
12703 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
12704 format how successful transformation is dumped.
12705 (gimple_mod_pow2_value_transform): Likewise.
12706 (gimple_mod_subtract_transform): Likewise.
12707 (gimple_stringops_transform): Likewise.
12709 2018-08-01 Martin Liska <mliska@suse.cz>
12711 PR value-prof/35543
12712 * value-prof.c (interesting_stringop_to_profile_p):
12713 Simplify the code and add BUILT_IN_MEMMOVE.
12714 (gimple_stringops_transform): Likewise.
12716 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
12718 * config/aarch64/aarch64-simd.md
12719 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
12720 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
12721 use GPI iterator instead of SI mode.
12723 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12725 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
12726 rs6000_speculation_barrier.
12727 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
12728 new barrier pattern name.
12730 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
12732 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
12733 (speculation_barrier): New insn.
12735 2018-08-01 Richard Biener <rguenther@suse.de>
12738 * graphite.h: Include isl/id.h and isl/space.h to allow build
12741 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
12744 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
12745 mode for COFF targets.
12746 * defaults.h (TARGET_COFF): Define.
12747 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
12748 TARGET_COFF): Define.
12749 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
12750 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
12751 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
12753 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12755 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
12756 * gimple-streamer-out.c (output_bb): Save it.
12757 * lto-streamer-in.c (input_struct_function_base): Restore
12758 instance discriminator if available. Create map on demand.
12759 * lto-streamer-out.c (output_struct_function_base): Save it if
12761 * final.c (decl_to_instance_map): Document LTO strategy.
12763 2018-07-31 Alexandre Oliva <oliva@adacore.com>
12764 Olivier Hainque <hainque@adacore.com>
12766 * debug.h (decl_to_instance_map_t): New type.
12767 (decl_to_instance_map): Declare.
12768 (maybe_create_decl_to_instance_map): New inline function.
12769 * final.c (bb_discriminator, last_bb_discriminator): New statics,
12770 to track basic block discriminators.
12771 (final_start_function_1): Initialize them.
12772 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
12774 (decl_to_instance_map): New variable.
12775 (map_decl_to_instance, maybe_set_discriminator): New functions.
12776 (notice_source_line): Set discriminator.
12778 2018-07-31 Ian Lance Taylor <iant@golang.org>
12780 * targhooks.c (default_have_speculation_safe_value): Add
12783 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12785 * dump-context.h: Include "pretty-print.h".
12786 (dump_context::refresh_dumps_are_enabled): New decl.
12787 (dump_context::emit_item): New decl.
12788 (class dump_context): Add fields "m_test_pp" and
12790 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12791 (temp_dump_context::get_dumped_text): New decl.
12792 (class temp_dump_context): Add field "m_pp".
12793 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
12794 (dump_context::refresh_dumps_are_enabled): ...and add a test for
12796 (set_dump_file): Update for above change.
12797 (set_alt_dump_file): Likewise.
12798 (dump_loc): New overload, taking a pretty_printer *.
12799 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
12800 to any test pretty-printer.
12801 (make_item_for_dump_gimple_stmt): New function, adapted from
12802 optinfo::add_gimple_stmt.
12803 (dump_context::dump_gimple_stmt): Call it, and use the result,
12804 eliminating the direct usage of dump_file and alt_dump_file in
12805 favor of indirectly using them via emit_item.
12806 (make_item_for_dump_gimple_expr): New function, adapted from
12807 optinfo::add_gimple_expr.
12808 (dump_context::dump_gimple_expr): Call it, and use the result,
12809 eliminating the direct usage of dump_file and alt_dump_file in
12810 favor of indirectly using them via emit_item.
12811 (make_item_for_dump_generic_expr): New function, adapted from
12813 (dump_context::dump_generic_expr): Call it, and use the result,
12814 eliminating the direct usage of dump_file and alt_dump_file in
12815 favor of indirectly using them via emit_item.
12816 (make_item_for_dump_printf_va): New function, adapted from
12817 optinfo::add_printf_va.
12818 (make_item_for_dump_printf): New function.
12819 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
12820 and use the result, eliminating the direct usage of dump_file and
12821 alt_dump_file in favor of indirectly using them via emit_item.
12822 (make_item_for_dump_dec): New function.
12823 (dump_context::dump_dec): Call it, and use the result,
12824 eliminating the direct usage of dump_file and alt_dump_file in
12825 favor of indirectly using them via emit_item.
12826 (make_item_for_dump_symtab_node): New function, adapted from
12827 optinfo::add_symtab_node.
12828 (dump_context::dump_symtab_node): Call it, and use the result,
12829 eliminating the direct usage of dump_file and alt_dump_file in
12830 favor of indirectly using them via emit_item.
12831 (dump_context::begin_scope): Reimplement, avoiding direct usage
12832 of dump_file and alt_dump_file in favor of indirectly using them
12834 (dump_context::emit_item): New member function.
12835 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12836 Set up test pretty-printer on the underlying context. Call
12837 refresh_dumps_are_enabled.
12838 (temp_dump_context::~temp_dump_context): Call
12839 refresh_dumps_are_enabled.
12840 (temp_dump_context::get_dumped_text): New member function.
12841 (selftest::verify_dumped_text): New function.
12842 (ASSERT_DUMPED_TEXT_EQ): New macro.
12843 (selftest::test_capture_of_dump_calls): Run all tests twice, with
12844 and then without optinfo enabled. Add uses of
12845 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
12846 * dumpfile.h: Update comment for the dump_* API.
12847 * optinfo-emit-json.cc
12848 (selftest::test_building_json_from_dump_calls): Update for new
12849 param for temp_dump_context ctor.
12850 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
12851 and "m_owned" field.
12852 (optinfo_item::~optinfo_item): Likewise.
12853 (optinfo::add_item): New member function.
12854 (optinfo::emit): Update comment.
12855 (optinfo::add_string): Delete.
12856 (optinfo::add_printf): Delete.
12857 (optinfo::add_printf_va): Delete.
12858 (optinfo::add_gimple_stmt): Delete.
12859 (optinfo::add_gimple_expr): Delete.
12860 (optinfo::add_tree): Delete.
12861 (optinfo::add_symtab_node): Delete.
12862 (optinfo::add_dec): Delete.
12863 * optinfo.h (class dump_context): New forward decl.
12864 (optinfo::add_item): New decl.
12865 (optinfo::add_string): Delete.
12866 (optinfo::add_printf): Delete.
12867 (optinfo::add_printf_va): Delete.
12868 (optinfo::add_gimple_stmt): Delete.
12869 (optinfo::add_gimple_expr): Delete.
12870 (optinfo::add_tree): Delete.
12871 (optinfo::add_symtab_node): Delete.
12872 (optinfo::add_dec): Delete.
12873 (optinfo::add_poly_int): Delete.
12874 (optinfo_item::optinfo_item): Remove "owned" param.
12875 (class optinfo_item): Remove field "m_owned".
12877 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12879 PR middle-end/86705
12880 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
12881 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
12882 requested variable alignment.
12883 (expand_one_ssa_partition): Likewise.
12884 (expand_one_var): Likewise.
12886 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12888 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
12889 to speculation_safe_value_not_needed.
12891 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12893 * targhooks.h (speculation_safe_value_not_needed): New prototype.
12894 * targhooks.c (speculation_safe_value_not_needed): New function.
12895 * target.def (have_speculation_safe_value): Update documentation.
12896 * doc/tm.texi: Regenerated.
12898 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12900 * config/aarch64/iterators.md (ALLI_TI): New iterator.
12901 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
12903 (despeculate_copy<ALLI:mode>_insn): New insn.
12904 (despeculate_copyti_insn): New insn.
12905 (despeculate_simple<ALLI:mode>): New insn
12906 (despeculate_simpleti): New insn.
12907 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
12909 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
12910 aarch64_speculation_safe_value.
12911 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
12913 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12915 * config/aarch64/aarch64-speculation.cc: New file.
12916 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
12917 before pass_reorder_blocks.
12918 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
12920 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
12921 X14 and X15 when tracking speculation.
12922 * config/aarch64/aarch64.md (register name constants): Add
12923 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
12924 (unspec): Add UNSPEC_SPECULATION_TRACKER.
12925 (speculation_barrier): New insn attribute.
12926 (cmp<mode>): Allow SP in comparisons.
12927 (speculation_tracker): New insn.
12928 (speculation_barrier): Add speculation_barrier attribute.
12929 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
12930 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
12931 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
12933 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12935 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
12936 aarch64_track_speculation is true.
12937 (tb<optab><mode>1): Likewise.
12938 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
12939 generate CB[N]Z when tracking speculation.
12940 (aarch64_split_compare_and_swap): Likewise.
12941 (aarch64_split_atomic_op): Likewise.
12943 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12945 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
12947 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12949 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
12950 (speculation_barrier): New insn.
12952 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12954 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
12955 * config/arm/arm.md (speculation_barrier): New expand.
12956 (speculation_barrier_insn): New pattern.
12958 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12960 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
12961 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
12962 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
12963 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
12965 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
12966 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
12967 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
12968 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
12969 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
12970 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
12971 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
12972 * builtins.c (expand_speculation_safe_value): New function.
12973 (expand_builtin): Call it.
12974 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
12975 * doc/extend.texi: Document __builtin_speculation_safe_value.
12976 * doc/md.texi: Document "speculation_barrier" pattern.
12977 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
12978 TARGET_HAVE_SPECULATION_SAFE_VALUE.
12979 * doc/tm.texi: Regenerated.
12980 * target.def (have_speculation_safe_value, speculation_safe_value): New
12982 * targhooks.c (default_have_speculation_safe_value): New function.
12983 (default_speculation_safe_value): New function.
12984 * targhooks.h (default_have_speculation_safe_value): Add prototype.
12985 (default_speculation_safe_value): Add prototype.
12987 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12989 * dump-context.h (dump_context::dump_loc): New decl.
12990 * dumpfile.c (dump_context::dump_loc): New member function.
12991 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
12992 and dump_gimple_stmt.
12993 (dump_context::dump_gimple_expr_loc): Likewise, using
12995 (dump_context::dump_generic_expr_loc): Likewise, using
12997 (dump_context::dump_printf_loc_va): Likewise, using
12999 (dump_context::begin_scope): Explicitly using the global function
13000 "dump_loc", rather than the member function.
13002 2018-07-31 Martin Sebor <msebor@redhat.com>
13004 PR tree-optimization/86741
13005 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
13007 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
13009 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
13010 depend on whether prefetch instructions will be emitted or not.
13011 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
13012 will be emitted or not.
13013 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
13014 (TARGET_SETMEM_PFD): New macros.
13016 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13018 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
13019 (NULL_STMT_VEC_INFO): Delete.
13020 (stmt_vec_info::operator*): Likewise.
13021 (stmt_vec_info::operator gimple *): Likewise.
13022 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
13023 of NULL_STMT_VEC_INFO.
13024 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13025 (vect_reassociating_reduction_p): Likewise.
13026 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13027 (vectorizable_store): Likewise.
13028 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
13029 (vec_info::free_stmt_vec_infos): Likewise.
13031 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13033 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
13034 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
13036 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13038 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
13039 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
13040 (vec_info::free_stmt_vec_info): New private member functions.
13041 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
13042 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
13043 * tree-parloops.c (gather_scalar_reductions): Remove calls to
13044 set_stmt_vec_info_vec and free_stmt_vec_infos.
13045 * tree-vect-loop.c (_loop_vec_info): Remove call to
13046 set_stmt_vec_info_vec.
13047 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
13048 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
13049 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
13050 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
13051 (vec_info::free_stmt_vec_info): ...these new functions. Remove
13052 assignments in {vec_info::,}new_stmt_vec_info that are redundant
13053 with the clearing in the xcalloc.
13054 (stmt_vec_info_vec): Delete.
13055 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
13056 (vectorize_loops): Likewise.
13057 (vec_info::~vec_info): Remove argument from call to
13058 free_stmt_vec_infos.
13059 (vec_info::add_stmt): Remove vinfo argument from call to
13062 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13064 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
13065 rather than a gimple stmt.
13066 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
13067 information for pattern statements when passed the original
13068 statement; instead wait to be passed the pattern statement itself.
13069 Don't call set_vinfo_for_stmt here.
13070 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
13071 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
13072 stmt_vec_infos here.
13073 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
13074 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
13075 stmt_vec_infos entry.
13077 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13079 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
13080 * tree-vectorizer.c (vec_info::replace_stmt): New function.
13081 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
13082 * tree-vect-stmts.c (vectorizable_call): Likewise.
13083 (vectorizable_simd_clone_call): Likewise.
13085 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13087 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
13088 * tree-vectorizer.c (vec_info::remove_stmt): New function.
13089 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
13090 * tree-vect-loop.c (vect_transform_loop): Likewise.
13091 * tree-vect-slp.c (vect_schedule_slp): Likewise.
13092 * tree-vect-stmts.c (vect_remove_stores): Likewise.
13094 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13096 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
13097 (vect_dr_stmt): Delete.
13098 * tree-vectorizer.c (vec_info::lookup_dr): New function.
13099 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
13101 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
13102 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
13103 (vect_verify_datarefs_alignment, vect_peeling_supportable)
13104 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
13105 (vect_analyze_data_refs): Likewise.
13106 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
13108 (vect_find_same_alignment_drs): Likewise.
13109 (vect_slp_analyze_node_dependences): Update calls accordingly.
13110 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
13111 instead of DR_VECT_AUX.
13112 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
13113 of a vector data references. Use vec_info::lookup_dr instead of
13115 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
13116 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
13117 instead of DR_VECT_AUX.
13119 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13121 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
13123 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
13125 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13126 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13127 (vect_gen_prolog_loop_niters): Likewise.
13129 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13131 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
13132 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
13133 (vect_known_alignment_in_bytes, vect_dr_behavior)
13134 (vect_get_scalar_dr_size): Take references as dr_vec_infos
13135 instead of data_references. Update calls to other routines for
13136 which the same change has been made.
13137 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
13138 dr_vec_infos instead of stmt_vec_infos.
13139 (vect_analyze_data_ref_dependence): Update call accordingly.
13140 (vect_slp_analyze_data_ref_dependence)
13141 (vect_record_base_alignments): Use DR_VECT_AUX.
13142 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
13143 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
13144 (vector_alignment_reachable_p, vect_get_data_access_cost)
13145 (vect_peeling_supportable, vect_analyze_group_access_1)
13146 (vect_analyze_group_access, vect_analyze_data_ref_access)
13147 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
13148 (vect_compile_time_alias, vect_small_gap_p)
13149 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
13150 (vect_supportable_dr_alignment): Take references as dr_vec_infos
13151 instead of data_references. Update calls to other routines for
13152 which the same change has been made.
13153 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
13154 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
13155 (vect_slp_analyze_and_verify_node_alignment)
13156 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
13157 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
13158 (vect_setup_realignment): Use dr_vec_infos. Update calls after
13160 (_vect_peel_info::dr): Replace with...
13161 (_vect_peel_info::dr_info): ...this new field.
13162 (vect_peeling_hash_get_most_frequent)
13163 (vect_peeling_hash_choose_best_peeling): Update accordingly.
13164 (vect_peeling_hash_get_lowest_cost):
13165 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
13166 routines for which the same change has been made.
13167 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
13169 * tree-vect-loop-manip.c (get_misalign_in_elems)
13170 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
13172 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13173 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
13174 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
13175 (get_group_load_store_type, get_negative_load_store_type)
13176 (vect_get_data_ptr_increment, vectorizable_store)
13177 (vectorizable_load): Likewise.
13178 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
13179 Update calls to other routines for which the same change has been made.
13181 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13183 * tree-vectorizer.h (vec_info::move_dr): New member function.
13184 (dataref_aux): Rename to...
13185 (dr_vec_info): ...this and add "dr" and "stmt" fields.
13186 (_stmt_vec_info::dr_aux): Update accordingly.
13187 (_stmt_vec_info::data_ref_info): Delete.
13188 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
13189 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
13190 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
13191 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
13193 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
13194 (STMT_VINFO_DR_INFO): New macro.
13195 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
13196 (set_dr_misalignment): Update after rename of dataref_aux.
13197 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
13198 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
13199 initialization of STMT_VINFO_DATA_REF.
13200 * tree-vectorizer.c (vec_info::move_dr): New function.
13201 * tree-vect-patterns.c (vect_recog_bool_pattern)
13202 (vect_recog_mask_conversion_pattern)
13203 (vect_recog_gather_scatter_pattern): Use it.
13204 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
13205 the "dr" and "stmt" fields of dr_vec_info instead of
13206 STMT_VINFO_DATA_REF.
13208 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13210 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
13211 (is_pattern_stmt_p): Use it.
13212 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
13213 on pattern statements.
13215 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13217 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
13218 original stmt as a stmt_vec_info rather than a gimple stmt.
13219 (vect_pattern_recog_1): Take the statement directly as a
13220 stmt_vec_info, rather than via a gimple_stmt_iterator.
13221 Update call to vect_mark_pattern_stmts.
13222 (vect_pattern_recog): Update calls accordingly.
13224 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13226 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
13227 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
13228 a vect_def_type for the first argument.
13229 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
13230 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
13231 operand if it isn't defined by a vectorized statement.
13232 (vect_build_gather_load_calls): Remove the mask_dt argument and
13233 update calls to vect_get_vec_def_for_stmt_copy.
13234 (vectorizable_bswap): Likewise the dt argument.
13235 (vectorizable_call): Update calls to vectorizable_bswap and
13236 vect_get_vec_def_for_stmt_copy.
13237 (vectorizable_simd_clone_call, vectorizable_assignment)
13238 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
13239 (vectorizable_comparison): Update calls to
13240 vect_get_vec_def_for_stmt_copy.
13241 (vectorizable_store): Likewise. Remove now-unnecessary calls to
13242 vect_is_simple_use.
13243 (vect_get_loop_based_defs): Remove dt argument and update call
13244 to vect_get_vec_def_for_stmt_copy.
13245 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
13246 and vect_get_vec_def_for_stmt_copy.
13247 (vectorizable_load): Update calls to vect_build_gather_load_calls
13248 and vect_get_vec_def_for_stmt_copy.
13249 * tree-vect-loop.c (vect_create_epilog_for_reduction)
13250 (vectorizable_reduction, vectorizable_live_operation): Update calls
13251 to vect_get_vec_def_for_stmt_copy.
13253 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13255 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
13256 and gimple stmt arguments with a stmt_vec_info.
13257 (vect_record_base_alignments): Update calls accordingly.
13258 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
13259 and gimple stmt arguments with a stmt_vec_info.
13260 (vect_build_slp_tree_1): Remove vinfo argument and update call
13261 to vect_record_max_nunits.
13262 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
13263 and vect_record_max_nunits.
13265 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13267 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
13268 file and take a stmt_vec_info instead of a gimple stmt.
13269 (supportable_widening_operation, vect_finish_replace_stmt)
13270 (vect_finish_stmt_generation, vect_get_store_rhs)
13271 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
13272 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
13273 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
13274 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
13275 (vect_create_data_ref_ptr, bump_vector_ptr)
13276 (vect_permute_store_chain, vect_setup_realignment)
13277 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
13278 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
13279 (vectorizable_reduction, vectorizable_induction)
13280 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
13281 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
13282 than gimple stmts as arguments.
13283 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
13284 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
13285 (can_group_stmts_p, vect_check_gather_scatter)
13286 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
13287 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
13288 (vect_permute_load_chain, vect_shift_permute_load_chain)
13289 (vect_transform_grouped_load)
13290 (vect_record_grouped_load_vectors): Likewise.
13291 * tree-vect-loop.c (vect_fixup_reduc_chain)
13292 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
13293 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
13294 (vectorizable_reduction, vectorizable_induction)
13295 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
13296 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
13297 (vect_get_load_store_mask): Likewise.
13298 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
13299 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
13300 * tree-vect-stmts.c (vect_mark_relevant)
13301 (is_simple_and_all_uses_invariant)
13302 (exist_non_indexing_operands_for_use_p, process_use)
13303 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
13304 (vect_get_vec_def_for_operand, vect_get_vec_defs)
13305 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
13306 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
13307 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
13308 (get_negative_load_store_type, get_load_store_type)
13309 (vect_check_load_store_mask, vect_check_store_rhs)
13310 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
13311 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13312 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13313 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13314 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
13315 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
13316 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
13317 (supportable_widening_operation): Likewise.
13319 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13321 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
13322 a stmt_vec_info instead of a gcall.
13323 (vect_check_gather_scatter): Update call accordingly.
13324 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
13326 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
13327 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
13328 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
13329 instead of a gimple stmt.
13330 (vect_transform_loop): Update calls accordingly.
13331 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
13332 stmt_vec_infos instead of gimple stmts.
13333 (vect_analyze_slp_instance): Update use accordingly.
13334 * tree-vect-stmts.c (read_vector_array, write_vector_array)
13335 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
13336 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
13337 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
13338 (vect_gen_widened_results_half, vect_get_loop_based_defs)
13339 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
13340 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
13341 down to subroutines.
13343 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13345 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
13346 of the worklist from a vector of gimple stmts to a vector of
13348 * tree-vect-stmts.c (vect_mark_relevant, process_use)
13349 (vect_mark_stmts_to_be_vectorized): Likewise
13351 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13353 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
13354 statement before passing it to vect_analyze_stmt.
13355 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
13356 the chain of phi vector definitions. Track the exit phi via its
13358 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
13359 STMT_VINFO_REDUC_DEF.
13360 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
13361 stmt_vec_infos to handle the statement chains.
13362 (vect_get_slp_defs): Record the first statement in the node
13363 using a stmt_vec_info.
13364 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
13365 statements here and pass their stmt_vec_info down to subroutines.
13366 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
13367 down to vect_finish_stmt_generation.
13368 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
13369 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
13370 stmt_vec_infos to subroutines.
13371 (vect_remove_stores): Use stmt_vec_infos to handle the statement
13374 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13376 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
13377 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
13378 (vect_permute_store_chain, vect_setup_realignment)
13379 (vect_permute_load_chain, vect_shift_permute_load_chain)
13380 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
13381 stmts internally, and when passing values to other vectorizer routines.
13382 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
13383 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
13384 (vect_analyze_loop_operations, get_initial_def_for_reduction)
13385 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
13386 (vectorizable_reduction, vectorizable_induction)
13387 (vectorizable_live_operation, vect_transform_loop_stmt)
13388 (vect_transform_loop): Likewise.
13389 * tree-vect-patterns.c (vect_reassociating_reduction_p)
13390 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
13391 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
13392 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13393 (vect_slp_analyze_node_operations_1): Likewise.
13394 * tree-vect-stmts.c (vect_mark_relevant, process_use)
13395 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
13396 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
13397 (vect_finish_stmt_generation_1, get_group_load_store_type)
13398 (get_load_store_type, vect_build_gather_load_calls)
13399 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13400 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13401 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13402 (vectorizable_store, vectorizable_load, vectorizable_condition)
13403 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
13404 (supportable_widening_operation): Likewise.
13405 (vect_get_vector_types_for_stmt): Likewise.
13406 * tree-vectorizer.h (vect_dr_behavior): Likewise.
13408 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13410 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
13411 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
13412 (vect_permute_store_chain, vect_permute_load_chain)
13413 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
13414 repeated stmt_vec_info lookups.
13415 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
13416 (vect_update_ivs_after_vectorizer): Likewise.
13417 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
13418 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13419 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
13420 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13421 (vect_bb_slp_scalar_cost): Likewise.
13422 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
13424 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13426 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
13427 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
13428 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
13429 to get gassigns and gcalls, rather than operating on generc gimple
13431 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
13432 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
13433 (vectorizable_load, vect_analyze_stmt): Likewise.
13434 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
13436 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13438 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
13439 return stmt_vec_infos rather than gimple stmts. Do not accept
13441 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
13443 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
13444 Update use of get_later_stmt.
13445 (vect_get_constant_vectors): Update call accordingly.
13446 (vect_schedule_slp_instance): Likewise
13447 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
13448 (vect_slp_analyze_instance_dependence): Likewise.
13449 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
13451 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13453 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
13454 (stmt_info_for_cost::stmt_info): ...this new field.
13455 (add_stmt_costs): Update accordingly.
13456 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13457 (vect_get_known_peeling_cost): Likewise.
13458 (vect_estimate_min_profitable_iters): Likewise.
13459 * tree-vect-stmts.c (record_stmt_cost): Likewise.
13461 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13463 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
13464 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13465 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
13467 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
13469 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13471 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
13472 a gimple stmt to a stmt_vec_info.
13473 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
13475 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13477 * tree-vectorizer.h (vec_info::grouped_stores): Change from
13478 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13479 (_loop_vec_info::reduction_chains): Likewise.
13480 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
13482 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13484 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13486 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
13487 a gimple stmt to a stmt_vec_info.
13488 (_stmt_vec_info::next_element): Likewise.
13489 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
13490 (vect_slp_analyze_and_verify_node_alignment)
13491 (vect_analyze_group_access_1, vect_analyze_group_access)
13492 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
13493 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
13494 (vect_supportable_dr_alignment): Update accordingly.
13495 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
13496 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
13497 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
13498 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
13499 * tree-vect-slp.c (vect_build_slp_tree_1)
13500 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
13501 (vect_split_slp_store_group, vect_analyze_slp_instance)
13502 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
13503 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
13504 (get_group_load_store_type, get_load_store_type)
13505 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
13506 (vect_transform_stmt, vect_remove_stores): Likewise.
13508 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13510 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
13511 than a gimple stmt.
13512 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
13513 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
13514 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
13515 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
13516 (vector_alignment_reachable_p, vect_get_data_access_cost)
13517 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
13518 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
13519 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
13520 (vect_analyze_group_access_1, vect_analyze_group_access)
13521 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
13522 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
13523 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
13524 result of vect_dr_stmt and use the stmt_vec_info instead of
13525 the associated gimple stmt.
13526 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
13527 (vect_gen_prolog_loop_niters): Likewise.
13528 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
13530 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13532 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
13533 to a vec<stmt_vec_info>.
13534 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
13535 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
13536 vec<stmt_vec_info>.
13537 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
13538 to a vec<stmt_vec_info>.
13539 (bst_traits::value_type, bst_traits::value_type): Likewise.
13540 (bst_traits::hash): Update accordingly.
13541 (vect_get_and_check_slp_defs): Change the stmts parameter from
13542 a vec<gimple *> to a vec<stmt_vec_info>.
13543 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
13544 (vect_build_slp_tree): Likewise.
13545 (vect_build_slp_tree_2): Likewise. Update uses of
13546 SLP_TREE_SCALAR_STMTS.
13547 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
13548 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
13549 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
13550 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
13551 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
13552 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
13553 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
13554 (vect_get_constant_vectors, vect_get_slp_defs)
13555 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
13556 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
13557 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
13558 instead of gimple stmts.
13559 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
13560 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
13561 (vect_slp_analyze_instance_dependence): Update uses of
13562 SLP_TREE_SCALAR_STMTS.
13563 (vect_slp_analyze_and_verify_node_alignment): Likewise.
13564 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
13565 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
13566 (get_initial_defs_for_reduction): Likewise.
13567 (vect_create_epilog_for_reduction): Likewise.
13568 (vectorize_fold_left_reduction): Likewise.
13569 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
13570 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
13571 (can_vectorize_live_stmts): Likewise.
13573 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13575 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
13576 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
13577 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
13579 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
13581 (gather_scalar_reductions): Update after above interface changes.
13582 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
13583 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
13585 (vect_force_simple_reduction): Likewise.
13586 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
13587 LOOP_VINFO_REDUCTIONS.
13588 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
13590 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13592 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
13593 a gimple stmt to a stmt_vec_info.
13594 * tree-vect-loop.c (vect_active_double_reduction_p)
13595 (vect_force_simple_reduction, vectorizable_reduction): Update
13598 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13600 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
13601 vec<gimple *> to a vec<stmt_vec_info>.
13602 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
13603 the reduction_phis argument from a vec<gimple *> to a
13604 vec<stmt_vec_info>.
13605 (vectorizable_reduction): Likewise the phis local variable that
13606 is passed to vect_create_epilog_for_reduction. Update for new type
13607 of SLP_TREE_VEC_STMTS.
13608 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
13609 (vectorizable_live_operation): Likewise.
13610 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
13611 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
13613 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13615 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
13616 a gimple stmt to a stmt_vec_info.
13617 (vectorizable_condition, vectorizable_live_operation)
13618 (vectorizable_reduction, vectorizable_induction): Pass back the
13619 vectorized statement as a stmt_vec_info.
13620 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
13621 use of STMT_VINFO_VEC_STMT.
13622 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
13623 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
13624 as stmt_vec_infos rather than gimple stmts.
13625 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
13626 to a stmt_vec_info.
13627 (vectorizable_live_operation): Likewise.
13628 (vectorizable_reduction, vectorizable_induction): Likewise,
13629 updating use of STMT_VINFO_VEC_STMT.
13630 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
13631 of STMT_VINFO_VEC_STMT.
13632 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
13633 (vectorizable_simd_clone_call, vectorizable_conversion)
13634 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13635 (vectorizable_store, vectorizable_load, vectorizable_condition)
13636 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
13637 from a gimple stmt to a stmt_vec_info.
13638 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
13639 pointer to a stmt_vec_info to the vectorizable_* routines.
13641 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13643 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
13644 a gimple stmt to a stmt_vec_info.
13645 (is_pattern_stmt_p): Update accordingly.
13646 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
13647 (vect_record_grouped_load_vectors): Likewise.
13648 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
13649 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
13650 (vect_model_reduction_cost): Likewise.
13651 (vect_create_epilog_for_reduction): Likewise.
13652 (vectorizable_reduction, vectorizable_induction): Likewise.
13653 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13654 Return the stmt_vec_info for the pattern statement.
13655 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
13656 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
13657 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13658 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
13659 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
13660 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
13661 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
13662 (free_stmt_vec_info, vect_is_simple_use): Likewise.
13664 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13666 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
13667 (vect_finish_stmt_generation): Likewise.
13668 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13669 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
13670 (vect_build_gather_load_calls): Use the return value of the above
13671 functions instead of a separate call to vinfo_for_stmt. Use narrow
13672 scopes for the input gimple stmt and wider scopes for the associated
13673 stmt_vec_info. Use vec_info::lookup_def when setting these
13674 stmt_vec_infos from an SSA_NAME definition.
13675 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
13676 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
13677 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
13678 (vectorizable_store, vectorizable_load, vectorizable_condition)
13679 (vectorizable_comparison): Likewise.
13680 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
13681 (vectorizable_reduction): Likewise.
13683 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13685 * tree-vectorizer.h (vect_is_simple_use): Add an optional
13686 stmt_vec_info * parameter before the optional gimple **.
13687 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13688 (process_use, vect_get_vec_def_for_operand_1): Update callers.
13689 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
13690 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13691 (vectorizable_live_operation): Likewise.
13692 * tree-vect-patterns.c (type_conversion_p): Likewise.
13693 (vect_look_through_possible_promotion): Likewise.
13694 (vect_recog_rotate_pattern): Likewise.
13695 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
13697 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13699 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
13700 a typedef to a wrapper class.
13701 (NULL_STMT_VEC_INFO): New macro.
13702 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
13703 (stmt_vec_info::operator*): New function.
13704 (stmt_vec_info::operator gimple *): Likewise.
13705 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
13706 (add_stmt_costs): Likewise.
13707 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
13708 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
13709 (vect_get_known_peeling_cost): Likewise.
13710 (vect_estimate_min_profitable_iters): Likewise.
13711 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13712 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
13713 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
13714 (vectorizable_store, free_stmt_vec_infos): Likewise.
13715 (new_stmt_vec_info): Change return type of xcalloc to
13718 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13720 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
13721 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
13722 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
13723 a single_imm_use-based sequence.
13724 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
13726 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13728 * tree-vectorizer.h (vec_info::lookup_def): Declare.
13729 * tree-vectorizer.c (vec_info::lookup_def): New function.
13730 * tree-vect-patterns.c (vect_get_internal_def): Use it.
13731 (vect_widened_op_tree): Likewise.
13732 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
13733 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
13734 (vectorizable_reduction): Likewise.
13735 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
13737 (vect_is_slp_reduction): Update calls accordingly. Use
13738 vec_info::lookup_def.
13739 (vect_is_simple_reduction): Likewise
13740 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
13742 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13744 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
13745 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
13746 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
13748 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
13749 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
13750 (vect_update_vf_for_slp, vect_analyze_loop_operations)
13751 (vect_is_slp_reduction, vectorizable_induction)
13752 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
13753 * tree-vect-patterns.c (vect_init_pattern_stmt):
13754 (vect_determine_min_output_precision_1, vect_determine_precisions)
13755 (vect_pattern_recog): Likewise.
13756 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
13757 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
13758 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
13759 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
13760 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
13761 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
13762 info field from a loop to a loop_vec_info.
13764 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13766 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
13767 (vec_info::add_stmt): Declare.
13768 * tree-vectorizer.c (vec_info::add_stmt): New function.
13769 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
13770 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
13771 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
13772 (vectorizable_induction): Likewise.
13773 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
13774 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
13775 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
13776 (vectorizable_load): Likewise.
13777 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
13778 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
13779 (vect_recog_gather_scatter_pattern): Likewise.
13780 (append_pattern_def_seq): Likewise. Remove a check that is
13781 performed by add_stmt itself.
13783 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13785 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
13786 which make_ssa_name was called with new_stmt before new_stmt
13789 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13791 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
13793 (vect_is_slp_reduction): ...here...
13794 (vect_is_simple_reduction): ...and here. Remove repetition of tests
13795 that are already known to be false.
13797 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13799 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
13800 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
13801 STMT_VINFO_NUM_SLP_USES when it's true.
13802 (vect_free_slp_instance): Add a final_p parameter and pass it to
13803 vect_free_slp_tree.
13804 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
13805 (vect_analyze_slp_instance): Likewise.
13806 (vect_slp_analyze_operations): Likewise.
13807 (vect_slp_analyze_bb_1): Likewise.
13808 * tree-vectorizer.c (vec_info): Likewise.
13809 * tree-vect-loop.c (vect_transform_loop): Likewise.
13811 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13813 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13814 function is not called for second and subsequent members of
13817 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13819 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
13820 cases for nested loops from here to ...
13821 (vect_create_epilog_for_reduction): ...here. Only call
13822 vect_is_simple_use for inner-loop reductions.
13824 2018-07-31 Martin Liska <mliska@suse.cz>
13826 PR gcov-profile/85338
13827 PR gcov-profile/85350
13828 PR gcov-profile/85372
13829 * profile.c (struct location_triplet): New.
13830 (struct location_triplet_hash): Likewise.
13831 (output_location): Do not output a BB that
13832 is already recorded for a line.
13833 (branch_prob): Use streamed_locations.
13835 2018-07-31 Martin Liska <mliska@suse.cz>
13837 PR gcov-profile/85370
13838 * coverage.c (coverage_begin_function): Do not mark target
13839 clones as artificial functions.
13841 2018-07-31 Martin Liska <mliska@suse.cz>
13843 PR gcov-profile/83813
13844 PR gcov-profile/84758
13845 PR gcov-profile/85217
13846 PR gcov-profile/85332
13847 * profile.c (branch_prob): Do not record GOTO expressions
13848 for GIMPLE statements which locations are already streamed.
13850 2018-07-31 Olivier Hainque <hainque@adacore.com>
13852 * gcc.c (handle_spec_function): Accept a soft_matched_part
13853 argument, as do_spec_1. Pass it down to ...
13854 (eval_spec_function): Accept a soft_matched_part argument,
13855 and pass it down to ...
13856 (do_spec_2): Accept a soft_matched_part argument, and pass
13857 it down to do_spec_1.
13858 (do_spec_1): Pass soft_matched_part to handle_spec_function.
13859 (handle_braces): Update call to handle_spec_function.
13860 (driver::set_up_specs): Update calls to do_spec_2.
13861 (compare_debug_dump_opt_spec_function): Likewise.
13862 (compare_debug_self_opt_spec_function): Likewise.
13864 2018-07-31 Olivier Hainque <hainque@adacore.com>
13866 * common.opt (nolibc): New option.
13867 * doc/invoke.texi (Link Options): Document it.
13868 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
13869 * config/alpha/linux.h: Likewise.
13870 * config/arc/elf.h: Likewise.
13871 * config/arm/uclinux-elf.h: Likewise.
13872 * config/arm/unknown-elf.h: Likewise.
13873 * config/avr/avrlibc.h: Likewise.
13874 * config/bfin/bfin.h: Likewise.
13875 * config/bfin/linux.h: Likewise.
13876 * config/bfin/uclinux.h: Likewise.
13877 * config/darwin.h: Likewise.
13878 * config/darwin10.h: Likewise.
13879 * config/darwin12.h: Likewise.
13880 * config/gnu-user.h: Likewise.
13881 * config/lm32/uclinux-elf.h: Likewise.
13882 * config/pa/pa-hpux11.h: Likewise.
13883 * config/pa/pa64-hpux.h: Likewise.
13884 * config/sparc/sparc.h: Likewise.
13886 2018-07-31 Olivier Hainque <hainque@adacore.com>
13888 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
13889 undefined variables.
13891 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13894 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
13895 instead of GEN_INT.
13897 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
13899 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
13900 terminated string literal.
13902 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13904 PR rtl-optimization/85160
13905 * combine.c (is_just_move): New function.
13906 (try_combine): Allow combining two instructions into two if neither of
13907 the original instructions was a move.
13909 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
13912 * doc/extend.texi (Global Register Variables): Discourage use of type
13914 (Local Register Variables): Likewise.
13916 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
13918 PR tree-optimization/86506
13919 * hwint.h (ceil_log2): Resync with hwint.c implementation.
13921 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
13924 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
13925 hard_regno, make sure no insn between `from` and `to` clobbers it.
13927 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
13928 Tom de Vries <tdevries@suse.de>
13930 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
13931 (PTX_DEFAULT_RUNTIME_DIM): ... this.
13932 (nvptx_goacc_validate_dims): Set default worker and gang dims to
13933 PTX_DEFAULT_RUNTIME_DIM.
13934 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
13936 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
13938 * config/pa/pa.c (pa_output_addr_vec): Align address table.
13939 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
13940 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
13942 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
13944 * config/rs6000/constraints.md (wG constraint): Delete, no longer
13946 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
13947 predicate to reflect toc fusion has been deleted.
13948 (toc_fusion_mem_raw): Delete, no longer used.
13949 (toc_fusion_mem_wrapped): Likewise.
13950 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
13952 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
13953 Delete, no longer used.
13954 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
13955 meant to be used for toc fusion.
13956 (rs6000_debug_print_mode): Delete toc fusion debugging.
13957 (rs6000_debug_reg_global): Likewise.
13958 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
13959 fusion and secondary reload support that were never used.
13960 (rs6000_option_override_internal): Delete TOC fusion, that was only
13961 partially defined, and it did not work unless you also used the
13963 (rs6000_legitimate_address_p): Delete TOC fusion support.
13964 (rs6000_opt_masks): Likewise.
13965 (fusion_wrap_memory_address): Delete function, no longer used.
13966 (fusion_split_address); Delete TOC fusion support.
13967 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
13968 longer used with toc fusion being deleted.
13969 (TARGET_TOC_FUSION_FP): Likewise.
13970 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
13972 (toc fusion spliter): Delete TOC fusion support.
13973 (toc_fusionload_<mode>): Likewise.
13974 (toc_fusionload_di): Likewise.
13975 (fusion_gpr_load_<mode>): Delete generator function, this insn no
13976 longer needs to be named. Rename predicate to delete TOC fusion.
13977 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
13978 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
13979 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
13980 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
13981 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
13983 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13985 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
13986 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
13987 __int128 in built-in function prototypes.
13988 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
13989 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
13991 2018-07-27 Martin Sebor <msebor@redhat.com>
13993 PR tree-optimization/86696
13994 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
13995 types, including enums.
13996 (handle_char_store): Be prepared for the above function to fail.
13998 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
14000 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
14001 when optimization level is lower than 2 or optimize for size.
14003 2018-07-26 Martin Sebor <msebor@redhat.com>
14005 PR tree-optimization/86043
14006 PR tree-optimization/86042
14007 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
14008 (get_string_cst_length): Rename...
14009 (get_min_string_length): ...to this. Add argument.
14010 (handle_char_store): Extend to handle multi-character stores by
14012 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
14013 * tree.h (initializer_zerop): Add argument.
14015 2018-07-26 Jakub Jelinek <jakub@redhat.com>
14017 PR middle-end/86660
14018 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
14019 declare target to variables if they have always,{to,from,tofrom} map
14022 2018-07-26 Martin Liska <mliska@suse.cz>
14025 * lto-wrapper.c: Add linker_output as prefix
14026 for ltrans_output_file.
14028 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
14030 PR rtl-optimization/85805
14031 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
14032 value for hard registers if that was written in the same mode.
14034 2018-07-26 Martin Liska <mliska@suse.cz>
14036 PR gcov-profile/86536
14037 * gcov.c (format_gcov): Use printf format %.*f directly
14038 and do not handle special values.
14040 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14042 * common/config/arc/arc-common.c (arc_option_optimization_table):
14043 Update default optimizations for size.
14045 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14047 * config/arc/arc.md (movsf_insn): Add short instruction selection.
14048 * config/arc/constraints.md (CfZ): New constraint.
14049 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
14050 (subsf3_fpu): Likewise.
14051 (cmpsf_fpu): Likewise.
14052 (cmpsf_fpu_uneq): Likewise.
14054 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14056 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
14058 (arc_is_uncached_mem_p): uncached applies to both the variable and
14061 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
14063 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
14066 2018-07-25 David Malcolm <dmalcolm@redhat.com>
14068 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
14069 field "m_scopes" from vec to auto_vec.
14071 2018-07-25 Martin Liska <mliska@suse.cz>
14073 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
14076 2018-07-25 Richard Biener <rguenther@suse.de>
14079 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
14080 special wrt context_die late.
14081 (gen_subprogram_die): Re-use DIEs in local scope.
14083 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
14085 PR tree-optimization/86644
14086 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
14088 2018-07-25 Martin Liska <mliska@suse.cz>
14090 PR middle-end/86645
14091 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
14092 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
14094 2018-07-25 Martin Liska <mliska@suse.cz>
14097 * params.def: Explain ASan abbreviation and provide
14098 a documentation link.
14100 2018-07-24 Martin Sebor <msebor@redhat.com>
14102 PR tree-optimization/86622
14103 PR tree-optimization/86532
14104 * builtins.h (string_length): Declare.
14105 * builtins.c (c_strlen): Correct handling of non-constant offsets.
14106 (check_access): Be prepared for non-constant length ranges.
14107 (string_length): Make extern.
14108 * expr.c (string_constant): Only handle the minor non-constant
14109 array index. Use string_constant to compute the length of
14110 a generic string constant.
14112 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
14114 PR tree-optimization/86618
14115 * tree-vect-stmts.c (vectorizable_call): Don't take the address
14116 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
14118 2018-07-24 David Malcolm <dmalcolm@redhat.com>
14120 PR tree-optimization/86636
14121 * json.cc (json::object::set): Fix comment. Add assertions.
14122 (json::array::append): Move here from json.h. Add comment and an
14124 (json::string::string): Likewise.
14125 * json.h (json::array::append): Move to json.cc.
14126 (json::string::string): Likewise.
14127 * optinfo-emit-json.cc
14128 (optrecord_json_writer::impl_location_to_json): Assert that we
14129 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
14130 wrapper around it. Expand the location once, rather than three
14132 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
14133 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
14135 (optrecord_json_writer::optinfo_to_json): Likewise, in four
14136 places. Fix some overlong lines.
14138 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
14140 * config/aarch64/aarch64-simd.md
14141 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
14142 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
14143 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
14144 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
14145 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
14146 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
14147 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
14148 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
14149 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
14151 2018-07-24 Jakub Jelinek <jakub@redhat.com>
14153 PR middle-end/86627
14154 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
14155 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
14156 and abs_d == d, do the power of two handling if profitable.
14158 2018-07-24 Richard Biener <rguenther@suse.de>
14160 * match.pd: Add BIT_FIELD_REF canonicalizations.
14162 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14165 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
14167 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14169 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
14170 terminated STRING_CST object.
14172 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14174 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
14176 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
14178 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
14179 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
14180 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
14181 the elements into a register.
14182 (rs6000_split_v4si_init_di_reg): Delete.
14183 (rs6000_split_v4si_init): Delete.
14184 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
14185 (vsx_init_v4si): Rewrite as a define_expand.
14187 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
14189 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
14190 zero_extend argument from memory): New.
14192 2018-07-22 Martin Sebor <msebor@redhat.com>
14195 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
14196 diagnosing calls with unknown arguments unless -Walloca-larger-than
14197 is restricted to less than PTRDIFF_MAX bytes.
14199 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
14201 * doc/gcov.texi (Invoking Gcov): Editorial changes.
14203 2018-07-20 David Malcolm <dmalcolm@redhat.com>
14205 * pretty-print.c (text_info::set_location): Remove redundant
14206 "line_table" parameter from call to rich_location::set_range.
14208 2018-07-20 Martin Sebor <msebor@redhat.com>
14210 PR middle-end/82063
14211 * builtins.c (expand_builtin_alloca): Adjust.
14212 * calls.c (alloc_max_size): Simplify.
14213 * cgraphunit.c (cgraph_node::expand): Adjust.
14214 * common.opt (larger_than_size, warn_frame_larger_than): Remove
14216 (frame_larger_than_size): Same.
14217 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
14218 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
14220 * doc/invoke.texi (GCC Command Options): Document option arguments.
14221 Explain byte-size arguments and suffixes.
14222 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
14223 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
14224 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
14225 * doc/options.texi (UInteger): Expand.
14226 (Host_Wide_Int, ByteSize): Document new properties.
14227 * final.c (final_start_function_1): Include sizes in an error message.
14228 * function.c (frame_offset_overflow): Same.
14229 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
14230 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
14231 Diagnose unbounded alloca calls only for limits of less than
14233 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
14234 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
14236 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
14237 only for limits of less than PTRDIFF_MAX.
14238 * langhooks-def.h (lhd_handle_option): Change function argument
14240 * langhooks.c (lhd_handle_option): Same.
14241 * langhooks.h (handle_option): Same.
14242 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
14244 (var_type, var_type_struct): Same.
14245 (var_set): Handle ByteSize flag.
14246 * optc-gen.awk: Add comments to output to ease debugging. Make
14247 use of HOST_WIDE_INT where appropriate.
14248 * opts-gen-save.awk: Use %lx to format unsigned long.
14249 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
14250 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
14251 arguments. Parse bytes-size suffixes.
14252 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
14253 (enum_value_to_arg): Same.
14254 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
14255 (handle_option): Adjust.
14256 (generate_option): Change function argument to HOST_WIDE_INT.
14257 (cmdline_handle_error): Adjust.
14258 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
14259 (set_option): Change function argument to HOST_WIDE_INT.
14260 (option_enabled): Handle cl_host_wide_int.
14261 (get_option_state): Handle CLVC_SIZE.
14262 (control_warning_option): Same.
14263 * opts.c (common_handle_option): Change function argument to
14264 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
14265 OPT_Wvla_larger_than_.
14266 * opts.h (enum cl_var_type): Add an enumerator.
14267 * stor-layout.c (layout_decl): Print a more meaningful warning.
14268 * toplev.c (output_stack_usage): Adjust.
14270 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
14272 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
14273 call to inline_expand_builtin_string_cmp.
14274 (expand_builtin_strcmp): Likewise.
14275 (expand_builtin_strncmp): Likewise.
14276 (inline_string_cmp): Delete the last parameter, change char_type_node
14277 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
14279 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
14280 the inlining expansion on target where the type of the call has same or
14281 narrower precision than unsigned char.
14283 2018-07-20 David Malcolm <dmalcolm@redhat.com>
14285 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
14286 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
14287 * common.opt (fsave-optimization-record): New option.
14288 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
14289 * doc/invoke.texi (-fsave-optimization-record): New option.
14290 * dumpfile.c: Include "optinfo-emit-json.h".
14291 (struct kv_pair): Move to coretypes.h.
14292 (optgroup_options): Make non-static.
14293 (dump_context::end_scope): Call
14294 optimization_records_maybe_pop_dump_scope.
14295 * dumpfile.h (optgroup_options): New decl.
14296 * json.cc: New file.
14297 * json.h: New file.
14298 * optinfo-emit-json.cc: New file.
14299 * optinfo-emit-json.h: New file.
14300 * optinfo.cc: Include "optinfo-emit-json.h".
14301 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
14302 (optinfo_enabled_p): Check optimization_records_enabled_p.
14303 (optinfo_wants_inlining_info_p): Likewise.
14304 * optinfo.h: Update comment.
14305 * profile-count.c (profile_quality_as_string): New function.
14306 * profile-count.h (profile_quality_as_string): New decl.
14307 (profile_count::quality): New accessor.
14308 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
14309 and optinfo_emit_json_cc_tests.
14310 * selftest.h (selftest::json_cc_tests): New decl.
14311 (selftest::optinfo_emit_json_cc_tests): New decl.
14312 * toplev.c: Include "optinfo-emit-json.h".
14313 (compile_file): Call optimization_records_finish.
14314 (do_compile): Call optimization_records_start.
14315 * tree-ssa-live.c: Include optinfo.h.
14316 (remove_unused_scope_block_p): Retain inlining information if
14317 optinfo_wants_inlining_info_p returns true.
14319 2018-07-20 Richard Biener <rguenther@suse.de>
14322 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
14323 to cover -flto-partition=none.
14325 2018-07-20 Martin Liska <mliska@suse.cz>
14327 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
14328 (get_decl_source_range): Remove unused function.
14330 2018-07-20 Richard Biener <rguenther@suse.de>
14332 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
14333 (struct vn_phi_s): Likewise.
14334 (struct vn_reference_s): Likewise.
14335 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
14336 for searching the slot of an entry known to be in the hash itself.
14337 (vn_phi_hasher::equal): Likewise.
14338 (vn_reference_hasher::equal): Likewise.
14339 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
14341 (optimistic_info, current_info): Remove, keeping only valid_info.
14342 (vn_reference_lookup_1): Remove fallback lookup.
14343 (vn_reference_lookup_2): Likewise.
14344 (vn_nary_op_lookup_1): Likewise.
14345 (vn_phi_lookup): Likewise.
14346 (vn_nary_build_or_lookup_1): Make sure to not chain the built
14348 (vn_reference_insert): Adjust, chain the inserted hash element
14349 at last_inserted_ref.
14350 (vn_reference_insert_pieces): Likewise.
14351 (visit_reference_op_call): Likewise.
14352 (vn_nary_op_insert_into): Chain the inserted hash element at
14353 last_inserted_nary.
14354 (vn_nary_op_insert_pieces): Adjust.
14355 (vn_nary_op_insert): Likewise.
14356 (vn_nary_op_insert_stmt): Likewise.
14357 (vn_phi_insert): Adjust, chain the inserted hash element at
14359 (process_scc): Remove clearing and copying the optimistic
14360 table. Instead remove elements inserted during an optimistic
14361 iteration from the single table we maintain.
14362 (init_scc_vn): Adjust.
14363 (free_scc_vn): Likewise.
14364 (sccvn_dom_walker::record_cond): Likewise.
14365 (sccvn_dom_walker::after_dom_children): Likewise.
14367 2018-07-19 Martin Sebor <msebor@redhat.com>
14369 PR tree-optimization/84047
14370 PR tree-optimization/83776
14371 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
14372 (check_array_bounds): Call it.
14374 2018-07-19 Martin Sebor <msebor@redhat.com>
14376 * align.h (align_flags): Use member initialization.
14378 2018-07-19 David Malcolm <dmalcolm@redhat.com>
14380 * Makefile.in (OBJS): Add optinfo.o.
14381 * coretypes.h (class symtab_node): New forward decl.
14382 (struct cgraph_node): New forward decl.
14383 (class varpool_node): New forward decl.
14384 * dump-context.h: New file.
14385 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
14387 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
14388 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
14389 (set_alt_dump_file): Likewise.
14390 (dump_context::~dump_context): New dtor.
14391 (dump_gimple_stmt): Move implementation to...
14392 (dump_context::dump_gimple_stmt): ...this new member function.
14393 Add the stmt to any pending optinfo, creating one if need be.
14394 (dump_gimple_stmt_loc): Move implementation to...
14395 (dump_context::dump_gimple_stmt_loc): ...this new member function.
14396 Start a new optinfo and add the stmt to it.
14397 (dump_gimple_expr): Move implementation to...
14398 (dump_context::dump_gimple_expr): ...this new member function.
14399 Add the stmt to any pending optinfo, creating one if need be.
14400 (dump_gimple_expr_loc): Move implementation to...
14401 (dump_context::dump_gimple_expr_loc): ...this new member function.
14402 Start a new optinfo and add the stmt to it.
14403 (dump_generic_expr): Move implementation to...
14404 (dump_context::dump_generic_expr): ...this new member function.
14405 Add the tree to any pending optinfo, creating one if need be.
14406 (dump_generic_expr_loc): Move implementation to...
14407 (dump_context::dump_generic_expr_loc): ...this new member
14408 function. Add the tree to any pending optinfo, creating one if
14410 (dump_printf): Move implementation to...
14411 (dump_context::dump_printf_va): ...this new member function. Add
14412 the text to any pending optinfo, creating one if need be.
14413 (dump_printf_loc): Move implementation to...
14414 (dump_context::dump_printf_loc_va): ...this new member function.
14415 Start a new optinfo and add the stmt to it.
14416 (dump_dec): Move implementation to...
14417 (dump_context::dump_dec): ...this new member function. Add the
14418 value to any pending optinfo, creating one if need be.
14419 (dump_context::dump_symtab_node): New member function.
14420 (dump_context::get_scope_depth): New member function.
14421 (dump_context::begin_scope): New member function.
14422 (dump_context::end_scope): New member function.
14423 (dump_context::ensure_pending_optinfo): New member function.
14424 (dump_context::begin_next_optinfo): New member function.
14425 (dump_context::end_any_optinfo): New member function.
14426 (dump_context::s_current): New global.
14427 (dump_context::s_default): New global.
14428 (dump_scope_depth): Delete global.
14429 (dumpfile_ensure_any_optinfo_are_flushed): New function.
14430 (dump_symtab_node): New function.
14431 (get_dump_scope_depth): Reimplement in terms of dump_context.
14432 (dump_begin_scope): Likewise.
14433 (dump_end_scope): Likewise.
14434 (selftest::temp_dump_context::temp_dump_context): New ctor.
14435 (selftest::temp_dump_context::~temp_dump_context): New dtor.
14436 (selftest::verify_item): New function.
14437 (ASSERT_IS_TEXT): New macro.
14438 (ASSERT_IS_TREE): New macro.
14439 (ASSERT_IS_GIMPLE): New macro.
14440 (selftest::test_capture_of_dump_calls): New test.
14441 (selftest::dumpfile_c_tests): Call it.
14442 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
14443 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
14444 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
14445 descriptive comment.
14446 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
14447 (dump_node, dump_bb): Move these unrelated decls.
14448 (class dump_manager): Add leading comment.
14449 * optinfo.cc: New file.
14450 * optinfo.h: New file.
14452 2018-07-19 Michael Collison <michael.collison@arm.com>
14453 Richard Henderson <rth@redhat.com>
14455 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
14456 (subti): Handle op1 zero.
14457 (subvti4, usub4ti4): New.
14458 (*sub<GPI>3_compare1_imm): New.
14459 (sub<GPI>3_carryinCV): New.
14460 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
14461 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
14463 2018-07-19 Michael Collison <michael.collison@arm.com>
14464 Richard Henderson <rth@redhat.com>
14466 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
14467 (addti3): Create simpler code if low part is already known to be 0.
14468 (addvti4, uaddvti4): New.
14469 (*add<GPI>3_compareC_cconly_imm): New.
14470 (*add<GPI>3_compareC_cconly): New.
14471 (*add<GPI>3_compareC_imm): New.
14472 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
14473 handle constants within this pattern..
14474 (*add<GPI>3_compareV_cconly_imm): New.
14475 (*add<GPI>3_compareV_cconly): New.
14476 (*add<GPI>3_compareV_imm): New.
14477 (add<GPI>3_compareV): New.
14478 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
14479 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
14480 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
14481 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
14482 with 'comparison' operator.
14483 (*add<GPI>3_compareV_cconly_imm): Ditto.
14484 (*add<GPI>3_compareV_cconly): Ditto.
14485 (*add<GPI>3_compareV_imm): Ditto.
14486 (add<GPI>3_compareV): Ditto.
14487 (add<mode>3_carryinC): Ditto.
14488 (*add<mode>3_carryinC_zero): Ditto.
14489 (*add<mode>3_carryinC): Ditto.
14490 (add<mode>3_carryinV): Ditto.
14491 (*add<mode>3_carryinV_zero): Ditto.
14492 (*add<mode>3_carryinV): Ditto.
14494 2018-07-19 Michael Collison <michael.collison@arm.com>
14495 Richard Henderson <rth@redhat.com>
14497 * config/aarch64/aarch64-modes.def (CC_V): New.
14498 * config/aarch64/aarch64-protos.h
14499 (aarch64_addti_scratch_regs): Declare
14500 (aarch64_subvti_scratch_regs): Declare.
14501 (aarch64_expand_subvti): Declare.
14502 (aarch64_gen_unlikely_cbranch): Declare
14503 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
14504 for signed overflow using CC_Vmode.
14505 (aarch64_get_condition_code_1): Handle CC_Vmode.
14506 (aarch64_gen_unlikely_cbranch): New function.
14507 (aarch64_addti_scratch_regs): New function.
14508 (aarch64_subvti_scratch_regs): New function.
14509 (aarch64_expand_subvti): New function.
14511 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14513 * config/aarch64/aarch64-option-extensions.def: New entry for profile
14515 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
14516 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
14519 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14522 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
14523 address check not strict.
14525 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
14527 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
14529 (store_pair_lanes<mode>): Likewise.
14530 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
14531 enum value 'ADDR_QUERY_LDP_STP_N'.
14532 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
14533 (aarch64_print_address_internal): Add declaration.
14534 (aarch64_print_ldpstp_address): Remove.
14535 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
14536 (aarch64_print_operand): Change printing of 'y'.
14537 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
14538 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
14539 'true' rather than '1'.
14540 * config/aarch64/constraints.md (Uml): Likewise.
14541 (Uml): Rename to Umn.
14544 2018-07-19 Richard Biener <rguenther@suse.de>
14546 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
14548 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
14549 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
14550 (vn_reference_hasher): Likewise.
14551 (struct vn_tables_s): Remove obstack and alloc-pool members.
14552 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
14553 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
14554 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
14555 (vn_reference_insert_pieces): Likewise.
14556 (alloc_vn_nary_op_noinit): Adjust.
14557 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
14558 (vn_phi_eq): Adjust.
14559 (shared_lookup_phiargs): Remove.
14560 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
14561 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
14562 (visit_reference_op_call): Likewise.
14563 (copy_nary, copy_phi, copy_reference): Remove.
14564 (process_scc): Rewind the obstack when iterating. Do not
14565 copy the elements to valid_info but just move them from one
14566 hashtable to the other.
14567 (allocate_vn_table): Adjust.
14568 (free_vn_table): Likewise.
14569 (init_scc_vn): Likewise.
14570 (free_scc_vn): Likewise.
14572 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
14575 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
14576 indirect_return as function type attribute.
14577 (ix86_attribute_table): Change indirect_return to function
14579 * doc/extend.texi: Update indirect_return attribute.
14581 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
14583 * wide-int.h (widest2_int): New.
14584 * gimple-fold.c (arith_overflowed_p): Use it.
14585 * tree.h (widest2_int_cst): New.
14586 * tree-vrp.c (wide_int_binop_overflow): Rename from
14587 vrp_int_const_binop.
14588 Rewrite to work on trees.
14589 (extract_range_from_multiplicative_op_1): Abstract code to...
14590 (wide_int_range_min_max): ...here.
14591 (wide_int_range_cross_product): ...and here.
14592 (extract_range_from_binary_expr_1): Abstract overflow code to...
14593 (wide_int_range_mult_wrapping): ...here.
14594 * tree-vrp.h (wide_int_range_cross_product): New.
14595 (wide_int_range_mult_wrapping): New.
14597 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
14598 Julia Koval <julia.koval@intel.com>
14600 * config/i386/x86-tune-costs.h (skylake_memcpy,
14601 skylake_memset): Replace rep_prefix with unrolling for size 512.
14603 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
14605 PR middle-end/86544
14606 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
14607 comparision with EQ_EXPR in last stmt.
14609 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
14611 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
14612 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
14613 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
14614 previously known as "PowerPC AltiVec Built-in Functions". Move
14615 some material to new subsubsections "PowerPC AltiVec Built-in
14616 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
14618 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
14619 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
14620 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
14621 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
14623 2018-07-18 Richard Biener <rguenther@suse.de>
14625 PR tree-optimization/86557
14626 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
14629 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
14631 * config/s390/s390.c (s390_function_profiler): Generate CFI.
14633 2018-07-17 Jeff Law <law@redhat.com>
14635 * config/arm/arm.c (get_label_padding): Update for recent
14636 changes to label_to_alignment.
14638 PR tree-optimization/86010
14639 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
14641 * config/mips/mips.c (vr4130_align_insns): Update for recent
14642 changes to label_to_alignment.
14644 * config/frv/frv.c (frv_label_align): Update for recent changes
14645 to label_to_alignment.
14647 * config/nios2/nios2.c (nios2_label_align): Update for recent
14648 changes which dropped ALIGN_LABELS_LOG.
14650 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
14652 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
14653 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
14655 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14657 * config/arc/arc.c (arc_label_align): Use align_labels instead of
14658 deprecated align_labels_log.
14660 2018-07-17 Richard Biener <rguenther@suse.de>
14663 * dwarf2out.c (init_sections_and_labels): Always generate
14664 a debug_line_str_section for early LTO debug.
14665 (dwarf2out_finish): Reset debug_line_str_hash output early.
14666 Bump counter for extra dwarf5 .debug_loc labels to not conflict
14668 (dwarf2out_early_finish): Output debug_line_str.
14670 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14672 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
14673 index register on z196 or later.
14675 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
14677 * config/s390/s390.c (s390_default_align): Set default function
14679 (s390_override_options_after_change): Call s390_default align.
14680 (s390_option_override_internal): Call s390_default_align.
14681 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
14683 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14685 PR middle-end/86542
14686 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
14687 to _looptemp_ clauses, other than the first two.
14689 2018-07-17 Martin Liska <mliska@suse.cz>
14691 * opts.c: Do not enable OPT_falign_* for -Os.
14693 2018-07-17 Martin Liska <mliska@suse.cz>
14695 * align.h (MAX_CODE_ALIGN): New.
14696 (MAX_CODE_ALIGN_VALUE): New.
14697 * common/config/i386/i386-common.c (ix86_handle_option):
14698 (MAX_CODE_ALIGN): Moved to align.h.
14699 * final.c (MAX_CODE_ALIGN): Likewise.
14700 * opts.c (parse_and_check_align_values):
14701 (MAX_CODE_ALIGN): Likewise.
14702 (MAX_CODE_ALIGN_VALUE): Likewise.
14704 2018-07-17 Martin Liska <mliska@suse.cz>
14706 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
14707 in order to fulfil coding style.
14708 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
14709 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14710 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14711 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
14712 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
14713 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
14714 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
14715 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
14716 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
14718 2018-07-17 Martin Liska <mliska@suse.cz>
14720 * align.h: New file.
14721 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
14723 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
14724 align_flags of label_to_alignment.
14725 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
14727 * config/m68k/m68k.c: Do not use removed align_labels_value and
14729 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
14730 (LOOP_ALIGN): Likewise.
14731 (LABEL_ALIGN): Likewise.
14732 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14733 Remove not used macro.
14734 (rs6000_loop_align): Change return type to align_flags.
14735 (rs6000_loop_align_max_skip): Remove.
14736 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
14737 Change return type to align_flags.
14738 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
14739 Remove not used macro.
14740 (rs6000_loop_align): Change return type to align_flags.
14741 (rs6000_loop_align_max_skip): Remove.
14742 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
14743 * config/rx/rx-protos.h (rx_align_for_label): Make it
14745 * config/rx/rx.c (rx_align_for_label): Change return type
14747 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
14749 into align_flags class.
14750 (LABEL_ALIGN): Likewise.
14751 (LOOP_ALIGN): Likewise.
14752 * config/s390/s390.c (s390_label_align): Use align_flags
14754 (s390_asm_output_function_label): Likewise.
14755 * config/sh/sh.c (sh_override_options_after_change):
14756 Use align_flags class directly without macros.
14757 (find_barrier): Likewise.
14758 (barrier_align): Likewise.
14759 (sh_loop_align): Likewise.
14760 * config/spu/spu.c (spu_option_override):
14761 Use align_flags_tuple::get_value instead of removed macros.
14762 (spu_sched_init): Likewise.
14763 * config/spu/spu.h (GTY): Likewise.
14764 * config/visium/visium.c (visium_option_override):
14765 Set "8" as default secondary alignment.
14766 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
14767 in order to guarantee secondary alignment of 8.
14768 * coretypes.h: Include align.h header file.
14769 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
14770 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
14771 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
14772 * doc/tm.texi.in: Likewise.
14773 * final.c (struct label_alignment): Remove not used structure.
14774 (LABEL_ALIGN): Change type to align_flags.
14775 (LOOP_ALIGN): Likewise.
14776 (JUMP_ALIGN): Likewise.
14777 (default_loop_align_max_skip): Remove.
14778 (default_label_align_max_skip): Likewise.
14779 (default_jump_align_max_skip): Likewise.
14780 (default_label_align_after_barrier_max_skip):
14781 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
14782 (LABEL_TO_MAX_SKIP): Remove.
14783 (label_to_alignment): Return align_flags type instead of integer.
14784 (label_to_max_skip): Remove.
14785 (align_fuzz): Use align_flags type.
14786 (compute_alignments): Use align_flags type and use align_flags::max
14787 to combine multiple alignments.
14788 (grow_label_align): Grow vec instead of C array.
14789 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
14790 (shorten_branches): Use align_flags type and use align_flags::max
14791 to combine multiple alignments.
14792 (final_scan_insn_1): Remove usage of secondary alignment that comes
14793 from label alignment, but instead use proper secondary alignment
14794 which is computed in grow_label_align.
14795 * flags.h (struct align_flags_tuple): Move to align.h.
14796 (struct align_flags): Likewise.
14797 (state_align_loops): Rename to align_loops.
14798 (state_align_jumps): Rename to align_jumps.
14799 (state_align_labels): Rename to align_labels.
14800 (state_align_functions): Rename to align_functions.
14801 (align_loops_log): Remove.
14802 (align_jumps_log): Remove.
14803 (align_labels_log): Remove.
14804 (align_functions_log): Remove.
14805 (align_loops_max_skip): Remove.
14806 (align_jumps_max_skip): Remove.
14807 (align_labels_max_skip): Remove.
14808 (align_functions_max_skip): Remove.
14809 (align_loops_value): Remove.
14810 (align_jumps_value): Remove.
14811 (align_labels_value): Remove.
14812 (align_functions_value): Remove.
14813 * output.h (label_to_alignment): Change return type to align_flags.
14814 (label_to_max_skip): Remove.
14815 * target.def: Remove loop_align_max_skip, label_align_max_skip,
14816 jump_align_max_skip macros.
14817 * targhooks.h (default_loop_align_max_skip): Remove.
14818 (default_label_align_max_skip): Likewise.
14819 (default_jump_align_max_skip): Likewise.
14820 (default_label_align_after_barrier_max_skip): Remove.
14821 * toplev.c (read_log_maxskip): Use ::normalize function.
14822 (parse_N_M): Remove not used argument and also call ::normalize.
14823 (parse_alignment_opts): Do not pass unused arguments.
14824 * varasm.c (assemble_start_function): Use directly align_functions
14825 instead of removed macros.
14826 * system.h: Do not poison removed macros.
14828 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14830 PR middle-end/86539
14831 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
14832 and cond temporaries don't have reference type if iterator has
14833 pointer type. For init use &for_pre_body instead of pre_p if
14834 for_pre_body is non-empty.
14836 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14838 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
14839 double-double modes to SFmode directly directly.
14840 (trunc<mode>sf2_fprs): Delete.
14842 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14844 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
14845 for conversions between IFmode and the decimal floating point modes.
14846 (init_float128_ieee): Use the correct names for conversions between
14847 KFmode and the decimal floating point modes.
14849 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14851 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
14852 for the conversions between TDmode and IFmode.
14853 (init_float128_ieee): Use more correct names for the conversions
14854 between TDmode and KFmode.
14856 2018-07-16 Jakub Jelinek <jakub@redhat.com>
14858 PR tree-optimization/86526
14859 * builtins.c (expand_builtin_memcmp): Formatting fixes.
14860 (inline_expand_builtin_string_cmp): Likewise.
14861 (inline_string_cmp): Likewise. Use c_readstr instead of
14862 builtin_memcpy_read_str. Add unit_mode temporary.
14864 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
14866 PR middle-end/86528
14867 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
14868 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
14870 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
14872 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
14873 Alphabetize prototypes of built-in functions, separating out
14874 built-in functions that are listed in this section but should be
14875 described elsewhere.
14877 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
14880 * expmed.c (emit_store_flag): Do not emit setcc followed by a
14881 conditional move when trapping comparison was split to a
14882 non-trapping one (and vice versa).
14884 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14886 * config/s390/s390.c (s390_function_profiler): Generate nops
14887 instead of profiler call sequences.
14888 * config/s390/s390.opt: Add the new option.
14890 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14892 * config/s390/s390.c (s390_function_profiler): Generate
14893 __mcount_loc section.
14894 * config/s390/s390.opt: Add the new option.
14896 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14898 * common.opt: Add the new warning.
14899 * config/s390/s390.c (s390_function_profiler): Emit "brasl
14900 %r0,__fentry__" when -mfentry is specified.
14901 (s390_option_override_internal): Disallow -mfentry for 31-bit
14903 * config/s390/s390.opt: Add the new option.
14905 2018-07-16 Richard Biener <rguenther@suse.de>
14908 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
14909 for function-local FUNCTION_DECL and RESULT_DECL immediately.
14911 2018-07-16 Martin Liska <mliska@suse.cz>
14914 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
14917 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14919 * config/arc/arcHS.md: Update ARCHS scheduling rules.
14921 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14923 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
14925 * config/arc/arc-cpus.def (hs4x): New cpu.
14927 * config/arc/arc-tables.opt: Regenerate.
14928 * config/arc/arc.c (arc_sched_issue_rate): New function.
14929 (TARGET_SCHED_ISSUE_RATE): Define.
14930 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
14931 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
14933 (attr tune): Add ARCHS4x tune values.
14934 (attr tune_dspmpy): Define.
14935 (*tst): Correct instruction type.
14936 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
14937 * config/arc/arcHS4x.md: New file.
14938 * config/arc/fpu.md: Update instruction type attributes.
14939 * config/arc/t-multilib: Regenerate.
14941 2018-07-16 Tom de Vries <tdevries@suse.de>
14944 * var-tracking.c (vt_initialize): Fix pre_dec handling.
14946 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14948 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
14951 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
14953 PR tree-optimization/86514
14954 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
14955 conversion to a boolean type from a type with greater precision.
14957 2018-07-16 Tom de Vries <tdevries@suse.de>
14959 * var-tracking.c (vt_initialize): Print adjusted insn slim if
14960 dump_flags request TDF_SLIM.
14962 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
14964 * fold-const.c (int_const_binop_1): Abstract...
14965 (wide_int_binop): ...wide int code here.
14966 (poly_int_binop): ...poly int code here.
14967 Abstract the rest of int_const_binop_1 into int_const_binop.
14968 * fold-const.h (wide_int_binop): New.
14969 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
14970 Remove useless PLUS/MINUS_EXPR case.
14971 (zero_nonzero_bits_from_vr): Move wide int code...
14972 (zero_nonzero_bits_from_bounds): ...here.
14973 (extract_range_from_binary_expr_1): Move mask optimization code...
14974 (range_easy_mask_min_max): ...here.
14975 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
14976 (range_easy_mask_min_max): New.
14978 2018-07-15 Jeff Law <law@redhat.com>
14981 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
14984 2018-07-14 Jim Wilson <jimw@sifive.com>
14986 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
14988 2018-07-14 Paul Koning <ni1d@arrl.net>
14990 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
14992 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
14994 * lto-streamer-out.c (copy_function_or_variable): Dump info about
14997 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
14998 Steve Munroe <munroesj52@gmail.com>
15000 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
15001 (_mm_andnot_si128): Likewise.
15002 (_mm_or_si128): Likewise.
15003 (_mm_xor_si128): Likewise.
15005 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
15007 PR middle-end/78809
15008 * builtins.c (expand_builtin_memcmp): Inline the calls first
15009 when result_eq is false.
15010 (expand_builtin_strcmp): Inline the calls first.
15011 (expand_builtin_strncmp): Likewise.
15012 (inline_string_cmp): New routine. Expand a string compare
15013 call by using a sequence of char comparison.
15014 (inline_expand_builtin_string_cmp): New routine. Inline expansion
15015 a call to str(n)cmp/memcmp.
15016 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
15018 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
15020 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
15022 * config/arm/driver-arm.c: Include arm-native.h.
15023 (host_detect_local_cpu): Use auto-generated data tables.
15024 (vendors, arm_cpu_table): Delete. Move part information to ...
15025 * config/arm/arm-cpus.in: ... here.
15026 * config/arm/parsecpu.awk (gen_native): New function.
15027 (vendor, part): New CPU fields.
15028 (END): Add support for building the native CPU detection tables.
15029 * config/arm/t-arm (arm-native.h): Add build rule.
15030 (driver-arm.o): Add dependency on arm-native.h.
15032 2018-07-13 Richard Biener <rguenther@suse.de>
15034 PR middle-end/85974
15035 * match.pd (addr1 - addr2): Allow either of the operand to
15038 2018-07-13 Tom de Vries <tdevries@suse.de>
15040 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
15043 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
15045 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
15046 arrays instead of numbered variables.
15048 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
15050 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
15051 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
15053 2018-07-13 Richard Biener <rguenther@suse.de>
15056 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
15057 instead of get_context_die.
15059 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
15060 Richard Biener <rguenther@suse.de>
15062 PR middle-end/86489
15063 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
15064 that the loop latch destination where phi is defined.
15066 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
15068 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
15069 (riscv_expand_epilogue): Add assertion to check interrupt mode.
15070 (riscv_set_current_function): Extract getting interrupt type to new
15072 (riscv_get_interrupt_type): New function.
15073 (riscv_merge_decl_attributes): New function, checking interrupt type is
15075 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
15077 2018-07-12 Paul Koning <ni1d@arrl.net>
15079 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
15082 2018-07-12 Paul Koning <ni1d@arrl.net>
15084 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
15085 zero reference, add doloop_end instead.
15086 * doc/md.texi (decrement_and_branch_until_zero): Remove.
15087 (Looping patterns): Remove decrement_and_branch_until_zero. Add
15088 detail for doloop_end.
15090 2018-07-12 Martin Sebor <msebor@redhat.com>
15093 * attribs.c (decl_attributes): Reject conflicting attributes before
15094 calling attribute handlers.
15096 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
15098 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
15100 (gcc::dump_manager::get_dump_file_name): likewise.
15101 (dump_begin): Likewise.
15102 * dumpfile.h (dump_begin): Update prototype.
15103 (gcc::dump_manager::get_dump_file_name,
15104 gcc::dump_manager::get_dump_file_name): Update prototype.
15106 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15108 * internal-fn.h (vectorizable_internal_fn_p): New function.
15109 * tree-vect-slp.c (compatible_calls_p): Likewise.
15110 (vect_build_slp_tree_1): Remove nops argument. Handle calls
15111 to internal functions.
15112 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
15114 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15116 * fold-const.h (inverse_conditions_p): Declare.
15117 * fold-const.c (inverse_conditions_p): New function.
15118 * match.pd: Use inverse_conditions_p. Add folds of view_converts
15119 that test the inverse condition of a conditional internal function.
15120 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
15121 * internal-fn.c (internal_fn_mask_index): Handle conditional
15122 internal functions.
15123 (vectorized_internal_fn_supported_p): New function.
15124 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
15125 (any_pred_load_store): Replace with...
15126 (need_to_predicate): ...this new variable.
15127 (redundant_ssa_names): New variable.
15128 (ifcvt_can_use_mask_load_store): Move initial checks to...
15129 (ifcvt_can_predicate): ...this new function. Handle tree codes
15130 for which a conditional internal function exists.
15131 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
15132 instead of ifcvt_can_use_mask_load_store. Update after variable
15134 (predicate_load_or_store): New function, split out from
15135 predicate_mem_writes.
15136 (check_redundant_cond_expr): New function.
15137 (value_available_p): Likewise.
15138 (predicate_rhs_code): Likewise.
15139 (predicate_mem_writes): Rename to...
15140 (predicate_statements): ...this. Use predicate_load_or_store
15141 and predicate_rhs_code.
15142 (combine_blocks, tree_if_conversion): Update after above name changes.
15143 (ifcvt_local_dce): Handle redundant_ssa_names.
15144 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
15145 general conditional functions.
15146 * tree-vect-stmts.c (vectorizable_call): Likewise.
15148 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15149 Alan Hayward <alan.hayward@arm.com>
15150 David Sherwood <david.sherwood@arm.com>
15152 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
15153 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
15154 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
15155 plus and minus and convert them into IFN_COND_FMA-based sequences.
15156 (convert_mult_to_fma): Handle conditional plus and minus.
15158 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15160 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
15161 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
15162 (cond_fnms_optab): New optabs.
15163 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
15164 internal functions.
15165 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
15166 * internal-fn.h (get_conditional_internal_fn): Declare.
15167 (get_unconditional_internal_fn): Likewise.
15168 * internal-fn.c (cond_ternary_direct): New macro.
15169 (expand_cond_ternary_optab_fn): Likewise.
15170 (direct_cond_ternary_optab_supported_p): Likewise.
15171 (FOR_EACH_COND_FN_PAIR): Likewise.
15172 (get_conditional_internal_fn): New function.
15173 (get_unconditional_internal_fn): Likewise.
15174 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
15175 (gimple_match_op::gimple_match_op): Add a new overload for 5
15177 (gimple_match_op::set_op): Likewise.
15178 (gimple_resimplify5): Declare.
15179 * genmatch.c (decision_tree::gen): Generate simplifications for
15181 * gimple-match-head.c (gimple_simplify): Define an overload for
15182 5 operands. Handle calls with 5 arguments in the top-level overload.
15183 (convert_conditional_op): Handle conversions from unconditional
15184 internal functions to conditional ones.
15185 (gimple_resimplify5): New function.
15186 (build_call_internal): Pass a fifth operand.
15187 (maybe_push_res_to_seq): Likewise.
15188 (try_conditional_simplification): Try converting conditional
15189 internal functions to unconditional internal functions.
15190 Handle 3-operand unconditional forms.
15191 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
15192 Define ternary equivalents of the current rules for binary conditional
15193 internal functions.
15194 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
15195 ternary operations.
15196 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
15197 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
15198 (optab): Handle them.
15199 (SVE_COND_FP_TERNARY): New int iterator.
15200 (sve_fmla_op, sve_fmad_op): New int attributes.
15201 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
15202 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
15203 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
15205 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
15207 * target.def (preferred_else_value): New target hook.
15208 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
15209 * doc/tm.texi: Regenerate.
15210 * targhooks.h (default_preferred_else_value): Declare.
15211 * targhooks.c (default_preferred_else_value): New function.
15212 * internal-fn.h (conditional_internal_fn_code): Declare.
15213 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
15214 (get_conditional_internal_fn): Use it.
15215 (conditional_internal_fn_code): New function.
15216 * gimple-match.h (gimple_match_cond): New struct.
15217 (gimple_match_op): Add a cond member function.
15218 (gimple_match_op::gimple_match_op): Update all forms to take a
15220 * genmatch.c (expr::gen_transform): Use the same condition as res_op
15221 for the suboperation, but don't specify a particular else_value.
15222 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
15223 (visit_nary_op, visit_reference_op_load): Pass
15224 gimple_match_cond::UNCOND to the gimple_match_op constructor.
15225 * gimple-match-head.c: Include tree-eh.h
15226 (convert_conditional_op): New function.
15227 (maybe_resimplify_conditional_op): Likewise.
15228 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
15229 (gimple_resimplify2): Likewise.
15230 (gimple_resimplify3): Likewise.
15231 (gimple_resimplify4): Likewise.
15232 (maybe_push_res_to_seq): Return null for conditional operations.
15233 (try_conditional_simplification): New function.
15234 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
15236 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
15238 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
15240 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
15242 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
15244 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
15246 (hash_tree): Do not hash DECL_FCONTEXT
15247 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
15248 Do not stream DECL_FCONTEXT.
15249 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
15250 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
15252 2018-07-12 Richard Biener <rguenther@suse.de>
15255 * dwarf2out.c (gen_block_die): Only output blocks when they have
15256 at least one !DECL_IGNORED_P variable.
15258 2018-07-12 Richard Biener <rguenther@suse.de>
15261 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
15262 Remove -mieee-fp handling.
15264 2018-07-12 Richard Biener <rguenther@suse.de>
15266 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
15267 left-over from last patch.
15269 2018-07-12 Jakub Jelinek <jakub@redhat.com>
15271 PR tree-optimization/86492
15272 * gimple-ssa-store-merging.c
15273 (imm_store_chain_info::coalesce_immediate_stores): Call
15274 check_no_overlap even for the merge_overlapping case. Formatting fix.
15276 2018-07-12 Richard Biener <rguenther@suse.de>
15278 PR middle-end/86479
15279 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
15280 move possibly trapping operations into the conditional.
15282 2018-07-12 Richard Biener <rguenther@suse.de>
15284 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
15285 (vn_lookup_simplify_result): Remove recursion limit applied
15287 (vn_nary_build_or_lookup_1): Adjust.
15288 (try_to_simplify): Likewise.
15289 * gimple-match-head.c (gimple_resimplify1): Instead apply one
15291 (gimple_resimplify2): Likewise.
15292 (gimple_resimplify3): Likewise.
15293 (gimple_resimplify4): Likewise.
15295 2018-07-11 Jakub Jelinek <jakub@redhat.com>
15297 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
15298 Use __mmask64 type instead of __mmask8 for __M argument.
15299 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
15300 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
15302 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
15303 __mmask16 for __M argument.
15304 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
15305 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
15306 to __mmask16 instead of __mmask8.
15307 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
15308 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
15309 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
15310 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
15311 instead of __mmask16 for __U argument.
15312 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
15313 __mmask16 instead of __mmask8 for __U argument.
15314 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
15316 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
15318 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
15320 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
15322 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
15324 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
15326 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
15328 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
15330 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
15332 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
15334 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
15335 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
15336 return type as well as __M argument type and all casts from __mmask8
15338 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
15339 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
15340 return type as well as __M argument type and all casts from __mmask8
15342 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
15343 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
15344 return type as well as __M argument type and all casts from __mmask8
15346 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
15347 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
15348 return type as well as __M argument type and all casts from __mmask8
15350 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
15351 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
15354 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
15356 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
15357 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
15360 2018-07-11 Paul Koning <ni1d@arrl.net>
15362 * doc/md.texi (define_subst): Document how multiple occurrences of
15363 the same argument in the replacement pattern are handled.
15365 2018-07-11 Paul Koning <ni1d@arrl.net>
15367 * doc/extend.texi (Common Variable Attributes): Move "mode" into
15368 alphabetical order.
15369 (Common Type Attributes): Add "mode" attribute.
15371 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
15373 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
15374 stream DECL_ORIGINAL_TYPE.
15375 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
15376 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
15377 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
15378 Do not walk original type.
15379 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
15381 (write_ts_decl_non_common_tree_pointers): Do not stream
15383 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
15384 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
15386 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
15388 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
15389 thread twice from the same starting edge.
15391 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
15393 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
15394 * gimple.c (gimple_call_nonnull_result_p): ...here...
15395 (gimple_call_nonnull_arg): ...and here.
15396 * gimple.h (gimple_call_nonnull_result_p): New.
15397 (gimple_call_nonnull_arg): New.
15399 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
15401 * config/arm/arm-cpus.in: Move information from fpu field of each
15402 cpu definition to the isa field.
15403 * config/arm/parsecpu.awk (fpu): Delete match rule.
15404 (gen_comm_data): Don't add bits from the CPU's FPU entry.
15406 2018-07-11 Richard Biener <rguenther@suse.de>
15409 * dwarf2out.c (init_sections_and_labels): Use
15410 output_asm_line_debug_info consistently.
15411 (dwarf2out_early_finish): Likewise.
15412 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
15415 2018-07-11 Richard Biener <rguenther@suse.de>
15417 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
15418 Rework father_bb setting in a way to avoid propagating constants
15419 multiple times on a loop body.
15421 2018-07-10 Mark Wielaard <mark@klomp.org>
15424 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
15426 2018-07-10 Richard Biener <rguenther@suse.de>
15428 * hash-map.h (hash_map::iterator::operator*): Return
15429 references to key and value.
15431 2018-07-10 Jakub Jelinek <jakub@redhat.com>
15434 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
15435 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
15436 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
15437 (gimplify_omp_for): For composite loops, move outer
15438 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
15439 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
15440 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
15441 TREE_LIST for both the original class iterator and the "last" helper
15442 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
15443 loop, remember has_decl_expr from outer composite loops for the
15444 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
15446 2018-07-09 Martin Sebor <msebor@redhat.com>
15448 PR middle-end/77357
15449 PR middle-end/86428
15450 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
15451 accessing implicitly initialized array elements.
15452 * expr.c (string_constant): Handle string initializers of
15453 character arrays within aggregates.
15454 * gimple-fold.c (fold_array_ctor_reference): Add argument.
15455 Store element offset. As a special case, handle zero size.
15456 (fold_nonarray_ctor_reference): Same.
15457 (fold_ctor_reference): Add argument. Store subobject offset.
15458 * gimple-fold.h (fold_ctor_reference): Add argument.
15460 2018-07-09 Paul Koning <ni1d@arrl.net>
15462 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
15463 (pdp11_insn_cost): New function.
15464 (pdp11_md_asm_adjust): New function.
15465 (TARGET_INVALID_WITHIN_DOLOOP): Define.
15466 (pdp11_rtx_costs): Update to match machine better.
15467 (output_addr_const_pdp11): Correct format mismatch warnings.
15468 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
15469 * config/pdp11/pdp11.md: General change to add base_cost and/or
15470 length attributes for use by new pdp11_insn_cost function.
15471 (MIN_BRANCH): Correct definition.
15473 (doloop_end): Use standard pattern name for looping pattern.
15474 (doloop_end_nocc): New.
15475 (movsf): Add another constraint alternative.
15476 (zero_extendqihi2): Add constraint alternatives for not in place
15478 (zero_extendhisi2): Remove.
15479 (shift patterns): Add CC handling variants.
15483 (define_peephole2): New peephole to recognize mov that sets CC for
15486 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15488 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
15489 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
15492 2018-07-09 Jakub Jelinek <jakub@redhat.com>
15495 * real.c (real_nextafter): Return true if result is denormal.
15497 2018-07-09 Martin Liska <mliska@suse.cz>
15499 * common.opt: Add back wrongly removed attribute.
15501 2018-07-09 Richard Biener <rguenther@suse.de>
15504 * dwarf2out.c (gen_block_die): For an early generated DIE
15505 always output high/low PC attributes.
15507 2018-07-09 Tom de Vries <tdevries@suse.de>
15509 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
15510 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
15511 onto VAR_DECL with abstract origin.
15513 2018-07-07 Jim Wilson <jimw@sifive.com>
15515 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
15517 2018-07-07 Tom de Vries <tdevries@suse.de>
15519 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
15521 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
15523 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
15525 (combine_bound): Use wide-int overflow calculation instead of
15527 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
15529 * fold-const.c (int_const_binop_2): Same.
15530 (extract_muldiv_1): Same.
15531 (fold_div_compare): Same.
15532 (fold_abs_const): Same.
15534 * poly-int.h (add): Same.
15538 * predict.c (predict_iv_comparison): Same.
15539 * profile-count.c (slow_safe_scale_64bit): Same.
15540 * simplify-rtx.c (simplify_const_binary_operation): Same.
15541 * tree-chrec.c (tree_fold_binomial): Same.
15542 * tree-data-ref.c (split_constant_offset_1): Same.
15543 * tree-if-conv.c (idx_within_array_bound): Same.
15544 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
15545 * tree-ssa-phiopt.c (minmax_replacement): Same.
15546 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
15547 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
15548 * vr-values.c (vr_values::adjust_range_with_scev): Same.
15549 * wide-int.cc (wi::add_large): Same.
15550 (wi::mul_internal): Same.
15551 (wi::sub_large): Same.
15552 (wi::divmod_internal): Same.
15553 * wide-int.h: Change overflow type to overflow_type for neg, add,
15554 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
15555 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
15556 mul_internal, divmod_internal.
15557 (overflow_type): New enum.
15558 (accumulate_overflow): New.
15560 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
15562 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
15563 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
15565 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
15567 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
15568 argument is checked for zero before entering loop, avoid checking again.
15570 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
15572 * gimplify.h (generic_expr_could_trap_p): Set as global function.
15573 * gimplify.h (generic_expr_could_trap_p): Likwise.
15574 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
15576 2018-07-06 Jakub Jelinek <jakub@redhat.com>
15578 PR tree-optimization/86401
15579 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
15580 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
15581 (fold_bit_and_mask): ... here. New helper function for match.pd.
15582 * fold-const.h (fold_bit_and_mask): Declare.
15583 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
15585 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
15588 * target.def (translate_mode_attribute): New hook.
15589 * targhooks.h (default_translate_mode_attribute): Declare.
15590 * targhooks.c (default_translate_mode_attribute): New function.
15591 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
15592 * doc/tm.texi: Regenerate.
15593 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
15594 (rs6000_translate_mode_attribute): New function.
15596 2018-07-06 Paul Koning <ni1d@arrl.net>
15598 * doc/md.texi (define_split): Document DONE and FAIL.
15599 (define_peephole2): Ditto.
15601 2018-07-05 Jeff Law <law@redhat.com>
15603 PR tree-optimization/86010
15604 * tree-ssa-dse.c (compute_trims): More aggressively trim at
15605 both the head and tail of mem* and str* calls.
15607 2018-07-05 Jim Wilson <jimw@sifive.com>
15609 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
15611 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
15613 * config/aarch64/aarch64-simd.md: correct flags text for
15614 MIN_EXPR replacement.
15616 2018-07-05 James Clarke <jrtc27@jrtc27.com>
15618 * configure: Regenerated.
15620 2018-07-05 Carl Love <cel@us.ibm.com>
15622 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
15623 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
15624 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
15625 VSX_BUILTIN_DOUBLEL_V4SF.
15627 2018-07-05 Martin Sebor <msebor@redhat.com>
15630 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
15631 than its domain to compute its the upper bound of a char array.
15633 2018-07-05 Nathan Sidwell <nathan@acm.org>
15635 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
15636 * doc/cpp.texi: Update comment.
15637 * doc/tm.texi: Rebuilt.
15638 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
15639 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
15640 * doc/extend.texi (Backwards Compatibility): Clarify it is system
15641 headers affected by extern "C".
15642 * system.h: Poison NO_IMPLICIT_EXTERN_C.
15643 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
15644 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
15645 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
15646 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
15647 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
15648 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
15649 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
15650 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
15651 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
15652 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
15653 config/sparc/sp64-elf.h, config/spu/spu.h,
15654 config/stormy16/stormy16.h, config/v850/v850.h,
15655 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
15656 define NO_IMPLICIT_EXTERN_C.
15657 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
15659 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15662 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
15663 instead of GET_MODE_SIZE when comparing Units.
15665 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15668 * rtlanal.c (set_noop_p): Constrain on mode change,
15669 include hard-reg-set.h
15671 2018-07-05 Tamar Christina <tamar.christina@arm.com>
15673 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
15675 2018-07-05 Jakub Jelinek <jakub@redhat.com>
15678 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15681 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15683 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15685 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
15688 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
15690 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
15692 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
15694 PR middle-end/86380
15695 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
15697 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
15699 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
15700 neg_*_op* variables.
15702 2018-07-04 Martin Liska <mliska@suse.cz>
15704 * tree-switch-conversion.c: Define
15705 max_ratio_for_speed and max_ratio_for_size constants.
15707 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15708 Martin Liska <mliska@suse.cz>
15710 PR middle-end/66240
15713 * common.opt: Rename align options with 'str_' prefix.
15714 * common/config/i386/i386-common.c (set_malign_value): New
15716 (ix86_handle_option): Use it to set -falign-* options/
15717 * config/aarch64/aarch64-protos.h (struct tune_params): Change
15718 type from int to string.
15719 * config/aarch64/aarch64.c: Update default values from int
15721 * config/alpha/alpha.c (alpha_override_options_after_change):
15723 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
15724 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15725 max skip conditionally.
15726 * config/i386/freebsd.h (SUBALIGN_LOG): New.
15727 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15728 max skip conditionally.
15729 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15730 max skip conditionally.
15731 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
15732 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15733 max skip conditionally.
15734 * config/i386/i386.c (struct ptt): Change type from int to
15736 (ix86_default_align): Set default values.
15737 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
15738 max skip conditionally.
15739 * config/i386/iamcu.h (SUBALIGN_LOG): New.
15740 (ASM_OUTPUT_MAX_SKIP_ALIGN):
15741 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
15742 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15743 max skip conditionally.
15744 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
15745 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
15746 * config/i386/x86-64.h (SUBALIGN_LOG): New.
15747 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
15748 max skip conditionally.
15749 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15750 * config/ia64/ia64.c (ia64_option_override): Set default values
15751 for alignment options.
15752 * config/m68k/m68k.c: Handle new str_align_* options.
15753 * config/mips/mips.c (mips_set_compression_mode): Change
15755 (mips_option_override): Set default values for options.
15756 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
15758 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15760 * config/rx/rx.c (rx_option_override): Likewise.
15761 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
15762 (LABEL_ALIGN): Use align_labels_log.
15763 (LOOP_ALIGN): Use align_loops_align.
15764 * config/s390/s390.c (s390_asm_output_function_label): Use new
15766 * config/sh/sh.c (sh_override_options_after_change):
15767 Change type of constants.
15768 * config/spu/spu.c (spu_sched_init): Likewise.
15769 * config/sparc/sparc.c (sparc_option_override): Set default
15770 values for options.
15771 * config/visium/visium.c (visium_option_override): Likewise.
15772 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
15773 emit p2align format with last argument if it's not needed.
15774 * doc/invoke.texi: Document extended format of -falign-*.
15775 * final.c: Use align_labels alignment.
15776 * flags.h (struct target_flag_state): Change type to use
15778 (struct align_flags_tuple): New.
15779 (struct align_flags): Likewise.
15780 (align_loops_log): Redefine macro to use new types.
15781 (align_loops_max_skip): Redefine macro to use new types.
15782 (align_jumps_log): Redefine macro to use new types.
15783 (align_jumps_max_skip): Redefine macro to use new types.
15784 (align_labels_log): Redefine macro to use new types.
15785 (align_labels_max_skip): Redefine macro to use new types.
15786 (align_functions_log): Redefine macro to use new types.
15787 (align_loops): Redefine macro to use new types.
15788 (align_jumps): Redefine macro to use new types.
15789 (align_labels): Redefine macro to use new types.
15790 (align_functions): Redefine macro to use new types.
15791 (align_functions_max_skip): Redefine macro to use new types.
15792 (align_loops_value): New macro.
15793 (align_jumps_value): New macro.
15794 (align_labels_value): New macro.
15795 (align_functions_value): New macro.
15796 * function.c (invoke_set_current_function_hook): Propagate
15797 alignment values from flags to global variables default in
15799 * ipa-icf.c (sem_function::equals_wpa): Use
15800 cl_optimization_option_eq instead of memcmp.
15801 * lto-streamer.h (cl_optimization_stream_out): Support streaming
15803 (cl_optimization_stream_in): Likewise.
15804 * optc-save-gen.awk: Support strings in cl_optimization.
15805 * opth-gen.awk: Likewise.
15806 * opts.c (finish_options): Remove error checking of invalid
15808 (MAX_CODE_ALIGN): Remove.
15809 (MAX_CODE_ALIGN_VALUE): Likewise.
15810 (parse_and_check_align_values): New function.
15811 (check_alignment_argument): Likewise.
15812 (common_handle_option): Use check_alignment_argument.
15813 * opts.h (parse_and_check_align_values): Declare.
15814 * toplev.c (init_alignments): Remove.
15815 (read_log_maxskip): New.
15816 (parse_N_M): Likewise.
15817 (parse_alignment_opts): Likewise.
15818 (backend_init_target): Remove usage of init_alignments.
15819 * toplev.h (parse_alignment_opts): Declare.
15820 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
15822 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
15823 * tree.c (cl_option_hasher::equal): New.
15824 * varasm.c: Use new global macros.
15826 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15828 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
15829 Use a simpler align directive also if MAXSKIP = ALIGN-1.
15830 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15831 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15832 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15833 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15834 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
15835 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
15836 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
15837 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15838 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15839 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15840 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15842 2018-07-04 Martin Liska <mliska@suse.cz>
15843 Jonathan Wakely <jwakely@redhat.com>
15845 * coverage.c: Use correct type.
15846 * doc/invoke.texi: Language correction.
15848 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
15851 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
15852 ENDBRANCH for non-tail call which may return via indirect branch.
15853 * doc/extend.texi: Document indirect_return attribute.
15855 2018-07-03 Martin Sebor <msebor@redhat.com>
15857 PR tree-optimization/86274
15858 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
15860 (format_floating): Correct handling of infinities and NaNs.
15862 2018-07-03 Martin Sebor <msebor@redhat.com>
15864 * print-tree.c (print_real_cst): New function.
15865 (print_node_brief): Call it.
15866 (print_node): Ditto.
15868 2018-07-03 Jeff Law <law@redhat.com>
15870 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
15871 into a single pattern.
15873 * config/h8300/h8300.md (ors code_iterator): New.
15874 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
15875 a single pattern and single splitter.
15876 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
15877 (iorqi3_1, xorqi3_1): Likewise.
15878 (iorqi3, xorqi3 expanders): Similarly.
15880 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
15881 (movmd_internal) into a single pattern using the P mode iterator.
15882 (movmd splitters): Similarly.
15883 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
15884 (movsd splitters): Similarly.
15886 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
15887 ADDB, ADDW and ADDL into a single ADD attribute which selects the
15888 right table based on the size of the operand.
15889 * config/h8300/h8300.md (length_table): Corresponding changes. All
15890 references to "addb", "addw" and "addl" changed to "add".
15891 (btst patterns): Merge two variants into a single pattern.
15892 (tstqi, tsthi): Likewise.
15893 (addhi3_incdec, addsi3_incdec): Likewise.
15894 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
15895 (mulhi3, mulsi3): Likewise.
15896 (udivhi3, udivsi3): Likewise.
15897 (divhi3, divsi3): Likewise.
15898 (andorqi3, andorhi3, andorsi3): Likewise.
15900 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
15903 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
15904 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
15906 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15908 PR tree-optimization/85694
15909 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
15910 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
15912 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
15913 (<u>avg<mode>3_ceil): New patterns.
15915 2018-07-03 David Malcolm <dmalcolm@redhat.com>
15917 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
15918 scan-tree-dump directive.
15919 * gcc.dg/vect/slp-perm-2.c: Likewise.
15920 * gcc.dg/vect/slp-perm-3.c: Likewise.
15921 * gcc.dg/vect/slp-perm-5.c: Likewise.
15922 * gcc.dg/vect/slp-perm-6.c: Likewise.
15923 * gcc.dg/vect/slp-perm-7.c: Likewise.
15924 * gcc.dg/vect/slp-perm-8.c: Likewise.
15926 2018-07-03 Marek Polacek <polacek@redhat.com>
15928 PR middle-end/86202
15929 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
15931 2018-07-03 Richard Biener <rguenther@suse.de>
15934 * tree-ssa-structalias.c (find_func_clobbers): Properly
15935 handle indirect calls.
15937 2018-07-03 Jeff Law <law@redhat.com>
15939 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
15940 (shifts): New code iterator.
15941 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
15942 expander. Fix HImode handling on H8/SX.
15943 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
15944 (subqi3, subhi3, subsi3 expanders): Likewise.
15945 (andqi3, andhi3, andsi3 expanders): Likewise.
15946 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
15947 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
15948 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
15949 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
15950 (zero_extendqihi2, zero_extendqisi2): Likewise.
15951 (extendqihi2, extendqisi2): Likewise.
15952 (rotlqi3, rotlhi3, rotlsi3): Likewise.
15953 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
15954 (rotlqi3_1, rotlhi3_1): Likewise.
15955 (logicalhi3_sn, logicalsi3_sn): Likewise.
15956 (logicalhi3, logicalsi3): Likewise.
15958 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15960 * tree-vect-patterns.c (vect_recog_rotate_pattern)
15961 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
15962 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
15963 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
15964 type to append_pattern_def_seq instead of creating a stmt_vec_info
15966 (build_mask_conversion): Likewise. Remove vinfo argument.
15967 (vect_add_conversion_to_patterm): Likewise, renaming to...
15968 (vect_add_conversion_to_pattern): ...this.
15969 (vect_recog_mask_conversion_pattern): Update call to
15970 build_mask_conversion. Pass the vector type to
15971 append_pattern_def_seq here too.
15972 (vect_recog_gather_scatter_pattern): Update call to
15973 vect_add_conversion_to_pattern.
15975 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15977 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
15978 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15979 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
15980 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
15981 STMT_VINFO_PATTERN_DEF_SEQ to null here.
15982 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
15983 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
15984 append_pattern_def_seq instead of new_pattern_def_seq.
15985 (vect_recog_divmod_pattern): Do both of the above.
15986 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
15989 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15991 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
15992 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
15993 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
15994 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
15995 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
15996 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
15997 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
15998 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
15999 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
16000 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
16001 parameter with a single stmt_vec_info.
16002 (vect_recog_func_ptr): Likewise.
16003 (vect_recog_gather_scatter_pattern): Likewise, folding in...
16004 (vect_try_gather_scatter_pattern): ...this.
16005 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
16006 the stmt_vec_info of the statement to be matched. Don't clear
16007 STMT_VINFO_RELATED_STMT.
16008 (vect_pattern_recog): Update call accordingly.
16010 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16012 PR tree-optimization/85694
16013 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
16014 (uavgM3_ceil): Document new optabs.
16015 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
16016 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
16018 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
16019 (savg_ceil_optab): New optabs.
16020 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
16021 (vect_vect_recog_func_ptrs): Add it.
16022 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
16023 constant directly from the associated lhs.
16025 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16027 * tree-vect-patterns.c (vect_split_statement): New function.
16028 (vect_convert_input): Use it to try to split an existing cast.
16030 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16032 * poly-int.h (print_hex): New function.
16033 * dumpfile.h (dump_dec, dump_hex): Declare.
16034 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
16035 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
16036 min_input_precision, operation_precision and operation_sign.
16037 * tree-vect-patterns.c (vect_get_range_info): New function.
16038 (vect_same_loop_or_bb_p, vect_single_imm_use)
16039 (vect_operation_fits_smaller_type): Delete.
16040 (vect_look_through_possible_promotion): Add an optional
16041 single_use_p parameter.
16042 (vect_recog_over_widening_pattern): Rewrite to use new
16043 stmt_vec_info infomration. Handle one operation at a time.
16044 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
16045 (vect_truncatable_operation_p, vect_set_operation_type)
16046 (vect_set_min_input_precision): New functions.
16047 (vect_determine_min_output_precision_1): Likewise.
16048 (vect_determine_min_output_precision): Likewise.
16049 (vect_determine_precisions_from_range): Likewise.
16050 (vect_determine_precisions_from_users): Likewise.
16051 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
16052 (vect_vect_recog_func_ptrs): Put over_widening first.
16054 (vect_pattern_recog): Call vect_determine_precisions.
16056 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
16058 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
16059 statements that have been replaced by further pattern statements.
16060 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
16062 2018-07-03 Richard Biener <rguenther@suse.de>
16064 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
16065 always set *dt. Dump vectype in vectype overload.
16066 * dumpfile.h (dump_gimple_expr): New function.
16067 (dump_gimple_expr_loc): Likewise.
16068 * dumpfile.c (dump_gimple_expr): New function.
16069 (dump_gimple_expr_loc): Likewise.
16071 2018-07-02 Jeff Law <law@redhat.com>
16073 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
16074 the H8/300, H8/300H and H8/S variants into a single pattern.
16075 (movhi_h8300, movqi_h8300hs): Similarly.
16076 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
16077 (QHI mode iterator): New.
16079 * config/h8300/h8300.md: Remove trailing whitespace.
16081 2018-07-02 Jim Wilson <jimw@sifive.com>
16083 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
16084 instead of emit_insn for interrupt returns.
16085 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
16086 (riscv_sret, riscv_uret): Likewise.
16088 2018-07-02 David Malcolm <dmalcolm@redhat.com>
16090 * pretty-print.c (selftest::test_pp_format): Move save and restore
16091 of quotes to class auto_fix_quotes, and add an instance.
16092 * selftest.c: Include "intl.h".
16093 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
16094 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
16095 * selftest.h (selftest::auto_fix_quotes): New class.
16097 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
16099 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
16100 (aarch64_sve_prepare_conditional_op): Remove.
16101 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
16102 Allow aarch64_simd_reg_or_zero as select operand; remove
16103 the aarch64_sve_prepare_conditional_op call.
16104 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
16105 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
16106 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
16107 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
16108 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
16109 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
16110 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
16111 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
16112 and a splitters to match all of the *_any patterns.
16113 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
16115 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
16116 (SVE_COND_FP_BINARY_REV): Remove.
16117 (sve_int_op_rev, sve_fp_op_rev): New.
16118 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
16119 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
16120 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
16121 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
16122 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
16123 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
16124 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
16125 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
16126 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
16128 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
16129 Remove match_dup 1 from the inner unspec.
16130 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
16132 * config/aarch64/aarch64.md (movprfx): New attr.
16133 (length): Default movprfx to 8.
16134 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
16135 (*madd<SVE_I>, *msub<SVE_I): Likewise.
16136 (*<su>mul<SVE_I>3_highpart): Likewise.
16137 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
16138 (*v<ASHIFT><SVE_I>3): Likewise.
16139 (*<su><MAXMIN><SVE_I>3): Likewise.
16140 (*<su><MAXMIN><SVE_F>3): Likewise.
16141 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
16142 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
16143 (*div<SVE_F>4): Likewise.
16145 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
16147 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
16150 2018-07-02 Richard Biener <rguenther@suse.de>
16152 PR tree-optimization/86363
16153 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
16154 memset argument refers to a non-variable address.
16156 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
16158 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
16159 {PLUS,MINUS}_EXPR code to...
16160 (adjust_symbolic_bound): ...here,
16161 (combine_bound): ...here,
16162 (set_value_range_with_overflow): ...and here.
16164 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
16166 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
16168 (extract_range_from_abs_expr): ...here.
16170 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
16172 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
16173 -fno-omit-frame-pointer when not optimizing.
16175 2018-07-02 Martin Liska <mliska@suse.cz>
16178 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
16179 (propagate_nothrow): Likewise.
16181 2018-07-02 Martin Liska <mliska@suse.cz>
16184 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
16186 2018-07-02 David Malcolm <dmalcolm@redhat.com>
16188 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
16189 function in r262149, changing "loc" param from source_location to
16190 const dump_location_t &.
16191 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
16192 declaration, as above.
16194 2018-07-01 Paul Koning <ni1d@arrl.net>
16196 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16197 -munit-asm, -mgnu-asm, -mdec-asm.
16198 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
16199 (pdp11_output_labelref): New.
16200 (pdp11_output_def): New.
16201 (pdp11_output_addr_vec_elt): New.
16202 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
16203 %# and %@ format codes.
16204 (pdp11_option_override): New.
16205 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
16206 (pdp11_output_ident): New.
16207 (pdp11_asm_named_section): New.
16208 (pdp11_asm_init_sections): New.
16209 (pdp11_file_start): New.
16210 (pdp11_file_end): New.
16211 (output_ascii): Use .ascii/.asciz for -mdec-asm.
16212 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
16213 %o, like %c but octal.
16214 (pdp11_option_override): New.
16215 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
16217 (DATA_SECTION_ASM_OP): Ditto.
16218 (READONLY_DATA_SECTION_ASM_OP): New.
16219 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
16220 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
16221 (ASM_OUTPUT_LABELREF): Ditto.
16222 (ASM_OUTPUT_DEF): Ditto.
16223 (ASM_OUTPUT_EXTERNAL): New.
16224 (ASM_OUTPUT_SOURCE_FILENAME): New.
16225 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
16226 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
16227 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
16228 %# and %@ format codes.
16229 * config/pdp11/pdp11.opt (mgnu-asm): New.
16230 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
16231 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
16232 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
16234 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
16236 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
16237 dereferencing path[] beyond its length.
16239 (debug_all_paths): New.
16240 (rewire_first_differing_edge): New.
16241 (adjust_paths_after_duplication): New.
16242 (duplicate_thread_path): Call adjust_paths_after_duplication.
16244 (thread_through_all_blocks): Add new argument to
16245 duplicate_thread_path.
16247 2018-06-30 Jim Wilson <jimw@sifive.com>
16249 * config/riscv/predicates.md (p2m1_shift_operand): New.
16250 (high_mask_shift_operand): New.
16251 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
16252 pattern using p2m1_shift_operand.
16253 (lshsi3_zero_extend_3+2): New combiner pattern using
16254 high_mask_shift_operand.
16256 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16258 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
16260 (vect_recog_rotate_pattern): ...here.
16261 (vect_convert_input): Try to insert casts of invariants in the
16263 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
16264 preheader to be empty.
16266 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16268 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
16269 vector type. If given, install it in the new statement's
16270 STMT_VINFO_VECTYPE.
16271 (vect_element_precision): New function.
16272 (vect_unpromoted_value): New struct.
16273 (vect_unpromoted_value::vect_unpromoted_value): New function.
16274 (vect_unpromoted_value::set_op): Likewise.
16275 (vect_look_through_possible_promotion): Likewise.
16276 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
16277 (vect_widened_op_tree, vect_convert_input): Likewise.
16278 (vect_convert_inputs, vect_convert_output): Likewise.
16279 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
16280 to handle the optional cast of the multiplication result and
16281 vect_widened_op_tree to detect the widened multiplication itself.
16282 Do not require the input and output of promotion casts to have
16283 the same sign, but base the signedness of the operation on the
16284 input rather than the result. If the pattern includes two
16285 promotions, check that those promotions have the same sign.
16286 Do not restrict the MULT_EXPR handling to a double-width result;
16287 handle quadruple-width results and wider. Use vect_convert_inputs
16288 to convert the inputs to the common type.
16289 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
16290 to handle the optional cast of the ABS result. Also allow a sign
16291 change or a sign extension between the ABS and MINUS.
16292 Use vect_widened_op_tree to detect the widened subtraction and use
16293 vect_convert_inputs to convert the inputs to the common type.
16294 (vect_handle_widen_op_by_const): Delete.
16295 (vect_recog_widen_op_pattern): New function.
16296 (vect_recog_widen_mult_pattern): Use it.
16297 (vect_recog_widen_shift_pattern): Likewise.
16298 (vect_recog_widen_sum_pattern): Use
16299 vect_look_through_possible_promotion to handle the promoted
16302 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16304 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
16305 the containing gimple_seq *.
16306 * gimple-iterator.h (gsi_for_stmt): Declare it.
16307 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
16308 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
16309 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
16310 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
16311 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
16313 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
16315 (vect_mark_pattern_stmts): ...here. Handle cases in which the
16316 statement being replaced is part of an existing pattern
16317 definition sequence, inserting the new pattern statements before
16319 (vect_pattern_recog_1): Don't return a bool. If the statement
16320 is already part of a pattern, instead apply pattern matching
16321 to the pattern definition statements. Don't clear the
16322 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
16323 (vect_pattern_recog): Don't break after the first match;
16324 continue processing the pattern definition statements instead.
16325 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
16327 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16329 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
16330 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16331 (vect_recog_widen_sum_pattern): Use it.
16333 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16335 * tree-vect-loop.c (vectorizable_reduction): Assert that the
16336 phi is not a pattern statement and has not been replaced by
16337 a pattern statement.
16338 * tree-vect-patterns.c (type_conversion_p): Don't check
16339 STMT_VINFO_IN_PATTERN_P.
16340 (vect_recog_vector_vector_shift_pattern): Likewise.
16341 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
16342 the pattern statement rather than the original statement; check
16343 directly for a WIDEN_MULT_EXPR here.
16344 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
16345 vect_is_simple_use to return the pattern statement rather
16346 than the original statement; use is_pattern_stmt_p to check
16347 for such a pattern statement.
16348 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
16349 to return the pattern statement rather than the original statement;
16350 don't do the same transformation here.
16351 (vect_is_simple_use): If the defining statement has been replaced
16352 by a pattern statement, return the pattern statement instead.
16353 Remove the corresponding (local) transformation from the vectype
16356 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
16358 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
16359 end and default to null.
16360 * tree-vect-loop.c (vect_create_epilog_for_reduction)
16361 (vectorizable_reduction): Update calls accordingly, dropping the
16362 gimple ** argument if the passed-back statement isn't needed.
16363 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
16364 (vect_recog_rotate_pattern): Likewise.
16365 (vect_recog_mask_conversion_pattern): Likewise.
16366 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
16367 (vect_mask_constant_operand_p): Likewise.
16368 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
16369 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
16370 (get_group_load_store_type, get_load_store_type): Likewise.
16371 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
16372 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
16373 (vectorizable_conversion, vectorizable_assignment): Likewise.
16374 (vectorizable_shift, vectorizable_operation): Likewise.
16375 (vectorizable_store, vect_is_simple_cond): Likewise.
16376 (vectorizable_condition, vectorizable_comparison): Likewise.
16377 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
16378 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
16379 and move it to the end. Cope with null def_stmt_outs.
16381 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
16383 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
16385 2018-06-29 Jeff Law <law@redhat.com>
16387 * config/v850/v850.c (v850_legitimate_address_p): Handle large
16388 displacements for TARGET_V850E2V3 and newer.
16389 (TARGET_LRA_P): Remove. Defaults to LRA now.
16390 * config/v850/v850.md (sign23byte_load): Remove.
16391 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
16392 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
16394 2018-06-29 Martin Liska <mliska@suse.cz>
16397 * coverage.c (coverage_init): Mangle full path name.
16398 * doc/invoke.texi: Document the change.
16399 * gcov-io.c (mangle_path): New.
16400 * gcov-io.h (mangle_path): Likewise.
16401 * gcov.c (mangle_name): Use mangle_path for path mangling.
16403 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16405 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
16406 if starting source register is not even.
16408 2018-06-29 Martin Liska <mliska@suse.cz>
16410 PR tree-optimization/86263
16411 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
16412 Make edge redirection.
16414 2018-06-29 David Malcolm <dmalcolm@redhat.com>
16416 * dumpfile.c (dump_loc): Add indentation based on scope depth.
16417 (dump_scope_depth): New variable.
16418 (get_dump_scope_depth): New function.
16419 (dump_begin_scope): New function.
16420 (dump_end_scope): New function.
16421 * dumpfile.h (get_dump_scope_depth): New declaration.
16422 (dump_begin_scope): New declaration.
16423 (dump_end_scope): New declaration.
16424 (class auto_dump_scope): New class.
16425 (AUTO_DUMP_SCOPE): New macro.
16426 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
16429 2018-06-29 Richard Biener <rguenther@suse.de>
16431 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
16432 compute_all_dependences succeeds.
16433 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
16434 exceed --param loop-max-datarefs-for-datadeps.
16436 2018-06-29 Jakub Jelinek <jakub@redhat.com>
16438 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
16440 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
16443 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
16444 alternative 0 in preferred_for_speed attribute.
16446 2018-06-28 Paul Koning <ni1d@arrl.net>
16448 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
16449 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
16450 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
16451 * config/pdp11/pdp11.md: Correct "length" attribute calculation
16452 for shift insn patterns.
16454 2018-06-28 David Malcolm <dmalcolm@redhat.com>
16456 * cgraph.c (cgraph_node::get_body): Replace assignments to
16457 "dump_file" with calls to set_dump_file.
16458 * dumpfile.c (alt_dump_file): Make static, and group with...
16459 (alt_flags): ...this definition.
16460 (dumps_are_enabled): New variable.
16461 (refresh_dumps_are_enabled): New function.
16462 (set_dump_file): New function.
16463 (set_alt_dump_file): New function.
16464 (gcc::dump_manager::dump_start): Replace assignments to
16465 "dump_file" and "alt_dump_file" with calls to set_dump_file and
16467 (gcc::dump_manager::dump_finish): Likewise.
16468 * dumpfile.h (alt_dump_file): Delete decl.
16469 (dumps_are_enabled): New variable decl.
16470 (set_dump_file): New function decl.
16471 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
16473 * tree-nested.c (lower_nested_functions): Replace assignments to
16474 "dump_file" with calls to set_dump_file.
16476 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
16478 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
16479 goto_locus of each outgoing edge of each basic block.
16481 2018-06-28 Richard Biener <rguenther@suse.de>
16483 * dwarf2out.c (decl_scope_table): Remove.
16484 (push_decl_scope): Likewise.
16485 (pop_decl_scope): Likewise.
16486 (gen_type_die_for_member): Do not call push/pop_decl_scope.
16487 (gen_struct_or_union_type_die): Likewise.
16488 (gen_tagged_type_die): Likewise.
16489 (dwarf2out_init): Do not initialize decl_scope_table.
16490 (dwarf2out_c_finalize): Do not free it.
16492 2018-06-28 Richard Biener <rguenther@suse.de>
16494 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
16495 deciding whether to not re-use a DIE.
16497 2018-06-28 Richard Biener <rguenther@suse.de>
16499 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
16500 DW_AT_abstract_origin attribute.
16502 2018-06-28 Martin Liska <mliska@suse.cz>
16504 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
16505 Use newly introduced constants.
16506 * tree-switch-conversion.h (struct jump_table_cluster):
16507 Define max_ratio_for_size and max_ratio_for_speed.
16509 2018-06-28 Martin Liska <mliska@suse.cz>
16511 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16512 Add new checking assert to catch invalid state.
16513 (jump_table_cluster::can_be_handled): Handle single case
16515 (jump_table_cluster::is_beneficial): Bail out for such case.
16516 (bit_test_cluster::find_bit_tests):
16517 Add new checking assert to catch invalid state.
16518 (bit_test_cluster::can_be_handled): Handle single case
16520 (bit_test_cluster::is_beneficial): Bail out for such case.
16521 (switch_decision_tree::analyze_switch_statement):
16524 2018-06-28 Martin Liska <mliska@suse.cz>
16526 * common.opt: Introduce -completion option.
16527 * gcc.c (driver_handle_option): Handle it.
16528 (driver::main): Print completions if completion
16530 * opt-suggestions.c (option_proposer::get_completions):
16532 (option_proposer::suggest_completion): Likewise.
16533 (option_proposer::find_param_completions): Likewise.
16534 (verify_autocompletions): Likewise.
16535 (test_completion_valid_options): Likewise.
16536 (test_completion_valid_params): Likewise.
16537 (in_completion_p): Likewise.
16538 (empty_completion_p): Likewise.
16539 (test_completion_partial_match): Likewise.
16540 (test_completion_garbage): Likewise.
16541 (opt_proposer_c_tests): Likewise.
16542 * opt-suggestions.h: Declare new functions.
16543 * opts.c (common_handle_option): Handle OPT__completion_.
16544 * selftest-run-tests.c (selftest::run_tests): Add
16545 opt_proposer_c_tests.
16546 * selftest.c (assert_str_startswith): New.
16547 * selftest.h (assert_str_startswith): Likewise.
16548 (opt_proposer_c_tests): New.
16549 (ASSERT_STR_STARTSWITH): Likewise.
16551 2018-06-28 Martin Liska <mliska@suse.cz>
16553 * Makefile.in: Add opt-suggestions.o.
16554 * gcc-main.c: Include opt-suggestions.h.
16555 * gcc.c (driver::driver): Likewise.
16556 (driver::~driver): Remove m_option_suggestions.
16557 (driver::build_option_suggestions): Moved to option_proposer.
16558 (driver::suggest_option): Likewise.
16559 (driver::handle_unrecognized_options): Use option_proposer.
16560 * gcc.h (class driver): Add new memver m_option_proposer.
16561 * opt-suggestions.c: New file.
16562 * opt-suggestions.h: New file.
16564 2018-06-28 Martin Liska <mliska@suse.cz>
16566 * vec.h (class auto_string_vec): New (moved from auto_argvec).
16567 (auto_string_vec::~auto_string_vec): Likewise.
16569 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
16571 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
16572 prevent_decl_creation_for_types fields up and add reset_location field.
16573 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
16574 statement if id->reset_location is true.
16575 (copy_edges_for_bb): Do not set goto_locus on the new edges if
16576 id->reset_location is true.
16577 (copy_phis_for_bb): Force input_location on the arguments if
16578 id->reset_location is true.
16579 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
16580 is set on the function to be inlined.
16582 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
16584 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
16586 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
16588 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
16589 registers for Pmode.
16590 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
16591 hard registers for the clobbered pseudo.
16593 2018-06-27 Paul Koning <ni1d@arrl.net>
16595 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
16596 mutually exclusive options.
16597 * config/pdp11/constraints.md (h): New constraint.
16598 (O): Update definition to match shift code generation.
16599 (D): New constraint.
16600 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
16602 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
16604 (output_jump): Change arguments.
16605 (pdp11_fixed_cc_regs): New function.
16606 (pdp11_cc_mode): Ditto.
16607 (pdp11_expand_shift): Ditto.
16608 (pdp11_assemble_shift): Ditto.
16609 (pdp11_small_shift): Ditto.
16610 (pdp11_branch_cost): Remove.
16611 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
16613 (pdp11_register_move_cost): Update for CC registers.
16614 (pdp11_rtx_costs): Add case for LSHIFTRT.
16615 (pdp11_output_jump): Add CCNZ mode conditional branches.
16616 (notice_update_cc_on_set): Remove.
16617 (pdp11_cc_mode): New function.
16618 (simple_memory_operand): Correct pre/post decrement case.
16619 (no_side_effect_operand): New function.
16620 (pdp11_regno_reg_class): Add CC_REGS class.
16621 (pdp11_fixed_cc_regs): New function.
16622 (pdp11_small_shift): New function.
16623 (pdp11_expand_shift): New function to expand shift insns.
16624 (pdp11_assemble_shift): New function to output shifts.
16625 (pdp11_branch_cost): Remove.
16626 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
16627 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
16628 (WCHAR_TYPE): Ditto.
16629 (PTRDIFF_TYPE): Ditto.
16630 (ADJUST_INSN_LENGTH): New macro.
16631 (FIXED_REGISTERS): Add CC registers.
16632 (CALL_USED_REGISTERS): Ditto.
16633 (reg_class): Ditto.
16634 (REG_CLASS_NAMES): Ditto.
16635 (REG_CLASS_CONTENTS): Ditto.
16636 (SELECT_CC_MODE): Use new function.
16637 (TARGET_FLAGS_REGNUM): New macro.
16638 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
16639 (cc0_reg_rtx): Remove.
16640 (CC_STATUS_MDEP): Remove.
16641 (CC_STATUS_MDEFP_INIT): Remove.
16642 (CC_IN_FPU): Remove.
16643 (NOTICE_UPDATE_CC): Remove.
16644 (REGISTER_NAMES): Add CC registers.
16645 (BRANCH_COST): Change to constant 1.
16646 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
16648 * config/pdp11/pdp11.opt (mbcopy): Remove.
16649 (mbcopy-builtin): Remove.
16650 (mbranch-cheap): Remove.
16651 (mbranch-expensive): Remove.
16652 * config/pdp11/predicates.md (expand_shift_operand): Update to
16653 match shift code generation.
16654 (ccnz_operator): New predicate.
16655 * doc/invoke.texi (PDP-11 Options): Remove deleted options
16656 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
16657 Remove non-existent option -mabshi, -mno-abshi. Document mutually
16659 * doc/md.texi (PDP-11): Document new D and h constraints. Update
16660 description of O constraint.
16662 2018-06-27 Jeff Law <law@redhat.com>
16663 Austin Law <austinklaw@gmail.com>
16665 * config/v850/v850.md (addsi3_set_flags): New pattern.
16666 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
16667 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
16668 (zero_extendhisi2_v850_set_flags): Likewise.
16669 (zero_extendqisi2_v850_set_flags): Likewise.
16670 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
16671 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
16672 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
16674 * config/v850/v850-protos.h (notice_update_cc): Remove.
16675 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
16676 (v850_print_operand): Handle 'D' and "d".
16677 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
16678 Add handling of arithmetic/logical operations compared against zero.
16679 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
16680 Do not look at v850_compare_op, instead get mode from last argument.
16681 (v850_gen_compare): Remove
16682 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
16683 after reload for prologue insns.
16684 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
16686 (construct_save_jarl): Likewise.
16687 (TARGET_FLAGS_REGNUM): Define.
16688 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
16689 (NOTICE_UPDATE_CC): Remove.
16690 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
16691 than cc0. Conditionalize on reload_completed.
16692 (cmpsi_insn, setfcc_insn): Likewise.
16693 (tst1 splitter): Turn into define_and_split which sets the flags
16695 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
16696 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
16697 (cstoresf4, cstoredf4): Clobber the flags.
16698 (cmpsi, cmpsf, cmpdf): Remove expanders.
16699 (setf_insn): Remove pattern.
16700 (addsi3): Turn into define_and_split which clobbers the flags after
16701 reload and a suitable pattern (addsi3_clobber_flags) for use after
16703 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
16704 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
16705 (ashrsi3, ashrsi3_v850e2): Likewise.
16706 (bins): Clobber the flags.
16707 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
16708 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
16709 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
16710 (call_value_internal_short, call_value_internal_long): Likewise.
16711 (callt_save_interrupt, callt_return_interrupt): Likewise.
16712 (save_interrupt, return_interrupt): Likewise.
16713 (callt_save_all_interrupt, save_all_interrupt): Likewise.
16714 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
16715 (restore_all_interrupt, _restore_all_interrupt): Likewise.
16716 (All FP comparisons): Only allow after reload has completed.
16718 (divh, divhu): Tweak output template.
16719 (branch_z_normal, branch_z_invert): Remove
16720 (branch_nz_normal, branch_nz_invert): Likewise.
16721 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
16723 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
16724 * config/v850/v850.c (notice_update_cc): Remove.
16725 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
16726 (CC_NO_CARRY): Likewise.
16727 (NOTICE_UPDATE_CC): Define to nothing.
16728 * config/v850/v850.md: Remove block comment on cc0 handling
16729 Remove "cc" attribute from all patterns. Remove cc_status handling
16730 from all patterns. Minor formatting fixes.
16732 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16734 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
16735 (cortex-a76.cortex-a55): Likewise.
16736 * config/aarch64/aarch64-tune.md: Regenerate.
16737 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
16738 cortex-a76.cortex-a55.
16740 2018-06-27 Jeff Law <law@redhat.com>
16742 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
16743 (MULTILIB_DIRNAMES): Similarly.
16745 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
16747 * gimple.h (gimple_return_retbnd): Delete.
16748 (gimple_return_set_retbnd): Likewise.
16749 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
16750 gimple_return_set_retbnd.
16751 * gimple-pretty-print.c (dump_gimple_return): Remove call to
16752 gimple_return_retbnd and adjust.
16753 * tree-inline.h (struct copy_body_data): Remove retbnd field.
16754 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
16755 Explicitly return NULL in a couple more cases. Move assertion
16756 on debug statements and remove unreachable code.
16757 (reset_debug_binding): Do not test id->retbnd.
16758 (expand_call_inline): Do not set it.
16760 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
16762 * configure.ac: Add --disable-gcov option.
16763 * configure: Regenerate.
16764 * Makefile.in: Honour @enable_gcov@.
16765 * doc/install.texi: Document --disable-gcov.
16767 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16769 * config/arm/arm-cpus.in (cortex-a76): New entry.
16770 (cortex-a76.cortex-a55): Likewise.
16771 * config/arm/arm-tables.opt: Regenerate.
16772 * config/arm/arm-tune.md: Likewise.
16773 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
16774 * doc/invoke.texi (ARM Options): Document cortex-a76 and
16775 cortex-a76.cortex-a55.
16777 2018-06-27 Tamar Christina <tamar.christina@arm.com>
16780 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
16782 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
16784 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
16786 (EPILOGUE_USES): Likewise.
16788 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
16790 * tree-inline.c (remap_location): New function extracted from...
16791 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
16792 (copy_phis_for_bb): ...here. Call remap_location.
16793 (copy_cfg_body): Adjust call to copy_edges_for_bb.
16795 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
16797 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
16798 unaligned vsx for 16B memset.
16800 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
16803 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
16804 ieee128_float_type_node to long_double_type_node unless
16805 TARGET_LONG_DOUBLE_128 is set.
16807 2018-06-26 David Malcolm <dmalcolm@redhat.com>
16809 * cfgloop.c (get_loop_location): Convert return type from
16810 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
16811 by implicit construction from rtx_insn *, and using
16812 dump_user_location_t::from_function_decl for the fallback case.
16813 * cfgloop.h (get_loop_location): Convert return type from
16814 location_t to dump_user_location_t.
16815 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
16816 dump_printf_loc to pass in a dump_location_t rather than a
16817 location_t, via the gimple stmt.
16818 * coverage.c (get_coverage_counts): Update calls to
16819 dump_printf_loc to pass in dump_location_t rather than a
16821 * doc/optinfo.texi (Dump types): Convert example of
16822 dump_printf_loc from taking "locus" to taking "insn". Update
16823 description of the "_loc" calls to cover dump_location_t.
16824 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
16826 (dump_user_location_t::dump_user_location_t): New constructors,
16827 from gimple *stmt and rtx_insn *.
16828 (dump_user_location_t::from_function_decl): New function.
16829 (dump_loc): Make static.
16830 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
16831 const dump_location_t &.
16832 (dump_generic_expr_loc): Delete.
16833 (dump_printf_loc): Convert param "loc" from location_t to
16834 const dump_location_t &.
16835 (selftest::test_impl_location): New function.
16836 (selftest::dumpfile_c_tests): New function.
16837 * dumpfile.h: Include "profile-count.h".
16838 (class dump_user_location_t): New class.
16839 (struct dump_impl_location_t): New struct.
16840 (class dump_location_t): New class.
16841 (dump_printf_loc): Convert 2nd param from source_location to
16842 const dump_location_t &.
16843 (dump_generic_expr_loc): Delete.
16844 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
16845 const dump_location_t &.
16846 * gimple-fold.c (fold_gimple_assign): Update call to
16847 dump_printf_loc to pass in a dump_location_t rather than a
16848 location_t, via the gimple stmt.
16849 (gimple_fold_call): Likewise.
16850 * gimple-loop-interchange.cc
16851 (loop_cand::analyze_iloop_reduction_var): Update for change to
16852 check_reduction_path.
16853 (tree_loop_interchange::interchange): Update for change to
16854 find_loop_location.
16855 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
16856 change in return-type of find_loop_location.
16857 (graphite_regenerate_ast_isl): Likewise.
16858 * graphite-optimize-isl.c (optimize_isl): Likewise.
16859 * graphite.c (graphite_transform_loops): Likewise.
16860 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
16861 pass in a dump_location_t rather than a location_t, via the
16863 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16864 * ipa.c (walk_polymorphic_call_targets): Likewise.
16865 * loop-unroll.c (report_unroll): Convert "locus" param from
16866 location_t to dump_location_t.
16867 (decide_unrolling): Update for change to get_loop_location's
16869 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
16870 location_t to dump_user_location_t.
16871 (grid_find_single_omp_among_assignments_1): Updates calls to
16872 dump_printf_loc to pass in a dump_location_t rather than a
16873 location_t, via the gimple stmt.
16874 (grid_parallel_clauses_gridifiable): Convert "tloc" from
16875 location_t to dump_location_t. Updates calls to dump_printf_loc
16876 to pass in a dump_location_t rather than a location_t, via the
16878 (grid_inner_loop_gridifiable_p): Likewise.
16879 (grid_dist_follows_simple_pattern): Likewise.
16880 (grid_gfor_follows_tiling_pattern): Likewise.
16881 (grid_target_follows_gridifiable_pattern): Likewise.
16882 (grid_attempt_target_gridification): Convert initialization
16883 of local "grid" from memset to zero-initialization; FIXME: does
16884 this require C++11? Update call to dump_printf_loc to pass in a
16885 optinfo_location rather than a location_t, via the gimple stmt.
16886 * profile.c (read_profile_edge_counts): Updates call to
16887 dump_printf_loc to pass in a dump_location_t rather than a
16889 (compute_branch_probabilities): Likewise.
16890 * selftest-run-tests.c (selftest::run_tests): Call
16892 * selftest.h (dumpfile_c_tests): New decl.
16893 * tree-loop-distribution.c (pass_loop_distribution::execute):
16894 Update for change in return type of find_loop_location.
16895 * tree-parloops.c (parallelize_loops): Likewise.
16896 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
16897 "locus" from location_t to dump_user_location_t.
16898 (canonicalize_loop_induction_variables): Likewise.
16899 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
16900 for change in return type of find_loop_location.
16901 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
16902 to dump_printf_loc to pass in a dump_location_t rather than a
16903 location_t, via the stmt.
16904 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
16906 * tree-vect-loop-manip.c (find_loop_location): Convert return
16907 type from source_location to dump_user_location_t.
16908 (vect_do_peeling): Update for above change.
16909 (vect_loop_versioning): Update for change in type of
16911 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
16912 from location_t to dump_user_location_t.
16913 (vect_estimate_min_profitable_iters): Update for change in type
16915 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
16916 location_t to dump_location_t.
16917 (vect_slp_bb): Update for change in type of vect_location.
16918 * tree-vectorizer.c (vect_location): Convert from source_location
16919 to dump_user_location_t.
16920 (try_vectorize_loop_1): Update for change in vect_location's type.
16921 (vectorize_loops): Likewise.
16922 (increase_alignment): Likewise.
16923 * tree-vectorizer.h (vect_location): Convert from source_location
16924 to dump_user_location_t.
16925 (find_loop_location): Convert return type from source_location to
16926 dump_user_location_t.
16927 (check_reduction_path): Convert 1st param from location_t to
16928 dump_user_location_t.
16929 * value-prof.c (check_counter): Update call to dump_printf_loc to
16930 pass in a dump_user_location_t rather than a location_t; update
16931 call to error_at for change in type of "locus".
16932 (check_ic_target): Update call to dump_printf_loc to
16933 pass in a dump_user_location_t rather than a location_t, via the
16936 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
16938 * config/s390/s390.h (enum processor_flags): Do not use
16939 default tune parameter when -march was specified.
16941 2018-06-26 Jakub Jelinek <jakub@redhat.com>
16944 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
16945 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
16948 2018-06-26 Richard Biener <rguenther@suse.de>
16950 PR tree-optimization/86287
16952 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
16953 (vect_analyze_loop): Initialize n_stmts.
16955 2018-06-26 Richard Biener <rguenther@suse.de>
16957 PR middle-end/86271
16958 * fold-const.c (fold_convertible_p): Pointer extension
16961 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
16964 * dwarf2out.c (loc_list_has_views): Adjust comments.
16965 (dw_loc_list): Split single cross-partition range with
16968 2018-06-25 Jeff Law <law@redhat.com>
16970 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
16971 on -mbig-switch by default.
16973 * config/v850/predicates.md (const_float_1_operand): Fix match_code
16975 (const_float_0_operand): Remove unused predicate.
16976 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
16977 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
16978 (recipsf2): New expander. Original pattern now called
16980 (recipdf2, recipdf2_insn): Similarly.
16981 (rsqrtsf2, rsqrtsf2_insn): Similarly
16982 (rsqrtdf2, rsqrtdf2_insn): Similarly
16984 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
16986 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
16987 Simplify logic for FreeBSD (twice).
16989 2018-06-25 Martin Sebor <msebor@redhat.com>
16991 PR tree-optimization/86204
16992 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
16993 a strnlen result if it's less than the length of the string.
16995 2018-06-25 Martin Sebor <msebor@redhat.com>
16997 PR tree-optimization/85700
16998 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
16999 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
17000 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
17002 2018-06-25 Martin Sebor <msebor@redhat.com>
17004 * doc/extend.texi (Zero-length arrays): Update and clarify.
17006 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
17008 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
17009 added IEEE/IBM long double multilib support on PowerPC little
17010 endian Linux systems.
17011 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
17012 (MULTILIB_DEFAULTS): Likewise.
17013 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17015 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
17016 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
17017 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
17019 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
17021 PR middle-end/86311
17022 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
17023 (REORDER_45): Likewise.
17025 2018-06-25 Jeff Law <law@redhat.com>
17027 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
17028 dividend to 32 bits. Adjust length.
17029 (udivmodhi4): Cleanup output template. Fix length.
17031 2018-06-25 Carl Love <cel@us.ibm.com>
17033 * config/rs6000/vsx.md: Change word selector to prefered location.
17035 2018-06-25 Richard Biener <rguenther@suse.de>
17037 PR tree-optimization/86304
17038 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
17039 epilogue-if-converted loops as well.
17041 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
17043 * lto-section-out.c (lto_begin_section): Do not print section
17044 name for noaddr and unnumbered dumps.
17046 2018-06-25 Richard Biener <rguenther@suse.de>
17048 * tree-vectorizer.h (struct vec_info_shared): New structure
17049 with parts split out from struct vec_info and loop_nest from
17050 struct _loop_vec_info.
17051 (struct vec_info): Adjust accordingly.
17052 (struct _loop_vec_info): Likewise.
17053 (LOOP_VINFO_LOOP_NEST): Adjust.
17054 (LOOP_VINFO_DATAREFS): Likewise.
17055 (LOOP_VINFO_DDRS): Likewise.
17056 (struct _bb_vec_info): Likewise.
17057 (BB_VINFO_DATAREFS): Likewise.
17058 (BB_VINFO_DDRS): Likewise.
17059 (struct _stmt_vec_info): Add dr_aux member.
17060 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
17061 (DR_MISALIGNMENT_UNINITIALIZED): New.
17062 (set_dr_misalignment): Adjust.
17063 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
17064 (vect_analyze_loop): Adjust prototype.
17065 (vect_analyze_loop_form): Likewise.
17066 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17067 Compute dependences lazily.
17068 (vect_record_base_alignments): Use shared datarefs/ddrs.
17069 (vect_verify_datarefs_alignment): Likewise.
17070 (vect_analyze_data_refs_alignment): Likewise.
17071 (vect_analyze_data_ref_accesses): Likewise.
17072 (vect_analyze_data_refs): Likewise.
17073 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
17074 constructor parameter for shared part.
17075 (vect_analyze_loop_form): Pass in shared part and adjust.
17076 (vect_analyze_loop_2): Pass in storage for the number of
17077 stmts. Move loop nest finding to the caller. Compute
17079 (vect_analyze_loop): Pass in shared part.
17080 (vect_transform_loop): Verify shared datarefs are unchanged.
17081 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
17082 constructor parameter for shared part.
17083 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
17084 (vect_slp_bb): Verify shared datarefs are unchanged before
17086 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
17088 (new_stmt_vec_info): Initialize DR_AUX misalignment to
17089 DR_MISALIGNMENT_UNINITIALIZED.
17090 * tree-vectorizer.c (vec_info::vec_info): Add constructor
17091 parameter for shared part.
17092 (vec_info::~vec_info): Adjust.
17093 (vec_info_shared::vec_info_shared): New.
17094 (vec_info_shared::~vec_info_shared): Likewise.
17095 (vec_info_shared::save_datarefs): Likewise.
17096 (vec_info_shared::check_datarefs): Likewise.
17097 (try_vectorize_loop_1): Construct shared part live for analyses
17098 of a single loop for multiple vector sizes.
17099 * tree-parloops.c (gather_scalar_reductions): Adjust.
17101 2018-06-25 Richard Biener <rguenther@suse.de>
17103 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
17104 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
17105 (vect_analyze_data_refs): Remove similar code from here and
17106 simplify accordingly.
17108 2018-06-25 Richard Biener <rguenther@suse.de>
17110 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
17111 for reverse storage order accesses rather than asserting
17112 they cannot happen here.
17114 2018-06-25 Tom de Vries <tdevries@suse.de>
17117 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
17118 Use data16 instead of .byte for insn prefix.
17120 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
17123 * parser.c (make_char_string_pack): Pass this literal chars
17124 through cpp_interpret_string.
17125 (cp_parser_userdef_numeric_literal): Check the result of
17126 make_char_string_pack.
17128 2018-06-24 Maya Rashish <coypu@sdf.org>
17130 * ginclude/stddef.h: Simplify conditions around avoiding
17131 re-definition of __size_t.
17133 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17135 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
17136 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
17138 2018-06-22 Maya Rashish <coypu@sdf.org>
17140 * doc/invoke.texi (mno-fancy-math-387): Update for changes
17141 made to OpenBSD and NetBSD through the years.
17143 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
17145 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
17146 behavior of vec_pack (vector double, vector double) to match
17147 behavior of vec_float2 (vector double, vector double).
17149 2018-06-22 Olivier Hainque <hainque@adacore.com>
17151 * gimplify.c (gimplify_function_tree): Prevent creation
17152 of a trampoline for the address of the current function
17153 passed to entry/exit instrumentation hooks.
17155 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
17158 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
17161 2018-06-22 Martin Liska <mliska@suse.cz>
17163 PR tree-optimization/86263
17164 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
17165 Bail out if is_enabled is false.
17166 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
17168 (jump_table_cluster::is_enabled): New function.
17170 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17172 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
17173 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
17174 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
17175 (lto_input_ts_binfo_tree_pointers): Likewise.
17176 * tree-streamer-out.c (streamer_write_tree_bitfields,
17177 write_ts_binfo_tree_pointers): Likewise.
17178 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
17180 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
17182 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
17184 2018-06-22 Martin Liska <mliska@suse.cz>
17186 * symbol-summary.h (get): Make it pure and inline move
17187 functionality from ::get function.
17188 (get): Remove and inline into ::get and ::get_create.
17189 (get_create): Move code from ::get function.
17191 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17194 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
17195 -x assembler-with-cpp.
17197 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17199 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
17200 _FILE_OFFSET_BITS=64 for C++.
17202 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
17204 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
17205 conversion insn that shows up when pr85657-3.c is compiled using
17206 IEEE 128-bit long double.
17207 (neg<mode>2_internal): Use the correct mode to check whether the
17208 mode is IBM extended.
17209 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
17210 multiply and divide external functions from being created more
17213 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17215 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
17217 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
17218 the edge can be forwarded.
17219 (cfg_layout_merge_blocks): Likewise.
17221 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17223 * except.c (finish_eh_generation): Commit edge insertions only after
17224 the EH edges have been redirected from post-landing to landing pads.
17226 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17228 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
17229 create_tmp_var_for to create the FRAME decl.
17230 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
17232 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
17234 * tree-inline.c (copy_edges_for_bb): Minor tweak.
17235 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
17236 debug statement when resetting its value.
17237 (expand_call_inline): Copy the locus of the call onto the assignment
17238 of the return value, if any. Use local variable in more cases.
17240 2018-06-21 Martin Liska <mliska@suse.cz>
17242 * ipa-pure-const.c (propagate_nothrow): Use
17243 funct_state_summaries->get.
17244 (dump_malloc_lattice): Likewise.
17245 (propagate_malloc): Likewise.
17247 2018-06-21 Richard Biener <rguenther@suse.de>
17249 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
17250 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
17251 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
17253 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
17254 BLOCK_ABSTRACT_ORIGIN unconditionally.
17256 2018-06-21 David Malcolm <dmalcolm@redhat.com>
17258 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
17260 * ipa-reference.c (ipa_reference_c_finalize): Delete
17261 ipa_ref_opt_sum_summaries and set it to NULL.
17263 2018-06-21 Tom de Vries <tdevries@suse.de>
17265 PR tree-optimization/85859
17266 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
17267 test with comment from bb_no_side_effects_p.
17269 2018-06-21 Richard Biener <rguenther@suse.de>
17271 PR tree-optimization/86232
17272 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
17273 max for constant niter.
17275 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
17277 * config/aarch64/aarch64-simd.md
17278 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
17280 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
17282 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
17283 Make opernads of the unspec commutative.
17285 2018-06-21 Richard Biener <rguenther@suse.de>
17287 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
17288 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
17289 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
17290 (vect_analyze_data_ref_dependence): Re-order checks to deal with
17292 (vect_record_base_alignments): Do not record base alignment
17293 for gathers or scatters.
17294 (vect_compute_data_ref_alignment): Drop return value that is always
17295 true. Bail out early for gathers or scatters.
17296 (vect_enhance_data_refs_alignment): Bail out early for gathers
17298 (vect_find_same_alignment_drs): Likewise.
17299 (vect_analyze_data_refs_alignment): Remove dead code.
17300 (vect_slp_analyze_and_verify_node_alignment): Likewise.
17301 (vect_analyze_data_refs): For possible gathers or scatters do
17302 not create an alternate DR, just check their possible validity
17303 and mark them. Adjust DECL_NONALIASED handling to not rely
17304 on DR_BASE_ADDRESS.
17305 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
17306 update inits of gathers or scatters.
17307 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
17308 Also copy gather/scatter flag to pattern vinfo.
17310 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17312 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
17313 behavior of vec_packsu (vector unsigned long long, vector unsigned
17314 long long) to match behavior of vec_packs with same signature.
17316 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
17317 Thomas Schwinge <thomas@codesourcery.com>
17318 Cesar Philippidis <cesar@codesourcery.com>
17320 * gimplify.c (gimplify_scan_omp_clauses): Add support for
17321 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
17322 (gimplify_adjust_omp_clauses): Likewise.
17323 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
17324 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
17325 (gimplify_omp_target_update): Update handling of acc update and
17327 * omp-low.c (install_var_field): Remove unused parameter
17328 base_pointers_restrict.
17329 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
17330 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
17332 (omp_target_base_pointers_restrict_p): Delete.
17333 (scan_omp_target): Update call to scan_sharing_clauses.
17334 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
17336 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
17337 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
17338 (convert_local_omp_clauses): Likewise.
17339 * tree-pretty-print.c (dump_omp_clause): Likewise.
17340 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
17342 (omp_clause_code_name): Likewise.
17344 2018-06-20 Jakub Jelinek <jakub@redhat.com>
17347 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
17350 PR tree-optimization/86231
17351 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
17352 anti-range don't overwrite *vr0min before using it to compute *vr0max.
17354 2018-06-20 Tom de Vries <tdevries@suse.de>
17356 PR tree-optimization/86097
17357 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
17358 iv type if signedness of iv type is not the same as that of *nit.
17360 2018-06-20 Jakub Jelinek <jakub@redhat.com>
17362 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
17363 EDGE_EH edges, verify they are all EDGE_EH.
17365 2018-06-20 Maya Rashish <coypu@sdf.org>
17367 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
17369 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17371 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
17372 * config/aarch64/aarch64.c (xgene1_tunings): Add
17373 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
17374 (aarch64_mode_valid_for_sched_fusion_p):
17375 Allow 16-byte modes.
17376 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
17377 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
17379 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
17381 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
17382 * config/aarch64/iterators.md (VQ2): New mode iterator.
17384 2018-06-20 Martin Liska <mliska@suse.cz>
17386 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
17387 Change default ratio from 10 to 8.
17389 2018-06-20 Martin Liska <mliska@suse.cz>
17391 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
17393 (bit_test_cluster::find_bit_tests): Likewise.
17394 (switch_decision_tree::analyze_switch_statement): Find clusters.
17395 * tree-switch-conversion.h (struct jump_table_cluster): Document
17398 2018-06-20 Martin Liska <mliska@suse.cz>
17400 * tree-switch-conversion.c (switch_conversion::collect):
17401 Record m_uniq property.
17402 (switch_conversion::expand): Bail out for special conditions.
17403 (group_cluster::~group_cluster): New.
17404 (group_cluster::group_cluster): Likewise.
17405 (group_cluster::dump): Likewise.
17406 (jump_table_cluster::emit): New.
17407 (switch_decision_tree::fix_phi_operands_for_edges): New.
17408 (struct case_node): Remove struct.
17409 (jump_table_cluster::can_be_handled): New.
17410 (case_values_threshold): Moved to header.
17411 (reset_out_edges_aux): Likewise.
17412 (jump_table_cluster::is_beneficial): New.
17413 (bit_test_cluster::can_be_handled): Likewise.
17414 (add_case_node): Remove.
17415 (bit_test_cluster::is_beneficial): New.
17416 (case_bit_test::cmp): New.
17417 (bit_test_cluster::emit): New.
17418 (expand_switch_as_decision_tree_p): Remove.
17419 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
17420 (fix_phi_operands_for_edge): Likewise.
17421 (switch_decision_tree::analyze_switch_statement): New.
17422 (compute_cases_per_edge): Move ...
17423 (switch_decision_tree::compute_cases_per_edge): ... here.
17424 (try_switch_expansion): Likewise.
17425 (switch_decision_tree::try_switch_expansion): Likewise.
17426 (record_phi_operand_mapping): Likewise.
17427 (switch_decision_tree::record_phi_operand_mapping): Likewise.
17428 (emit_case_decision_tree): Likewise.
17429 (switch_decision_tree::emit): Likewise.
17430 (balance_case_nodes): Likewise.
17431 (switch_decision_tree::balance_case_nodes): Likewise.
17432 (dump_case_nodes): Likewise.
17433 (switch_decision_tree::dump_case_nodes): Likewise.
17434 (emit_jump): Likewise.
17435 (switch_decision_tree::emit_jump): Likewise.
17436 (emit_cmp_and_jump_insns): Likewise.
17437 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
17438 (emit_case_nodes): Likewise.
17439 (switch_decision_tree::emit_case_nodes): Likewise.
17440 (conditional_probability): Remove.
17441 * tree-switch-conversion.h (enum cluster_type): New.
17443 (struct cluster): Likewise.
17444 (cluster::cluster): Likewise.
17445 (struct simple_cluster): Likewise.
17446 (simple_cluster::simple_cluster): Likewise.
17447 (struct group_cluster): Likewise.
17448 (struct jump_table_cluster): Likewise.
17449 (struct bit_test_cluster): Likewise.
17450 (struct min_cluster_item): Likewise.
17451 (struct case_tree_node): Likewise.
17452 (case_tree_node::case_tree_node): Likewise.
17453 (jump_table_cluster::case_values_threshold): Likewise.
17454 (struct case_bit_test): Likewise.
17455 (struct switch_decision_tree): Likewise.
17456 (struct switch_conversion): Likewise.
17457 (switch_decision_tree::reset_out_edges_aux): Likewise.
17459 2018-06-20 Martin Liska <mliska@suse.cz>
17461 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
17462 (hoist_edge_and_branch_if_true): Likewise.
17463 (expand_switch_using_bit_tests_p): Likewise.
17464 (struct case_bit_test): Likewise.
17465 (case_bit_test_cmp): Likewise.
17466 (emit_case_bit_tests): Likewise.
17467 (switch_conversion::switch_conversion): New class.
17468 (struct switch_conv_info): Remove old struct.
17469 (collect_switch_conv_info): More to ...
17470 (switch_conversion::collect): ... this.
17471 (check_range): Likewise.
17472 (switch_conversion::check_range): Likewise.
17473 (check_all_empty_except_final): Likewise.
17474 (switch_conversion::check_all_empty_except_final): Likewise.
17475 (check_final_bb): Likewise.
17476 (switch_conversion::check_final_bb): Likewise.
17477 (create_temp_arrays): Likewise.
17478 (switch_conversion::create_temp_arrays): Likewise.
17479 (free_temp_arrays): Likewise.
17480 (gather_default_values): Likewise.
17481 (switch_conversion::gather_default_values): Likewise.
17482 (build_constructors): Likewise.
17483 (switch_conversion::build_constructors): Likewise.
17484 (constructor_contains_same_values_p): Likewise.
17485 (switch_conversion::contains_same_values_p): Likewise.
17486 (array_value_type): Likewise.
17487 (switch_conversion::array_value_type): Likewise.
17488 (build_one_array): Likewise.
17489 (switch_conversion::build_one_array): Likewise.
17490 (build_arrays): Likewise.
17491 (switch_conversion::build_arrays): Likewise.
17492 (gen_def_assigns): Likewise.
17493 (switch_conversion::gen_def_assigns): Likewise.
17494 (prune_bbs): Likewise.
17495 (switch_conversion::prune_bbs): Likewise.
17496 (fix_phi_nodes): Likewise.
17497 (switch_conversion::fix_phi_nodes): Likewise.
17498 (gen_inbound_check): Likewise.
17499 (switch_conversion::gen_inbound_check): Likewise.
17500 (process_switch): Use the newly created class.
17501 (switch_conversion::expand): New.
17502 (switch_conversion::~switch_conversion): New.
17503 * tree-switch-conversion.h: New file.
17505 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17507 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
17508 tree-vect-patterns.c.
17509 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
17510 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
17511 (vect_recog_sad_pattern): Likewise.
17512 (vect_recog_widen_sum_pattern): Likewise.
17513 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
17514 (vect_recog_widen_shift_pattern): Remove the type_in argument.
17515 (vect_recog_rotate_pattern): Likewise.
17516 (vect_recog_mult_pattern): Likewise.
17517 (vect_recog_vector_vector_shift_pattern): Likewise.
17518 (vect_recog_divmod_pattern): Likewise.
17519 (vect_recog_mixed_size_cond_pattern): Likewise.
17520 (vect_recog_bool_pattern): Likewise.
17521 (vect_recog_mask_conversion_pattern): Likewise.
17522 (vect_try_gather_scatter_pattern): Likewise.
17523 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
17524 (vect_recog_over_widening_pattern): Likewise.
17525 (vect_recog_gather_scatter_pattern): Likewise.
17526 (vect_recog_func_ptr): Move from tree-vectorizer.h
17527 (vect_vect_recog_func_ptrs): Move further down the file.
17528 (vect_recog_func): Likewise. Remove the third argument.
17529 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
17530 (vect_pattern_recog_1): Expect the pattern function to do any
17531 necessary target tests. Also expect it to provide a vector type.
17532 Remove the type_in handling.
17534 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17536 * tree-vect-patterns.c (vect_pattern_detected): New function.
17537 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
17538 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
17539 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
17540 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
17541 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
17542 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
17543 (vect_recog_mask_conversion_pattern)
17544 (vect_try_gather_scatter_pattern): Likewise.
17546 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17548 * tree-vect-patterns.c (vect_get_internal_def): New function.
17549 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
17550 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
17551 (search_type_for_mask_1): Use it.
17553 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17555 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
17556 redundant WIDEN_SUM_EXPR handling.
17557 (vect_recog_sad_pattern): Likewise.
17559 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17561 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
17562 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
17563 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
17564 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
17565 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
17567 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17569 * tree-vect-stmts.c (vectorizable_call): Make sure that we
17570 use the stmt_vec_info of the original bb statement for the
17571 new zero assignment, even if the call is part of a pattern.
17573 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17575 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
17576 that the sequence is attached to the original statement rather
17577 than the pattern statement.
17578 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
17579 PATTERN_DEF_SEQ from the original statement rather than
17580 the main pattern statement.
17581 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
17582 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
17583 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
17585 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
17587 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
17588 definition statements before the early exit for statements that aren't
17590 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
17592 (vect_transform_loop): ...here. Process pattern definition
17593 statements without first checking whether the main pattern
17594 statement is live or relevant.
17596 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17598 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
17599 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
17601 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
17603 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
17604 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
17605 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
17606 (expand_block_compare): Change select_block_compare_mode call.
17607 (expand_strncmp_align_check): Use new functions, fix comment.
17608 (emit_final_str_compare_gpr): New function.
17609 (expand_strn_compare): Refactor and clean up code.
17610 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
17612 2018-06-19 Tony Reix <tony.reix@atos.com>
17613 Damien Bergamini <damien.bergamini@atos.com>
17614 David Edelsohn <dje.gcc@gmail.com>
17616 * collect2.c (static_obj): New variable.
17617 (static_libs): New variable.
17618 (is_in_list): Uncomment declaration.
17619 (main): Track AIX libraries linked statically.
17620 (is_in_list): Uncomment definition.
17621 (scan_prog_file): Don't add AIX shared libraries initializer
17622 to constructor list if linking statically.
17624 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
17626 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
17628 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
17630 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
17632 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
17635 2018-06-19 Martin Liska <mliska@suse.cz>
17637 * config/i386/i386.c (ix86_can_inline_p): Do not use
17638 ipa_fn_summaries::get_create.
17639 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
17641 (devirtualization_time_bonus): Likewise.
17642 (ipcp_propagate_stage): Likewise.
17643 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
17644 (edge_set_predicate): Likewise.
17645 (evaluate_conditions_for_known_args): Likewise.
17646 (evaluate_properties_for_edge): Likewise.
17647 (ipa_call_summary::reset): Tranform to ...
17648 (ipa_call_summary::~ipa_call_summary): ... this.
17649 (ipa_fn_summary::reset): Transform to ...
17650 (ipa_fn_summary::~ipa_fn_summary): ... this.
17651 (ipa_fn_summary_t::remove): Rename to ...
17652 (ipa_fn_summary_t::remove_callees): ... this.
17653 (ipa_fn_summary_t::duplicate): Use placement new
17654 instead of memory copy.
17655 (ipa_call_summary_t::duplicate): Likewise.
17656 (ipa_call_summary_t::remove): Remove.
17657 (dump_ipa_call_summary): Change get_create to get.
17658 (ipa_dump_fn_summary): Dump only when summary exists.
17659 (analyze_function_body): Use symbol_summary::get instead
17661 (compute_fn_summary): Likewise.
17662 (estimate_edge_devirt_benefit): Likewise.
17663 (estimate_edge_size_and_time): Likewise.
17664 (inline_update_callee_summaries): Likewise.
17665 (remap_edge_change_prob): Likewise.
17666 (remap_edge_summaries): Likewise.
17667 (ipa_merge_fn_summary_after_inlining): Likewise.
17668 (write_ipa_call_summary): Likewise.
17669 (ipa_fn_summary_write): Likewise.
17670 (ipa_free_fn_summary): Likewise.
17671 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
17672 (struct ipa_call_summary): Likewise.
17673 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
17675 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
17676 (estimate_size_after_inlining): Likewise.
17677 (estimate_growth): Likewise.
17678 (growth_likely_positive): Likewise.
17679 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
17680 (inline_call): Likewise.
17681 * ipa-inline.c (caller_growth_limits): Likewise.
17682 (can_inline_edge_p): Likewise.
17683 (can_inline_edge_by_limits_p): Likewise.
17684 (compute_uninlined_call_time): Likewise.
17685 (compute_inlined_call_time): Likewise.
17686 (want_inline_small_function_p): Likewise.
17687 (edge_badness): Likewise.
17688 (update_caller_keys): Likewise.
17689 (update_callee_keys): Likewise.
17690 (inline_small_functions): Likewise.
17691 (inline_to_all_callers_1): Likewise.
17692 (dump_overall_stats): Likewise.
17693 (early_inline_small_functions): Likewise.
17694 (early_inliner): Likewise.
17695 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
17696 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
17697 * ipa-pure-const.c (malloc_candidate_p): Likewise.
17698 * ipa-split.c (execute_split_functions): Likewise.
17699 * symbol-summary.h: Likewise.
17700 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
17702 2018-06-19 Richard Biener <rguenther@suse.de>
17704 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
17705 (vectorize_loops): ... here. Fix dbgcnt handling.
17706 (try_vectorize_loop): Wrap try_vectorize_loop_1.
17708 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
17711 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
17712 ieee128 argument takes up only one (vector) register, not two (floating
17715 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
17717 * gimplify.c (gimplify_init_constructor): Really never clear for an
17718 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
17720 2018-06-19 Richard Biener <rguenther@suse.de>
17722 PR tree-optimization/86179
17723 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
17724 after failed recognition.
17726 2018-06-18 Martin Sebor <msebor@redhat.com>
17728 PR middle-end/85602
17729 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
17730 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
17731 Handle integer subtraction.
17732 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
17733 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
17735 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17737 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
17738 param from rtx to rtx_insn *.
17739 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
17741 (frv_ifcvt_modify_insn): Likwise.
17742 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
17743 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
17744 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
17745 as_a <rtx_insn *> cast to local "unprotected_region" once
17746 it's been established that it's not NULL or pc_rtx.
17747 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
17748 param "sethi" from rtx to rtx_insn *.
17749 (nds32_group_float_insns): Likewise for param "insn".
17750 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
17752 (vax_output_int_subtract): Likewise.
17753 * config/vax/vax.c (vax_output_int_add): Likewise for param
17755 (vax_output_int_subtract): Likewise.
17756 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
17757 (emit_pattern_after): Likewise for param "after".
17758 (emit_insn_after): Likewise.
17759 (emit_jump_insn_after): Likewise.
17760 (emit_call_insn_after): Likewise.
17761 (emit_debug_insn_after): Likewise.
17762 (emit_pattern_before): Likewise for param "before".
17763 (emit_insn_before): Likewise.
17764 (emit_jump_insn_before): Likewise.
17765 * final.c (get_insn_template): Likewise for param "insn", removing
17767 * output.h (get_insn_template): Likewise for 2nd param.
17768 * rtl.h (emit_insn_before): Likewise.
17769 (emit_jump_insn_before): Likewise.
17770 (emit_debug_insn_before_noloc): Likewise.
17771 (emit_insn_after): Likewise.
17772 (emit_jump_insn_after): Likewise.
17773 (emit_call_insn_after): Likewise.
17774 (emit_debug_insn_after): Likewise.
17775 (set_insn_deleted): Likewise for param.
17777 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
17780 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
17781 floating point modes, so that IFmode is numerically greater than
17782 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
17783 to declare the ordering. This prevents IFmode from being
17784 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
17785 machine. Include rs6000-modes.h to share the fractional values
17786 between genmodes* and the rest of the compiler.
17787 (IFmode): Likewise.
17788 (KFmode): Likewise.
17789 (TFmode): Likewise.
17790 * config/rs6000/rs6000-modes.h: New file.
17791 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
17792 meaning of rs6000_long_double_size so that 126..128 selects an
17793 appropriate 128-bit floating point type.
17794 (rs6000_option_override_internal): Likewise.
17795 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
17796 (TARGET_LONG_DOUBLE_128): Change the meaning of
17797 rs6000_long_double_size so that 126..128 selects an appropriate
17798 128-bit floating point type.
17799 (LONG_DOUBLE_TYPE_SIZE): Update comment.
17800 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
17801 source and destination to match the standard usage.
17802 (truncifkf2): Likewise.
17803 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
17804 ISA 2.07 to use an explicit clobber, instead of passing in a
17806 (copysign<mode>3_soft): Likewise.
17808 2018-06-18 David Malcolm <dmalcolm@redhat.com>
17810 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
17811 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
17812 (vect_slp_analyze_instance_dependence): Likewise.
17813 (vect_enhance_data_refs_alignment): Likewise.
17814 (vect_analyze_data_refs_alignment): Likewise.
17815 (vect_slp_analyze_and_verify_instance_alignment
17816 (vect_analyze_data_ref_accesses): Likewise.
17817 (vect_prune_runtime_alias_test_list): Likewise.
17818 (vect_analyze_data_refs): Likewise.
17819 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
17820 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17821 (vect_analyze_scalar_cycles_1): Likewise.
17822 (vect_get_loop_niters): Likewise.
17823 (vect_analyze_loop_form_1): Likewise.
17824 (vect_update_vf_for_slp): Likewise.
17825 (vect_analyze_loop_operations): Likewise.
17826 (vect_analyze_loop): Likewise.
17827 (vectorizable_induction): Likewise.
17828 (vect_transform_loop): Likewise.
17829 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
17830 * tree-vect-slp.c (vect_analyze_slp): Likewise.
17831 (vect_make_slp_decision): Likewise.
17832 (vect_detect_hybrid_slp): Likewise.
17833 (vect_slp_analyze_operations): Likewise.
17834 (vect_slp_bb): Likewise.
17835 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
17836 (vectorizable_bswap): Likewise.
17837 (vectorizable_call): Likewise.
17838 (vectorizable_simd_clone_call): Likewise.
17839 (vectorizable_conversion): Likewise.
17840 (vectorizable_assignment): Likewise.
17841 (vectorizable_shift): Likewise.
17842 (vectorizable_operation): Likewise.
17843 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
17845 2018-06-18 Martin Sebor <msebor@redhat.com>
17847 PR tree-optimization/81384
17848 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
17849 * builtins.c (expand_builtin_strnlen): New function.
17850 (expand_builtin): Call it.
17851 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
17852 * builtins.def (BUILT_IN_STRNLEN): New.
17853 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
17854 Warn for bounds in excess of maximum object size.
17855 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
17856 single-value ranges. Handle strnlen.
17857 (handle_builtin_strlen): Handle strnlen.
17858 (strlen_check_and_optimize_stmt): Same.
17859 * doc/extend.texi (Other Builtins): Document strnlen.
17861 2018-06-18 Maya Rashish <coypu@sdf.org>
17863 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
17864 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
17865 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
17867 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
17869 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
17871 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17873 * tree.c (escaped_string::escape): Replace cast to char * by
17874 const_cast<char *> (unescaped).
17876 2018-06-18 Nick Clifton <nickc@redhat.com>
17879 * tree.c (escaped_string): New class. Converts an unescaped
17880 string into its escaped equivalent.
17881 (warn_deprecated_use): Use the new class to convert the
17882 deprecation message, if present.
17883 (test_escaped_strings): New self test.
17884 (test_c_tests): Add test_escaped_strings.
17885 * doc/extend.texi (deprecated): Add a note that the
17886 deprecation message is affected by the -fmessage-length
17887 option, and that control characters will be escaped.
17888 (#pragma GCC error): Document this pragma.
17889 (#pragma GCC warning): Likewise.
17890 * doc/invoke.texi (-fmessage-length): Document this option's
17891 effect on the #warning and #error preprocessor directives and
17892 the deprecated attribute.
17894 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
17896 * tree.c (decl_value_expr_lookup): Revert latest change.
17897 (decl_value_expr_insert): Likewise.
17899 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
17901 * gimplify.c (nonlocal_vlas): Delete.
17902 (nonlocal_vla_vars): Likewise.
17903 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
17905 (gimplify_body): Do not create and destroy nonlocal_vlas.
17906 * tree-nested.c: Include diagnostic.h.
17907 (use_pointer_in_frame): Tweak.
17908 (lookup_field_for_decl): Add assertion and declare the transformation.
17909 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
17910 internal error when the reference is in a wrong context. Do not
17911 create a debug decl by default.
17912 (note_nonlocal_block_vlas): Delete.
17913 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
17914 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
17915 create a debug decl by default.
17916 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
17917 call is in a wrong context.
17918 (fixup_vla_decls): New function.
17919 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
17920 debug variables were created.
17921 * tree.c (decl_value_expr_lookup): Add checking assertion.
17922 (decl_value_expr_insert): Likewise.
17924 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17926 PR middle-end/82479
17927 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
17928 * tree-scalar-evolution.c (interpret_expr): Likewise.
17929 (expression_expensive_p): Likewise.
17930 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
17931 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
17932 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
17933 (ssa_defined_by_minus_one_stmt_p): New.
17935 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17937 PR middle-end/64946
17938 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
17939 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
17940 * dojump.c (do_jump): Likewise.
17941 * expr.c (expand_expr_real_2): Check operand type's sign.
17942 * fold-const.c (const_unop): Handle ABSU_EXPR.
17943 (fold_abs_const): Likewise.
17944 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
17945 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
17946 (strip_sign_op_1): Likesise.
17947 * match.pd: Add new pattern to generate ABSU_EXPR.
17948 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
17949 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17950 * tree-eh.c (operation_could_trap_helper_p): Likewise.
17951 * tree-inline.c (estimate_operator_cost): Likewise.
17952 * tree-pretty-print.c (dump_generic_node): Likewise.
17953 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
17954 * tree.def (ABSU_EXPR): New.
17956 2018-06-16 Jakub Jelinek <jakub@redhat.com>
17958 PR middle-end/86095
17959 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
17960 documented as preserved for backward compatibility only.
17961 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
17963 PR rtl-optimization/86108
17964 * bb-reorder.c (create_forwarder_block): Renamed to ...
17965 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
17966 jump from new landing pad to the second part.
17967 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
17970 2018-06-15 Jakub Jelinek <jakub@redhat.com>
17972 PR middle-end/85878
17973 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
17974 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
17975 Only call store_expr for halves if the mode is the same.
17977 PR middle-end/86123
17978 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
17979 Fix up comment formatting.
17981 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
17983 * typed-splay-tree.h (typed_splay_tree::remove): New function.
17984 (typed_splay_tree::closure,
17985 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
17986 (typed_splay_tree::typed_splay_tree,
17987 typed_splay_tree::operator =): Declared private.
17988 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
17989 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
17990 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
17991 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
17992 typed_splay_tree::splay_tree_splay,
17993 typed_splay_tree::splay_tree_foreach_helper,
17994 typed_splay_tree::splay_tree_insert,
17995 typed_splay_tree::splay_tree_remove,
17996 typed_splay_tree::splay_tree_lookup,
17997 typed_splay_tree::splay_tree_predecessor,
17998 typed_splay_tree::splay_tree_successor,
17999 typed_splay_tree::splay_tree_min,
18000 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
18001 (typed_splay_tree::root, typed_splay_tree::comp,
18002 typed_splay_tree::delete_key,
18003 typed_splay_tree::delete_value): New data members.
18004 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
18005 typed_splay_tree::remove.
18007 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
18009 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
18010 -mginv and -mno-ginv to the assembler.
18011 * config/mips/mips.opt (-mcrc): New option.
18012 (-mginv): Likewise.
18013 * doc/invoke.text (-mcrc): Document.
18014 (-mginv): Likewise.
18016 2018-06-15 Nick Clifton <nickc@redhat.com>
18019 * tree.c (escaped_string): New class. Converts an unescaped
18020 string into its escaped equivalent.
18021 (warn_deprecated_use): Use the new class to convert the
18022 deprecation message, if present.
18023 (test_escaped_strings): New self test.
18024 (test_c_tests): Add test_escaped_strings.
18025 * doc/extend.texi (deprecated): Add a note that the
18026 deprecation message is affected by the -fmessage-length
18027 option, and that control characters will be escaped.
18028 (#pragma GCC error): Document this pragma.
18029 (#pragma GCC warning): Likewise.
18030 * doc/invoke.texi (-fmessage-length): Document this option's
18031 effect on the #warning and #error preprocessor directives and
18032 the deprecated attribute.
18034 2018-06-15 Richard Biener <rguenther@suse.de>
18036 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
18037 here, also noting vector size used.
18038 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
18039 size used in MSG_OPTIMIZED_LOCATIONS dump.
18040 (pass_slp_vectorize::execute): Adjust.
18042 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
18045 * config/arc/arc.c (arc_return_address_register): Fix
18048 2018-06-15 Richard Biener <rguenther@suse.de>
18050 PR middle-end/86159
18051 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
18052 leave useless conversion stripping to force_gimple_operand_gsi.
18053 (gimplify_build2): Likewise.
18054 (gimplify_build1): Likewise.
18056 2018-06-15 Richard Biener <rguenther@suse.de>
18058 PR middle-end/86076
18059 * tree-cfg.c (move_stmt_op): unshare invariant addresses
18060 before adjusting their block.
18062 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
18064 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
18065 multilibs for *-*-rtems*.
18066 * config/riscv/t-rtems: New file.
18068 2018-06-14 Jakub Jelinek <jakub@redhat.com>
18070 PR middle-end/86122
18071 * match.pd ((A +- CST1) +- CST2): Punt if last resort
18072 unsigned_type_for returns NULL.
18075 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
18076 subregs of multi-word pseudos unless the float mode has word size.
18078 2018-06-14 Richard Biener <rguenther@suse.de>
18080 PR middle-end/86139
18081 * tree-vect-generic.c (build_word_mode_vector_type): Remove
18082 duplicate and harmful type_hash_canon.
18083 * tree.c (type_hash_canon): Assert we didn't find ourselves.
18085 2018-06-14 Richard Biener <rguenther@suse.de>
18088 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
18091 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
18093 * config/rtems.h (STDINT_LONG32): Define.
18095 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
18096 Prachi Godbole <prachi.godbole@imgtec.com>
18098 * config/mips/mips-cpus.def: Define P6600.
18099 * config/mips/mips-tables.opt: Regenerate.
18100 * config/mips/mips.c (mips_ucbranch_type): New enum.
18101 (mips_rtx_cost_data): Add support for P6600.
18102 (mips_issue_rate): Likewise.
18103 (mips_multipass_dfa_lookahead): Likewise.
18104 (mips_avoid_hazard): Likewise.
18105 (mips_reorg_process_insns): Likewise.
18106 (mips_classify_branch_p6600): New function.
18107 * config/mips/mips.h (TUNE_P6600): New define.
18108 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
18109 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
18110 * config/mips/mips.md: Include p6600.md.
18111 (processor): Add p6600.
18112 * config/mips/p6600.md: New file.
18113 * doc/invoke.texi: Add p6600 to supported architectures.
18115 2018-06-13 Martin Sebor <msebor@redhat.com>
18117 PR tree-optimization/86114
18118 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
18120 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
18122 2018-06-13 Richard Biener <rguenther@suse.de>
18124 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
18125 Properly set vector type of the intermediate stmt.
18126 * tree-vect-stmts.c (vectorizable_operation): The destination
18127 var always has vectype_out type.
18129 2018-06-13 Jeff Law <law@redhat.com>
18131 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
18132 integer 0 for argument to print_rtl_with_bb.
18133 (rl78_reorg): Likewise.
18135 2018-06-13 David Malcolm <dmalcolm@redhat.com>
18137 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
18138 from rtx to rtx_insn *.
18139 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
18141 (add_sched_insns_for_speculation): Likewise for local "target",
18142 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
18143 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
18144 from rtx_insn ** to rtx_code_label **.
18145 (reorg_emit_nops): Likewise.
18146 (c6x_reorg): Likewise for local "call_labels".
18147 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
18149 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
18150 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
18151 the loops over LABEL_REFS.
18152 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
18154 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
18155 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
18156 (split_branches): Strengthen local "olabel" from rtx to
18157 rtx_insn *, adding a safe_as_a cast.
18158 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
18160 (add_insn_after): Likewise for first two params.
18161 (add_insn_before): Likewise.
18162 (remove_insn): Likewise for param.
18163 (emit_pattern_before_noloc): Likewise for second and third params.
18164 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
18165 (emit_call_insn_before_noloc): Likewise.
18166 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
18168 (emit_barrier_before): Likewise.
18169 (emit_label_before): Strengthen "label" param from "rtx" to
18170 "rtx_code_label *". Strengthen "before" param from "rtx" to
18172 (emit_insn_after_1): Strengthen "after" param from "rtx" to
18174 (emit_pattern_after_noloc): Likewise.
18175 (emit_insn_after_noloc): Likewise.
18176 (emit_jump_insn_after_noloc): Likewise.
18177 (emit_call_insn_after_noloc): Likewise.
18178 (emit_debug_insn_after_noloc): Likewise.
18179 (emit_barrier_after): Likewise.
18180 (emit_label_after): Likewise for both params.
18181 (emit_pattern_after_setloc): Likewise for "after" param. Convert
18182 "loc" param from "int" to "location_t".
18183 (emit_insn_after_setloc): Likewise.
18184 (emit_jump_insn_after_setloc): Likewise.
18185 (emit_call_insn_after_setloc): Likewise.
18186 (emit_debug_insn_after_setloc): Likewise.
18187 (emit_pattern_before_setloc): Likewise for "before" param. Convert
18188 "loc" param from "int" to "location_t".
18189 (emit_pattern_before): Convert NULL_RTX to NULL.
18190 (emit_insn_before_setloc): Convert "loc" param from "int" to
18192 (emit_jump_insn_before_setloc): Likewise.
18193 (emit_call_insn_before_setloc): Likewise.
18194 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
18195 rtx_insn *. Convert "loc" param from "int" to "location_t".
18196 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
18197 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
18198 Convert 3rd param from "int" to "location_t".
18199 (emit_barrier_before, emit_barrier_after, next_real_insn):
18200 Strengthen param from rtx to rtx_insn *.
18201 (emit_label_before): Strengthen 1st param from "rtx" to
18202 "rtx_code_label *". Strengthen 2nd param from "rtx" to
18204 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
18205 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
18206 Strengthen 2nd param from "rtx" to "rtx_insn *".
18207 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
18208 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
18209 Likewise. Convert 3rd param from "int" to "location_t".
18210 (emit_label_after): Strengthen 1st param from "rtx" to
18211 "rtx_code_label *".
18212 (next_real_insn, remove_insn): Strengthen param from "rtx" to
18214 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
18215 from "rtx" to "rtx_insn *".
18217 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
18219 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
18220 bodies streamed in with -Q.
18221 * dumpfile.c (dump_files): Add lto-stream-out dump file.
18222 * dumpfile.h (tree_dump_index): Add lto_stream_out.
18223 * gimple-streamer-out.c: Include gimple-pretty-print.h
18224 (output_bb): Dump stmts streamed.
18225 * lto-section-out.c: Include print-tree.h
18226 (lto_begin_section): Dump sections created.
18227 (lto_output_decl_index): Dump decl encoded.
18228 * lto-streamer-out.c: Include print-tree.h
18229 (create_output_block): Dump output block created.
18230 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
18231 (output_function): Dump function output.
18232 (output_constructor): Dump constructor streamed.
18233 (write_global_stream): Output indexes encoded.
18234 (produce_asm_for_decls): Dump streams encoded.
18235 * lto-streamer.c (streamer_dump_file): New global var.
18236 * lto-streamer.h (streamer_dump_file): Declare.
18237 * passes.c (ipa_write_summaries): Initialize streamer dump.
18238 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
18241 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
18244 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
18245 offsets for register save directives. Emit a second batch of save
18246 directives, if need be, when the function accesses prior frames.
18248 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18250 * config/arc/fpu.md (fmasf4): Force operand to register.
18251 (fnmasf4): Likewise.
18253 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18255 * config/arc/arc-protos.h (arc_pad_return): Remove.
18256 * config/arc/arc.c (machine_function): Remove force_short_suffix
18258 (arc_print_operand): Adjust printing of '&'.
18259 (arc_verify_short): Remove conditional printing of short suffix.
18260 (arc_final_prescan_insn): Remove reference to size_reason.
18261 (pad_return): New function.
18262 (arc_reorg): Call pad_return.
18263 (arc_pad_return): Remove.
18264 (arc_init_machine_status): Remove reference to force_short_suffix.
18265 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
18266 (attr length): When attribute iscompact is true force to 2
18267 regardless; in the case of maybe check if we want to force the
18268 instruction to have 4 bytes length.
18269 (nopv): Change it to generate 4 byte long nop as well.
18270 (blockage): New pattern.
18271 (simple_return): Remove call to arc_pad_return.
18272 (p_return_i): Likewise.
18274 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18276 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
18278 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18280 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
18283 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18285 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
18286 for ARC700 and ARCv2.
18288 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
18291 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
18292 operands[2] instead of operands[1].
18295 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18297 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
18298 case, check whether the outer register overlaps an unallocatable
18299 register, not just whether it fits the required class.
18301 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
18303 * poly-int.h (can_div_trunc_p): Add new overload in which all values
18305 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
18306 (memrefs_conflict_p): Likewise.
18307 (init_alias_analysis): Likewise.
18308 * cfgexpand.c (expand_debug_expr): Likewise.
18309 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
18310 * cse.c (fold_rtx): Likewise.
18311 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
18312 * expr.c (emit_block_move_hints): Likewise.
18313 (clear_storage_hints, push_block, emit_push_insn): Likewise.
18314 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
18315 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
18316 (emit_group_store): Likewise.
18317 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
18318 to read the PRE/POST_MODIFY increment.
18319 * calls.c (store_one_arg): Use strip_offset.
18320 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
18322 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
18324 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
18325 (simplify_binary_operation_1): Extend CONST_INT handling to
18327 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
18328 than a HOST_WIDE_INT.
18329 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
18331 (adjust_mems, add_stores): Update accodingly.
18332 (vt_canonicalize_addr): Track polynomial offsets.
18333 (emit_note_insn_var_location): Likewise.
18334 (vt_add_function_parameter): Likewise.
18335 (vt_initialize): Likewise.
18337 2018-06-12 Jeff Law <law@redhat.com>
18339 * config.gcc (alpha*-*-freebsd*): Remove.
18340 * config/alpha/freebsd.h: Remove.
18342 2018-06-12 David Malcolm <dmalcolm@redhat.com>
18345 * spellcheck-tree.c (levenshtein_distance): Rename to...
18346 (get_edit_distance): ...this, and update for underlying renaming.
18347 * spellcheck-tree.h (levenshtein_distance): Rename to...
18348 (get_edit_distance): ...this.
18349 * spellcheck.c (levenshtein_distance): Rename to...
18350 (get_edit_distance): ...this. Convert from Levenshtein distance
18351 to Damerau-Levenshtein distance by supporting transpositions of
18352 adjacent characters. Rename "v1" to "v_next" and "v0" to
18354 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
18355 (selftest::test_edit_distance_unit_test_oneway): ...this, and
18356 update for underlying renaming.
18357 (selftest::levenshtein_distance_unit_test): Rename to...
18358 (selftest::test_get_edit_distance_unit): ...this, and update for
18359 underlying renaming.
18360 (selftest::test_find_closest_string): Add example from PR 69968
18361 where transposition helps
18362 (selftest::test_metric_conditions): Update for renaming.
18363 (selftest::test_metric_conditions): Likewise.
18364 (selftest::spellcheck_c_tests): Likewise.
18365 * spellcheck.h (levenshtein_distance): Rename both overloads to...
18366 (get_edit_distance): ...this.
18367 (best_match::consider): Update for renaming.
18369 2018-06-12 Martin Sebor <msebor@redhat.com>
18371 PR tree-optimization/85259
18372 * builtins.c (compute_objsize): Handle constant offsets.
18373 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
18374 true iff a warning has been issued.
18375 * gimple.h (gimple_nonartificial_location): New function.
18376 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
18377 gimple_nonartificial_location and handle -Wno-system-headers.
18378 (handle_builtin_stxncpy): Same.
18380 2018-06-12 Martin Sebor <msebor@redhat.com>
18383 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
18385 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18387 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18388 BUILTIN_VEC_XST entries for pointer to double and long long.
18390 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
18393 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
18395 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
18398 2018-06-12 Martin Liska <mliska@suse.cz>
18400 * doc/options.texi: Document IntegerRange.
18402 2018-06-12 Martin Liska <mliska@suse.cz>
18404 * config/i386/i386.opt: Make MPX-related options as Deprecated.
18405 * opt-functions.awk: Handle Deprecated flag.
18406 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
18408 (read_cmdline_option): Report warning for a deprecated option.
18409 * opts.h (struct cl_option): Add new field cl_deprecated.
18410 (CL_ERR_DEPRECATED): New.
18412 2018-06-12 Martin Liska <mliska@suse.cz>
18414 * doc/options.texi: Document Deprecated option flag.
18416 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18418 * config/arc/arc-arch.h (arc_extras): New enum.
18419 (arc_cpu_t):Add field extra.
18420 (arc_cpu_types): Consider the extras.
18421 * config/arc/arc-cpus.def: Add extras info.
18422 * config/arc/arc-opts.h (processor_type): Consider extra field.
18423 * config/arc/arc.c (arc_override_options): Handle extra field.
18425 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
18427 * config/arc/arc-arch.h: Update ARC_OPTX macro.
18428 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
18430 * config/arc/arc.c (arc_init): Update pic warning.
18431 (irq_range): Update irq range parsing warnings.
18432 (arc_override_options): Update various warning messages.
18433 (arc_handle_aux_attribute): Likewise.
18435 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
18437 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
18439 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18441 * doc/sourcebuild.texi: Document usage of line number 0 in verify
18442 compiler messages directives.
18444 2018-06-12 Matthew Fortune <mfortune@gmail.com>
18446 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
18447 * config/mips/mips-tables.opt: Regenerate.
18448 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
18450 * doc/invoke.texi: Document -march=i6500.
18452 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
18454 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
18455 (i6400_gpmul): Add cpu_unit.
18456 (i6400_gpdiv): Likewise.
18457 (i6400_msa_add_d): Update reservations.
18458 (i6400_msa_int_add) Likewise.
18459 (i6400_msa_short_logic3) Likewise.
18460 (i6400_msa_short_logic2) Likewise.
18461 (i6400_msa_short_logic) Likewise.
18462 (i6400_msa_move) Likewise.
18463 (i6400_msa_cmp) Likewise.
18464 (i6400_msa_short_float2) Likewise.
18465 (i6400_msa_div_d) Likewise.
18466 (i6400_msa_long_logic1) Likewise.
18467 (i6400_msa_long_logic2) Likewise.
18468 (i6400_msa_mult) Likewise.
18469 (i6400_msa_long_float2) Likewise.
18470 (i6400_msa_long_float4) Likewise.
18471 (i6400_msa_long_float5) Likewise.
18472 (i6400_msa_long_float8) Likewise.
18473 (i6400_fpu_fadd): Include frint type.
18474 (i6400_fpu_store): New define_insn_reservation.
18475 (i6400_fpu_load): Likewise.
18476 (i6400_fpu_move): Likewise.
18477 (i6400_fpu_fcmp): Likewise.
18478 (i6400_fpu_fmadd): Likewise.
18479 (i6400_int_mult): Include imul3nc type and update reservation.
18480 (i6400_int_div): Include idiv3 type and update reservation.
18481 (i6400_int_load): Update to check type not move_type.
18482 (i6400_int_store): Likewise.
18483 (i6400_int_prefetch): Set zero latency.
18485 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
18487 * gcc.c: Document new %@{...} sequence.
18488 (LINK_COMMAND_SPEC): Use it for the -L switches.
18489 (cpp_unique_options): Use it for the -I switches.
18490 (at_file_argbuf): New global variable.
18491 (in_at_file): Likewise.
18492 (alloc_args): Create at_file_argbuf.
18493 (clear_args): Truncate at_file_argbuf.
18494 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
18495 (open_at_file): New function.
18496 (close_at_file): Likewise.
18497 (create_at_file): Delete.
18498 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
18501 (validate_switches_from_spec): Deal with %@{...} sequence.
18502 (validate_switches): Likewise.
18503 (driver::finalize): Call clear_args.
18505 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
18507 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
18509 2018-06-11 Martin Sebor <msebor@redhat.com>
18511 * doc/invoke.texi (-Wall): List -Wc++17-compat.
18512 (Wno-class-memaccess): Add @opindex.
18513 (Wno-templates, Wno-multiple-inheritance): Same.
18514 (Wno-virtual-inheritance, Wno-namespaces): Same.
18515 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
18516 (Wno-format-overflow, Wno-format-truncation): Same.
18517 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
18518 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
18519 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
18520 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
18521 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
18522 (Wno-misspelled-isr): Same.
18524 2018-06-11 Martin Sebor <msebor@redhat.com>
18526 * PR tree-optimization/86083
18527 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
18529 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
18531 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
18533 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
18536 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
18537 on the correct operand.
18538 (*movdi_internal64): Ditto.
18540 2018-06-11 Martin Liska <mliska@suse.cz>
18542 PR tree-optimization/86089
18543 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
18545 2018-06-11 Julia Koval <julia.koval@intel.com>
18547 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
18548 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
18549 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
18551 2018-06-11 Olivier Hainque <hainque@adacore.com>
18553 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
18554 for Ada with strict dwarf2.
18556 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
18559 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
18562 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
18564 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
18566 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
18568 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
18571 2018-06-08 Martin Liska <mliska@suse.cz>
18573 * tree-cfg.h (debug_function): Fix argument type to match
18576 2018-06-08 Martin Liska <mliska@suse.cz>
18578 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
18579 Remove usage of MPX-related (and removed) fields.
18580 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
18582 2018-06-08 David Malcolm <dmalcolm@redhat.com>
18584 * cfg.c (debug): Use TDF_NONE rather than 0.
18585 * cfghooks.c (debug): Likewise.
18586 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
18587 (struct dump_option_value_info): Convert to...
18588 (struct kv_pair): ...this template type.
18589 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
18591 (optinfo_verbosity_options): Likewise.
18592 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
18594 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
18595 than int for "optgroup_flags" param.
18596 (dump_generic_expr_loc): Use dump_flags_t rather than int for
18598 (dump_dec): Likewise.
18599 (dump_finish): Use TDF_NONE rather than 0.
18600 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
18601 rather than int for "optgroup_flags" param. Use TDF_NONE rather
18602 than 0. Update for change to option_ptr.
18603 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
18604 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
18605 0. Update for changes to optinfo_verbosity_options and
18607 (opt_info_switch_p): Convert optgroup_flags from int to
18609 (dump_basic_block): Use dump_flags_t rather than int
18610 for "dump_kind" param.
18611 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
18612 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
18613 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
18614 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
18615 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
18616 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
18617 TDF_NONE): Convert from macros to...
18618 (enum dump_flag): ...this new enum.
18619 (dump_flags_t): Update to use enum.
18620 (operator|, operator&, operator~, operator|=, operator&=):
18621 Implement for dump_flags_t.
18622 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
18623 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
18624 Convert from macros to...
18625 (enum optgroup_flag): ...this new enum.
18626 (optgroup_flags_t): New typedef.
18627 (operator|, operator|=): Implement for optgroup_flags_t.
18628 (struct dump_file_info): Convert field "alt_flags" to
18629 dump_flags_t. Convert field "optgroup_flags" to
18631 (dump_basic_block): Use dump_flags_t rather than int for param.
18632 (dump_generic_expr_loc): Likewise.
18633 (dump_dec): Likewise.
18634 (dump_register): Convert param "optgroup_flags" to
18636 (opt_info_enable_passes): Likewise.
18637 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
18639 * gimple-pretty-print.c (debug): Likewise.
18640 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
18641 (merged_store_group::apply_stores): Likewise.
18642 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
18643 * gimple.c (verify_gimple_pp): Likewise.
18644 * graphite-poly.c (print_pbb_body): Likewise.
18645 * passes.c (pass_manager::register_one_dump_file): Convert
18646 local "optgroup_flags" to optgroup_flags_t.
18647 * print-tree.c (print_node): Use TDF_NONE rather than 0.
18649 (debug_body): Likewise.
18650 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
18651 to optgroup_flags_t.
18652 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
18654 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
18655 (convert_mult_to_fma): Likewise.
18656 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
18657 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
18658 * tree-vect-data-refs.c (dump_lower_bound): Convert param
18659 "dump_kind" to dump_flags_t.
18661 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18663 * config/rs6000/rs6000.c (min, max): Delete.
18665 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
18667 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
18668 -mabi=spe and -mabi=no-spe.
18670 2018-06-08 Martin Liska <mliska@suse.cz>
18672 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
18673 where we expect an existing summary.
18675 2018-06-08 Martin Liska <mliska@suse.cz>
18677 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
18678 * ipa-inline.h (estimate_edge_growth): Likewise.
18680 2018-06-08 Martin Liska <mliska@suse.cz>
18682 * cgraph.c (function_version_hasher::hash): Use
18683 cgraph_node::get_uid ().
18684 (function_version_hasher::equal):
18685 * cgraph.h (cgraph_node::get_uid): New method.
18686 * ipa-inline.c (update_caller_keys): Use
18687 cgraph_node::get_uid ().
18688 (update_callee_keys): Likewise.
18689 * ipa-utils.c (searchc): Likewise.
18690 (ipa_reduced_postorder): Likewise.
18691 * lto-cgraph.c (input_node): Likewise.
18692 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
18693 * symbol-summary.h (symtab_insertion): Likewise.
18694 (symtab_removal): Likewise.
18695 (symtab_duplication): Likewise.
18696 * tree-pretty-print.c (dump_function_header): Likewise.
18697 * tree-sra.c (convert_callers_for_node): Likewise.
18699 2018-06-08 Martin Liska <mliska@suse.cz>
18701 * cgraph.c (symbol_table::create_edge): Always assign a new
18703 (symbol_table::free_edge): Do not recycle numbers.
18704 * cgraph.h (cgraph_edge::get): New method.
18705 * symbol-summary.h (symtab_removal): Use it.
18706 (symtab_duplication): Likewise.
18707 (call_summary::hashable_uid): Remove.
18709 2018-06-08 Martin Liska <mliska@suse.cz>
18711 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
18712 (initialize_growth_caches): Remove.
18713 (free_growth_caches): Likewise.
18714 (do_estimate_edge_time): Use edge_growth_cache.
18715 (do_estimate_edge_size): Likewise.
18716 (do_estimate_edge_hints): Likewise.
18717 * ipa-inline.c (reset_edge_caches): Likewise.
18718 (recursive_inlining): Likewise.
18719 (inline_small_functions): Likewise.
18720 * ipa-inline.h (initialize_growth_caches): Remove.
18721 (estimate_edge_size): Likewise.
18722 (estimate_edge_time): Likewise.
18723 (estimate_edge_hints): Likewise.
18724 (reset_edge_growth_cache): Likewise.
18725 * symbol-summary.h (call_summary::remove): New method.
18727 2018-06-08 Martin Liska <mliska@suse.cz>
18729 * ipa-cp.c (class edge_clone_summary): New summary.
18730 (grow_edge_clone_vectors): Remove.
18731 (ipcp_edge_duplication_hook): Remove.
18732 (class edge_clone_summary_t): New call_summary class.
18733 (ipcp_edge_removal_hook): Remove.
18734 (edge_clone_summary_t::duplicate): New function.
18735 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
18736 (create_specialized_node): Likewise.
18737 (ipcp_driver): Initialize edge_clone_summaries and do not
18740 2018-06-08 Martin Liska <mliska@suse.cz>
18742 * symbol-summary.h (get): New function.
18743 (call_summary::m_initialize_when_cloning): New class member.
18745 2018-06-08 Martin Liska <mliska@suse.cz>
18747 * cgraph.c (cgraph_node::remove): Do not recycle uid.
18748 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
18749 (symbol_table::allocate_cgraph_symbol): Do not set uid.
18750 * passes.c (uid_hash_t): Record removed_nodes by their uids.
18751 (remove_cgraph_node_from_order): Use the removed_nodes set.
18752 (do_per_function_toporder): Likwise.
18753 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
18754 instead of summary_uid.
18755 (symtab_removal): Likewise.
18756 (symtab_duplication): Likewise.
18758 2018-06-08 Martin Liska <mliska@suse.cz>
18760 * ipa-cp.c (ipcp_store_bits_results): Use
18761 ipcp_transformation_sum.
18762 (ipcp_store_vr_results): Likewise.
18763 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
18765 (ipcp_transformation_initialize): ... this.
18766 (ipa_set_node_agg_value_chain):
18767 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
18768 (write_ipcp_transformation_info): Likewise.
18769 (read_ipcp_transformation_info): Likewise.
18770 (ipcp_update_bits): Likewise.
18771 (ipcp_update_vr): Likewise.
18772 (ipcp_transform_function): Likewise.
18773 * ipa-prop.h: Rename ipcp_transformation_summary to
18774 ipcp_transformation.
18775 (class ipcp_transformation_t): New function summary.
18776 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
18777 (ipa_get_agg_replacements_for_node): Likewise.
18779 2018-06-08 Martin Liska <mliska@suse.cz>
18781 * ipa-pure-const.c (struct funct_state_d): Do it class instead
18783 (class funct_state_summary_t): New function_summary class.
18784 (has_function_state): Remove.
18785 (get_function_state): Likewise.
18786 (set_function_state): Likewise.
18787 (add_new_function): Likewise.
18788 (funct_state_summary_t::insert): New function.
18789 (duplicate_node_data): Remove.
18790 (remove_node_data): Remove.
18791 (funct_state_summary_t::duplicate): New function.
18792 (register_hooks): Create new funct_state_summaries.
18793 (pure_const_generate_summary): Use it.
18794 (pure_const_write_summary): Likewise.
18795 (pure_const_read_summary): Likewise.
18796 (propagate_pure_const): Likewise.
18797 (propagate_nothrow): Likewise.
18798 (dump_malloc_lattice): Likewise.
18799 (propagate_malloc): Likewise.
18800 (execute): Do not register hooks, just remove summary
18802 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
18805 2018-06-08 Martin Liska <mliska@suse.cz>
18807 * ipa-reference.c (remove_node_data): Remove.
18808 (duplicate_node_data): Likewise.
18809 (class ipa_ref_var_info_summary_t): New class.
18810 (class ipa_ref_opt_summary_t): Likewise.
18811 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
18812 (get_reference_optimization_summary): Use
18813 ipa_ref_opt_sum_summaries.
18814 (set_reference_vars_info): Remove.
18815 (set_reference_optimization_summary): Likewise.
18816 (ipa_init): Create summaries.
18817 (init_function_info): Use function summary.
18818 (ipa_ref_opt_summary_t::duplicate): New function.
18819 (ipa_ref_opt_summary_t::remove): New function.
18820 (get_read_write_all_from_node): Fix GNU coding style.
18821 (propagate): Use function summary.
18822 (write_node_summary_p): Fix GNU coding style.
18823 (stream_out_bitmap): Likewise.
18824 (ipa_reference_read_optimization_summary): Use function summary.
18825 (ipa_reference_c_finalize): Do not release hooks.
18827 2018-06-08 Martin Liska <mliska@suse.cz>
18829 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
18830 (analyze_function_body): Extract multiple calls of get_create.
18831 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18832 * ipa-inline.c (recursive_inlining): Use ::get method.
18833 * ipa-inline.h (estimate_edge_growth): Likewise.
18835 2018-06-08 Martin Liska <mliska@suse.cz>
18837 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
18839 (hsa_function_summary::hsa_function_summary): Use the new enum
18841 (hsa_gpu_implementation_p): Use hsa_summaries::get.
18842 * hsa-gen.c (hsa_get_host_function): Likewise.
18843 (get_brig_function_name): Likewise.
18844 * ipa-hsa.c (process_hsa_functions): Likewise.
18845 (ipa_hsa_write_summary): Likewise.
18846 * symbol-summary.h (symtab_duplication): Use ::get function/
18847 (get): New function.
18849 2018-06-08 Martin Liska <mliska@suse.cz>
18851 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
18853 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
18854 (hsa_register_kernel): Likewise.
18855 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
18856 * hsa-gen.c (hsa_get_host_function): Likewise.
18857 (get_brig_function_name): Likewise.
18858 (generate_hsa): Likewise.
18859 (pass_gen_hsail::execute): Likewise.
18860 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
18861 (devirtualization_time_bonus): Likewise.
18862 (ipcp_propagate_stage): Likewise.
18863 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
18864 (edge_set_predicate): Likewise.
18865 (evaluate_conditions_for_known_args): Likewise.
18866 (evaluate_properties_for_edge): Likewise.
18867 (ipa_fn_summary::reset): Likewise.
18868 (ipa_fn_summary_t::duplicate): Likewise.
18869 (dump_ipa_call_summary): Likewise.
18870 (ipa_dump_fn_summary): Likewise.
18871 (analyze_function_body): Likewise.
18872 (compute_fn_summary): Likewise.
18873 (estimate_edge_devirt_benefit): Likewise.
18874 (estimate_edge_size_and_time): Likewise.
18875 (estimate_calls_size_and_time): Likewise.
18876 (estimate_node_size_and_time): Likewise.
18877 (inline_update_callee_summaries): Likewise.
18878 (remap_edge_change_prob): Likewise.
18879 (remap_edge_summaries): Likewise.
18880 (ipa_merge_fn_summary_after_inlining): Likewise.
18881 (ipa_update_overall_fn_summary): Likewise.
18882 (read_ipa_call_summary): Likewise.
18883 (inline_read_section): Likewise.
18884 (write_ipa_call_summary): Likewise.
18885 (ipa_fn_summary_write): Likewise.
18886 (ipa_free_fn_summary): Likewise.
18887 * ipa-hsa.c (process_hsa_functions): Likewise.
18888 (ipa_hsa_write_summary): Likewise.
18889 (ipa_hsa_read_section): Likewise.
18890 * ipa-icf.c (sem_function::merge): Likewise.
18891 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18892 (do_estimate_edge_time): Likewise.
18893 (estimate_size_after_inlining): Likewise.
18894 (estimate_growth): Likewise.
18895 (growth_likely_positive): Likewise.
18896 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
18897 (inline_call): Likewise.
18898 * ipa-inline.c (caller_growth_limits): Likewise.
18899 (can_inline_edge_p): Likewise.
18900 (can_inline_edge_by_limits_p): Likewise.
18901 (compute_uninlined_call_time): Likewise.
18902 (compute_inlined_call_time): Likewise.
18903 (want_inline_small_function_p): Likewise.
18904 (edge_badness): Likewise.
18905 (update_caller_keys): Likewise.
18906 (update_callee_keys): Likewise.
18907 (recursive_inlining): Likewise.
18908 (inline_small_functions): Likewise.
18909 (inline_to_all_callers_1): Likewise.
18910 (dump_overall_stats): Likewise.
18911 (early_inline_small_functions): Likewise.
18912 (early_inliner): Likewise.
18913 * ipa-inline.h (estimate_edge_growth): Likewise.
18914 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18915 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18916 * ipa-prop.h (IPA_NODE_REF): Likewise.
18917 (IPA_EDGE_REF): Likewise.
18918 * ipa-pure-const.c (malloc_candidate_p): Likewise.
18919 (propagate_malloc): Likewise.
18920 * ipa-split.c (execute_split_functions): Likewise.
18921 * symbol-summary.h: Rename get to get_create.
18923 (get_create): Likewise.
18924 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
18926 2018-06-08 Martin Liska <mliska@suse.cz>
18928 * symbol-summary.h (release): Move definition out of class
18930 (symtab_removal): Likewise.
18931 (symtab_duplication): Likewise.
18933 2018-06-08 Martin Liska <mliska@suse.cz>
18935 * symbol-summary.h (function_summary): Move constructor
18936 implementation out of class declaration.
18937 (release): Likewise.
18938 (symtab_insertion): Likewise.
18939 (symtab_removal): Likewise.
18940 (symtab_duplication): Likewise.
18943 2018-06-08 Martin Liska <mliska@suse.cz>
18945 * Makefile.in: Remove support for MPX (macros, related functions,
18946 fields in cgraph_node, ...).
18947 * builtin-types.def (BT_BND): Likewise.
18948 (BT_FN_BND_CONST_PTR): Likewise.
18949 (BT_FN_CONST_PTR_BND): Likewise.
18950 (BT_FN_VOID_PTR_BND): Likewise.
18951 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
18952 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
18953 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
18954 (expand_builtin_mempcpy_with_bounds): Likewise.
18955 (expand_builtin_memset_with_bounds): Likewise.
18956 (expand_builtin_memset_args): Likewise.
18957 (std_expand_builtin_va_start): Likewise.
18958 (expand_builtin): Likewise.
18959 (expand_builtin_with_bounds): Likewise.
18960 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
18961 (DEF_LIB_BUILTIN_CHKP): Likewise.
18962 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
18963 (DEF_CHKP_BUILTIN): Likewise.
18964 (BUILT_IN_MEMCPY): Likewise.
18965 (BUILT_IN_MEMMOVE): Likewise.
18966 (BUILT_IN_MEMPCPY): Likewise.
18967 (BUILT_IN_MEMSET): Likewise.
18968 (BUILT_IN_STPCPY): Likewise.
18969 (BUILT_IN_STRCAT): Likewise.
18970 (BUILT_IN_STRCHR): Likewise.
18971 (BUILT_IN_STRCPY): Likewise.
18972 (BUILT_IN_STRLEN): Likewise.
18973 (BUILT_IN_MEMCPY_CHK): Likewise.
18974 (BUILT_IN_MEMMOVE_CHK): Likewise.
18975 (BUILT_IN_MEMPCPY_CHK): Likewise.
18976 (BUILT_IN_MEMSET_CHK): Likewise.
18977 (BUILT_IN_STPCPY_CHK): Likewise.
18978 (BUILT_IN_STRCAT_CHK): Likewise.
18979 (BUILT_IN_STRCPY_CHK): Likewise.
18980 * calls.c (store_bounds): Likewise.
18981 (emit_call_1): Likewise.
18982 (special_function_p): Likewise.
18983 (maybe_warn_nonstring_arg): Likewise.
18984 (initialize_argument_information): Likewise.
18985 (finalize_must_preallocate): Likewise.
18986 (compute_argument_addresses): Likewise.
18987 (expand_call): Likewise.
18988 * cfgexpand.c (expand_call_stmt): Likewise.
18989 (expand_return): Likewise.
18990 (expand_gimple_stmt_1): Likewise.
18991 (pass_expand::execute): Likewise.
18992 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18993 (cgraph_node::remove): Likewise.
18994 (cgraph_node::dump): Likewise.
18995 (cgraph_node::verify_node): Likewise.
18996 * cgraph.h (chkp_function_instrumented_p): Likewise.
18997 (symtab_node::get_alias_target): Likewise.
18998 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
18999 (cgraph_local_p): Likewise.
19000 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
19001 (cgraph_edge::rebuild_references): Likewise.
19002 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
19003 (walk_polymorphic_call_targets): Likewise.
19004 (cgraph_node::expand_thunk): Likewise.
19005 (symbol_table::output_weakrefs): Likewise.
19006 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
19007 (ix86_handle_option): Likewise.
19008 * config/i386/constraints.md: Likewise.
19009 * config/i386/i386-builtin-types.def (BND): Likewise.
19013 * config/i386/i386-builtin.def (BDESC_END): Likewise.
19014 (BDESC_FIRST): Likewise.
19016 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
19017 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
19018 * config/i386/i386.c (enum reg_class): Likewise.
19019 (ix86_target_string): Likewise.
19020 (ix86_option_override_internal): Likewise.
19021 (ix86_conditional_register_usage): Likewise.
19022 (ix86_valid_target_attribute_inner_p): Likewise.
19023 (ix86_set_indirect_branch_type): Likewise.
19024 (ix86_set_current_function): Likewise.
19025 (ix86_function_arg_regno_p): Likewise.
19026 (init_cumulative_args): Likewise.
19027 (ix86_function_arg_advance): Likewise.
19028 (ix86_function_arg): Likewise.
19029 (ix86_pass_by_reference): Likewise.
19030 (ix86_function_value_regno_p): Likewise.
19031 (ix86_function_value_1): Likewise.
19032 (ix86_function_value_bounds): Likewise.
19033 (ix86_return_in_memory): Likewise.
19034 (ix86_setup_incoming_vararg_bounds): Likewise.
19035 (ix86_va_start): Likewise.
19036 (indirect_thunk_need_prefix): Likewise.
19037 (print_reg): Likewise.
19038 (ix86_print_operand): Likewise.
19039 (ix86_expand_call): Likewise.
19040 (ix86_output_function_return): Likewise.
19041 (reg_encoded_number): Likewise.
19042 (BDESC_VERIFYS): Likewise.
19043 (ix86_init_mpx_builtins): Likewise.
19044 (ix86_init_builtins): Likewise.
19045 (ix86_emit_cmove): Likewise.
19046 (ix86_emit_move_max): Likewise.
19047 (ix86_expand_builtin): Likewise.
19048 (ix86_builtin_mpx_function): Likewise.
19049 (ix86_get_arg_address_for_bt): Likewise.
19050 (ix86_load_bounds): Likewise.
19051 (ix86_store_bounds): Likewise.
19052 (ix86_load_returned_bounds): Likewise.
19053 (ix86_store_returned_bounds): Likewise.
19054 (ix86_class_likely_spilled_p): Likewise.
19055 (ix86_hard_regno_mode_ok): Likewise.
19056 (x86_order_regs_for_local_alloc): Likewise.
19057 (ix86_mitigate_rop): Likewise.
19058 (ix86_bnd_prefixed_insn_p): Likewise.
19059 (ix86_mpx_bound_mode): Likewise.
19060 (ix86_make_bounds_constant): Likewise.
19061 (ix86_initialize_bounds): Likewise.
19062 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
19063 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
19064 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
19065 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
19066 (TARGET_CHKP_BOUND_MODE): Likewise.
19067 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
19068 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
19069 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
19070 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
19071 * config/i386/i386.h (TARGET_MPX): Likewise.
19072 (TARGET_MPX_P): Likewise.
19073 (VALID_BND_REG_MODE): Likewise.
19074 (FIRST_BND_REG): Likewise.
19075 (LAST_BND_REG): Likewise.
19076 (enum reg_class): Likewise.
19077 (BND_REG_P): Likewise.
19078 (BND_REGNO_P): Likewise.
19079 (BNDmode): Likewise.
19080 (ADJUST_INSN_LENGTH): Likewise.
19081 * config/i386/i386.md: Likewise.
19082 * config/i386/i386.opt: Likewise.
19083 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
19084 (defined): Likewise.
19085 (LINK_MPX): Likewise.
19086 (MPX_SPEC): Likewise.
19087 (LIBMPX_SPEC): Likewise.
19088 (LIBMPXWRAPPERS_SPEC): Likewise.
19089 (CHKP_SPEC): Likewise.
19090 * config/i386/predicates.md: Likewise.
19091 * dbxout.c (dbxout_type): Likewise.
19092 * doc/extend.texi: Likewise.
19093 * doc/invoke.texi: Likewise.
19094 * doc/md.texi: Likewise.
19095 * doc/tm.texi: Likewise.
19096 * doc/tm.texi.in: Likewise.
19097 * dwarf2out.c (is_base_type): Likewise.
19098 (gen_formal_types_die): Likewise.
19099 (gen_subprogram_die): Likewise.
19100 (gen_type_die_with_usage): Likewise.
19101 (gen_decl_die): Likewise.
19102 (dwarf2out_late_global_decl): Likewise.
19103 * expr.c (expand_assignment): Likewise.
19104 (emit_storent_insn): Likewise.
19105 (store_expr_with_bounds): Likewise.
19106 (store_expr): Likewise.
19107 (expand_expr_real_1): Likewise.
19108 * expr.h (store_expr_with_bounds): Likewise.
19109 * function.c (use_register_for_decl): Likewise.
19110 (struct bounds_parm_data): Likewise.
19111 (assign_parms_augmented_arg_list): Likewise.
19112 (assign_parm_find_entry_rtl): Likewise.
19113 (assign_parm_is_stack_parm): Likewise.
19114 (assign_parm_load_bounds): Likewise.
19115 (assign_bounds): Likewise.
19116 (assign_parms): Likewise.
19117 (expand_function_start): Likewise.
19118 * gcc.c (CHKP_SPEC): Likewise.
19119 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
19120 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
19121 (wrestrict_dom_walker::check_call): Likewise.
19122 * gimple.c (gimple_build_call_from_tree): Likewise.
19123 * gimple.h (enum gf_mask): Likewise.
19124 (gimple_call_with_bounds_p): Likewise.
19125 (gimple_call_set_with_bounds): Likewise.
19126 * gimplify.c (gimplify_init_constructor): Likewise.
19127 * ipa-cp.c (initialize_node_lattices): Likewise.
19128 (propagate_constants_across_call): Likewise.
19129 (find_more_scalar_values_for_callers_subset): Likewise.
19130 * ipa-hsa.c (process_hsa_functions): Likewise.
19131 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
19132 * ipa-icf.c (sem_function::merge): Likewise.
19133 * ipa-inline.c (early_inliner): Likewise.
19134 * ipa-pure-const.c (warn_function_noreturn): Likewise.
19135 (warn_function_cold): Likewise.
19136 (propagate_pure_const): Likewise.
19137 * ipa-ref.h (enum GTY): Likewise.
19138 * ipa-split.c (find_retbnd): Likewise.
19139 (consider_split): Likewise.
19140 (split_function): Likewise.
19141 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
19142 * ipa.c (walk_polymorphic_call_targets): Likewise.
19143 (symbol_table::remove_unreachable_nodes): Likewise.
19144 (process_references): Likewise.
19145 (cgraph_build_static_cdtor_1): Likewise.
19146 * lto-cgraph.c (lto_output_node): Likewise.
19147 (output_refs): Likewise.
19148 (compute_ltrans_boundary): Likewise.
19149 (input_overwrite_node): Likewise.
19150 (input_node): Likewise.
19151 (input_cgraph_1): Likewise.
19152 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
19153 * passes.c (pass_manager::execute_early_local_passes): Likewise.
19154 (class pass_chkp_instrumentation_passes): Likewise.
19155 (make_pass_chkp_instrumentation_passes): Likewise.
19156 * passes.def: Likewise.
19157 * rtl.h (struct GTY): Likewise.
19158 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
19159 * stor-layout.c (layout_type): Likewise.
19160 * symtab.c: Likewise.
19161 * target.def: Likewise.
19162 * targhooks.c (default_chkp_bound_type): Likewise.
19163 (default_chkp_bound_mode): Likewise.
19164 (default_builtin_chkp_function): Likewise.
19165 (default_chkp_function_value_bounds): Likewise.
19166 (default_chkp_make_bounds_constant): Likewise.
19167 (default_chkp_initialize_bounds): Likewise.
19168 * targhooks.h (default_chkp_bound_type): Likewise.
19169 (default_chkp_bound_mode): Likewise.
19170 (default_builtin_chkp_function): Likewise.
19171 (default_chkp_function_value_bounds): Likewise.
19172 (default_chkp_make_bounds_constant): Likewise.
19173 (default_chkp_initialize_bounds): Likewise.
19174 * toplev.c (compile_file): Likewise.
19175 (process_options): Likewise.
19176 * tree-core.h (DEF_BUILTIN): Likewise.
19177 (DEF_BUILTIN_CHKP): Likewise.
19178 * tree-inline.c (declare_return_variable): Likewise.
19179 (remap_gimple_stmt): Likewise.
19180 (copy_bb): Likewise.
19181 (initialize_inlined_parameters): Likewise.
19182 (expand_call_inline): Likewise.
19183 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
19184 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
19185 (make_pass_ipa_chkp_produce_thunks): Likewise.
19186 (make_pass_chkp): Likewise.
19187 (make_pass_chkp_opt): Likewise.
19188 (make_pass_chkp_instrumentation_passes): Likewise.
19189 * tree-pretty-print.c (dump_generic_node): Likewise.
19190 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
19191 * tree-ssa-dce.c (propagate_necessity): Likewise.
19192 (eliminate_unnecessary_stmts): Likewise.
19193 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
19194 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
19195 * tree-ssa-sccvn.h: Likewise.
19196 * tree-ssa-strlen.c (get_string_length): Likewise.
19197 (valid_builtin_call): Likewise.
19198 (adjust_last_stmt): Likewise.
19199 (handle_builtin_strchr): Likewise.
19200 (handle_builtin_strcpy): Likewise.
19201 (handle_builtin_stxncpy): Likewise.
19202 (handle_builtin_memcpy): Likewise.
19203 (handle_builtin_strcat): Likewise.
19204 (strlen_check_and_optimize_stmt): Likewise.
19205 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
19206 * tree-streamer-in.c: Likewise.
19207 * tree-streamer.c (record_common_node): Likewise.
19208 * tree.c (tree_code_size): Likewise.
19209 (wide_int_to_tree_1): Likewise.
19210 (type_contains_placeholder_1): Likewise.
19211 (build_common_tree_nodes): Likewise.
19212 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
19213 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
19214 (POINTER_BOUNDS_P): Likewise.
19215 (BOUNDED_TYPE_P): Likewise.
19216 (BOUNDED_P): Likewise.
19217 (CALL_WITH_BOUNDS_P): Likewise.
19218 (pointer_bounds_type_node): Likewise.
19219 * value-prof.c (gimple_ic): Likewise.
19220 * var-tracking.c (vt_add_function_parameters): Likewise.
19221 * varasm.c (make_decl_rtl): Likewise.
19222 (assemble_start_function): Likewise.
19223 (output_constant): Likewise.
19224 (maybe_assemble_visibility): Likewise.
19225 * varpool.c (ctor_for_folding): Likewise.
19226 * chkp-builtins.def: Remove.
19227 * ipa-chkp.c: Remove.
19228 * ipa-chkp.h: Remove.
19229 * rtl-chkp.c: Remove.
19230 * rtl-chkp.h: Remove.
19231 * tree-chkp-opt.c: Remove.
19232 * tree-chkp.c: Remove.
19233 * tree-chkp.h: Remove.
19235 2018-06-07 Carl Love <cel@us.ibm.com>
19237 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19238 vextract_fp_from_shortl): Add BE support.
19240 2018-06-07 Paul Koning <ni1d@arrl.net>
19242 * compare-elim.c (try_merge_compare): Don't merge compare if
19243 address contains a side effect.
19244 (try_eliminate_compare): Likewise.
19246 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
19248 * config.gcc: Support "tremont".
19249 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
19250 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19252 * config/i386/i386.c (m_TREMONT): Define.
19253 (processor_target_table): Add "tremont".
19254 (PTA_TREMONT): Define.
19255 (ix86_lea_outperforms): Add TARGET_TREMONT.
19256 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
19257 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
19258 and M_INTEL_GOLDMONT_PLUS.
19259 (fold_builtin_cpu): Add "tremont".
19260 (ix86_add_stmt_cost): Add TARGET_TREMONT.
19261 (ix86_option_override_internal): Add "tremont".
19262 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
19263 (processor_type): Add PROCESSOR_TREMONT.
19264 * config/i386/x86-tune.def: Add m_TREMONT.
19265 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
19267 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19269 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
19270 symbol defined for msp430i* devices to be lower case.
19272 2018-06-07 Richard Biener <rguenther@suse.de>
19274 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
19275 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
19276 Properly wrap signed arithmetic if overflow wraps.
19278 2018-06-07 Jakub Jelinek <jakub@redhat.com>
19280 PR tree-optimization/69615
19281 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
19282 of a cast from a same precision integral SSA_NAME in a bb dominated
19283 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
19284 cast to utype if rhs2 has already a compatible type.
19286 2018-06-07 Richard Biener <rguenther@suse.de>
19288 PR tree-optimization/85935
19289 * graphite-scop-detection.c (find_params_in_bb): Analyze
19290 condition operands with respect to the correct loop. Assert
19291 the analysis doesn't fail.
19293 2018-06-04 Carl Love <cel@us.ibm.com>
19295 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
19296 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
19297 as it is slightly cheaper.
19298 (first_match_or_eos_index_<mode>):
19299 Calculate index using natural element order.
19300 (first_match_index_<mode>):
19301 Calculate index using natural element order.
19302 (first_match_or_eos_index_<mode>):
19303 Calculate index using natural order.
19304 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
19305 for BE and LE modes.
19306 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
19307 P9V_BUILTIN_VCLZLSBB_V16QI.
19308 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
19311 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
19313 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
19314 indentation and line wrap for many prototypes. Add missing
19315 @smallexample directives around block of prototypes for vec_xl and
19318 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
19320 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
19321 track if we pass or return IEEE 128-bit floating point.
19322 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
19323 C++ mangling that is compatible with GCC 8.1.
19324 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
19325 (init_cumulative_args): Note if we pass or return IEEE 128-bit
19326 floating point types.
19327 (rs6000_function_arg_advance_1): Likewise.
19328 (rs6000_mangle_type): Optionally generate mangled names that match
19329 what GCC 8.1 generated for IEEE 128-bit floating point types.
19330 (rs6000_globalize_decl_name): If we have an external function that
19331 passes or returns IEEE 128-bit types, generate a weak reference
19332 from the mangled name used in GCC 8.1 to the current mangled
19334 (rs6000_init_builtins): Make __ibm128 use the long double type if
19335 long double is IBM extended double. Make __float128 use the long
19336 double type if long double is IEEE 128-bit.
19339 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
19340 macro for __ibm128 built-in functions.
19341 (PACK_IF): Add __ibm128 pack/unpack functions.
19342 (UNPACK_IF): Likewise.
19343 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
19344 enable long double built-in functions if long double is IEEE
19345 128-bit floating point.
19346 (rs6000_invalid_builtin): Update long double built-in function
19348 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
19349 functions, adjust the built-in function to use the long double
19350 built-in function if __ibm128 and long double are the same type.
19351 * doc/extend.texi (PowerPC builtins): Update documention for
19352 __builtin_{,un}pack_longdouble. Add documentation for
19353 __builtin_{,un}pack_ibm128.
19355 2018-06-06 Jim Wilson <jimw@sifive.com>
19357 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
19358 (struct machine_function): New field interrupt_mode.
19359 (riscv_handle_type_attribute): New function. Add forward declaration.
19360 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
19361 (riscv_expand_epilogue): Check interrupt_mode field.
19362 (riscv_set_current_function): Check interrupt attribute args and
19363 set interrupt_mode field.
19364 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
19365 (riscv_sret, riscv_uret): New.
19366 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
19367 new arguments to interrupt attribute.
19369 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
19372 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
19373 Don't handle -mcpu=power8 if -mpower9-vector is also used.
19375 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
19377 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
19378 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
19379 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19380 several redundant entries.
19382 2018-06-06 David Malcolm <dmalcolm@redhat.com>
19384 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
19385 type from "rtx" to "rtx_insn *".
19386 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
19387 for local "call_insn", removing cast.
19388 (ix86_expand_call): Likewise, introducing a "call_insn" local.
19390 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
19392 PR tree-optimization/86066
19393 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
19394 for BIT_INSERT_EXPR stores.
19396 2018-06-06 Richard Biener <rguenther@suse.de>
19398 PR tree-optimization/86062
19399 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
19400 component refs ontop
19401 of to be offsetted base.
19403 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19405 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
19406 to be static and remove check on interrupt attribute name.
19408 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
19410 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
19411 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
19413 2018-06-05 Steve Ellcey <sellcey@cavium.com>
19416 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
19418 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
19419 Remove second argument, change how error is called.
19420 (aarch64_layout_arg): Remove second argument from
19421 aarch64_err_no_fpadvsimd call.
19422 (aarch64_init_cumulative_args): Ditto.
19423 (aarch64_gimplify_va_arg_expr): Ditto.
19424 * config/aarch64/aarch64.md (mov<mode>): Ditto.
19426 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
19428 * config/i386/i386.md (simple_return_indirect_internal): New expander.
19429 (*simple_return_indirect_internal<mode>): Rename from
19430 simple_return_indirect_internal. Use W mode iterator.
19431 (rstorssp): New expander.
19432 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
19433 (clrssbsy): New expander.
19434 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
19436 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19438 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
19440 (cmse_check_pointed_object): Likewise.
19442 2018-06-05 Martin Liska <mliska@suse.cz>
19444 PR gcov-profile/47618
19445 * doc/invoke.texi: Document how -fprofile-dir format
19448 2018-06-05 Richard Biener <rguenther@suse.de>
19450 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
19451 removal pretend DOM info isn't available so we do not update
19452 it and only remove edges, not dominated blocks. Actually free
19453 DOM info in case we removed something. Remove unreachable blocks.
19454 (mfb_keep_latches): Work with either DOM info or marked backedges.
19455 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
19456 first. Mark backedges if DOM info isn't available.
19457 (Re-)compute DOM info after cleanup_control_flow_pre.
19459 2018-06-05 Richard Biener <rguenther@suse.de>
19461 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
19462 (locus_discrim_hasher::hash): Adjust.
19463 (locus_discrim_hasher::equal): Likewise.
19464 (next_discriminator_for_locus): Work on line directly.
19465 (same_line_p): Pass in expanded locus1 as well.
19466 (assign_discriminators): Avoid redundant location expansions.
19468 2018-06-05 Richard Biener <rguenther@suse.de>
19470 PR tree-optimization/86046
19471 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
19472 if required after clearing TREE_ADDRESSABLE.
19474 2018-06-05 Richard Biener <rguenther@suse.de>
19476 PR tree-optimization/86047
19477 * tree-ssa-loop.c (for_each_index): Glob handling of all
19478 decls and constants and really handle all of them.
19480 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19483 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
19484 qualifier_void_pointer and qualifier_const_void_pointer.
19485 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
19486 (arm_init_builtins): Handle the above.
19487 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
19488 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
19489 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
19492 2018-06-05 Martin Liska <mliska@suse.cz>
19494 * auto-profile.c (read_autofdo_file): Do not use
19495 gcov_ctr_summary struct.
19496 (afdo_callsite_hot_enough_for_early_inline): Likewise.
19497 * coverage.c (struct counts_entry): Likewise.
19498 (read_counts_file): Read just single summary entry.
19499 (get_coverage_counts): Use gcov_summary struct.
19500 * coverage.h (get_coverage_counts): Likewise.
19501 * gcov-dump.c (dump_working_sets): Likewise.
19502 (tag_summary): Dump just single summary.
19503 * gcov-io.c (gcov_write_summary): Write just histogram
19505 (gcov_read_summary): Read just single summary.
19506 (compute_working_sets): Use gcov_summary struct.
19507 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
19508 of GCOV_COUNTERS_SUMMABLE.
19509 (GCOV_COUNTERS_SUMMABLE): Remove.
19510 (GCOV_FIRST_VALUE_COUNTER): Replace with
19511 GCOV_COUNTER_V_INTERVAL.
19512 (struct gcov_ctr_summary): Remove.
19513 (struct gcov_summary): Directly use fields of former
19515 (compute_working_sets): Use gcov_summary struct.
19516 * gcov.c (read_count_file): Do not use ctrs fields.
19517 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
19519 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
19521 * profile.c: Likewise.
19522 * profile.h: Likewise.
19524 2018-06-05 Martin Liska <mliska@suse.cz>
19526 PR gcov-profile/84846
19527 * gcov.c (output_lines): Print working directory only
19528 in intermediate format.
19530 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
19532 * config/s390/s390-builtin-types.def: Add void function type.
19533 * config/s390/s390-builtins.def: Use the function type for the
19536 2018-06-04 Jim Wilson <jimw@sifive.com>
19538 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
19540 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
19541 and maybe_eh_return. Change regno to unsigned int. Use new args to
19542 handle EH_RETURN_DATA_REGNO registers properly.
19543 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
19544 (riscv_expand_epilogue): Update comment. Change argument name and
19545 type. Update code to use new name and type. Pass new args to
19546 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
19548 * config/riscv/riscv.md (NORMAL_RETURN): New.
19549 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
19550 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
19551 (eh_return): Call gen_eh_return_internal and emit barrier.
19552 (eh_return_internal): Call riscv_expand_epilogue.
19554 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
19556 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
19557 bit_insertion field and declare can_be_merged_into method.
19558 (merged_store_group::can_be_merged_into): New method.
19559 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
19560 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
19561 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
19563 2018-06-04 Richard Biener <rguenther@suse.de>
19565 PR tree-optimization/85955
19566 * builtins.c (fold_builtin_sincos): Convert pointers to
19567 destination to appropriate type before dereferencing.
19569 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
19571 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
19573 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
19575 * expr.c (expand_expr_real_1): Force the operand into memory if
19576 its TYPE_MODE is BLKmode and if there is no integer mode for
19577 the number of bits being extracted.
19579 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19583 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
19584 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
19586 2018-06-04 Richard Biener <rguenther@suse.de>
19588 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
19589 (cleanup_tree_cfg_noloop): ... single caller. Do
19590 start_recording_case_labels later.
19592 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
19594 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
19595 to _IMMINTRIN_H_INCLUDED.
19596 * config/i386/pconfigintrin.h: Ditto.
19597 * config/i386/waitpkgintrin.h: Ditto.
19598 * config/i386/immintrin.h: Add includes for sgxintrin.h,
19599 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
19600 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
19601 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
19602 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
19603 waitpkgintrin.h and cldemoteintrin.h.
19605 2018-06-04 Richard Biener <rguenther@suse.de>
19607 PR tree-optimization/86038
19608 * tracer.c (find_best_successor): Check probability for
19609 being initialized, bail out if not.
19611 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
19614 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
19615 of bits to ignore when comparing architectures.
19617 2018-06-04 Jakub Jelinek <jakub@redhat.com>
19619 PR tree-optimization/69615
19620 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
19621 maximum or minimum of the type, try to merge it also as if
19622 range1 is + [-, x - 1] or + [x + 1, -].
19625 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
19627 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
19629 PR tree-optimization/86034
19630 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
19631 the unsigned bitfield type in a bit insertion sequence if it does not
19632 have a larger precision than the bitfield size.
19633 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
19635 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
19637 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
19639 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19641 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
19642 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
19643 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
19644 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
19646 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
19648 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
19649 Disable -fdelete-null-pointer-checks for ELF toolchain.
19651 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19652 Kito Cheng <kito.cheng@gmail.com>
19654 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
19655 (nds32le-*-*, nds32be-*-*): Integrate checking process.
19656 (nds32*-*-*): Add glibc and uclibc conditions.
19657 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
19658 (TARGET_EXCEPT_UNWIND_INFO): Define.
19659 * config/nds32/elf.h: New file.
19660 * config/nds32/linux.h: New file.
19661 * config/nds32/nds32-elf.opt: New file.
19662 * config/nds32/nds32-linux.opt: New file.
19663 * config/nds32/nds32-fp-as-gp.c
19664 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
19665 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
19667 (nds32_asm_file_end): Ditto.
19668 (nds32_print_operand): Ditto.
19669 (nds32_insert_attributes): Ditto.
19670 (nds32_init_libfuncs): New function.
19671 (TARGET_HAVE_TLS): Define.
19672 (TARGET_INIT_LIBFUNCS): Define.
19673 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
19675 (TARGET_ELF): Apply different mcmodel setting.
19676 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
19677 been migrated into elf.h and linux.h files.
19678 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
19679 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
19680 (mcmodel): The content has been migrated into nds32-elf.opt and
19681 nds32-linux.opt files.
19682 * config/nds32/t-elf: New file.
19683 * config/nds32/t-linux: New file.
19685 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
19686 Shiva Chen <shiva0217@gmail.com>
19688 * config/nds32/constants.md (unspec_volatile_element): Add
19689 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
19690 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
19692 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
19693 (make_pass_nds32_fp_as_gp): Declare.
19694 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
19696 (nds32_asm_function_end_prologue): Remove unused asm output.
19697 (nds32_asm_function_begin_epilogue): Remove unused asm output.
19698 (nds32_asm_file_start): Output necessary fp_as_gp information.
19699 (nds32_option_override): Adjust register usage.
19700 (nds32_expand_prologue): Consider fp_as_gp situation.
19701 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
19702 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
19705 (simple_return): Ditto.
19706 (omit_fp_begin): Output special directive for fp_as_gp.
19707 (omit_fp_end): Output special directive for fp_as_gp.
19708 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
19709 mforbid-fp-as-gp): New options.
19711 2018-06-01 Mark Wielaard <mark@klomp.org>
19713 * dwarf2out.c (dwarf2out_finish): Remove generation of
19714 DW_AT_loclists_base.
19716 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
19718 * gimple-ssa-store-merging.c: Include gimple-fold.h.
19719 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
19720 (struct merged_store_group): Add bit_insertion field.
19721 (dump_char_array): Use standard hexadecimal format.
19722 (merged_store_group::merged_store_group): Set bit_insertion to false.
19723 (merged_store_group::apply_stores): Use optimal buffer size. Deal
19724 with BIT_INSERT_EXPR stores. Move up code updating the mask and
19725 also print the mask in the dump file.
19726 (pass_store_merging::gate): Minor tweak.
19727 (imm_store_chain_info::coalesce_immediate): Fix wrong association
19728 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
19729 stores with INTEGER_CST stores.
19730 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
19731 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
19732 and use it throughout. Generate bit insertion sequences if need be.
19733 (pass_store_merging::process_store): Remove redundant condition.
19734 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
19736 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
19738 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
19739 the 128-bit floating point types. Fix function comment.
19741 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19743 * config/aarch64/aarch64-simd.md
19744 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
19746 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
19749 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19751 PR tree-optimization/85989
19752 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
19754 (backprop::intersect_uses): Check it when deciding whether this
19755 is a backedge reference.
19756 (backprop::process_block): Add each phi to m_visited_phis
19757 after visiting it, then clear it at the end.
19759 2018-06-01 Richard Biener <rguenther@suse.de>
19761 * tree-vectorizer.h (vect_dr_stmt): New function.
19762 (vect_get_load_cost): Adjust.
19763 (vect_get_store_cost): Likewise.
19764 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
19765 Use vect_dr_stmt instead of DR_SMTT.
19766 (vect_record_base_alignments): Likewise.
19767 (vect_calculate_target_alignment): Likewise.
19768 (vect_compute_data_ref_alignment): Likewise and make static.
19769 (vect_update_misalignment_for_peel): Likewise.
19770 (vect_verify_datarefs_alignment): Likewise.
19771 (vector_alignment_reachable_p): Likewise.
19772 (vect_get_data_access_cost): Likewise. Pass down
19773 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
19774 (vect_get_peeling_costs_all_drs): Likewise.
19775 (vect_peeling_hash_get_lowest_cost): Likewise.
19776 (vect_enhance_data_refs_alignment): Likewise.
19777 (vect_find_same_alignment_drs): Likewise.
19778 (vect_analyze_data_refs_alignment): Likewise.
19779 (vect_analyze_group_access_1): Likewise.
19780 (vect_analyze_group_access): Likewise.
19781 (vect_analyze_data_ref_access): Likewise.
19782 (vect_analyze_data_ref_accesses): Likewise.
19783 (vect_vfa_segment_size): Likewise.
19784 (vect_small_gap_p): Likewise.
19785 (vectorizable_with_step_bound_p): Likewise.
19786 (vect_prune_runtime_alias_test_list): Likewise.
19787 (vect_analyze_data_refs): Likewise.
19788 (vect_supportable_dr_alignment): Likewise.
19789 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
19790 (vect_gen_prolog_loop_niters): Likewise.
19791 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
19792 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
19794 (vect_recog_mask_conversion_pattern): Likewise.
19795 (vect_try_gather_scatter_pattern): Likewise.
19796 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
19797 to vect_get_store_cost.
19798 (vect_get_store_cost): Get stmt_info instead of DR.
19799 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
19800 (vect_get_load_cost): Get stmt_info instead of DR.
19802 2018-06-01 Richard Biener <rguenther@suse.de>
19804 PR middle-end/86017
19805 * gimple-fold.c (var_decl_component_p): Also allow offsetted
19806 vars wrapped in MEM_REFs.
19808 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
19810 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19811 Fix subreg tests so that we only return a choice between
19812 GENERAL_REGS and FP_REGS if the original classes included both.
19814 2018-06-01 Richard Biener <rguenther@suse.de>
19817 * tree-ssa-structalias.c (get_function_part_constraint):
19818 Handle NULL fi->decl.
19819 (find_func_aliases_for_call): Properly handle indirect
19820 fi from direct call.
19821 (find_func_clobbers): Likewise.
19822 (ipa_pta_execute): Likewise.
19823 (create_variable_info_for): For functions that are ifunc_resolver
19824 resolve to a varinfo that contains the result of the resolver call.
19825 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
19828 2018-05-31 Michael Collison <michael.collison@arm.com>
19830 * config/aarch64/aarch64.md:
19831 (*fix_to_zero_extenddfdi2): New pattern.
19832 * gcc.target/aarch64/fix_extend1.c: New testcase.
19834 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
19836 PR middle-end/78809
19837 PR middle-end/83026
19838 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
19839 and BUILT_IN_STRNCMP_EQ.
19840 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
19841 BUILT_IN_STRNCMP_EQ.
19842 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
19843 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19844 (gimple_fold_builtin): Likewise.
19845 * tree-ssa-strlen.c (compute_string_length): New function.
19846 (determine_min_obsize): New function.
19847 (handle_builtin_string_cmp): New function to handle calls to
19848 string compare functions.
19849 (strlen_optimize_stmt): Add handling to builtin string compare
19851 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19852 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19853 * tree.c (build_common_builtin_nodes): Add new defines of
19854 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
19856 2018-05-31 Jakub Jelinek <jakub@redhat.com>
19859 * bb-reorder.c (pass_partition_blocks::gate): Return false for
19860 functions with naked attribute.
19862 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
19864 * config/i386/sse.md (avx_vec_concat<mode>):
19865 Substitute concat_tg_mode mode attribute with xtg_mode.
19866 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
19867 (concat_tg_mode): Remove mode attribute.
19869 2018-05-31 Martin Sebor <msebor@redhat.com>
19872 * calls.c (alloc_max_size): Correct a logic error/typo.
19873 Treat excessive arguments as infinite. Warn for invalid arguments.
19874 * doc/invoke.texi (-Walloc-size-larger-than): Update.
19876 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
19879 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
19880 and movx for Haswell.
19882 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
19883 Cesar Philippidis <cesar@codesourcery.com>
19885 PR middle-end/85879
19886 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
19887 when emitting error on private/firstprivate reductions.
19888 * omp-low.c (lower_omp_target): Avoid reference-type processing
19889 on pointers for firstprivate clause.
19891 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
19893 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
19896 * config/aarch64/aarch64-simd.md
19897 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
19898 (aarch64_ld1_x3_<mode>): Likewise
19899 (aarch64_st1x2<VALLDIF:mode>): Likewise
19900 (aarch64_st1_x2_<mode>): Likewise
19901 (aarch64_st1x3<VALLDIF:mode>): Likewise
19902 (aarch64_st1_x3_<mode>): Likewise
19903 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
19904 (vld1_s8_x3): Likewise.
19905 (vld1_u16_x3): Likewise.
19906 (vld1_s16_x3): Likewise.
19907 (vld1_u32_x3): Likewise.
19908 (vld1_s32_x3): Likewise.
19909 (vld1_u64_x3): Likewise.
19910 (vld1_s64_x3): Likewise.
19911 (vld1_f16_x3): Likewise.
19912 (vld1_f32_x3): Likewise.
19913 (vld1_f64_x3): Likewise.
19914 (vld1_p8_x3): Likewise.
19915 (vld1_p16_x3): Likewise.
19916 (vld1_p64_x3): Likewise.
19917 (vld1q_u8_x3): Likewise.
19918 (vld1q_s8_x3): Likewise.
19919 (vld1q_u16_x3): Likewise.
19920 (vld1q_s16_x3): Likewise.
19921 (vld1q_u32_x3): Likewise.
19922 (vld1q_s32_x3): Likewise.
19923 (vld1q_u64_x3): Likewise.
19924 (vld1q_s64_x3): Likewise.
19925 (vld1q_f16_x3): Likewise.
19926 (vld1q_f32_x3): Likewise.
19927 (vld1q_f64_x3): Likewise.
19928 (vld1q_p8_x3): Likewise.
19929 (vld1q_p16_x3): Likewise.
19930 (vld1q_p64_x3): Likewise.
19931 (vst1_s64_x2): Likewise.
19932 (vst1_u64_x2): Likewise.
19933 (vst1_f64_x2): Likewise.
19934 (vst1_s8_x2): Likewise.
19935 (vst1_p8_x2): Likewise.
19936 (vst1_s16_x2): Likewise.
19937 (vst1_p16_x2): Likewise.
19938 (vst1_s32_x2): Likewise.
19939 (vst1_u8_x2): Likewise.
19940 (vst1_u16_x2): Likewise.
19941 (vst1_u32_x2): Likewise.
19942 (vst1_f16_x2): Likewise.
19943 (vst1_f32_x2): Likewise.
19944 (vst1_p64_x2): Likewise.
19945 (vst1q_s8_x2): Likewise.
19946 (vst1q_p8_x2): Likewise.
19947 (vst1q_s16_x2): Likewise.
19948 (vst1q_p16_x2): Likewise.
19949 (vst1q_s32_x2): Likewise.
19950 (vst1q_s64_x2): Likewise.
19951 (vst1q_u8_x2): Likewise.
19952 (vst1q_u16_x2): Likewise.
19953 (vst1q_u32_x2): Likewise.
19954 (vst1q_u64_x2): Likewise.
19955 (vst1q_f16_x2): Likewise.
19956 (vst1q_f32_x2): Likewise.
19957 (vst1q_f64_x2): Likewise.
19958 (vst1q_p64_x2): Likewise.
19959 (vst1_s64_x3): Likewise.
19960 (vst1_u64_x3): Likewise.
19961 (vst1_f64_x3): Likewise.
19962 (vst1_s8_x3): Likewise.
19963 (vst1_p8_x3): Likewise.
19964 (vst1_s16_x3): Likewise.
19965 (vst1_p16_x3): Likewise.
19966 (vst1_s32_x3): Likewise.
19967 (vst1_u8_x3): Likewise.
19968 (vst1_u16_x3): Likewise.
19969 (vst1_u32_x3): Likewise.
19970 (vst1_f16_x3): Likewise.
19971 (vst1_f32_x3): Likewise.
19972 (vst1_p64_x3): Likewise.
19973 (vst1q_s8_x3): Likewise.
19974 (vst1q_p8_x3): Likewise.
19975 (vst1q_s16_x3): Likewise.
19976 (vst1q_p16_x3): Likewise.
19977 (vst1q_s32_x3): Likewise.
19978 (vst1q_s64_x3): Likewise.
19979 (vst1q_u8_x3): Likewise.
19980 (vst1q_u16_x3): Likewise.
19981 (vst1q_u32_x3): Likewise.
19982 (vst1q_u64_x3): Likewise.
19983 (vst1q_f16_x3): Likewise.
19984 (vst1q_f32_x3): Likewise.
19985 (vst1q_f64_x3): Likewise.
19986 (vst1q_p64_x3): Likewise.
19988 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19990 * config/msp430/msp430.c (msp430_output_labelref): Prepend
19991 user_label_prefix to name.
19993 * tree-core.h: Update comment about the format of NAME string
19994 passed to handler in attribute_spec.
19996 * config/msp430/msp430.md: Remove erroneous subreg expression from
19997 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
19998 zero_extend{q,h}isi2.
20000 2018-05-30 Borislav Petkov <bp@suse.de>
20002 * doc/extend.texi: Document some architecture specific
20003 constraints and sort entries.
20005 2018-05-30 Martin Sebor <msebor@redhat.com>
20007 PR middle-end/85369
20008 * builtins.c (expand_builtin_stpcpy_1): New function.
20009 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
20010 only if the former succeeds.
20012 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
20014 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
20017 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20019 * doc/invoke.texi (-flinker-output): Document
20021 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20023 * passes.c (ipa_write_summaries): Only modify statements if body
20025 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
20026 incrementally linking.
20027 (ipa_passes): Likewise.
20028 * lto-cgraph.c (lto_output_node): When incrementally linking do not
20029 pass down resolution info.
20030 * common.opt (flag_incremental_link): Update info.
20031 * gcc.c (plugin specs): Turn flinker-output=* to
20032 -plugin-opt=-linker-output-known
20033 * toplev.c (compile_file): Also cut compilation when doing incremental
20035 * flag-types. (enum lto_partition_model): Add
20036 LTO_LINKER_OUTPUT_NOLTOREL.
20037 (invoke.texi): Add -flinker-output docs.
20038 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
20039 link same way as WPA; do not stream in dead initializers.
20041 * dwarf2out.c (dwarf2out_die_ref_for_decl,
20042 darf2out_register_external_decl): Support incremental link.
20044 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20046 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
20048 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
20050 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
20051 it down to simple_object_copy_lto_debug_sections.
20052 (run_gcc): Determine incremental LTO link time and configure
20053 lto1 into non-wpa mode, disable renaming of debug sections.
20055 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
20057 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
20058 descriptions of various incorrectly documented functions.
20060 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
20063 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20064 address check not strict.
20066 2018-05-30 Richard Biener <rguenther@suse.de>
20068 PR tree-optimization/85964
20069 * tracer.c (better_p): Drop initialized count check, we only
20070 call the function with initialized counts now.
20071 (find_best_successor): Do find a best edge if one
20072 has uninitialized count.
20073 (find_best_predecessor): Likewise. Do BB frequency check only
20074 if count is initialized.
20076 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
20078 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
20079 (aarch64_ldrstr_offset_compare): New.
20080 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
20081 load/store orderings.
20082 (aarch64_gen_adjusted_ldpstp): Likewise.
20084 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20086 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
20087 Check for subset of GENERAL_REGS and FP_REGS.
20088 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
20091 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
20093 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
20094 and wi::to_poly_offset. Add the current offset and then check
20095 whether the sum fits, rather than using an unchecked addition of
20096 a checked term. Check for a shwi rather than a uhwi.
20097 * expr.c (get_bit_range): Use tree_to_poly_uint64.
20098 (store_constructor): Use poly_int_tree_p.
20099 (expand_expr_real_1): Likewise.
20100 * function.c (assign_temp): Likewise.
20101 * fold-const.c (const_binop): Use poly_int_tree_p and
20102 wi::to_poly_offset.
20103 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
20105 * ipa-icf-gimple.c (func_checker::compare_operand): Use
20106 to_poly_offset for MEM offsets.
20107 * ipa-icf.c (sem_variable::equals): Likewise.
20108 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
20109 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
20110 wi::to_poly_offset for BIT_FIELD_REF offsets.
20111 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
20112 wi::to_poly_offset.
20113 * var-tracking.c (emit_note_insn_var_location): Use
20114 tree_to_poly_uint64.
20116 2018-05-29 Jim Wilson <jimw@sifive.com>
20118 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
20120 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
20123 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
20124 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
20126 (sse4_1_round<mode>2): Use nonimmediate_operand
20127 for operand 1 predicate.
20129 2018-05-29 Martin Sebor <msebor@redhat.com>
20130 Richard Biener <rguenther@suse.de>
20133 * calls.c (get_size_range): Call determine_value_range instead
20134 of get_value_range..
20135 * tree-vrp.h (determine_value_range): Declared new function.
20136 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
20138 2018-05-29 Richard Biener <rguenther@suse.de>
20140 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
20141 sure to use non-pattern stmts for get_earlier_stmt arguments.
20142 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
20143 called on pattern stmts.
20144 (get_later_stmt): Likewise.
20146 2018-05-29 Martin Liska <mliska@suse.cz>
20148 PR gcov-profile/85759
20149 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
20152 2018-05-29 Jakub Jelinek <jakub@redhat.com>
20154 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
20156 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
20160 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
20161 VEC_PACK_FLOAT_EXPR): New tree codes.
20162 * tree-pretty-print.c (op_code_prio): Handle
20163 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
20164 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
20165 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
20166 * tree-inline.c (estimate_operator_cost): Likewise.
20167 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
20168 * fold-const.c (const_binop): Likewise.
20169 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
20170 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
20171 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
20172 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
20173 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
20174 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
20175 * expr.c (expand_expr_real_2): Likewise.
20176 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
20177 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
20178 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
20180 * optabs.c (expand_widen_pattern_expr): For
20181 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
20182 sign from result type rather than operand's type.
20183 (expand_binop_directly): For vec_packu_float_optab and
20184 vec_packs_float_optab allow result type to be different from operand's
20186 * optabs-tree.c (optab_for_tree_code): Handle
20187 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
20188 VEC_PACK_FLOAT_EXPR. Formatting fixes.
20189 * tree-vect-generic.c (expand_vector_operations_1): Handle
20190 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
20191 VEC_PACK_FLOAT_EXPR.
20192 * tree-vect-stmts.c (supportable_widening_operation): Handle
20194 (supportable_narrowing_operation): Handle FLOAT_EXPR.
20195 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
20196 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
20197 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
20198 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
20200 (vec_pack<floatprefix>_float_<mode>): New expander.
20201 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
20203 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
20204 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
20205 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
20206 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
20207 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
20209 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
20210 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
20211 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
20212 VEC_PACK_FLOAT_EXPR): Document.
20214 2018-05-29 Richard Biener <rguenther@suse.de>
20216 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
20218 (stmt_vec_info_vec): Make pointer.
20219 (init_stmt_vec_info_vec): Remove.
20220 (free_stmt_vec_info_vec): Likewise.
20221 (set_stmt_vec_info_vec): New function.
20222 (free_stmt_vec_infos): Likewise.
20223 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
20224 (set_vinfo_for_stmt): Likewise.
20225 (get_earlier_stmt): Likewise.
20226 (get_later_stmt): Likewise.
20227 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
20228 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
20229 (vec_info::~vec_info): Free stmt_vec_infos.
20230 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
20231 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
20232 (pass_slp_vectorize::execute): Likewise.
20233 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
20234 (free_stmt_vec_info_vec): Likewise.
20235 (set_stmt_vec_info_vec): New function.
20236 (free_stmt_vec_infos): Likewise.
20237 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
20238 the global stmt_vec_info_vec.
20239 * tree-parloops.c (gather_scalar_reductions): Use
20240 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
20243 2018-05-29 Richard Biener <rguenther@suse.de>
20245 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
20247 2018-05-29 Martin Liska <mliska@suse.cz>
20248 David Malcolm <dmalcolm@redhat.com>
20250 * vec.c (test_reverse): New.
20251 (vec_c_tests): Add new test.
20252 * vec.h (vl_ptr>::reverse): New function.
20254 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
20256 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
20258 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
20261 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20263 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
20265 2018-05-28 Richard Biener <rguenther@suse.de>
20267 PR tree-optimization/85933
20268 * tree-vect-data-refs.c (vect_record_base_alignments): Only
20269 look at stmts marked as vectorizable.
20271 2018-05-28 Richard Biener <rguenther@suse.de>
20273 PR tree-optimization/85934
20274 * tree-vect-generic.c (expand_vector_operations_1): Hoist
20275 vector boolean check before scalar optimization.
20277 2018-05-28 Jakub Jelinek <jakub@redhat.com>
20279 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
20282 2018-05-28 Mark Wielaard <mark@klomp.org>
20284 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
20285 if it is an expression containing a minus sign.
20287 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
20289 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
20291 2018-05-27 Paul Koning <ni1d@arrl.net>
20293 * config/pdp11/pdp11.md (truncsihi2): Remove.
20295 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
20296 Chung-Ju Wu <jasonwucj@gmail.com>
20298 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
20300 (unaligned_store_dw): Ditto.
20301 * config/nds32/nds32-memory-manipulation.c
20302 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
20303 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
20304 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
20305 (emit_setmem_word_loop): Rename to ...
20306 (emit_setmem_doubleword_loop): ... this.
20307 (nds32_gen_dup_4_byte_to_word_value): New function.
20308 (nds32_gen_dup_8_byte_to_double_word_value): New function.
20309 (nds32_expand_setmem_loop): Refine implementation.
20310 (nds32_expand_setmem_loop_v3m): Ditto.
20311 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
20314 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
20316 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
20318 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
20320 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
20321 (nds32_init_machine_status): Initialize machine->attr_naked_p and
20322 machine->attr_no_prologue_p.
20323 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
20324 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
20325 (nds32_expand_epilogue): Consider attr_naked_p.
20326 (nds32_expand_epilogue_v3pop): Likewise.
20327 (nds32_can_use_return_insn): Likewise.
20328 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
20329 attr_no_prologue_p fields.
20330 * config/nds32/nds32.opt (mret-in-naked-func): New option.
20332 2018-05-27 Jakub Jelinek <jakub@redhat.com>
20335 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
20337 * config/i386/sse.md
20338 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
20340 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
20342 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
20344 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
20346 (*<floatsuffix>floatv2div2sf2): Rename to ...
20347 (*float<floatunssuffix>v2div2sf2): ... this.
20348 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
20349 (float<floatunssuffix>v2div2sf2_mask): ... this.
20350 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
20351 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
20352 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
20354 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
20356 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
20358 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
20360 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
20362 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
20364 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
20365 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
20366 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
20367 gen_ufix_truncv8dfv8si2.
20368 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
20369 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
20370 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
20371 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
20372 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
20373 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
20374 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
20375 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
20377 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20381 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
20383 2018-05-25 Jim Wilson <jimw@sifive.com>
20385 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
20386 * config/riscv/riscv.c (struct machine_function): Add
20387 interrupt_handler_p and attribute_checked_p fields.
20388 (riscv_attribute_table): Add interrupt.
20389 (riscv_interrupt_type_p): New.
20390 (riscv_save_reg_p): Save extra regs for interrupt handler.
20391 (riscv_use_save_libcall): Return false for interrupt handler.
20392 (riscv_first_stack_step): Add forward declaration.
20393 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
20394 for interrupt handler with large frame. Use it for saved reg list.
20395 (riscv_expand_prologue): Move flag_stack_usage_info support to
20396 eliminate duplication.
20397 (riscv_expand_epilogue): Generate mret for interrupt handler.
20398 (riscv_epilogue_uses): New.
20399 (riscv_can_use_return_insn): Return false for interrupt handler.
20400 (riscv_function_ok_for_sibcall): Likewise.
20401 (riscv_set_current_function): Add interrupt handler support.
20402 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
20403 * config/riscv/riscv.md (UNSPECV_MRET): New.
20405 (riscv_frflags, riscv_fsflags): Use tab after opcode.
20407 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
20409 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
20411 PR tree-optimization/85712
20412 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
20413 this candidate has already been replaced in-situ by a copy.
20415 2018-05-25 Jason Merrill <jason@redhat.com>
20417 PR c++/80485 - inline function non-zero address.
20418 * symtab.c (nonzero_address): Check DECL_COMDAT.
20420 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
20423 * config/alpha/alpha.md (ashlsi3): New insn pattern.
20424 (*ashlsi_se): Rename from *ashldi_se. Define as sign
20425 extension of SImode operation. Use const123_operand predicate.
20426 (*saddsi_1): Remove.
20427 (*saddl_se_1): Ditto.
20428 (*ssubsi_1): Ditto.
20429 (*ssubl_se_1): Ditto.
20430 * config/alpha/predicates.md (const123_operand): New predicate.
20431 * config/alpha/constraints.md (P): Use IN_RANGE.
20433 2018-05-25 Richard Biener <rguenther@suse.de>
20435 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
20437 (ref_maybe_used_by_stmt_p): Likewise.
20438 (stmt_may_clobber_ref_p): Likewise.
20439 (stmt_may_clobber_ref_p_1): Likewise.
20440 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
20442 (ref_maybe_used_by_stmt_p): Likewise.
20443 (stmt_may_clobber_ref_p): Likewise.
20444 (stmt_may_clobber_ref_p_1): Likewise.
20445 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
20446 the alias oracle to disambiguate DRs with stmts DR analysis
20448 (vect_analyze_data_refs): Do not give up on not analyzable
20449 DRs for BB vectorization. Remove code truncating the dataref
20452 2018-05-25 Jakub Jelinek <jakub@redhat.com>
20455 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
20456 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
20457 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
20459 2018-05-25 Richard Biener <rguenther@suse.de>
20461 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
20462 function, combining stmt data ref gathering and fatal analysis
20464 (vect_analyze_data_refs): Remove now redudnant code and simplify.
20465 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
20466 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
20467 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
20468 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
20470 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20472 PR tree-optimization/85720
20473 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
20474 SCC if all partitions are builtins.
20475 (version_loop_by_alias_check): New parameter. Generate cancelable
20476 runtime alias check if all partitions are builtins.
20477 (distribute_loop): Update call to above function.
20479 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20481 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
20482 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
20483 (parm_default_def_partition_arg): Ditto.
20484 (set_parm_default_def_partition): Ditto.
20485 (get_parm_default_def_partitions): Ditto and make it static.
20486 (get_undefined_value_partitions): Ditto and make it static.
20487 (remove_ssa_form): Refactor call to init_var_map here.
20488 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
20489 computation for loop region.
20490 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
20491 (register_default_def): Delete.
20492 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
20493 (parm_default_def_partition_arg): Ditto.
20494 (set_parm_default_def_partition): Ditto.
20495 (get_parm_default_def_partitions): Ditto and make it static.
20496 (get_undefined_value_partitions): Ditto and make it static.
20497 (coalesce_with_default, coalesce_with_default): Update comment.
20498 (create_coalesce_list_for_region): New func factored out from
20499 create_outofssa_var_map.
20500 (populate_coalesce_list_for_outofssa): New func factored out from
20501 create_outofssa_var_map and coalesce_ssa_name.
20502 (create_outofssa_var_map): Delete.
20503 (coalesce_ssa_name): Refactor to support live range computation.
20504 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
20505 (get_parm_default_def_partitions): Delete.
20506 (get_undefined_value_partitions): Ditto.
20507 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
20508 computation for loop region.
20509 (new_tree_live_info, loe_visit_block): Ditto.
20510 (live_worklist, set_var_live_on_entry): Ditto.
20511 (calculate_live_on_exit, verify_live_on_entry): Ditto.
20512 * tree-ssa-live.h (struct _var_map): New fields.
20513 (init_var_map): Change decl.
20514 (region_contains_p): New.
20516 2018-05-25 Bin Cheng <bin.cheng@arm.com>
20518 * tree-ssa-live.h (live_merge_and_clear): Delete.
20520 2018-05-25 Richard Biener <rguenther@suse.de>
20523 * tree-dump.c (dequeue_and_dump): Remove access to removed
20524 operand 2 of a SWITCH_EXPR.
20526 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20528 * doc/sourcebuild.texi (vect_double_cond_arith): Include
20529 multiplication and division.
20530 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
20531 (cond_udiv@var{m}, cond_umod@var{m}): Document.
20532 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
20533 (cond_udiv_optab, cond_umod_optab): New optabs.
20534 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
20535 (IFN_COND_RDIV): New internal functions.
20536 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
20537 TRUNC_MOD_EXPR and RDIV_EXPR.
20538 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
20539 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
20541 (SVE_INT_BINARY): Include mult.
20542 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
20543 (optab, sve_int_op): Handle mult.
20544 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
20546 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
20547 for SVE_INT_BINARY_SD.
20549 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20551 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
20552 (optab, sve_int_op): Handle div and udiv.
20553 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
20554 for SVE_INT_BINARY_SD.
20555 (*<optab><mode>3): New insn for the same.
20557 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20559 * tree-vect-patterns.c: Include predict.h.
20560 (vect_recog_divmod_pattern): Restrict check for division support
20561 to when optimizing for size.
20563 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20565 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
20566 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
20567 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
20568 (gimple_match_op::set_op): Likewise.
20569 (gimple_resimplify4): Declare.
20570 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
20571 (expr::gen_transform): Likewise.
20572 (decision_tree::gen): Generate a simplification routine for 4 operands.
20573 * gimple-match-head.c (gimple_simplify): Add an overload for
20574 4 operands. In the top-level function, handle up to 4 call
20575 arguments and call gimple_resimplify4.
20576 (gimple_resimplify4): New function.
20577 (build_call_internal): Pass a fourth operand.
20578 (maybe_push_to_seq): Likewise.
20579 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
20580 Fold VEC_COND_EXPRs of an operation and a default value into
20581 an IFN_COND_* function if possible.
20582 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
20584 (SVE_COND_FP_BINARY): Include them.
20585 (optab, sve_fp_op): Handle them.
20586 (SVE_INT_BINARY_REV): New code iterator.
20587 (SVE_COND_FP_BINARY_REV): New int iterator.
20588 (commutative): New int attribute.
20589 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
20591 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
20593 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
20594 (*cond_<optab><mode>): New patterns for reversed operands.
20596 2018-05-25 Richard Biener <rguenther@suse.de>
20598 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
20599 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
20600 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
20601 (STMT_VINFO_GROUPED_ACCESS): Adjust.
20602 * tree-vect-data-refs.c (everywhere): Adjust users.
20603 * tree-vect-loop.c (everywhere): Likewise.
20604 * tree-vect-slp.c (everywhere): Likewise.
20605 * tree-vect-stmts.c (everywhere): Likewise.
20606 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
20608 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20610 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
20612 (gcc_cv_as_section_exclude): ... this.
20613 Try Solaris as #exclude syntax.
20614 * configure: Regenerate.
20615 * config.in: Regenerate.
20616 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
20618 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
20619 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
20621 * varasm.c (default_elf_asm_named_section): Don't check if
20622 HAVE_GAS_SECTION_EXCLUDE is defined.
20624 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20626 * doc/md.texi: Update the documentation of the cond_* optabs
20627 to mention the new final operand. Fix GET_MODE_NUNITS call.
20628 Describe the scalar case too.
20629 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
20630 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
20632 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
20633 (get_conditional_internal_fn): Update comment.
20634 * tree-vect-loop.c (vectorizable_reduction): Pass the original
20635 accumulator value as a final argument to conditional functions.
20636 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
20637 a define_expand and add an "else" operand. Assert for now that
20638 the else operand is equal to operand 2. Use SVE_INT_BINARY and
20639 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
20640 (*cond_<optab><mode>): New patterns.
20641 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
20642 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
20643 (UNSPEC_COND_EOR): Delete.
20644 (optab): Remove associated mappings.
20645 (SVE_INT_BINARY): New code iterator.
20646 (sve_int_op): Remove int attribute and add "minus" to the code
20648 (SVE_COND_INT_OP): Delete.
20649 (SVE_COND_FP_OP): Rename to...
20650 (SVE_COND_FP_BINARY): ...this.
20652 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
20654 * optabs.c (can_reuse_operands_p): New function.
20655 (maybe_legitimize_operands): Try to reuse the results for
20658 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20660 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
20661 Add {q} suffix to insn mnemonic.
20663 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
20665 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
20666 (msp430_warn_func_return): New.
20668 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
20670 * fold-const.c (tree_nonzero_bits): New function.
20671 * fold-const.h (tree_nonzero_bits): Likewise.
20672 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
20673 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
20675 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
20679 * varasm.c (assemble_alias): Check ifunc_resolver only on
20682 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
20685 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
20686 when memory input operand is handled.
20688 2018-05-24 Luis Machado <luis.machado@linaro.org>
20690 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
20692 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
20694 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20696 * match.pd: Delay FMA folds until after vectorization.
20698 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
20701 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
20702 address check not strict.
20704 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
20706 * gimple-match.h (gimple_match_op): New class.
20707 (mprts_hook): Replace parameters with a gimple_match_op *.
20708 (maybe_build_generic_op): Likewise.
20709 (gimple_simplified_result_is_gimple_val): Replace parameters with
20710 a const gimple_match_op *.
20711 (gimple_simplify): Replace code_helper * and tree * parameters with
20712 a gimple_match_op * parameter.
20713 (gimple_resimplify1): Replace code_helper *, tree and tree *
20714 parameters with a gimple_match_op * parameter.
20715 (gimple_resimplify2): Likewise.
20716 (gimple_resimplify3): Likewise.
20717 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
20718 parameters with a gimple_match_op * parameter.
20719 * gimple-match-head.c (gimple_simplify): Change prototypes of
20720 auto-generated functions to take a gimple_match_op * instead of
20721 separate code_helper * and tree * parameters. Make the same
20722 change in the top-level overload and update calls to the
20723 gimple_resimplify routines. Update calls to the auto-generated
20724 functions and to maybe_push_res_to_seq in the publicly-facing
20725 operation-specific gimple_simplify overloads.
20726 (gimple_match_op::MAX_NUM_OPS): Define.
20727 (gimple_resimplify1): Replace rcode and ops with a single res_op
20728 parameter. Update call to gimple_simplify.
20729 (gimple_resimplify2): Likewise.
20730 (gimple_resimplify3): Likewise.
20731 (mprts_hook): Replace parameters with a gimple_match_op *.
20732 (maybe_build_generic_op): Likewise.
20733 (build_call_internal): Replace type, nargs and ops with
20734 a gimple_match_op *.
20735 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
20736 with a single gimple_match_op *. Update calls to mprts_hook,
20737 build_call_internal and gimple_simplified_result_is_gimple_val.
20738 Factor out code that is common to the tree_code and combined_fn cases.
20739 * genmatch.c (expr::gen_transform): Replace tem_code and
20740 tem_ops with a gimple_match_op called tem_op. Update calls
20741 to the gimple_resimplify functions and maybe_push_res_to_seq.
20742 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
20743 res_ops. Update call to the gimple_resimplify functions.
20744 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
20745 (decision_tree::gen): Make the functions take a gimple_match_op *
20746 called res_op instead of separate res_code and res_ops parameters.
20747 Update call accordingly.
20748 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
20749 and ops with a single res_op parameter. Update calls to
20750 maybe_build_generic_op and maybe_push_res_to_seq.
20751 (fold_stmt_1): Update calls to gimple_simplify and
20752 replace_stmt_with_simplification.
20753 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
20754 and gimple_simplified_result_is_gimple_val.
20755 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
20757 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
20758 with a gimple_match_op *.
20759 (vn_nary_build_or_lookup): Likewise. Update call to
20760 vn_nary_build_or_lookup_1.
20761 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
20762 gimple_match_op *. Update calls to the gimple_resimplify routines
20763 and to gimple_simplified_result_is_gimple_val.
20764 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
20765 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
20766 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
20767 (visit_nary_op): Likewise.
20768 (visit_reference_op_load): Likewise.
20770 2018-05-23 Luis Machado <luis.machado@linaro.org>
20772 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
20773 modifier for printing the step amount.
20775 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20778 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
20781 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
20783 * doc/sourcebuild.texi (Endianness): New subsubsection.
20785 2018-05-23 Luis Machado <luis.machado@linaro.org>
20787 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20788 <prefetch_dynamic_strides>: New const bool field.
20789 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20790 prefetch_dynamic_strides.
20791 (exynosm1_prefetch_tune): Likewise.
20792 (thunderxt88_prefetch_tune): Likewise.
20793 (thunderx_prefetch_tune): Likewise.
20794 (thunderx2t99_prefetch_tune): Likewise.
20795 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
20797 (aarch64_override_options_internal): Update to set
20798 PARAM_PREFETCH_DYNAMIC_STRIDES.
20799 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
20800 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
20801 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
20802 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
20803 prefetch-dynamic-strides setting.
20805 2018-05-23 Luis Machado <luis.machado@linaro.org>
20807 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
20808 <minimum_stride>: New const int field.
20809 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
20810 minimum_stride field defaulting to -1.
20811 (exynosm1_prefetch_tune): Likewise.
20812 (thunderxt88_prefetch_tune): Likewise.
20813 (thunderx_prefetch_tune): Likewise.
20814 (thunderx2t99_prefetch_tune): Likewise.
20815 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
20816 <default_opt_level>: Set to 3.
20817 (aarch64_override_options_internal): Update to set
20818 PARAM_PREFETCH_MINIMUM_STRIDE.
20819 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20820 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20821 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20822 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20823 stride is constant and is below the minimum stride threshold.
20825 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20827 * config/arm/arm-cpus.in (mode26): Delete.
20828 (armv4): Delete mode26 reference.
20829 * config/arm/arm.c (arm_configure_build_target): Delete use of
20832 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20834 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
20836 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
20837 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
20839 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
20840 (floatunsdidf2): Ditto.
20842 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20844 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
20845 (fixuns_trunc<mode>si2_avx512f): Ditto.
20846 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
20847 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
20848 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
20850 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20852 PR rtl-optimization/79985
20853 * df-scan.c (df_insn_refs_collect): Remove special case for
20854 global registers and asm statements.
20856 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20858 * extend.texi (Global Register Variables): Rewrite the bullet list.
20859 Note that the register is available for allocation. Note that access
20860 via inline asm must use constraints. Add note about async-signal
20861 handlers. Remove paragraph about automagic register selection.
20863 2018-05-23 Richard Biener <rguenther@suse.de>
20865 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
20866 of fixed offset from memset VN.
20868 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
20870 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
20871 first_interp field.
20872 (alloc_cand_and_find_basis): Initialize first_interp field.
20873 (slsr_process_mul): Modify first_interp field.
20874 (slsr_process_add): Likewise.
20875 (slsr_process_cast): Modify first_interp field for each new
20877 (slsr_process_copy): Likewise.
20878 (dump_candidate): Dump first_interp field.
20879 (replace_mult_candidate): Process all interpretations, not just
20881 (replace_rhs_if_not_dup): Likewise.
20882 (replace_one_candidate): Likewise.
20884 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
20886 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
20888 (aarch64_needs_frame_chain): New function.
20889 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
20891 2018-05-23 Sudakshina Das <sudi.das@arm.com>
20894 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
20895 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
20896 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
20897 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
20898 as true for strict-align.
20899 (aarch64_can_inline_p): Perform checks even when callee has no
20900 attributes to check for strict alignment.
20901 * doc/extend.texi (AArch64 Function Attributes): Document
20903 * doc/invoke.texi: (AArch64 Options): Likewise.
20905 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
20907 PR tree-optimization/85853
20908 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
20909 the handling of the root of the node to...
20910 (vect_slp_analyze_node_operations_1): ...this new function,
20911 and run the whole thing with the child nodes' def types
20912 set according to their SLP node's def type.
20914 2018-05-23 Richard Biener <rguenther@suse.de>
20916 PR middle-end/85874
20917 * tree-data-ref.c (create_runtime_alias_checks): Defer
20918 and ignore overflow warnings.
20920 2018-05-23 Yury Gribov <tetra2005@gmail.com>
20922 PR tree-optimization/85822
20923 * tree-vrp.c (is_masked_range_test): Fix handling of negative
20926 2018-05-23 Richard Biener <rguenther@suse.de>
20928 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
20929 memset constants via native_interpret_expr.
20931 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
20934 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
20936 (cgraph_node::create_alias): Likewise.
20937 (cgraph_node::get_availability): Check ifunc_resolver instead
20938 of looking up ifunc attribute.
20939 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
20940 * varasm.c (do_assemble_alias): Likewise.
20941 (assemble_alias): Likewise.
20942 (default_binds_local_p_3): Likewise.
20943 * cgraph.h (cgraph_node): Add ifunc_resolver.
20944 (cgraph_node::only_called_directly_or_aliased_p): Return false
20945 for IFUNC resolver.
20946 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
20948 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
20949 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
20950 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
20951 instead of looking up ifunc attribute.
20953 2018-05-22 Luis Machado <luis.machado@linaro.org>
20955 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
20957 2018-05-22 Martin Sebor <msebor@redhat.com>
20959 PR middle-end/85359
20960 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
20961 only when expasion succeeds.
20962 (expand_builtin_strcmp): Same.
20963 (expand_builtin_strncmp): Same.
20965 2018-05-22 Martin Sebor <msebor@redhat.com>
20967 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
20969 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
20970 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20972 * config/aarch64/aarch64-ldpstp.md: Replace uses of
20973 aarch64_mem_pair_operand with memory_operand and delete operand swapping
20975 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
20976 Add check for legitimate_address.
20977 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
20978 (aarch64_swap_ldrstr_operands): New.
20979 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
20982 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
20983 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20985 * config/aarch64/aarch64.md: New patterns to generate stp
20987 (store_pair_sw, store_pair_dw): New patterns to generate stp for
20988 single words and double words.
20989 (load_pair_sw, load_pair_dw): Likewise.
20990 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
20992 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
20994 * config/aarch64/aarch64-ldpstp.md: Modify peephole
20995 for different mode ldpstp and add peephole for merged zero stores.
20996 Likewise for loads.
20997 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
20999 (aarch64_gen_store_pair): Rename calls to match new patterns.
21000 (aarch64_gen_load_pair): Rename calls to match new patterns.
21001 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
21002 (load_pair<DREG:mode><DREG2:mode>): ... This.
21003 (store_pair<mode>): Rename to...
21004 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
21005 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
21006 New mode iterators.
21007 (V_INT_EQUIV): Handle SImode.
21008 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
21011 2018-05-22 Martin Sebor <msebor@redhat.com>
21014 * calls.c (maybe_warn_nonstring_arg): Use string length to set
21015 or ajust the presumed bound on an operation to avoid unnecessary
21018 2018-05-22 Martin Sebor <msebor@redhat.com>
21020 PR tree-optimization/85826
21021 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
21022 assuming that a DECL necesarily has a constant size.
21024 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
21026 PR middle-end/85862
21027 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
21029 2018-05-22 Richard Biener <rguenther@suse.de>
21031 PR tree-optimization/85834
21032 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
21033 non-constant and non-zero memset arguments.
21035 2018-05-22 Martin Liska <mliska@suse.cz>
21038 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
21040 2018-05-22 Richard Biener <rguenther@suse.de>
21042 PR tree-optimization/85863
21043 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
21044 comparisons when vectype is specified.
21045 (vectorizable_condition): Do not specify vectype for
21046 vect_is_simple_cond when SLP vectorizing.
21048 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
21051 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
21052 define __ibm128 as long double.
21053 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
21054 as a distinct type when IEEE 128-bit support is enabled.
21055 (init_float128_ieee): Fix up conversions between IFmode and IEEE
21056 128-bit types to use the correct functions.
21057 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
21058 convert between 128-bit floating point types that have different
21059 modes but the same representation, instead of using gen_lowpart to
21061 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
21063 (IFKF_reg): New attributes to give the register constraints for
21065 (extend<mode>tf2_internal): New insns to mark an explicit
21066 conversion between 128-bit floating point types that have a
21067 different mode but share the same representation.
21069 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
21071 PR tree-optimization/85814
21072 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
21073 a null return from get_strinfo when unsharing the next
21074 strinfo in the chain.
21076 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21079 * varasm.c (weak_finish): Clean up weak_decls.
21081 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21083 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
21084 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
21085 UNSPEC_UADALP values.
21086 * config/aarch64/iterators.md (ABAL): New int iterator.
21089 (sur): Add mappings for the above.
21090 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
21092 (aarch64_<sur>abal<mode>_4): Likewise.
21093 (aarch64_<sur>adalp<mode>_3): Likewise.
21094 (<sur>sadv16qi): New define_expand.
21096 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
21098 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
21099 (*movdf_internal): Ditto.
21100 (*rcpsf2_sse): Ditto.
21101 (*rsqrtsf2_sse): Ditto.
21102 (*sqrt<mode>2_sse): Ditto.
21104 2018-05-21 Tamar Christina <tamar.christina@arm.com>
21106 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
21108 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
21109 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
21110 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
21111 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
21113 * config/aarch64/arm_neon.h: Likewise.
21114 * config/aarch64/iterators.md (VQ_I): New.
21116 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
21118 * config.gcc: Add arc/t-multilib-linux to tmake_file for
21120 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
21123 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
21125 * config/nds32/constraints.md (S): New constraint.
21126 * config/nds32/nds32.md (call_internal): Use constraint S.
21127 (call_value_internal): Likewise.
21128 (sibcall_internal): Likewise.
21129 (sibcall_value_internal): Likewise.
21131 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
21132 Chung-Ju Wu <jasonwucj@gmail.com>
21134 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
21135 into consideration.
21137 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
21138 Chung-Ju Wu <jasonwucj@gmail.com>
21140 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
21141 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
21142 (nds32_rtx_costs_impl): Simplify.
21143 (nds32_address_cost_impl): Simplify.
21144 (nds32_init_rtx_costs): New function.
21145 (nds32_rtx_costs_speed_prefer): Likewise.
21146 (nds32_rtx_costs_size_prefer): Likewise.
21147 (nds32_address_cost_speed_prefer): Likewise.
21148 (nds32_address_cost_speed_fwprop): Likewise.
21149 (nds32_address_cost_size_prefer): Likewise.
21150 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
21151 * config/nds32/nds32.c (nds32_option_override): Use
21152 nds32_init_rtx_costs function.
21154 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
21156 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
21157 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
21158 (TARGET_PIPELINE_N8): Likewise.
21159 (TARGET_PIPELINE_N10): Likewise.
21160 (TARGET_PIPELINE_N13): Likewise.
21161 (TARGET_PIPELINE_GRAYWOLF): Likewise.
21163 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
21165 * config/nds32/nds32-fpu.md: Update copyright year.
21167 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21169 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
21171 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21173 * config/nds32/nds32.c
21174 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
21175 * config/nds32/nds32.opt (minline-asm-r15): New option.
21177 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21179 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
21181 * config/nds32/nds32.md (abssi2): New pattern.
21183 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
21185 * config/i386/i386.md (rex64namesuffix): New mode attribute.
21186 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
21187 Merge insn pattern from sse_cvtsi2ss<round_name> and
21188 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
21189 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
21190 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
21191 using SWI48 mode iterator.
21192 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
21193 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
21194 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
21195 pattern from sse_cvttss2si<round_saeonly_name>
21196 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
21197 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
21198 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
21199 using SWI48 mode iterator.
21200 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
21201 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
21202 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
21203 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
21204 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
21205 using SWI48 mode iterator.
21206 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
21207 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
21208 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
21209 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
21210 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
21211 SWI48 mode iterator.
21212 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
21213 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
21214 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
21215 pattern from sse_cvttsd2si<round_saeonly_name>
21216 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
21218 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21220 * config/nds32/nds32-md-auxiliary.c
21221 (nds32_valid_smw_lwm_base_p): Refine.
21222 (nds32_output_smw_single_word): Refine.
21223 (nds32_output_smw_double_word): New.
21224 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
21226 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21228 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
21229 (nds32_output_stack_pop): Refine.
21230 (nds32_expand_unaligned_load): Refine.
21231 (nds32_expand_unaligned_store): Refine.
21233 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21234 Chung-Ju Wu <jasonwucj@gmail.com>
21236 * config/nds32/constants.md: Add TP_REGNUM constant.
21237 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
21238 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
21240 * config/nds32/nds32-doubleword.md: Consider flag_pic.
21241 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
21242 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
21243 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
21244 and PIC code generation.
21245 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
21247 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
21249 * config/nds32/nds32.md: Support TLS and PIC.
21250 * config/nds32/nds32.c: Support TLS and PIC.
21251 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
21252 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
21255 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
21257 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
21258 mode with E_ prefix.
21260 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21261 Chung-Ju Wu <jasonwucj@gmail.com>
21263 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
21264 * config/nds32/nds32-md-auxiliary.c
21265 (symbolic_reference_mentioned_p): New.
21266 (nds32_legitimize_ict_address): New.
21267 (nds32_expand_ict_move): New.
21268 (nds32_indirect_call_referenced_p): New.
21269 (nds32_symbol_binds_local_p): Delete.
21270 (nds32_long_call_p): Modify.
21271 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
21272 * config/nds32/nds32-protos.h
21273 (symbolic_reference_mentioned_p): Declare.
21274 (nds32_legitimize_ict_address): Declare.
21275 (nds32_expand_ict_move): Declare.
21276 (nds32_indirect_call_referenced_p): Declare.
21277 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
21278 (nds32_relax_group): Use nds32_ict_const_p as condition.
21279 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
21280 (nds32_asm_file_start): Output ict_model directive in asm code.
21281 (nds32_legitimate_address_p): Consider indirect call.
21282 (nds32_print_operand): Consider indirect call.
21283 (nds32_print_operand_address): Consider indirect call.
21284 (nds32_insert_attributes): Handle "indirect_call" attribute.
21285 (TARGET_LEGITIMATE_ADDRESS_P): Define.
21286 (TARGET_LEGITIMATE_CONSTANT_P): Define.
21287 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
21288 (TARGET_DELEGITIMIZE_ADDRESS): Define.
21289 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
21290 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
21291 (TARGET_ICT_MODEL_SMALL): Define.
21292 (TARGET_ICT_MODEL_LARGE): Define.
21293 * config/nds32/nds32.md (movsi): Consider ict model.
21294 (call, call_value): Consider ict model.
21295 (sibcall, sibcall_value): Consider ict model.
21296 * config/nds32/nds32.opt (mict-model): New option.
21297 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
21300 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
21301 Monk Chiang <sh.chiang04@gmail.com>
21302 Jim Wilson <jimw@sifive.com>
21304 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
21305 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
21306 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
21307 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
21308 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
21309 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
21310 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
21311 compute save_libcall_adjustment properly.
21312 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
21313 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
21314 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
21315 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
21316 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
21317 (ABI_SPEC): Handle mabi=ilp32e.
21318 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
21319 (RVE): Add RVE mask.
21320 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
21321 <-march>: Add rv32e as an example.
21323 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
21326 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
21327 (intra_create_variable_infos): Handle C++ constructors.
21329 2018-05-18 Martin Liska <mliska@suse.cz>
21331 * passes.def: Remove a redundant pass.
21333 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
21336 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
21338 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21340 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
21342 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
21344 (armv2, armv2a, armv3, armv3m): Delete architectures.
21345 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
21346 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
21347 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
21349 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
21350 (*mulsidi3adddi): Likewise.
21351 (mulsidi3): Likewise.
21352 (*mulsidi3_nov6): Likewise.
21353 (umulsidi3): Likewise.
21354 (umulsidi3_nov6): Likewise.
21355 (umaddsidi4): Likewise.
21356 (*umulsidi3adddi): Likewise.
21357 (smulsi3_highpart): Likewise.
21358 (*smulsi3_highpart_nov6): Likewise.
21359 (umulsi3_highpart): Likewise.
21360 (*umulsi3_highpart_nov6): Likewise.
21361 * config/arm/arm.h (arm_arch3m): Delete.
21362 * config/arm/arm.c (arm_arch3m): Delete.
21363 (arm_option_override_internal): Update armv3-related comment.
21364 (arm_configure_build_target): Delete use of isa_bit_mode32.
21365 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
21366 (arm_rtx_costs_internal): Delete check of arm_arch3m.
21367 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
21368 (mulsa3): Likewise.
21369 (mulusa3): Likewise.
21370 * config/arm/arm-protos.h (arm_arch3m): Delete.
21371 * config/arm/arm-tables.opt: Regenerate.
21372 * config/arm/arm-tune.md: Likewise.
21373 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
21374 deleted architectures.
21376 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21378 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
21379 (armv5t, armv5te): New features.
21380 (ARMv5, ARMv5e): Delete fgroups.
21381 (ARMv5t, ARMv5te): Adjust for above changes.
21382 (ARMv6m): Likewise.
21383 (armv5, armv5e): Delete arches.
21384 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
21386 (*call_reg_arm): Likewise.
21387 (*call_value_reg_armv5): Likewise.
21388 (*call_value_reg_arm): Likewise.
21389 (*call_symbol): Likewise.
21390 (*call_value_symbol): Likewise.
21391 (*sibcall_insn): Likewise.
21392 (*sibcall_value_insn): Likewise.
21393 (clzsi2): Likewise.
21394 (prefetch): Likewise.
21395 (define_split and define_peephole2 dependent on arm_arch5):
21397 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
21399 (TARGET_ARM_QBIT): Likewise.
21400 (TARGET_DSP_MULTIPLY): Likewise.
21401 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
21402 (arm_arch5, arm_arch5e): Delete.
21403 (arm_arch5t, arm_arch5te): Declare.
21404 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
21405 (arm_arch5t): Declare.
21406 (arm_option_reconfigure_globals): Update for the above.
21407 (arm_options_perform_arch_sanity_checks): Update comment, replace
21408 use of arm_arch5 with arm_arch5t.
21409 (use_return_insn): Likewise.
21410 (arm_emit_call_insn): Likewise.
21411 (output_return_instruction): Likewise.
21412 (arm_final_prescan_insn): Likewise.
21413 (arm_coproc_builtin_available): Likewise.
21414 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
21415 arm_arch5e with arm_arch5t and arm_arch5te.
21416 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
21417 (arm_arch5t, arm_arch5te): Declare.
21418 * config/arm/arm-tables.opt: Regenerate.
21419 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
21420 * config/arm/t-multilib: Likewise.
21421 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
21422 instead of arm_arch5.
21423 (*call_reg_thumb1): Likewise.
21424 (*call_value_reg_thumb1_v5): Likewise.
21425 (*call_value_reg_thumb1): Likewise.
21426 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
21428 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
21430 2018-05-18 Martin Liska <mliska@suse.cz>
21432 PR gcov-profile/84846
21433 * doc/gcov.texi: Document -t option of gcov tool.
21435 2018-05-18 Martin Liska <mliska@suse.cz>
21437 PR gcov-profile/84846
21438 * gcov.c (print_usage): Add new -t option.
21439 (process_args): Handle the option.
21440 (generate_results): Use stdout as output when requested by
21443 2018-05-18 Martin Liska <mliska@suse.cz>
21445 PR gcov-profile/84846
21446 * coverage.c (coverage_init): Write PWD to .gcno file.
21447 * doc/gcov.texi: Document how working directory is printed.
21448 * gcov-dump.c (dump_gcov_file): Print PWD.
21449 * gcov.c (output_intermediate_file): Likewise.
21450 (read_graph_file): Read PWD string.
21451 (output_lines): Print PWD.
21453 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21455 PR middle-end/85817
21456 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
21457 for retval and return false if all args to phi are zero.
21459 2018-05-18 Richard Biener <rguenther@suse.de>
21461 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
21463 (evrp_dom_walker::before_dom_children): Call it.
21465 2018-05-18 Richard Biener <rguenther@suse.de>
21467 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
21468 results when processing array refs with variable index.
21470 2018-05-18 Toon Moene <toon@moene.org>
21472 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
21473 directly after that of -floop-interchange. Indicate that both
21474 options are enabled by default when specifying -O3.
21476 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21478 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
21479 iterator. Delete separate integer-mode vec_set<mode> expander.
21480 (aarch64_simd_vec_setv2di): Delete.
21481 (vec_setv2di): Delete.
21482 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
21483 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
21484 the "w, r" alternative.
21486 2018-05-18 Martin Liska <mliska@suse.cz>
21488 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
21489 * tree-pass.h (make_pass_lower_switch_O0): New function.
21490 * tree-switch-conversion.c (node_has_low_bound): Remove.
21491 (node_has_high_bound): Likewise.
21492 (node_is_bounded): Likewise.
21493 (class pass_lower_switch): Make it a template type and create
21495 (pass_lower_switch::execute): Add template argument.
21496 (make_pass_lower_switch): New function.
21497 (make_pass_lower_switch_O0): New function.
21498 (do_jump_if_equal): Remove.
21499 (emit_case_nodes): Simplify to just handle all 3 cases and leave
21500 all the hard work to tree optimization passes.
21502 2018-05-18 Martin Liska <mliska@suse.cz>
21504 * dbgcnt.c (limit_low): Renamed from limit.
21505 (limit_high): New variable.
21506 (dbg_cnt_is_enabled): Check for upper limit.
21507 (dbg_cnt): Adjust dumping.
21508 (dbg_cnt_set_limit_by_index): Add new argument for high
21510 (dbg_cnt_set_limit_by_name): Likewise.
21511 (dbg_cnt_process_single_pair): Parse new format.
21512 (dbg_cnt_process_opt): Use strtok.
21513 (dbg_cnt_list_all_counters): Remove 'value' and add
21515 * doc/invoke.texi: Document changes.
21517 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
21519 * doc/sourcebuild.texi (scalar_all_fma): Document.
21520 * tree.def (FMA_EXPR): Delete.
21521 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
21522 * internal-fn.c (ternary_direct): New macro.
21523 (expand_ternary_optab_fn): Likewise.
21524 (direct_ternary_optab_supported_p): Likewise.
21525 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
21526 * builtins.c (fold_builtin_fma): Delete.
21527 (fold_builtin_3): Don't call it.
21528 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
21529 * expr.c (expand_expr_real_2): Likewise.
21530 * fold-const.c (operand_equal_p): Likewise.
21531 (fold_ternary_loc): Likewise.
21532 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
21533 * gimple.c (DEFTREECODE): Likewise.
21534 * gimplify.c (gimplify_expr): Likewise.
21535 * optabs-tree.c (optab_for_tree_code): Likewise.
21536 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
21537 * tree-eh.c (operation_could_trap_p): Likewise.
21538 (stmt_could_throw_1_p): Likewise.
21539 * tree-inline.c (estimate_operator_cost): Likewise.
21540 * tree-pretty-print.c (dump_generic_node): Likewise.
21541 (op_code_prio): Likewise.
21542 * tree-ssa-loop-im.c (stmt_cost): Likewise.
21543 * tree-ssa-operands.c (get_expr_operands): Likewise.
21544 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
21545 * fold-const-call.h (fold_fma): Delete.
21546 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
21547 CFN_FNMA and CFN_FNMS.
21548 (fold_fma): Delete.
21549 * genmatch.c (combined_fn): New enum.
21550 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
21551 (commutative_op): New function.
21552 (commutate): Use it. Handle more than 2 operands.
21553 (dt_operand::gen_gimple_expr): Use commutative_op.
21554 (parser::parse_expr): Allow :c to be used with non-binary
21555 operators if the commutative operand is known.
21556 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
21557 CFN_FMS, CFN_FNMA and CFN_FNMS.
21558 (backprop::process_assign_use): Remove FMA_EXPR handling.
21559 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
21560 (gen_hsa_fma): New function.
21561 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
21562 IFN_FNMA and IFN_FNMS.
21563 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
21564 * gimple-fold.h (follow_all_ssa_edges): Declare.
21565 * gimple-fold.c (follow_all_ssa_edges): New function.
21566 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
21567 gimple_build interface and use follow_all_ssa_edges to fold the result.
21568 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
21569 instead of checking for optabs directly.
21570 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
21571 rather than FMA_EXPRs.
21572 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
21573 call to IFN_FMA instead of an FMA_EXPR.
21575 2018-05-17 Jim Wilson <jimw@sifive.com>
21577 * expr.c (do_tablejump): When converting index to Pmode, if we have a
21578 sign extended promoted subreg, and the range does not have the sign bit
21579 set, then do a sign extend.
21581 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
21582 test, check for sign extended subreg and/or constant operands, and
21583 do a sign extend in that case.
21585 2018-05-17 Steve Ellcey <sellcey@cavium.com>
21587 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
21588 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
21590 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
21591 Change logics_shift_reg to logics_shift_imm.
21592 (thunderx2t99_fp_loadpair_basic): Delete.
21593 (thunderx2t99_fp_storepair_basic): Delete.
21594 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
21595 (thunderx2t99_asimd_polynomial): Delete.
21596 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
21597 and neon_fp_mul_d_scalar_q.
21598 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
21599 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
21600 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
21601 (thunderx2t99_asimd_lut): Add missing tbl types.
21602 (thunderx2t99_asimd_ext): Delete.
21603 (thunderx2t99_asimd_load1_1_mult): Delete.
21604 (thunderx2t99_asimd_load1_2_mult): Delete.
21605 (thunderx2t99_asimd_load1_ldp): New.
21606 (thunderx2t99_asimd_load1): New.
21607 (thunderx2t99_asimd_load2): Add missing *load2* types.
21608 (thunderx2t99_asimd_load3): New.
21609 (thunderx2t99_asimd_load4): New.
21610 (thunderx2t99_asimd_store1_1_mult): Delete.
21611 (thunderx2t99_asimd_store1_2_mult): Delete.
21612 (thunderx2t99_asimd_store2_mult): Delete.
21613 (thunderx2t99_asimd_store2_onelane): Delete.
21614 (thunderx2t99_asimd_store_stp): New.
21615 (thunderx2t99_asimd_store1): New.
21616 (thunderx2t99_asimd_store2): New.
21617 (thunderx2t99_asimd_store3): New.
21618 (thunderx2t99_asimd_store4): New.
21620 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
21622 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
21623 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
21625 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
21626 Segher Boessenkool <segher@kernel.crashing.org>
21629 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
21632 2018-05-17 Richard Biener <rguenther@suse.de>
21634 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
21635 for pruning loop and prune defs feeding only already visited PHIs.
21637 2018-05-17 Richard Biener <rguenther@suse.de>
21639 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
21641 2018-05-17 Bin Cheng <bin.cheng@arm.com>
21642 Richard Biener <rguenther@suse.de>
21644 PR tree-optimization/85793
21645 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
21646 for VMAT_ELEMENTWISE.
21648 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21650 * internal-fn.h (lookup_internal_fn): Declare
21651 * internal-fn.c (lookup_internal_fn): New function.
21652 * gimple.c (gimple_build_call_from_tree): Handle calls to
21653 internal functions.
21654 * gimple-pretty-print.c (dump_gimple_call): Print "." before
21655 internal function names.
21656 * tree-pretty-print.c (dump_generic_node): Likewise.
21657 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
21659 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
21661 * gimple-fold.h (gimple_build): Make the function forms take
21662 combined_fn rather than built_in_function.
21663 (gimple_simplify): Likewise.
21664 * gimple-match-head.c (gimple_simplify): Likewise.
21665 * gimple-fold.c (gimple_build): Likewise.
21666 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
21667 rather than gimple_build_call_internal.
21668 (get_initial_defs_for_reduction): Likewise.
21669 (vect_create_epilog_for_reduction): Likewise.
21670 (vectorizable_live_operation): Likewise.
21672 2018-05-17 Martin Liska <mliska@suse.cz>
21674 * gimple-ssa-sprintf.c (format_directive): Do not use
21675 space in between 'G_' and '('.
21677 2018-05-17 Jakub Jelinek <jakub@redhat.com>
21680 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
21681 even if the mask is not all ones.
21684 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
21686 (ix86_gimple_fold_builtin): Likewise.
21689 * config/i386/i386.c: Include tree-vector-builder.h.
21690 (ix86_vector_shift_count): New function.
21691 (ix86_fold_builtin): Fold shift builtins by scalar count.
21692 (ix86_gimple_fold_builtin): Likewise.
21694 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
21695 _mm512_setzero): New intrinsics.
21697 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
21698 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21700 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
21701 code generation for cases where splatting a value is not useful.
21702 * simplify-rtx.c (simplify_ternary_operation): Simplify
21703 vec_merge across a vec_duplicate and a paradoxical subreg forming
21704 a vector mode to a vec_concat.
21706 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
21708 * config.gcc: Support "goldmont-plus".
21709 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21711 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21712 PROCESSOR_GOLDMONT_PLUS.
21713 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
21714 (processor_target_table): Add "goldmont-plus".
21715 (PTA_GOLDMONT_PLUS): Define.
21716 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
21717 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
21718 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
21719 (fold_builtin_cpu): Add "goldmont-plus".
21720 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
21721 (ix86_option_override_internal): Add "goldmont-plus".
21722 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
21723 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
21724 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
21725 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
21727 2018-05-17 Richard Biener <rguenther@suse.de>
21729 PR tree-optimization/85757
21730 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
21731 remove defs that only feed that PHI from further processing.
21733 2018-05-16 Jim Wilson <jimw@sifive.com>
21735 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
21737 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
21739 2018-05-16 Mark Wielaard <mark@klomp.org>
21741 * dwarf2out.c (count_index_strings): New function.
21742 (output_indirect_strings): Call count_index_strings and generate
21743 header for dwarf_version >= 5.
21745 2018-05-16 Mark Wielaard <mark@klomp.org>
21747 * dwarf2out.c (dwarf_FORM): New function.
21748 (set_indirect_string): Use dwarf_FORM.
21749 (reset_indirect_string): Likewise.
21750 (size_of_die): Likewise.
21751 (value_format): Likewise.
21752 (output_die): Likewise.
21753 (add_skeleton_AT_string): Likewise.
21754 (output_macinfo_op): Likewise.
21755 (index_string): Likewise.
21756 (output_index_string_offset): Likewise.
21757 (output_index_string): Likewise.
21758 (count_index_strings): Likewise.
21760 2018-05-16 Carl Love <cel@us.ibm.com>
21762 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
21763 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
21765 2018-05-16 Martin Jambor <mjambor@suse.cz>
21767 * ipa-prop.c (ipa_free_all_edge_args): Remove.
21768 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
21770 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
21772 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
21773 (fnma<mode>4): Likewise.
21774 (fms<mode>4): Likewise.
21775 (fnms<mode>4): Likewise.
21776 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
21777 (aarch64_fnma<mode>4): Likewise.
21778 (aarch64_fms<mode>4): Likewise.
21779 (aarch64_fnms<mode>4): Likewise.
21780 (aarch64_fnmadd<mode>4): Likewise.
21782 2018-05-16 Jason Merrill <jason@redhat.com>
21784 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
21786 2018-05-16 Richard Biener <rguenther@suse.de>
21788 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
21789 (dump_stmt_cost): Declare.
21790 (add_stmt_cost): Dump cost we add.
21791 (add_stmt_costs): New function.
21792 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
21793 No longer exported.
21794 (vect_analyze_stmt): Adjust prototype.
21795 (vectorizable_condition): Likewise.
21796 (vectorizable_live_operation): Likewise.
21797 (vectorizable_reduction): Likewise.
21798 (vectorizable_induction): Likewise.
21799 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
21800 cost vector to pass to vectorizable_ and record afterwards.
21801 (vect_model_reduction_cost): Take cost vector argument and adjust.
21802 (vect_model_induction_cost): Likewise.
21803 (vectorizable_reduction): Likewise.
21804 (vectorizable_induction): Likewise.
21805 (vectorizable_live_operation): Likewise.
21806 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
21807 SLP_TREE_NUMBER_OF_VEC_STMTS.
21808 (vect_analyze_slp_cost_1): Remove.
21809 (vect_analyze_slp_cost): Likewise.
21810 (vect_slp_analyze_node_operations): Take visited args and
21811 a target cost vector. Avoid processing already visited stmt sets.
21812 (vect_slp_analyze_operations): Use a local cost vector to gather
21813 costs and register those of non-discarded instances.
21814 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
21815 (vect_schedule_slp_instance): Remove copying of
21816 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
21818 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
21819 adding cost. Record cost entry location.
21820 (vect_prologue_cost_for_slp_op): Function to compute cost of
21821 a constant or invariant generated for SLP vect in the prologue,
21822 split out from vect_analyze_slp_cost_1.
21823 (vect_model_simple_cost): Make static. Adjust for SLP costing.
21824 (vect_model_promotion_demotion_cost): Likewise.
21825 (vect_model_store_cost): Likewise, make static.
21826 (vect_model_load_cost): Likewise.
21827 (vectorizable_bswap): Add cost vector arg and adjust.
21828 (vectorizable_call): Likewise.
21829 (vectorizable_simd_clone_call): Likewise.
21830 (vectorizable_conversion): Likewise.
21831 (vectorizable_assignment): Likewise.
21832 (vectorizable_shift): Likewise.
21833 (vectorizable_operation): Likewise.
21834 (vectorizable_store): Likewise.
21835 (vectorizable_load): Likewise.
21836 (vectorizable_condition): Likewise.
21837 (vectorizable_comparison): Likewise.
21838 (can_vectorize_live_stmts): Likewise.
21839 (vect_analyze_stmt): Likewise.
21840 (vect_transform_stmt): Adjust calls to vectorizable_*.
21841 * tree-vectorizer.c: Include gimple-pretty-print.h.
21842 (dump_stmt_cost): New function.
21844 2018-05-16 Richard Biener <rguenther@suse.de>
21846 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
21847 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
21848 * tree-ssa-dse.c: Include tree-ssa-loop.h.
21849 (check_name): New callback.
21850 (dse_classify_store): Track cycles via a visited bitmap of PHI
21851 defs and simplify handling of in-loop and across loop dead stores
21852 and properly fail for loop-variant refs. Handle byte-tracking with
21853 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
21856 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
21858 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
21859 (vect_get_mask_type_for_stmt): Likewise.
21860 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
21862 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
21863 to determine the statement's vector type and the vector type that
21864 should be used for calculating nunits. Deal with cases in which
21865 the type has to be deferred.
21866 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
21867 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
21868 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
21869 (vect_determine_vf_for_stmt): New functions, split out from...
21870 (vect_determine_vectorization_factor): ...here.
21871 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
21872 (vect_get_mask_type_for_stmt): New functions, split out from
21873 vect_determine_vectorization_factor.
21875 2018-05-16 Richard Biener <rguenther@suse.de>
21877 * tree-cfg.c (verify_gimple_assign_ternary): Properly
21878 verify the [VEC_]COND_EXPR embedded comparison.
21880 2018-05-15 Martin Sebor <msebor@redhat.com>
21882 PR tree-optimization/85753
21883 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
21884 RECORD_TYPE in addition to ARRAY_TYPE.
21886 2018-05-15 Martin Sebor <msebor@redhat.com>
21888 PR middle-end/85643
21889 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
21891 2018-05-15 Richard Biener <rguenther@suse.de>
21893 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
21894 add by_clobber_p one. Change algorithm to collect all defs
21895 representing uses we need to walk and try reducing them to
21896 a single one before failing.
21897 (dse_dom_walker::dse_optimize_stmt): Adjust.
21899 2018-05-13 Mark Wielaard <mark@klomp.org>
21901 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
21902 (size_of_loc_descr): Likewise.
21903 (output_loc_operands): Likewise.
21904 (output_loc_operands_raw): Likewise.
21905 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
21906 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
21907 (hash_loc_operands): Likewise.
21908 (compare_loc_operands): Likewise.
21910 2018-05-14 Mark Wielaard <mark@klomp.org>
21912 * dwarf2out.c (count_index_addrs): New function.
21913 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
21915 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21917 PR tree-optimization/83648
21918 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
21919 return value as malloc candidate.
21921 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21924 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
21925 param as true in call to suggest_attribute.
21927 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
21929 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
21930 -mreadonly-in-sdata.
21932 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21934 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
21936 (aarch64_crypto_aesd_fused): Likewise.
21938 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
21940 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
21941 (movsi_aarch64): Likewise.
21942 (load_pairsi): Likewise.
21943 (load_pairdi): Likewise.
21944 (store_pairsi): Likewise.
21945 (store_pairdi): Likewise.
21946 (load_pairsf): Likewise.
21947 (load_pairdf): Likewise.
21948 (store_pairsf): Likewise.
21949 (store_pairdf): Likewise.
21950 (zero_extend): Likewise.
21951 (trunc): Swap alternatives.
21952 (fcvt_target): Add '?' to prefer w over r.
21954 2018-05-14 Jakub Jelinek <jakub@redhat.com>
21957 * config/i386/i386.md: Disallow non-commutative arithmetics in
21958 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
21959 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
21960 in the peephole2 before it.
21962 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
21964 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
21965 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
21966 (ix86_handle_option): Handle -mcldemote.
21967 * config.gcc: New header.
21968 * config/i386/cldemoteintrin.h: New file.
21969 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
21970 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21972 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21973 OPTION_MASK_ISA_CLDEMOTE.
21974 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
21975 (ix86_valid_target_attribute_inner_p): Ditto.
21976 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
21977 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
21978 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
21979 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
21980 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
21982 * config/i386/i386.opt: Add -mcldemote.
21983 * config/i386/x86intrin.h: New header.
21984 * doc/invoke.texi: Add -mcldemote.
21986 2018-05-14 Richard Biener <rguenther@suse.de>
21988 * doc/match-and-simplify.texi: Adjust :s documentation.
21990 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
21992 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
21993 intended memcpy size.
21994 (REORDER_45): Likewise.
21996 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
21998 * sort.cc: New file.
21999 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
22000 * vec.c (qsort_chk): Use gcc_qsort.
22001 * Makefile.in (OBJS-libcommon): Add sort.o.
22002 (build/sort.o): New target. Use it...
22003 (BUILD_RTL): ... here, and...
22004 (build/gencfn-macros): ... here, and...
22005 (build/genmatch): ... here.
22007 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22008 Chung-Ju Wu <jasonwucj@gmail.com>
22010 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
22011 * config/nds32/nds32-graywolf.md: New file.
22012 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
22013 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
22015 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
22016 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
22017 * config/nds32/nds32.md (pipeline_model): Add graywolf.
22018 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
22019 * config/nds32/pipelines.md: Include n15 settings.
22021 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22022 Chung-Ju Wu <jasonwucj@gmail.com>
22024 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
22025 * config/nds32/nds32-n13.md: New file.
22026 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
22027 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
22029 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
22030 * config/nds32/nds32.md (pipeline_model): Add n13.
22031 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
22032 * config/nds32/pipelines.md: Include n13 settings.
22034 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
22035 Chung-Ju Wu <jasonwucj@gmail.com>
22037 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
22038 * config/nds32/nds32-n10.md: New file.
22039 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
22040 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
22042 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
22043 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
22044 * config/nds32/nds32.md (pipeline_model): Add n10.
22045 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
22046 * config/nds32/pipelines.md: Include n10 settings.
22048 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
22049 Kito Cheng <kito.cheng@gmail.com>
22050 Chung-Ju Wu <jasonwucj@gmail.com>
22052 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
22053 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
22054 Add enum values for DSP extension instructions.
22055 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
22057 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
22058 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
22059 New code iterators.
22060 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
22061 * config/nds32/nds32-dspext.md: New file for DSP implementation.
22062 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
22063 * config/nds32/nds32-intrinsic.md: Likewise.
22064 * config/nds32/nds32_intrinsic.h: Likewise.
22065 * config/nds32/nds32-md-auxiliary.c: Likewise.
22066 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
22067 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
22068 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
22069 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
22070 * config/nds32/nds32-protos.h: New declarations for DSP extension.
22071 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
22072 TYPE_DMAC in switch statement.
22073 * config/nds32/nds32.c: New checking and implementation for DSP
22074 extension instructions.
22075 * config/nds32/nds32.h: Likewise.
22076 * config/nds32/nds32.md: Likewise.
22077 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
22078 * config/nds32/predicates.md: Implement new predicates for DSP
22081 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22083 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
22084 Reformat alternatives and attributes so it is easier to identify
22085 which constraints/attributes go with which instruction.
22086 (mov<mode>_hardfloat32, FMOVE64): Likewise.
22087 (mov<mode>_softfloat32, FMOVE64): Likewise.
22088 (mov<mode>_hardfloat64, FMOVE64): Likewise.
22089 (mov<mode>_softfloat64, FMOVE64): Likewise.
22091 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22093 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22095 (Basic PowerPC Built-in Functions): The new name of the
22096 subsection previously known as "PowerPC Built-in Functions".
22097 (Basic PowerPC Built-in Functions Available on all Configurations):
22099 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
22100 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
22101 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
22102 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
22104 2018-05-11 Martin Jambor <mjambor@suse.cz>
22107 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
22110 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
22113 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
22115 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
22117 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
22118 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
22119 (ix86_handle_option): Handle -mwaitpkg.
22120 * config.gcc: New header.
22121 * config/i386/cpuid.h (bit_WAITPKG): New bit.
22122 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
22123 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
22125 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22126 OPTION_MASK_ISA_WAITPKG.
22127 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
22128 (ix86_option_override_internal): Add PTA_WAITPKG.
22129 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
22130 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
22131 IX86_BUILTIN_TPAUSE.
22132 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
22133 __builtin_ia32_umwait and __builtin_ia32_tpause.
22134 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
22135 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
22136 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
22137 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
22138 UNSPECV_TPAUSE): New.
22139 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
22140 * config/i386/i386.opt: Add -mwaitpkg.
22141 * config/i386/waitpkgintrin.h: New file.
22142 * config/i386/x86intrin.h: New header.
22143 * doc/invoke.texi: Add -mwaitpkg.
22145 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
22148 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
22150 (cortex-m0): Use armv6s-m isa.
22151 (cortex-m0plus): Likewise.
22152 (cortex-m1): Likewise.
22153 (cortex-m0.small-multiply): Likewise.
22154 (cortex-m0plus.small-multiply): Likewise.
22155 (cortex-m1.small-multiply): Likewise.
22157 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
22158 Jakub Jelinek <jakub@redhat.com>
22160 PR tree-optimization/85692
22161 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
22162 source permute as well.
22164 2018-05-11 Martin Liska <mliska@suse.cz>
22167 * doc/extend.texi: Document LLVM style format for no_sanitize
22170 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
22172 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
22173 mode_supports_vsx_dform_quad to mode_supports_dq_form.
22174 (mode_supports_vsx_dform_quad): Likewise.
22175 (mode_supports_vmx_dform): Move these functions to be next to the
22176 other mode_supports functions.
22177 (mode_supports_dq_form): Likewise.
22178 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
22179 mode_supports_dq_form.
22180 (reg_offset_addressing_ok_p): Likewise.
22181 (offsettable_ok_by_alignment): Likewise.
22182 (rs6000_legitimate_offset_address_p): Likewise.
22183 (legitimate_lo_sum_address_p): Likewise.
22184 (rs6000_legitimize_address): Likewise.
22185 (rs6000_legitimize_reload_address): Likewise.
22186 (rs6000_secondary_reload_inner): Likewise.
22187 (rs6000_preferred_reload_class): Likewise.
22188 (rs6000_output_move_128bit): Likewise.
22190 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
22192 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
22193 Generate SImode target register for null target.
22194 <case IX86_BUILTIN_XGETBV>: Ditto.
22195 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
22196 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
22198 2018-05-10 Carl Love <cel@us.ibm.com>
22200 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
22201 dcbtt and dcbtstt if operands[2] is 0.
22203 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
22206 * config/i386/sse.md (usadv64qi): New expander.
22208 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
22210 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
22211 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
22212 -maltivec=be support.
22213 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
22214 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
22215 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22216 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
22217 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
22218 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
22219 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
22220 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
22221 altivec_vsumsws): Adjust.
22222 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
22223 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
22224 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
22226 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
22227 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
22228 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
22229 (altivec_lve<VI_char>x): Delete expand.
22230 (*altivec_lve<VI_char>x_internal): Rename to...
22231 (altivec_lve<VI_char>x): ... this.
22232 (altivec_lvxl_<mode>): Delete expand.
22233 (*altivec_lvxl_<mode>_internal): Rename to ...
22234 (altivec_lvxl_<mode>): ... this.
22235 (altivec_stvxl_<mode>): Delete expand.
22236 (*altivec_stvxl_<mode>_internal): Rename to ...
22237 (altivec_stvxl_<mode>): ... this.
22238 (altivec_stve<VI_char>x): Delete expand.
22239 (*altivec_stve<VI_char>x_internal): Rename to ...
22240 (altivec_stve<VI_char>x): ... this.
22241 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
22242 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
22243 reduc_plus_scal_<mode>): Adjust.
22244 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
22246 (rs6000_cpu_cpp_builtins): Adjust.
22247 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
22248 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
22249 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
22250 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
22251 -maltivec=be support.
22252 (rs6000_split_vec_extract_var): Adjust.
22253 (rs6000_split_v4si_init): Adjust.
22254 (swap_selector_for_mode): Delete.
22255 (altivec_expand_lvx_be, altivec_expand_stvx_be,
22256 altivec_expand_stvex_be): Delete.
22257 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
22258 -maltivec=be support.
22259 (rs6000_gimple_fold_builtin): Ditto.
22260 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
22262 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
22263 (TARGET_DIRECT_MOVE_64BIT): Adjust.
22264 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
22265 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
22266 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
22267 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
22268 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
22269 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
22270 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
22271 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
22272 anonymous split): Adjust.
22273 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
22274 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
22276 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
22278 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
22279 when --with-gxx-include-dir is also specified.
22280 * configure: Regenerate.
22282 2018-05-09 Jim Wilson <jimw@sifive.com>
22285 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
22286 * config/riscv/t-withmultilib: New.
22287 * config/riscv/withmultilib.h: New.
22288 * doc/install.texi: Document RISC-V --with-multilib-list support.
22290 2018-05-09 Richard Biener <rguenther@suse.de>
22292 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
22294 (vect_bb_vectorization_profitable_p): Adjust. Compute
22295 actual scalar cost using the cost vector and the add_stmt_cost
22298 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22300 PR rtl-optimization/85645
22301 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
22302 in the REG_CFA_REGISTER note for LR, don't leave it empty.
22304 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22306 PR rtl-optimization/85645
22307 * shrink-wrap.c (spread_components): Return a boolean saying if
22308 anything was changed.
22309 (try_shrink_wrapping_separate): Iterate spread_components until
22310 nothing changes anymore.
22312 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22314 PR rtl-optimization/85645
22315 * regrename.c (build_def_use): Also kill the chains that include the
22316 destination of a REG_CFA_REGISTER note.
22318 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
22320 PR rtl-optimization/85645
22321 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
22322 insn that has a REG_CFA_REGISTER note.
22324 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
22326 * cfgexpand.c (expand_clobber): New function.
22327 (expand_gimple_stmt_1): Use it.
22328 * tree-vect-stmts.c (vect_clobber_variable): New function,
22330 (vectorizable_simd_clone_call): ...here.
22331 (vectorizable_store): Emit a clobber either side of an
22332 IFN_STORE_LANES sequence.
22333 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
22335 2018-05-09 Tom de Vries <tom@codesourcery.com>
22338 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
22339 (define_insn "trap_if_false"): Add exit after trap.
22341 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
22343 PR rtl-optimization/85638
22344 * bb-reorder.c: Include common/common-target.h.
22345 (create_forwarder_block): New function extracted from...
22346 (fix_up_crossing_landing_pad): ...here. Rename into...
22347 (dw2_fix_up_crossing_landing_pad): ...this.
22348 (sjlj_fix_up_crossing_landing_pad): New function.
22349 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
22350 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
22351 from both partitions and exit the loop after one iteration.
22353 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22356 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22358 (Basic PowerPC Built-in Functions): The new name of the
22359 subsection previously known as "PowerPC Built-in Functions".
22360 (Basic PowerPC Built-in Functions Available on all Configurations):
22362 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
22364 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
22366 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
22368 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
22371 2018-05-08 Jim Wilson <jimw@sifive.com>
22373 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
22374 (LD_EMUL_SUFFIX): New.
22375 (LINK_SPEC): Use it.
22377 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22379 * doc/extend.texi (PowerPC Built-in Functions): Rename this
22381 (Basic PowerPC Built-in Functions): The new name of the
22382 subsection previously known as "PowerPC Built-in Functions".
22383 (Basic PowerPC Built-in Functions Available on all Configurations):
22385 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
22387 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
22389 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
22391 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
22394 2018-05-08 Jakub Jelinek <jakub@redhat.com>
22397 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
22398 after cmpelim optimization.
22400 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
22402 * config.gcc: Support "goldmont".
22403 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
22404 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
22405 PROCESSOR_GOLDMONT.
22406 * config/i386/i386.c (m_GOLDMONT): Define.
22407 (processor_target_table): Add "goldmont".
22408 (PTA_GOLDMONT): Define.
22409 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
22410 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
22411 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
22412 (fold_builtin_cpu): Add "goldmont".
22413 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
22414 (ix86_option_override_internal): Add "goldmont".
22415 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
22416 (processor_type): Add PROCESSOR_GOLDMONT.
22417 * config/i386/i386.md: Add CPU "glm".
22418 * config/i386/glm.md: New file.
22419 * config/i386/x86-tune.def: Add m_GOLDMONT.
22420 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
22422 2018-05-08 Jakub Jelinek <jakub@redhat.com>
22425 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
22427 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
22428 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
22429 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
22432 * config/i386/i386.c (ix86_fold_builtin): Handle
22433 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
22436 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
22437 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
22439 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
22442 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
22443 (check_arch): Likewise.
22444 (check_fpu): Return the result rather than printing it.
22445 (end arch): Fix operator precedence.
22446 (end cpu): Likewise.
22447 (END): Print the result from check_fpu.
22449 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
22450 Alan Hayward <alan.hayward@arm.com>
22451 David Sherwood <david.sherwood@arm.com>
22453 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
22454 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
22455 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
22456 (*fcmuo<mode>_and): New patterns.
22458 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
22460 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
22461 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
22462 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
22463 (cmp_op, sve_imm_con): New code attributes.
22464 (SVE_COND_INT_CMP, imm_con): Delete.
22465 (cmp_op): Remove above unspecs from int attribute.
22466 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
22468 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
22469 comparison-specific unspecs.
22470 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
22471 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
22472 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
22473 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
22474 (*vec_fcm<cmp_op><mode>): Rename to...
22475 (*fcm<cmp_op><mode>): ...this and adjust likewise.
22476 (*vec_fcmuo<mode>): Rename to...
22477 (*fcmuo<mode>): ...this and adjust likewise.
22478 (*pred_fcm<cmp_op><mode>): New pattern.
22479 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
22480 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
22482 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
22484 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
22485 (aarch64_emit_sve_predicated_cond): New function.
22486 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
22487 (aarch64_emit_unspec_cond_or): Replace with...
22488 (aarch64_emit_sve_or_conds): ...this new function. Use
22489 aarch64_emit_sve_ptrue_op for the individual comparisons and
22490 aarch64_emit_binop to OR them together.
22491 (aarch64_emit_inverted_unspec_cond): Replace with...
22492 (aarch64_emit_sve_inverted_cond): ...this new function. Use
22493 aarch64_emit_sve_ptrue_op for the comparison and
22494 aarch64_emit_unop to invert the result.
22495 (aarch64_expand_sve_vec_cmp_float): Update after the above
22496 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
22498 2018-05-07 Nathan Sidwell <nathan@acm.org>
22500 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
22501 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
22502 (Backwards Compatibility): Likewise.
22504 2018-05-07 Luis Machado <luis.machado@linaro.org>
22508 2018-05-07 Luis Machado <luis.machado@linaro.org>
22510 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22511 <prefetch_dynamic_strides>: New const bool field.
22512 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22513 prefetch_dynamic_strides.
22514 (exynosm1_prefetch_tune): Likewise.
22515 (thunderxt88_prefetch_tune): Likewise.
22516 (thunderx_prefetch_tune): Likewise.
22517 (thunderx2t99_prefetch_tune): Likewise.
22518 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
22520 (aarch64_override_options_internal): Update to set
22521 PARAM_PREFETCH_DYNAMIC_STRIDES.
22522 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
22523 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
22524 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
22525 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
22526 prefetch-dynamic-strides setting.
22528 2018-05-07 Luis Machado <luis.machado@linaro.org>
22530 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22531 <minimum_stride>: New const int field.
22532 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22533 minimum_stride field.
22534 (exynosm1_prefetch_tune): Likewise.
22535 (thunderxt88_prefetch_tune): Likewise.
22536 (thunderx_prefetch_tune): Likewise.
22537 (thunderx2t99_prefetch_tune): Likewise.
22538 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
22539 (aarch64_override_options_internal): Update to set
22540 PARAM_PREFETCH_MINIMUM_STRIDE.
22541 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
22542 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
22543 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22544 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22545 stride is constant and is below the minimum stride threshold.
22547 2018-05-07 Luis Machado <luis.machado@linaro.org>
22549 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
22552 2018-05-07 Luis Machado <luis.machado@linaro.org>
22554 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22555 <prefetch_dynamic_strides>: New const bool field.
22556 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22557 prefetch_dynamic_strides.
22558 (exynosm1_prefetch_tune): Likewise.
22559 (thunderxt88_prefetch_tune): Likewise.
22560 (thunderx_prefetch_tune): Likewise.
22561 (thunderx2t99_prefetch_tune): Likewise.
22562 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
22564 (aarch64_override_options_internal): Update to set
22565 PARAM_PREFETCH_DYNAMIC_STRIDES.
22566 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
22567 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
22568 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
22569 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
22570 prefetch-dynamic-strides setting.
22572 2018-05-07 Luis Machado <luis.machado@linaro.org>
22574 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
22575 <minimum_stride>: New const int field.
22576 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
22577 minimum_stride field.
22578 (exynosm1_prefetch_tune): Likewise.
22579 (thunderxt88_prefetch_tune): Likewise.
22580 (thunderx_prefetch_tune): Likewise.
22581 (thunderx2t99_prefetch_tune): Likewise.
22582 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
22583 (aarch64_override_options_internal): Update to set
22584 PARAM_PREFETCH_MINIMUM_STRIDE.
22585 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
22586 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
22587 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
22588 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
22589 stride is constant and is below the minimum stride threshold.
22591 2018-05-06 Jakub Jelinek <jakub@redhat.com>
22594 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
22595 the type is addressable. Don't force op into register if it has
22598 2018-05-05 Roland McGrath <mcgrathr@google.com>
22601 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
22602 any section for which we don't know a specific type it should have,
22603 regardless of name. Previously this was done only for the exact
22604 names ".init_array", ".fini_array", and ".preinit_array".
22605 (default_elf_asm_named_section): Add comment about
22606 relationship with default_section_type_flags and SECTION_NOTYPE.
22607 (get_section): Don't consider it a type conflict if one side has
22608 SECTION_NOTYPE and the other doesn't, as long as neither has the
22609 SECTION_BSS et al used in the default_section_type_flags logic.
22611 2018-05-05 Tom de Vries <tom@codesourcery.com>
22614 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
22615 (workaround_barsyncs): New function.
22616 (nvptx_reorg): Use workaround_barsyncs.
22617 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
22618 (define_expand "nvptx_membar_cta"): New define_expand.
22619 (define_insn "*nvptx_membar_cta"): New insn.
22621 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22623 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
22624 To improve optimization opportunities.
22625 * builtin-types.def: The new needed builtin types for the above.
22627 2018-05-04 Richard Biener <rguenther@suse.de>
22629 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
22630 * gimple-ssa-store-merging.c
22631 (imm_store_chain_info::output_merged_store): Remove redundant create,
22632 release split_store vector contents on failure.
22633 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
22634 scalar stmt vector on cache hit.
22636 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
22638 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
22640 * config.gcc (powerpc-xilinx-eabi*): Remove.
22641 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
22643 (fusion_addis_mem_combo_load): Ditto.
22644 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
22646 (rs6000_cpu_cpp_builtins): Ditto.
22647 * config/rs6000/rs6000-linux.c
22648 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
22649 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
22650 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
22652 (rs6000_setup_reg_addr_masks): Ditto.
22653 (rs6000_init_hard_regno_mode_ok): Ditto.
22654 (rs6000_option_override_internal): Ditto.
22655 (legitimate_lo_sum_address_p): Ditto.
22656 (rs6000_legitimize_address): Ditto.
22657 (rs6000_legitimize_reload_address): Ditto.
22658 (rs6000_legitimate_address_p): Ditto.
22659 (abi_v4_pass_in_fpr): Ditto.
22660 (setup_incoming_varargs): Ditto.
22661 (rs6000_gimplify_va_arg): Ditto.
22662 (rs6000_split_multireg_move): Ditto.
22663 (rs6000_savres_strategy): Ditto.
22664 (rs6000_emit_prologue_components): Ditto.
22665 (rs6000_emit_epilogue_components): Ditto.
22666 (rs6000_emit_prologue): Ditto.
22667 (rs6000_emit_epilogue): Ditto.
22668 (rs6000_elf_file_end): Ditto.
22669 (rs6000_function_value): Ditto.
22670 (rs6000_libcall_value): Ditto.
22671 * config/rs6000/rs6000.h: Ditto.
22672 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
22673 (TARGET_MINMAX): ... this. New.
22674 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
22675 * config/rs6000/rs6000.md: Remove Xilinx FP support.
22676 (*movsi_internal1_single): Delete.
22677 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
22678 mfpu=, mxilinx-fpu): Delete.
22679 * config/rs6000/singlefp.h: Delete.
22680 * config/rs6000/sysv4.h: Remove Xilinx FP support.
22681 * config/rs6000/t-rs6000: Ditto.
22682 * config/rs6000/t-xilinx: Delete.
22683 * config/rs6000/titan.md: Adjust for fp_type removal.
22684 * config/rs6000/vsx.md: Remove Xilinx FP support.
22685 (VStype_simple): Delete.
22686 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
22687 * config/rs6000/xfpu.h: Delete.
22688 * config/rs6000/xfpu.md: Delete.
22689 * config/rs6000/xilinx.h: Delete.
22690 * config/rs6000/xilinx.opt: Delete.
22691 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
22692 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
22694 2018-05-04 Tom de Vries <tom@codesourcery.com>
22697 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
22700 2018-05-04 Richard Biener <rguenther@suse.de>
22702 PR middle-end/85627
22703 * tree-complex.c (update_complex_assignment): We are always in SSA form.
22704 (expand_complex_div_wide): Likewise.
22705 (expand_complex_operations_1): Likewise.
22706 (expand_complex_libcall): Preserve EH info of the original stmt.
22707 (tree_lower_complex): Handle removed blocks.
22708 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
22709 on complex multiplication and division libcall builtins.
22711 2018-05-04 Richard Biener <rguenther@suse.de>
22713 PR middle-end/85574
22714 * fold-const.c (negate_expr_p): Restrict negation of operand
22715 zero of a division to when we know that can happen without
22717 (fold_negate_expr_1): Likewise.
22719 2018-05-04 Jakub Jelinek <jakub@redhat.com>
22722 * real.h (real_nextafter): Declare.
22723 * real.c (real_nextafter): New function.
22724 * fold-const-call.c (fold_const_nextafter): New function.
22725 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
22726 CASE_CFN_NEXTTOWARD.
22727 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
22728 even when arg1_mode is different from arg0_mode.
22730 2018-05-03 Nathan Sidwell <nathan@acm.org>
22732 * doc/extend.texi (Deprecated Features): Remove
22733 -ffriend-injection.
22734 (Backwards Compatibility): Likewise.
22735 * doc/invoke.texi (C++ Language Options): Likewise.
22736 (C++ Dialect Options): Likewise.
22738 2018-05-03 Jakub Jelinek <jakub@redhat.com>
22741 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
22742 _mm512_mask_mullox_epi64): New intrinsics.
22744 2018-05-03 Tom de Vries <tom@codesourcery.com>
22747 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22748 dump files): Add offload-tree.
22750 2018-05-03 Richard Biener <rguenther@suse.de>
22752 PR tree-optimization/85615
22753 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
22754 to loops not nested in BBs loop father to avoid creating multi-entry
22757 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22759 PR tree-optimization/70291
22760 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
22761 arguments. Change return type to tree. Emit libcall as a new
22762 statement rather than replacing existing one when inplace_p is true.
22763 (expand_complex_multiplication_components): New function.
22764 (expand_complex_multiplication): Expand floating-point complex
22765 multiplication using the above.
22766 (expand_complex_division): Rename inner_type parameter to type.
22767 Update expand_complex_libcall call-site.
22768 (expand_complex_operations_1): Update expand_complex_multiplication
22769 and expand_complex_division call-sites.
22771 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22774 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22775 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22776 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
22777 the highest significant bit of the shift count mask is clear. In
22778 check whether and[sq]i3 is needed verify that all significant bits
22779 of the shift count other than the highest are set.
22781 2018-05-02 Tom de Vries <tom@codesourcery.com>
22784 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
22785 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
22786 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
22787 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
22788 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
22789 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
22790 __builtin_goacc_parlevel_size.
22792 2018-05-02 Richard Biener <rguenther@suse.de>
22794 PR tree-optimization/85597
22795 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
22796 do not use split vect_get_vec_defs call but call vect_get_slp_defs
22799 2018-05-02 Tom de Vries <tom@codesourcery.com>
22802 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22803 dump files): Add ltrans-tree.
22805 2018-05-02 Tom de Vries <tom@codesourcery.com>
22808 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
22809 dump files): Add wpa-ipa.
22811 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
22813 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
22814 powerpc*-*-linux*paired* target.
22815 * config/rs6000/750cl.h: Delete.
22816 * config/rs6000/paired.h: Delete.
22817 * config/rs6000/paired.md: Delete.
22818 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
22820 * config/rs6000/rs6000-builtin.def: Remove paired float support.
22821 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
22822 comment. Remove paired float support.
22823 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
22824 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
22826 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
22827 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
22829 * config/rs6000/rs6000.c: Remove paired float support.
22830 (paired_expand_vector_init, paired_expand_vector_move,
22831 paired_emit_vector_compare, paired_emit_vector_cond_expr,
22832 (paired_expand_lv_builtin, paired_expand_stv_builtin,
22833 paired_expand_builtin, paired_expand_predicate_builtin,
22834 paired_init_builtins): Delete.
22835 * config/rs6000/rs6000.h: Remove paired float support.
22836 * config/rs6000/rs6000.md: Remove paired float support.
22837 (move_from_CR_ov_bit): Delete.
22838 * config/rs6000/rs6000.opt (mpaired): Delete.
22839 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
22840 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
22842 2018-05-02 Richard Biener <rguenther@suse.de>
22844 PR middle-end/85567
22845 * gimplify.c (gimplify_save_expr): When in SSA form allow
22846 SAVE_EXPRs to compute to SSA vars.
22848 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22851 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22852 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22853 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
22854 clobber operands[2], instead use a new pseudo. Formatting fixes.
22856 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
22858 PR tree-optimization/85586
22859 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
22860 exit early for statements in the same group if the accesses are
22863 2018-05-02 Tom de Vries <tom@codesourcery.com>
22866 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
22869 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
22871 PR tree-optimization/85143
22872 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
22874 2018-05-01 Tom de Vries <tom@codesourcery.com>
22877 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
22878 not found" error message.
22880 2018-05-01 Tom de Vries <tom@codesourcery.com>
22883 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
22884 * vec.c (test_ordered_remove_if): New function.
22885 (vec_c_tests): Call test_ordered_remove_if.
22886 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
22887 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
22888 * tree-vect-patterns.c (vect_pattern_recog_1): Use
22889 VEC_ORDERED_REMOVE_IF.
22891 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22893 PR tree-optimization/82665
22894 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
22895 pointer subtraction where arguments come from a memchr call.
22897 2018-05-01 Jakub Jelinek <jakub@redhat.com>
22899 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
22900 --push-state --as-needed and --pop-state instead of --as-needed and
22901 --no-as-needed if ld supports it.
22902 * configure: Regenerated.
22905 * doc/install.texi2html: Replace _002d with - and _002a with * in
22906 generated html files using sed.
22908 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22911 * gcc-rich-location.c (blank_line_before_p): New function.
22912 (use_new_line): New function.
22913 (gcc_rich_location::add_fixit_insert_formatted): New function.
22914 * gcc-rich-location.h
22915 (gcc_rich_location::add_fixit_insert_formatted): New function.
22917 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22919 * selftest.c (assert_streq): Rename "expected" and "actual" to
22920 "val1" and "val2". Extend NULL-handling to cover both inputs
22921 symmetrically, while still requiring both to be non-NULL for a pass.
22922 * selftest.h (assert_streq): Rename "expected" and "actual" to
22924 (ASSERT_EQ): Likewise.
22925 (ASSERT_EQ_AT): Likewise.
22926 (ASSERT_KNOWN_EQ): Likewise.
22927 (ASSERT_KNOWN_EQ_AT): Likewise.
22928 (ASSERT_NE): Likewise.
22929 (ASSERT_MAYBE_NE): Likewise.
22930 (ASSERT_MAYBE_NE_AT): Likewise.
22931 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
22932 the assertion to pass.
22933 (ASSERT_STREQ_AT): Likewise.
22935 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
22937 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
22938 interaction with -pie.
22940 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22942 * selftest.h: Fix alphabetization of per-source-file selftest
22945 2018-04-30 Jason Merrill <jason@redhat.com>
22947 PR c++/61982 - dead stores to destroyed objects.
22948 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
22951 2018-04-30 Jason Merrill <jason@redhat.com>
22953 * tree.c (build_clobber): New.
22954 * tree.h: Declare it.
22955 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
22957 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22959 * diagnostic-show-locus.c (layout::layout): Update for
22960 location_get_source_line returning a char_span.
22961 (struct char_span): Move to input.h.
22962 (struct correction): Update for fields in char_span becoming
22964 (struct source_line): Update for location_get_source_line
22965 returning a char_span.
22966 (layout::print_line): Likewise.
22967 * edit-context.c (edited_file::print_content): Likewise.
22968 (edited_file::print_diff_hunk): Likewise.
22969 (edited_file::print_run_of_changed_lines): Likewise.
22970 (edited_file::get_num_lines): Likewise.
22971 (edited_line::edited_line): Likewise.
22972 * final.c (asm_show_source): Likewise.
22973 * input.c (location_get_source_line): Convert return type
22974 from const char * to char_span, losing the final "line_len"
22976 (dump_location_info): Update for the above.
22977 (get_substring_ranges_for_loc): Likewise. Use a char_span
22978 when handling the literal within the line.
22979 (test_reading_source_line): Update for location_get_source_line
22980 returning a char_span.
22981 * input.h (class char_span): Move here from
22982 diagnostic-show-locus.c, converting from a struct to a class.
22983 Make data members private.
22984 (char_span::operator bool): New.
22985 (char_span::length): New.
22986 (char_span::get_buffer): New.
22987 (char_span::operator[]): New.
22988 (char_span::subspan): Make const.
22989 (char_span::xstrdup): New.
22990 (location_get_source_line): Convert return type from const char *
22991 to char_span, losing the final "line_size" param.
22993 2018-04-30 Jan Hubicka <jh@suse.cz>
22995 * lto-wrapper.c (ltrans_priorities): New static var.
22996 (cmp_priority): New.
22997 (run_gcc): Read priorities and if doing parallel build order
22998 the Makefile by them.
23000 2018-04-30 David Malcolm <dmalcolm@redhat.com>
23002 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
23004 2018-04-30 Richard Biener <rguenther@suse.de>
23006 * tree-cfg.c (verify_address): Remove base argument, add
23007 flag whether to check TREE_ADDRESSABLE and do that.
23008 (verify_expr): Remove.
23009 (verify_types_in_gimple_reference): Add pieces from verify_expr.
23010 (verify_gimple_assign_single): Likewise.
23011 (verify_gimple_switch): Likewise.
23012 (verify_expr_location_1): Dereference tp once. Add (disabled)
23013 piece from verify_expr.
23014 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
23016 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23018 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
23020 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23022 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
23023 (small_data_pattern): Likewise.
23024 (arc_rewrite_small_data): Likewise.
23025 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
23026 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
23027 (get_symbol_alignment): New function.
23028 (legitimate_small_data_address_p): Likewise.
23029 (legitimate_scaled_address): Update, call
23030 legitimate_small_data_address_p.
23031 (output_sdata): New static variable.
23032 (arc_print_operand): Update how we handle small data operands.
23033 (arc_print_operand_address): Likewise.
23034 (arc_legitimate_address_p): Update, use
23035 legitimate_small_data_address_p.
23036 (arc_rewrite_small_data_p): Remove.
23037 (arc_rewrite_small_data_1): Likewise.
23038 (arc_rewrite_small_data): Likewise.
23039 (small_data_pattern): Likewise.
23040 (compact_sda_memory_operand): Update to use
23041 legitimate_small_data_address_p and get_symbol_alignment.
23042 (prepare_move_operands): Don't rewite sdata pattern.
23043 (prepare_extend_operands): Remove.
23044 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
23046 (zero_extendqisi2): Likewise.
23047 (zero_extendhisi2): Likewise.
23048 (extendqihi2): Likewise.
23049 (extendqisi2): Likewise.
23050 (extendhisi2): Likewise.
23051 (addsi3): Likewise.
23052 (subsi3): Likewise.
23053 (andsi3): Likewise.
23054 * config/arc/constraints.md (Usd): Change it to memory constraint.
23056 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
23058 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
23059 as source of std instructions.
23060 * config/arc/arc.md (movsi_insn): Update pattern predicate to
23061 allow 6-bit constants as source for store instructions.
23062 (movdi_insn): Update instruction pattern to allow 6-bit constants
23063 as source for store instructions.
23065 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
23067 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
23069 2018-04-30 Nathan Sidwell <nathan@acm.org>
23070 Sandra Loosemore <sandra@codesourcery.com>
23072 * dumpfile.c (dump_open): Allow '-' for stdout.
23073 * doc/invoke.texi (Developer Options): Document dump filename
23074 determination early. Document stdin/stdout selection.
23076 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
23078 Microblaze Target: PIC data text relative
23080 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
23081 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
23083 * config/microblaze/microblaze.h (microblaze_constant_address_p):
23084 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
23085 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
23086 New addressing mode for data-text relative position indepenedent code.
23087 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
23088 'ADDRESS_SYMBOLIC_TXT_REL'.
23089 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
23090 (microblaze_legitimate_pic_operand): Exclude function calls from
23091 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
23092 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
23094 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
23095 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
23096 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
23097 for 'address + offset'.
23098 (microblaze_expand_prologue): Add new function prologue call for
23100 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
23101 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
23102 table in case of TARGET_PIC_DATA_TEXT_REL.
23103 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
23104 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
23105 Add new macros 'UNSPEC_TEXT',
23106 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
23107 + exclude function calls from 'UNSPEC_PLT' in case of data text
23109 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
23110 new target hook for generating address diff vector tables in case of
23112 * doc/tm.texi : Regenerate.
23113 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
23114 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
23115 of addr diff vector generation.
23116 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
23117 target hook definition.
23118 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
23119 Add default function for generate_pic_addr_diff_vec -> flag_pic.
23120 * doc/invoke.texi (Add new pic option): Add new microblaze pic
23121 option for data text relative.
23123 2018-04-30 Richard Biener <rguenther@suse.de>
23125 * tree-chrec.h (evolution_function_is_constant_p): Remove
23127 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
23129 2018-04-30 Richard Biener <rguenther@suse.de>
23132 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
23134 2018-04-30 Richard Biener <rguenther@suse.de>
23136 PR tree-optimization/28364
23137 PR tree-optimization/85275
23138 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
23139 copying first exit test.
23141 2018-04-28 Mark Wielaard <mark@klomp.org>
23143 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
23144 dwarf_version >= 5.
23145 (dwarf_AT): Handle DW_AT_addr_base.
23146 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
23148 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
23151 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
23152 (*ashl<dwi>3_doubleword_mask_1): Ditto.
23153 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
23154 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
23156 2018-04-28 Richard Biener <rguenther@suse.de>
23158 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
23159 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
23160 to reflect use. Only add interesting stmts.
23162 2018-04-27 Martin Jambor <mjambor@suse.cz>
23165 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
23166 the jump function allows for passing through aggregate values.
23168 2018-04-27 David Malcolm <dmalcolm@redhat.com>
23170 * input.h (in_system_header_at): Convert from macro to inline
23172 (from_macro_expansion_at): Likewise.
23173 (from_macro_definition_at): Likewise.
23175 2018-04-27 Jeff Law <law@redhat.com>
23177 * config.gcc: Mark tile* targets as deprecated/obsolete.
23179 2018-04-27 Richard Biener <rguenther@suse.de>
23181 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
23184 2018-04-27 Richard Biener <rguenther@suse.de>
23186 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
23188 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
23190 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
23191 with Yd constraint. Set "preferred_for_speed" attribute from
23192 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
23193 with Yd constraint.
23194 (*movdi_internal): Ditto.
23195 (movti_interunit splitters): Remove
23196 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
23197 (movdi_interunit splitters): Ditto.
23198 * config/i386/constraints.md (Ye): Remove.
23199 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
23201 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23204 * config/aarch64/constraints.md (Usg): Limit to 31.
23205 (Usj): Limit to 63.
23207 2018-04-27 Jakub Jelinek <jakub@redhat.com>
23209 PR tree-optimization/85529
23210 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
23211 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
23212 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
23213 zero extension or masking of the MSB bit.
23214 (optimize_range_tests): Add FIRST_BB argument, pass it through
23215 to optimize_range_tests_var_bound.
23216 (maybe_optimize_range_tests, reassociate_bb): Adjust
23217 optimize_range_tests callers.
23219 2018-04-26 Richard Biener <rguenther@suse.de>
23220 Jakub Jelinek <jakub@redhat.com>
23222 * cgraph.h (symbol_table): Just declare debug method here.
23223 * symtab.c (symbol_table::debug): Define.
23225 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
23227 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
23229 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
23231 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
23232 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
23233 (*movdi_internal): Substitute Yi and Yj constraint with x
23234 and Ym and Yn constraint with y constraint. Update "isa"
23235 attribute and set "preferred_for_speed" attribute from
23236 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
23237 (*movsi_internal): Ditto.
23238 (*movdf_internal): Ditto.
23239 (*movsf_internal): Ditto.
23240 (*zero_extendsidi2): Ditto.
23241 * config/i386/sse.md (vec_set<mode>_0): Ditto.
23242 (sse2_loadld): Ditto.
23243 (*vec_extract<ssevecmodelower>_0): Ditto.
23244 (*vec_extractv4si_0_zext_sse4): Ditto.
23245 (vec_concatv2di): Ditto.
23246 (*vec_dup<mode>): Ditto.
23247 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
23248 * config/i386/constraints.md (Yi): Remove.
23253 2018-04-26 Nathan Sidwell <nathan@acm.org>
23255 * dumpfile.c (dump_open): New.
23256 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
23257 (dump_finish): Detect stdio/stderr by value not name.
23259 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
23261 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
23263 2018-04-26 Tom de Vries <tom@codesourcery.com>
23266 * config/nvptx/nvptx.c (verify_neutering_jumps)
23267 (verify_neutering_labels): New function
23268 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
23270 2018-04-26 Tom de Vries <tom@codesourcery.com>
23273 * config/nvptx/nvptx.c (needs_neutering_p): New function.
23274 (nvptx_single): Use needs_neutering_p to skip over insns that do not
23277 2018-04-26 Richard Biener <rguenther@suse.de>
23278 Tom de Vries <tom@codesourcery.com>
23281 * lto-streamer-out.c (output_function): Fixup loops if required to match
23282 discovery done in the reader.
23284 2018-04-26 Richard Biener <rguenther@suse.de>
23286 PR tree-optimization/85116
23287 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
23288 have a loop exit from the single latch predecessor. Remove
23289 case of header with just condition.
23290 (ch_base::copy_headers): Exclude infinite loops from any
23292 (pass_ch::execute): Record exits.
23294 2018-04-26 Richard Biener <rguenther@suse.de>
23296 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
23297 prologue cost vector and pass it to vect_get_load_cost.
23298 (vect_get_peeling_costs_all_drs): Likewise.
23299 (vect_peeling_hash_get_lowest_cost): Likewise.
23300 (vect_enhance_data_refs_alignment): Likewise.
23302 2018-04-26 Richard Biener <rguenther@suse.de>
23304 PR middle-end/85450
23305 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
23306 checking of integer<->pointer conversions.
23307 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
23308 sign-/zero-extending pointer types.
23309 (expand_omp_for_static_chunk): Likewise.
23311 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
23312 Jean Lee <xiaoyur347@gmail.com>
23314 * config/mips/mips.c (mips_asan_shadow_offset): New function.
23315 (TARGET_ASAN_SHADOW_OFFSET): Define.
23316 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
23317 true for -fsanitize=address.
23319 2018-04-25 Mark Wielaard <mark@klomp.org>
23321 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
23324 2018-04-25 Jakub Jelinek <jakub@redhat.com>
23326 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
23327 than "alu", remove explicit "memory" and "imm_disp" attributes.
23328 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
23330 PR middle-end/85414
23331 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
23332 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
23333 gen_lowpart_no_emit.
23335 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
23338 * config/i386/i386.c (ix86_expand_builtin): Change memory
23339 operand to XI, extend p0 to Pmode.
23340 * config/i386/i386.md: Change unspec volatile and operand
23341 1 mode to XI, change operand 0 mode to P.
23343 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23345 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
23346 GET_MODE_MASK before any checking.
23347 (nds32_can_use_bset_p): Likewise.
23348 (nds32_can_use_btgl_p): Likewise.
23350 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23352 * config/nds32/nds32-doubleword.md: New define_split pattern for
23353 illegal register number.
23355 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23357 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
23359 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
23361 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
23363 2018-04-25 Richard Biener <rguenther@suse.de>
23365 * lto-streamer.h (LTO_major_version): Bump to 8.
23367 2018-04-25 Jakub Jelinek <jakub@redhat.com>
23369 * BASE-VER: Set to 9.0.0.
23371 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
23373 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
23374 in __abskf2 and __powikf2.
23376 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23379 * config/aarch64/constraints.md (Usg, Usj): New constraints.
23380 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
23381 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
23382 Use the above on operand 2. Reindent.
23383 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
23385 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
23388 * common/config/i386/i386-common.c (ix86_handle_option): Don't
23390 * config/i386/i386.opt (mcet): Removed.
23391 * doc/install.texi: Remove -mcet documentation.
23392 * doc/invoke.texi: Likewise.
23394 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
23397 * doc/install.texi: Remove -mcet from bootstrap-cet.
23399 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23402 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
23403 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
23407 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
23408 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
23409 containing a CONST_VECTOR.
23411 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
23413 * doc/install.texi: Update newlib dependency for nvptx.
23415 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23418 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
23419 instead of INTVAL when shifting x left.
23421 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
23423 PR tree-optimization/85478
23424 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
23425 vect_grouped_store_supported for single element vectors.
23427 2018-04-24 Richard Biener <rguenther@suse.de>
23430 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
23431 load cost increase to the case of non-constant step.
23433 2018-04-24 Jakub Jelinek <jakub@redhat.com>
23436 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
23437 destination if any_malformed_asm.
23439 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
23441 PR middle-end/85496
23442 * expr.c (store_field): In the bitfield case, if the value comes from
23443 a function call and is returned in registers by means of a PARALLEL,
23444 do not change the mode of the temporary unless BLKmode and VOIDmode.
23446 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
23448 PR rtl-optimization/85423
23449 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
23450 dependencies to debug insns when the previous insn is non-debug.
23452 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
23454 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
23455 enums into a single definition.
23456 (fls): Fix predicates and printing.
23459 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
23461 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
23462 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
23463 and short u6 immediate.
23464 (check_if_valid_sleep_operand): Remove.
23465 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
23467 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23469 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
23470 flag_always_save_lp condition.
23471 * config/nds32/nds32.opt (malways-save-lp): New option.
23473 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23475 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
23476 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
23477 * config/nds32/nds32.h
23478 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
23479 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
23481 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23483 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
23484 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
23486 2018-04-22 Shiva Chen <shiva0217@gmail.com>
23487 Chung-Ju Wu <jasonwucj@gmail.com>
23489 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
23491 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
23492 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
23494 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23496 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
23498 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23500 * config/nds32/nds32-protos.h (nds32_data_alignment,
23501 nds32_local_alignment): Declare.
23502 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
23503 nds32_local_alignment): New functions.
23504 (TARGET_CONSTANT_ALIGNMENT): Define.
23505 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
23507 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23509 * config/nds32/nds32.c
23510 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
23511 (TARGET_MODES_TIEABLE_P): Likewise.
23513 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
23515 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
23516 level Ofast and Og.
23518 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
23519 Chung-Ju Wu <jasonwucj@gmail.com>
23521 * config/nds32/constants.md (unspec_volatile_element): Add enum values
23522 for unaligned access.
23523 * config/nds32/nds32-intrinsic.c: Implementation of expanding
23525 * config/nds32/nds32-intrinsic.md: Likewise.
23526 * config/nds32/nds32_intrinsic.h: Likewise.
23527 * config/nds32/nds32.h (nds32_builtins): Likewise.
23528 * config/nds32/nds32.opt (munaligned-access): New option.
23529 * config/nds32/nds32.c (nds32_asm_file_start): Display
23530 flag_unaligned_access status.
23532 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
23534 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
23535 -mno-relax is present.
23536 * config/riscv/linux.h (LINK_SPEC): Ditto.
23538 2018-04-20 Martin Sebor <msebor@redhat.com>
23541 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
23543 (gimple_fold_builtin_stxcpy_chk): Same.
23544 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
23546 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
23549 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
23550 __powikf2 when long double is IEEE 128-bit.
23552 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
23554 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
23555 step to make sure stack always aligned.
23557 2018-04-20 Carl Love <cel@us.ibm.com>
23560 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
23561 size check for arg0.
23563 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
23564 Tom de Vries <tom@codesourcery.com>
23567 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
23568 Emit insns for calls too.
23569 (nvptx_find_par): Always look for worker-level predecessor insn.
23570 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
23572 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
23573 (nvptx_process_pars): Propagate frames for calls.
23575 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
23578 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
23580 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
23581 (ix86_handle_option): Don't handle OPT_mibt.
23582 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
23584 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
23586 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
23587 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
23588 (ix86_target_macros): Define __CET__ with flag_cf_protection
23589 for -fcf-protection.
23590 * config/i386/i386.c (isa2_opts): Remove -mibt.
23591 * config/i386/i386.h (TARGET_IBT): Removed.
23592 (TARGET_IBT_P): Likewise.
23593 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
23594 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
23595 * config/i386/i386.opt (mcet): Update help message.
23596 (mshstk): Likewise.
23598 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
23599 -mcet as an alias for -mshstk.
23601 2018-04-20 Richard Biener <rguenther@suse.de>
23603 PR middle-end/85475
23604 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
23605 complexity by forcing a single use of the multiply operand.
23607 2018-04-20 Martin Jambor <mjambor@suse.cz>
23610 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
23611 recursion dependency to only apply to non-clones.
23613 2018-04-20 Martin Jambor <mjambor@suse.cz>
23616 * ipa-cp.c (create_specialized_node): Check that clones of
23617 self-recursive edges exist during IPA-CP.
23619 2018-04-19 Toon Moene <toon@moene.org>
23621 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
23624 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23626 PR tree-optimization/85467
23627 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
23628 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
23629 VECTOR_CST element to type.
23631 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23634 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
23635 * config/i386/i386.md (builtin_setjmp_setup): Removed.
23636 (builtin_longjmp): Likewise.
23637 (save_stack_nonlocal): New pattern.
23638 (restore_stack_nonlocal): Likewise.
23640 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23643 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23644 Replace ASM_OUTPUT_LABEL with fprintf.
23646 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
23649 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
23650 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
23651 * config/i386/i386-c.c (ix86_target_macros_internal): Also
23652 define __IBT__ and __SHSTK__ for -fcf-protection.
23653 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
23655 (ix86_trampoline_init): Likewise.
23656 (x86_output_mi_thunk): Likewise.
23657 (ix86_notrack_prefixed_insn_p): Likewise.
23658 (ix86_option_override_internal): Don't disallow -fcf-protection.
23659 * config/i386/i386.md (rdssp<mode>): Also enable for
23661 (incssp<mode>): Likewise.
23662 (nop_endbr): Likewise.
23663 * config/i386/i386.opt (mcet): Change help message to built-in
23666 (mshstk): Likewise.
23667 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
23668 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
23669 enable CET built-in functions.
23671 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
23673 * common/config/i386/i386-common.c
23674 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
23675 OPTION_MASK_ISA_MOVDIRI_UNSET,
23676 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
23677 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
23678 * config.gcc (movdirintrin.h): New header.
23679 * config/i386/cpuid.h (bit_MOVDIRI,
23680 bit_MOVDIR64B): New bits.
23681 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
23683 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
23684 (VOID, PVOID, PCVOID)): New function types.
23685 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
23686 __builtin_ia32_directstoreu_u64,
23687 __builtin_ia32_movdir64b): New builtins.
23688 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
23689 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
23691 (ix86_valid_target_attribute_inner_p): Ditto.
23692 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
23693 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
23694 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
23695 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
23696 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
23697 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
23698 (movdiri<mode>, movdir64b_<mode>): New.
23699 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
23700 * config/i386/immintrin.h: Include movdirintrin.h.
23701 * config/i386/movdirintrin.h: New file.
23702 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
23704 2018-04-19 Richard Biener <rguenther@suse.de>
23706 PR middle-end/85455
23707 * cfg.c (clear_bb_flags): When loop state says we have
23708 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
23710 2018-04-19 Richard Biener <rguenther@suse.de>
23712 PR tree-optimization/84737
23713 * tree-vect-data-refs.c (vect_copy_ref_info): New function
23714 copying restrict info.
23715 (vect_setup_realignment): Use it.
23716 * tree-vectorizer.h (vect_copy_ref_info): Declare.
23717 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
23718 the first DR to all generated stores.
23719 (vectorizable_load): Likewise for loads.
23721 2018-04-19 Jakub Jelinek <jakub@redhat.com>
23723 PR tree-optimization/85446
23724 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
23725 the integral and pointer types to have the same precision.
23727 * doc/install.texi: Document --disable-cet being the default and
23730 2018-04-18 Martin Liska <mliska@suse.cz>
23732 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
23735 2018-04-18 Martin Liska <mliska@suse.cz>
23738 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
23741 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
23742 arguments if they are comparable.
23744 2018-04-18 Martin Liska <mliska@suse.cz>
23747 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
23750 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
23753 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
23756 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
23757 ENDBR after calling __morestack.
23759 2018-04-18 David Malcolm <dmalcolm@redhat.com>
23762 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
23763 by using gcc_base_ver to generate a gcc_driver_version, and use
23764 it when generating GCC_DRIVER_NAME.
23765 * configure: Regenerate.
23767 2018-04-18 Jakub Jelinek <jakub@redhat.com>
23770 * config.gcc: Obsolete powerpc*-*-*spe*.
23772 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23775 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
23776 (stabstr_D): Change type of unum from unsigned int to
23777 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
23780 2018-04-17 Jim Wilson <jimw@sifive.com>
23783 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
23784 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
23785 Set arg_pointer_offset after using pretend_args_size.
23787 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23789 PR rtl-optimization/85431
23790 * dse.c (record_store): Ignore zero width stores.
23793 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
23794 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
23795 __builtin_stack_restore rather than after it.
23796 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
23797 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
23798 argument instead of virtual_dynamic_stack_rtx.
23800 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
23802 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
23804 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23805 Add note to error message to explain internal mapping of overloaded
23806 built-in function name to non-overloaded built-in function name.
23807 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
23810 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
23813 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
23814 where the inputs overlap with the output.
23816 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23818 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
23819 (=v, v) alternative and explicit "memory" attribute.
23820 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
23821 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23823 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
23824 "sselog1" type instead of "sselog".
23825 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
23826 "sselog". Remove explicit "memory" attribute.
23827 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
23828 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23830 (vec_extract_hi_v32hi): Merge all alternatives into one, use
23831 "sselog1" type instead of "sselog". Remove explicit "memory"
23833 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
23834 use "sselog1" type instead of "sselog". Remove explicit "memory"
23836 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
23837 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23839 (vec_extract_hi_v64qi): Merge all alternatives into one, use
23840 "sselog1" type instead of "sselog". Remove explicit "memory"
23842 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
23843 use "sselog1" type instead of "sselog". Remove explicit "memory"
23847 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
23849 PR middle-end/85414
23850 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
23853 2018-04-17 Martin Jambor <mjambor@suse.cz>
23856 * ipa-cp.c (create_specialized_node): Call
23857 expand_all_artificial_thunks if necessary.
23859 2018-04-17 Martin Liska <mliska@suse.cz>
23862 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
23863 in message, remote space in between '_G' and '('.
23865 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23868 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
23869 avx512f_vmcmp<mode>3<round_saeonly_name>,
23870 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
23871 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
23872 avx512f_rndscale<mode><round_saeonly_name>,
23873 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
23874 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
23875 Use %<iptr>2 instead of %2 for -masm=intel.
23876 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
23877 avx512f_vcvttss2usi<round_saeonly_name>,
23878 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
23880 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
23881 avx512f_vcvttsd2usi<round_saeonly_name>,
23882 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
23883 Use %q1 instead of %1 for -masm=intel.
23884 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
23885 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
23886 of %3 for -masm=intel.
23887 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
23889 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
23891 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
23893 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
23894 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
23896 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
23898 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
23899 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
23900 %g1 and one with %0 and %1.
23901 (avx512er_vmrcp28<mode><round_saeonly_name>,
23902 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
23903 %1 for -masm=intel.
23904 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
23905 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
23906 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
23907 of %0 and %{%4%} for -masm=intel.
23908 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
23909 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
23910 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
23911 order of %0 and %{%5%}%{z%} for -masm=intel.
23913 2018-04-17 Jan Hubicka <jh@suse.cz>
23916 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
23918 2018-04-17 Martin Liska <mliska@suse.cz>
23921 * multiple_target.c (create_dispatcher_calls): Set apostrophes
23922 for target_clone error message. Make default implementation
23923 clone to be a local declaration.
23924 (separate_attrs): Add new argument and check for an empty
23926 (expand_target_clones): Handle it.
23927 (ipa_target_clone): Make redirection just for target_clones
23930 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
23931 Tom de Vries <tom@codesourcery.com>
23933 PR middle-end/84955
23934 * omp-expand.c (expand_oacc_for): Add dummy false branch for
23935 tiled basic blocks without omp continue statements.
23937 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
23940 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
23941 vec_extract expression as having side effects to make sure it gets
23944 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
23947 * config/i386/i386.c (get_builtin_code_for_version): Check
23950 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
23953 * config.gcc: Support "skylake".
23954 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
23956 * config/i386/i386.c (m_SKYLAKE): Define.
23957 (processor_target_table): Add "skylake".
23958 (ix86_option_override_internal): Add "skylake".
23959 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
23960 PROCESSOR_CANNONLAKE.
23961 (get_builtin_code_for_version): Fix priority for
23962 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
23963 PROCESSOR_SKYLAKE-AVX512.
23964 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
23965 (processor_type): Add PROCESSOR_SKYLAKE.
23967 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
23968 Jason Merrill <jason@redhat.com>
23971 * convert.c (convert_to_integer_1): Use direct recursion for
23972 enumeral types and types with a precision less than the number
23973 of bits in their mode.
23975 2018-04-16 Julia Koval <julia.koval@intel.com>
23978 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
23979 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
23981 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
23984 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
23985 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
23986 and -mno-direct-move.
23988 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
23991 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
23992 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
23993 Ensure negative shifts result in {0}.
23995 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
23997 PR rtl-optimization/79916
23998 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
23999 regs (if any) to define how to gnerate SD moves when LRA is in
24002 2018-04-13 Jakub Jelinek <jakub@redhat.com>
24004 PR rtl-optimization/85393
24005 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
24006 * except.c (expand_dw2_landing_pad_for_region): Make static.
24007 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
24008 a label and unconditional jump to old_bb, rather than
24009 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
24012 PR rtl-optimization/85376
24013 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
24014 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
24015 instead of a specific value.
24017 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
24018 Bin Cheng <bin.cheng@arm.com>
24020 PR tree-optimization/82965
24021 PR tree-optimization/83991
24022 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
24023 by_profile_only parameter.
24024 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
24025 information if the loop was predicted to iterate too many times.
24026 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
24028 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
24031 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
24034 2018-04-13 Martin Liska <mliska@suse.cz>
24035 Jakub Jelinek <jakub@redhat.com>
24037 PR middle-end/81657
24038 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
24039 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
24040 * builtins.c (expand_builtin_memory_copy_args): Use
24041 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
24042 handle dest_addr == pc_rtx.
24044 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
24047 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
24048 asked to not generate direct moves.
24049 (fix_trunc<mode>si2_stfiwx): Similar.
24050 (fix_trunc<mode>si2_internal): Similar.
24052 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24055 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
24056 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
24057 lookup if dest in some wider mode is known to be const0_rtx and
24058 if so, record permanent equivalence for it to be ZERO_EXTEND of
24059 the narrower mode destination.
24061 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
24063 * lto-streamer-out.c (output_function): Revert 259346.
24064 * omp-expand.c (expand_oacc_for): Likewise.
24066 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
24068 PR rtl-optimization/85354
24069 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
24070 * sel-sched.c (sel_global_init): ... here.
24072 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
24075 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
24076 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
24077 mode for PE-COFF targets.
24078 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
24079 (i386_pe_asm_lto_end): Likewise.
24080 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
24081 (TARGET_ASM_LTO_END): Likewise.
24082 * config/i386/winnt.c (saved_debug_info_level): New static variable.
24083 (i386_pe_asm_lto_start): New function.
24084 (i386_pe_asm_lto_end): Likewise.
24086 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
24087 Richard Biener <rguenther@suse.de>
24089 PR middle-end/84955
24090 * lto-streamer-out.c (output_function): Fix CFG loop state before
24092 * omp-expand.c (expand_oacc_for): Handle calls to internal
24093 functions like regular functions.
24095 2018-04-12 Richard Biener <rguenther@suse.de>
24098 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
24099 for the early LTO debug to properly generate references to it
24100 during DIE emission. Do not re-use that for the skeleton for
24102 (dwarf2out_early_finish): Likewise.
24104 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24107 * config/i386/sse.md
24108 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
24109 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
24110 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
24111 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
24112 and output is a reg, avoid creating invalid lowpart subreg, but
24113 instead split into a 512-bit move. Don't split if not AVX512VL,
24114 input is xmm16+ reg and output is a mem.
24115 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
24116 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
24117 xmm16+ reg and output is a mem.
24119 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24121 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
24122 also for flag_dwarf2_cfi_asm.
24124 2018-04-12 Jakub Jelinek <jakub@redhat.com>
24126 PR rtl-optimization/85342
24127 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
24128 a bool scalar var inside of the loop instead. Don't try to update
24129 recog_data.operand after failed apply_change_group.
24131 2018-04-12 Tom de Vries <tom@codesourcery.com>
24134 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
24135 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
24136 array with flexible array member as array without given dimension.
24137 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
24138 argument for undefined param to true.
24140 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
24143 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
24144 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
24145 from PowerPC section.
24146 * config/rs6000/sysv4.opt (mcall-): Improve help text.
24147 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
24148 help text that is too long.
24149 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
24150 help text that is too long.
24151 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
24152 help text that is too long.
24154 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
24156 * config/alpha/alpha.md (stack_probe_internal): Rename
24157 from "probe_stack". Update all callers.
24159 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24161 PR rtl-optimization/84566
24162 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
24163 sched_macro_fuse_insns.
24165 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24168 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
24169 (compute_block_dependences): ... from here.
24171 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24173 PR tree-optimization/85331
24174 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
24175 from int to HOST_WIDE_INT.
24177 2018-04-11 Martin Jambor <mjambor@suse.cz>
24180 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
24181 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
24182 not the same as the source val.
24183 (cgraph_edge_brings_value_p): New parameter.
24184 (gather_edges_for_value): Pass destination value to
24185 cgraph_edge_brings_value_p.
24186 (perhaps_add_new_callers): Likewise.
24187 (get_info_about_necessary_edges): Likewise and exclude values brought
24188 only by self-recursive edges.
24189 (create_specialized_node): Redirect only clones of self-calling edges.
24190 (+self_recursive_pass_through_p): New function.
24191 (find_more_scalar_values_for_callers_subset): Use it.
24192 (find_aggregate_values_for_callers_subset): Likewise.
24193 (known_aggs_to_agg_replacement_list): Removed.
24194 (decide_whether_version_node): Re-calculate known constants for all
24195 remaining context clones.
24197 2018-04-11 Richard Biener <rguenther@suse.de>
24200 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
24201 from early DWARF output.
24202 (dwarf2out_early_finish): Output line info unconditionally into
24203 early DWARF and add reference to it.
24205 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24208 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
24209 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
24210 other than V2DFmode using iptr mode attribute.
24211 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
24213 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
24215 PR rtl-optimization/84659
24216 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
24218 2018-04-11 Jakub Jelinek <jakub@redhat.com>
24221 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
24223 (output_loc_list): Pass address of a dummy size variable even in the
24224 locview handling loop.
24225 (index_location_lists): Add comment on why skip_loc_list_entry can't
24228 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
24231 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
24234 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
24237 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
24238 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
24239 and -mstring-compare-inline-limit.
24241 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24244 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
24245 for stack clash protection in a register whenever we need it to be in
24248 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24250 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
24251 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
24253 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
24256 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
24258 (mlong-double-): Ditto.
24259 * config/rs6000/sysv4.opt (msdata=): Ditto.
24260 (mtls-size=): Ditto.
24262 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24264 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
24265 erroneous entries for
24266 "vector int vec_ldl (int, long int *)", and
24267 "vector unsigned int vec_ldl (int, unsigned long int *)".
24268 Add comments and entries for
24269 "vector bool char vec_ldl (int, bool char *)",
24270 "vector bool short vec_ldl (int, bool short *)",
24271 "vector bool int vec_ldl (int, bool int *)",
24272 "vector bool long long vec_ldl (int, bool long long *)",
24273 "vector pixel vec_ldl (int, pixel *)",
24274 "vector long long vec_ldl (int, long long *)",
24275 "vector unsigned long long vec_ldl (int, unsigned long long *)".
24276 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
24277 type tree bool_long_long_type_node and correct definition of
24278 bool_V2DI_type_node to make reference to this new type tree.
24279 (rs6000_mangle_type): Replace erroneous reference to
24280 bool_long_type_node with bool_long_long_type_node.
24281 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
24282 comments to emphasize sign distinctions for char and int types and
24283 replace RS6000_BTI_bool_long constant with
24284 RS6000_BTI_bool_long_long constant. Also add comment to restrict
24285 use of RS6000_BTI_pixel.
24286 (bool_long_type_node): Remove this macro definition.
24287 (bool_long_long_type_node): New macro definition
24289 2018-04-10 Jakub Jelinek <jakub@redhat.com>
24291 PR rtl-optimization/85300
24292 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
24293 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
24294 simplify_unary_operation fails.
24296 2018-04-10 Martin Liska <mliska@suse.cz>
24298 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
24299 cgraph_edge and ipa_ref.
24301 2018-04-10 Jakub Jelinek <jakub@redhat.com>
24305 * config/i386/sse.md
24306 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
24307 computation of the VEC_MERGE selector from mask.
24308 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
24309 Fix decoding of the VEC_MERGE selector into mask.
24311 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
24313 PR tree-optimization/85286
24314 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
24316 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
24318 * final.c (final_1): Set insn_last_address as well as
24319 insn_current_address.
24321 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24324 * explow.c (emit_stack_probe): Call validize_mem on memory location
24325 before passing it to gen_probe_stack. Create address operand and
24326 legitimize it for the probe_stack_address case.
24328 2018-04-09 Jan Hubicka <jh@suse.cz>
24331 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
24332 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
24333 * tree.c (free_lang_data_in_type): Fix handling of binfos;
24335 (free_lang_data): Rebuild type inheritance graph.
24337 2018-04-09 Martin Sebor <msebor@redhat.com>
24339 * invoke.texi (-finline-small-functions): Mention other optimization
24341 (-findirect-inlining, -fpartial-inlining): Same.
24342 (-finline-functions-called-once): Same.
24343 (-freorder-blocks-and-partition): Same.
24345 2018-04-09 Jan Hubicka <jh@suse.cz>
24348 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
24349 jumps; choose last target that matches the criteria (i.e.
24350 no partition changes for non-crossing jumps).
24351 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
24352 support for redirecting crossing jumps to non-crossing.
24354 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
24356 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
24357 also for naked functions.
24359 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
24361 * config/arc/arc.md (add_shift): New pattern.
24362 (add_shift2): Likewise.
24363 (sub_shift): Likewise.
24364 (sub_shift_cmp0_noout): Likewise.
24365 (compare_si_ashiftsi): Likewise.
24366 (xbfu_cmp0_noout): New combine pattern.
24367 (xbfu_cmp0"): Likewise.
24368 (movsi_set_cc_insn): Place the predicable variant first.
24369 (commutative_binary_cmp0_noout): Remove clobber.
24370 (commutative_binary_cmp0): New pattern.
24371 (noncommutative_binary_cmp0): Likewise.
24372 (noncommutative_binary_cmp0_noout): Likewise.
24373 (noncommutative_binary_comparison_result_used): Removed.
24374 (rsub_cmp0): New pattern.
24375 (rsub_cmp0_noout): Likewise.
24376 (extzvsi): Changed, keep only meaningful variants.
24377 (SQH, SEZ): New iterators.
24378 (SQH_postfix): New mode attribute.
24379 (SEZ_prefix): New code attribute.
24380 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
24381 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
24382 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
24383 of numerical value.
24384 (noncommutative_operator): Check the availability of barrel
24387 2018-04-09 Richard Biener <rguenther@suse.de>
24389 PR tree-optimization/85284
24390 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
24391 Only use the niter constraining form of simple_iv when the exit
24392 is always executed.
24394 2018-04-09 Tom de Vries <tom@codesourcery.com>
24397 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
24398 (define_expand "*memory_barrier"): New define_expand.
24399 (define_insn "memory_barrier"): New insn.
24401 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24403 PR rtl-optimization/80463
24404 PR rtl-optimization/83972
24405 PR rtl-optimization/83480
24407 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
24408 correct producer for the insn.
24409 (tidy_control_flow): Fixup seqnos in case of debug insns.
24411 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24413 PR rtl-optimization/83913
24415 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
24416 different sched-times when merging exprs.
24418 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24420 PR rtl-optimization/83962
24422 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
24423 tidy_fallthru_edge and tidy_control_flow.
24425 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
24427 PR rtl-optimization/83530
24429 * sel-sched.c (force_next_insn): New global variable.
24430 (remove_insn_for_debug): When force_next_insn is true, also leave only
24431 next insn in the ready list.
24432 (sel_sched_region): When the region wasn't scheduled, make another pass
24433 over it with force_next_insn set to 1.
24435 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
24437 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
24439 * config/nds32/constants.md (unspec_volatile_element): Add enum values
24440 for interrupt control.
24441 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
24442 functions for interrupt control.
24443 * config/nds32/nds32-intrinsic.md: Likewise.
24444 * config/nds32/nds32_intrinsic.h: Likewise.
24445 * config/nds32/nds32.h (nds32_builtins): Likewise.
24447 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
24449 * config/nds32/nds32.c (nds32_init_machine_status,
24450 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
24451 strict_aligned_p field.
24452 (nds32_expand_to_rtl_hook): New function.
24453 (TARGET_EXPAND_TO_RTL_HOOK): Define.
24454 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
24456 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24457 Chung-Ju Wu <jasonwucj@gmail.com>
24459 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
24460 * config/nds32/nds32-n7.md: New file.
24461 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
24462 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
24464 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
24465 * config/nds32/nds32.md (pipeline_model): Add n7.
24466 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
24467 * config/nds32/pipelines.md: Include n7 settings.
24469 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24470 Chung-Ju Wu <jasonwucj@gmail.com>
24472 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
24473 * config/nds32/nds32-e8.md: New file.
24474 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
24475 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
24477 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
24478 * config/nds32/nds32.md (pipeline_model): Add e8.
24479 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
24480 * config/nds32/pipelines.md: Include e8 settings.
24482 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24483 Chung-Ju Wu <jasonwucj@gmail.com>
24485 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
24486 * config/nds32/nds32-n8.md: New file.
24487 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
24488 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
24490 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
24491 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
24492 * config/nds32/nds32.md (pipeline_model): Add n8.
24493 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
24494 * config/nds32/pipelines.md: Include n8 settings.
24496 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
24497 Chung-Ju Wu <jasonwucj@gmail.com>
24499 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
24500 * config/nds32/nds32-n9-2r1w.md: New file.
24501 * config/nds32/nds32-n9-3r2w.md: New file.
24502 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
24503 nds32_register_ports): New or modify for cpu n9.
24504 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
24506 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
24507 * config/nds32/nds32-utils.c: New file.
24508 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
24509 TARGET_MUL_SLOW): Define.
24510 * config/nds32/nds32.md (pipeline_model): New attribute.
24511 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
24512 New options that support cpu n9.
24513 * config/nds32/pipelines.md: Include n9 settings.
24514 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
24516 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
24518 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
24519 information if necessary.
24520 (output_cond_branch_compare_zero): Likewise.
24521 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
24522 (nds32_target_alignment): Refine for alignment.
24523 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
24524 (FUNCTION_BOUNDARY): Modify.
24525 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
24527 * config/nds32/nds32.opt (malways-align, malign-functions): New.
24529 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
24531 * config/nds32/constants.md (unspec_volatile_element): Add values for
24532 TLB operation and data prefetch.
24533 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
24534 functions for TLB operation and data prefetch.
24535 * config/nds32/nds32-intrinsic.md: Likewise.
24536 * config/nds32/nds32_intrinsic.h: Likewise.
24537 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
24538 (nds32_print_operand): Likewise.
24539 * config/nds32/nds32.h (nds32_builtins): Likewise.
24541 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
24542 Andrew Pinski <pinsika@gcc.gnu.org>
24544 PR middle-end/82976
24545 * match.pd: Use constant_boolean_node of correct type instead of
24546 boolean_true_node or boolean_false_node for simplifying
24547 pointer comparisons to zero.
24549 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24551 PR tree-optimization/80021
24552 * tree.c (verify_type_variant): Make error call in verify_variant_match
24553 translatable and remove final full stop.
24555 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24557 * config/nds32/constants.md (unspec_volatile_element): Add
24558 UNSPEC_VOLATILE_EH_RETURN.
24559 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
24560 nds32_output_stack_pop): Support dwarf exception handling process.
24561 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
24562 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
24563 exception handling process.
24564 (nds32_compute_stack_frame): Likewise.
24565 (nds32_return_addr_rtx): Likewise.
24566 (nds32_initial_elimination_offset): Likewise.
24567 (nds32_expand_prologue): Likewise.
24568 (nds32_expand_epilogue): Likewise.
24569 (nds32_dynamic_chain_address): New function.
24570 * config/nds32/nds32.h (machine_function): Add fields for dwarf
24571 exception handling.
24572 (DYNAMIC_CHAIN_ADDRESS): Define.
24573 (EH_RETURN_DATA_REGNO): Define.
24574 (EH_RETURN_STACKADJ_RTX): Define.
24575 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
24576 patterns for dwarf exception handling.
24578 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24580 * config/nds32/nds32.h: Clean up obsolete macros.
24582 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24584 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24585 Add enum values for particular instructions.
24586 * config/nds32/nds32-intrinsic.c: Implementation of expanding
24587 particular intrinsic functions.
24588 * config/nds32/nds32-intrinsic.md: Likewise.
24589 * config/nds32/nds32_intrinsic.h: Likewise.
24590 * config/nds32/nds32.h (nds32_builtins): Likewise.
24591 * config/nds32/nds32.md (type): Add pbsad and pbsada.
24592 (btst, ave): New patterns for particular instructions.
24594 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24596 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
24597 Add enum values for atomic load/store and memory sync.
24598 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
24600 * config/nds32/nds32-intrinsic.md: Likewise.
24601 * config/nds32/nds32_intrinsic.h: Likewise.
24602 * config/nds32/nds32.h (nds32_builtins): Likewise.
24604 2018-04-07 Jakub Jelinek <jakub@redhat.com>
24606 PR tree-optimization/85257
24607 * fold-const.c (native_encode_vector): If not all elts could fit
24608 and off is -1, return 0 rather than offset.
24609 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
24610 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
24611 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
24612 adjust buffer in native_interpret_expr call.
24614 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24616 * config/nds32/constants.md (unspec_volatile_element): Add cache
24617 control enum values.
24618 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
24619 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
24620 * config/nds32/nds32.c (nds32_cctl_names): New.
24621 (nds32_print_operand): Handle cache control register names.
24622 * config/nds32/nds32.h (nds32_builtins): New enum values.
24623 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
24625 * config/nds32/nds32.md (type): Add mmu.
24626 * config/nds32/pipelines.md (simple_insn): Add mmu.
24628 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24630 * config/nds32/nds32.md (type): Remove call.
24631 * config/nds32/pipelines.md (simple_insn): Likewise.
24633 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24635 * config/nds32/constants.md (unspec_volatile_element): Add
24636 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
24637 UNSPEC_VOLATILE_FMFCFG.
24638 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
24639 description for fmfcfg and fmfcsr.
24640 (bdesc_1arg): Add fmtcsr.
24641 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
24642 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
24643 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
24644 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
24645 unspec_fmfcfg): New patterns.
24646 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
24647 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
24648 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
24649 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
24650 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
24651 __nds32__fmfcfg): Define.
24653 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
24655 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
24656 intrinsic register names.
24657 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
24658 intrinsic register enum values and macros.
24660 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
24662 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
24663 for load/store addressing form.
24664 (nds32_print_operand_address): Likewise.
24666 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
24669 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
24670 based on LABEL_REF. Remove useless assertion.
24671 (pic_address_needs_scratch): Fix formatting.
24672 (sparc_legitimize_pic_address): Minor tweaks.
24673 (sparc_delegitimize_address): Adjust assertion accordingly.
24674 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
24675 into symbolic_operand.
24676 (movsi_high_pic_label_ref): Likewise.
24677 (movsi_lo_sum_pic_label_ref): Likewise.
24678 (movdi_pic_label_ref): Likewise.
24679 (movdi_high_pic_label_ref): Likewise.
24680 (movdi_lo_sum_pic_label_ref): Likewise.
24682 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
24684 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
24685 custom LIB_SPEC setup.
24687 2018-04-06 Ruslan Bukin <br@bsdpad.com>
24688 Kito Cheng <kito.cheng@gmail.com>
24690 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
24691 * config/riscv/freebsd.h: New.
24693 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24695 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
24696 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
24699 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24700 Kito Cheng <kito.cheng@gmail.com>
24702 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
24703 nds32_output_call, nds32_symbol_binds_local_p): New functions.
24704 * config/nds32/nds32-protos.h (nds32_output_call,
24705 nds32_output_return): Declare.
24706 * config/nds32/nds32.md: Refine all the call and return patterns.
24708 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24711 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
24712 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
24714 PR rtl-optimization/84872
24715 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
24716 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
24717 EDGE_CROSSING edge.
24719 2018-04-06 Tamar Christina <tamar.christina@arm.com>
24721 * expr.c (copy_blkmode_to_reg): Revert 254862.
24722 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
24724 2018-04-06 Richard Biener <rguenther@suse.de>
24726 PR middle-end/85244
24727 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
24728 after seeing a component reference with an adjacent field. Treat
24729 refs to arrays at struct end of external decls similar to
24730 refs to unconstrained commons.
24732 2018-04-06 Jakub Jelinek <jakub@redhat.com>
24735 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
24736 look through SAVE_EXPRs with non-side-effects argument. Adjust
24738 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
24741 2018-04-06 Richard Biener <rguenther@suse.de>
24743 PR middle-end/85180
24744 * alias.c (find_base_term): New wrapper around find_base_term
24745 unwinding CSELIB_VAL_PTR changes.
24746 (find_base_term): Do not restore CSELIB_VAL_PTR during the
24749 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24751 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
24753 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
24754 constant definitions.
24755 ("nop"): lr 0,0 -> nopr r0
24756 ("nop_lr0", "nop_lr1"): New insn definitions.
24758 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
24760 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
24761 NDS32_V3PUSH_AVAILABLE_P macro.
24763 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
24764 Chung-Ju Wu <jasonwucj@gmail.com>
24766 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
24767 (nds32*-*-*): Add float and fpu_config into supported_defaults.
24768 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
24769 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
24770 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
24771 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
24772 * config/nds32/constraints.md: New constraints and checking for hard
24773 float configuration.
24774 * config/nds32/iterators.md: New mode iterator and attribute for hard
24775 float configuration.
24776 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
24778 * config/nds32/nds32-fpu.md: New file.
24779 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
24780 deal with hard float code generation.
24781 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
24783 (abi_type, float_reg_number): New enum type.
24784 * config/nds32/nds32-predicates.c: New predicates for hard float.
24785 * config/nds32/nds32-protos.h: Declare functions for hard float.
24786 * config/nds32/nds32.c: Implementation for hard float configuration.
24787 * config/nds32/nds32.h: Definitions for hard float configuration.
24788 * config/nds32/nds32.md: Include hard float machine description and
24789 modify patterns for hard float configuration.
24790 * config/nds32/nds32.opt: New options for hard float configuration.
24791 * config/nds32/predicates.md: New predicates for hard float
24794 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24796 * common/config/nds32/nds32-common.c
24797 (nds32_option_optimization_table): Enable -mreleax-hint by default.
24799 2018-04-05 Jakub Jelinek <jakub@redhat.com>
24801 PR middle-end/85195
24802 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
24803 CONSTRUCTOR_ELT (ctor, ...)->value.
24805 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
24808 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
24810 2018-04-05 Tom de Vries <tom@codesourcery.com>
24813 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
24816 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24817 Kito Cheng <kito.cheng@gmail.com>
24819 * config/nds32/constraints.md (U33): Fine-tune checking condition.
24820 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
24821 * config/nds32/nds32.h (nds32_16bit_address_type): Add
24822 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
24824 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24825 Kito Cheng <kito.cheng@gmail.com>
24827 * config/nds32/constraints.md (Ufe): New memory constraint.
24828 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
24829 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
24830 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
24832 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
24833 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
24835 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24837 * config/nds32/nds32.md: Use optimize_size in the condition for
24838 alu-shift instructions.
24840 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24842 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
24844 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24846 * config/nds32/nds32.md (negsi2): Refine pattern.
24848 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24849 Chung-Ju Wu <jasonwucj@gmail.com>
24851 * config/nds32/iterators.md (shift_rotate): New code iterator.
24852 (shift): New code attribute.
24853 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
24854 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
24855 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
24856 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
24857 bit-wise operations.
24858 (andsi3, *andsi3): Ditto.
24859 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
24860 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
24861 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
24862 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
24863 nds32_ior_operand, nds32_xor_operand): New predicates.
24865 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24867 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
24868 (addsi3, subsi3): ... this.
24870 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24872 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
24874 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24876 * config/nds32/nds32.md: Adjust indention.
24878 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24880 * config/nds32/nds32.md (feature): New attribute.
24882 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24884 * config/nds32/nds32.md (subtype): New attribute.
24886 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24889 * config/arm/arm-builtins.c (arm_expand_builtin): Change
24890 expansion to perform a bitwise AND of the argument followed by a
24891 boolean negation of the result.
24893 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
24895 PR rtl-optimization/84878
24896 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
24897 the basic block. Assert the use reference is not artificial and that
24898 it has an associated insn.
24900 2018-04-04 Michael Matz <matz@suse.de>
24902 * builtins.c (compute_objsize): Pass correct operand
24903 to array_at_struct_end_p.
24905 2018-04-04 Richard Biener <rguenther@suse.de>
24908 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
24909 from contexts for DINFO_LEVEL_TERSE and below.
24911 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24913 * config/nds32/nds32-doubleword.md (move_<mode>): Require
24914 resiter_operand condition.
24915 * config/nds32/nds32.md (*move<mode>): Ditto.
24917 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24918 Monk Chiang <sh.chiang04@gmail.com>
24920 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
24922 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24924 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
24926 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24927 Kito Cheng <kito.cheng@gmail.com>
24929 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
24930 nds32_cond_code_str, output_cond_branch,
24931 output_cond_branch_compare_zero, nds32_expand_cbranch,
24932 nds32_expand_cstore, nds32_expand_movcc,
24933 nds32_output_cbranchsi4_equality_zero,
24934 nds32_output_cbranchsi4_equality_reg,
24935 nds32_output_cbranchsi4_equality_reg_or_const_int,
24936 nds32_output_cbranchsi4_greater_less_zero: New functions.
24937 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
24938 nds32_expand_cstore, nds32_expand_movcc,
24939 nds32_output_cbranchsi4_equality_zero,
24940 nds32_output_cbranchsi4_equality_reg,
24941 nds32_output_cbranchsi4_equality_reg_or_const_int,
24942 nds32_output_cbranchsi4_greater_less_zero): Declare.
24943 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
24944 nds32_rimm11s_operand): New predicates.
24945 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
24946 * config/nds32/nds32.md: Rewrite all the branch and conditional move
24949 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24951 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
24952 * config/nds32/nds32.md: Ditto.
24953 * config/nds32/pipelines.md: Ditto.
24955 2018-04-04 Richard Biener <rguenther@suse.de>
24957 PR tree-optimization/85168
24958 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
24959 propagating abnormals.
24961 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24963 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
24965 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24966 Kito Cheng <kito.cheng@gmail.com>
24968 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
24969 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
24970 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
24971 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
24972 * config/nds32/nds32.md (sibcall_internal): New.
24973 (sibcall_register): Remove.
24974 (sibcall_immediate): Remove.
24975 (sibcall_value_internal): New.
24976 (sibcall_value_register): Remove.
24977 (sibcall_value_immediate): Remove.
24978 * config/nds32/predicates.md (nds32_general_register_operand): New.
24979 (nds32_call_address_operand): New.
24981 2018-04-03 Jakub Jelinek <jakub@redhat.com>
24983 PR rtl-optimization/85167
24984 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
24985 bb_defs if *split_p, instead preinitialize it to NULL.
24987 PR tree-optimization/85156
24988 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
24989 evaluating the argument multiple times.
24991 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
24993 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
24995 (_mm_cvtpd_ps): Likewise.
24996 (_mm_cvttpd_epi32): Likewise.
24997 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
24998 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
24999 vector, pixel, and bool following altivec.h include.
25001 2018-04-03 Martin Sebor <msebor@redhat.com>
25003 * doc/extend.texi (Common Function Attributes): Clarify.
25004 (const attribute): Likewise.
25005 (pure attribute): Likewise.
25007 2018-04-03 Jakub Jelinek <jakub@redhat.com>
25010 * config/i386/i386.c (ix86_expand_vector_set): Use
25011 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
25013 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
25015 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
25016 instructions when changing rounding bits to preserve precision bits
25017 in the x87 control word.
25019 2018-04-03 Martin Liska <mliska@suse.cz>
25021 PR tree-optimization/82491
25022 * rtl.h (strip_offset_and_add): Replace += suboffset with
25023 poly_uint64 () + suboffset.
25025 2018-03-29 Martin Liska <mliska@suse.cz>
25026 Martin Jambor <mjambor@suse.cz>
25029 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
25030 param_type is not an integral or pointer type.
25032 2018-04-03 Richard Biener <rguenther@suse.de>
25034 * sese.h (recompute_all_dominators): Remove.
25036 2018-04-02 Martin Sebor <msebor@redhat.com>
25038 * doc/invoke.texi (-Wrestrict): Fix typos.
25040 2018-04-02 Jim Wilson <jimw@sifive.com>
25042 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
25043 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
25044 (<optab>di3, <optab>si3_extend): Likewise.
25045 (<optab>si3_mask, <optab>si3_mask_1): New.
25046 (<optab>di3_mask, <optab>di3_mask_1): New.
25047 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
25048 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
25049 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
25051 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
25053 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
25056 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
25058 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
25059 (nds32_canonicalize_comparison): New function.
25061 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25062 Kito Cheng <kito.cheng@gmail.com>
25063 Kuan-Lin Chen <kuanlinchentw@gmail.com>
25065 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
25066 * config/nds32/constants.md (unspec_volatile_element): Add
25067 UNSPEC_VOLATILE_RELAX_GROUP.
25068 * config/nds32/nds32-relax-opt.c: New file.
25069 * config/nds32/nds32-predicates.c
25070 (nds32_symbol_load_store_p): New function.
25071 * config/nds32/nds32-protos.h
25072 (nds32_symbol_load_store_p): Declare function.
25073 (make_pass_nds32_relax_opt): Declare new rtl pass function.
25074 * config/nds32/nds32.c
25075 (nds32_register_pass): New function to register pass.
25076 (nds32_register_passes): New function to register passes.
25077 * config/nds32/nds32.md (relax_group): New pattern.
25078 * config/nds32/nds32.opt (mrelax-hint): New option.
25079 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
25081 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
25083 * config/nds32/t-nds32: Modify files dependency.
25085 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25087 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
25088 (PROFILE_HOOK): Define its implementation.
25090 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
25092 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
25093 type and 32-bit size.
25095 2018-04-01 Jakub Jelinek <jakub@redhat.com>
25097 PR middle-end/85090
25098 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
25099 (V_128_256): New mode iterator.
25100 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
25101 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
25102 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
25104 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
25105 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
25107 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25110 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
25111 NaN inputs correctly.
25113 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
25116 * config/rs6000/vsx.md (??r): New mode attribute.
25117 (*vsx_mov<mode>_64bit): Use it.
25118 (*vsx_mov<mode>_32bit): Likewise.
25120 2018-03-30 Martin Sebor <msebor@redhat.com>
25122 PR tree-optimization/84818
25123 * builtins.c (check_access): Use warning_n.
25125 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25128 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
25130 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
25133 2018-03-30 Julia Koval <julia.koval@intel.com>
25136 * x86-tune.def (movx, partial_reg_dependency): Enable for
25139 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
25141 PR inline-asm/84985
25142 * lra-constraints.c (process_alt_operands): Move setting
25143 this_alternative_matches below.
25145 2018-03-29 Martin Liska <mliska@suse.cz>
25148 * doc/invoke.texi: Document how LTO works with debug info.
25149 Describe auto-load support of binutils. Mention 'x86-64'
25150 as valid option value of -march option.
25152 2018-03-29 Jakub Jelinek <jakub@redhat.com>
25154 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
25157 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
25158 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
25159 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
25162 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
25165 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
25166 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
25167 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
25168 for RS6000_BTM_POWERPC64.
25169 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
25170 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
25171 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
25174 (DIVDEU): Likewise.
25176 2018-03-28 Carl Love <cel@us.ibm.com>
25179 2017-09-27 Carl Love <cel@us.ibm.com>
25181 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
25182 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
25183 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
25186 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25188 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
25189 instead of __vector bool.
25190 (_mm_max_pu8): Likewise.
25191 (_mm_min_pi16): Likewise.
25193 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
25196 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
25197 (DIVWEUO): Likewise.
25198 (DIVDEO): Likewise.
25199 (DIVDEUO): Likewise.
25200 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
25201 DIVWEUO and DIVDEUO.
25202 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
25203 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
25204 (div_extend): Likewise.
25205 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
25207 (__builtin_divweuo): Likewise.
25208 (__builtin_divdeo): Likewise.
25209 (__builtin_divdeuo): Likewise.
25211 2018-03-28 Jakub Jelinek <jakub@redhat.com>
25214 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
25215 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
25217 PR tree-optimization/82004
25218 * gimple-match-head.c (optimize_pow_to_exp): New function.
25219 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
25220 Don't fold to exp if optimize_pow_to_exp is false.
25222 2018-03-28 Martin Liska <mliska@suse.cz>
25225 * calls.c (initialize_argument_information): Fix trailing space.
25226 * common.opt: Fix typo and provide better explanation for
25227 -fsanitize-coverage option.
25228 * config/i386/i386.opt: Fix typo.
25230 2018-03-28 Jakub Jelinek <jakub@redhat.com>
25231 Martin Liska <mliska@suse.cz>
25234 * gimplify.c (asan_poison_variable): Don't do the check for
25235 gimplify_omp_ctxp here.
25236 (gimplify_decl_expr): Do it here.
25237 (gimplify_target_expr): Likewise.
25239 2018-03-28 Martin Liska <mliska@suse.cz>
25242 * config/i386/i386.c (ix86_function_arg_advance): Do not call
25243 chkp_type_bounds_count if MPX is not enabled.
25245 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
25247 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
25249 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25252 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
25253 function to create the function decl for complex long double
25254 multiply and divide for -mabi=ieeelongdouble.
25255 (init_float128_ieee): Call it.
25257 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
25260 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
25261 -fcf-protection=branch -mibt.
25262 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
25264 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25267 * config/arm/arm.c (arm_valid_symbolic_address): Handle
25268 arm_word_relocations.
25270 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
25273 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
25274 extern array declarations.
25276 2018-03-27 Richard Biener <rguenther@suse.de>
25278 PR middle-end/84067
25279 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
25280 explicit single_use checks.
25282 2018-03-27 Richard Biener <rguenther@suse.de>
25284 PR tree-optimization/85082
25285 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
25288 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25290 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
25291 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
25292 Turn on fasynchronous-unwind-tables and funwind-tables.
25294 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
25297 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
25298 (*bmi_blsr_<mode>_ccz): Ditto.
25300 2018-03-26 Tom de Vries <tom@codesourcery.com>
25302 PR tree-optimization/85063
25303 * omp-general.c (offloading_function_p): New function. Factor out
25305 * omp-offload.c (pass_omp_target_link::gate): ... here.
25306 * omp-general.h (offloading_function_p): Declare.
25307 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
25308 with attribute omp declare target for offloading functions.
25310 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
25312 PR tree-optimization/84005
25313 * tree-data-ref.h (get_base_for_alignment): Declare.
25314 * tree-data-ref.c (get_base_for_alignment_1): New function.
25315 (get_base_for_alignment): Likewise.
25316 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
25317 get_base_for_alignment to find a suitable base object, instead
25318 of always using drb->base_address.
25320 2018-03-23 Jakub Jelinek <jakub@redhat.com>
25322 PR inline-asm/85022
25323 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
25324 known size by default.
25326 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
25328 PR inline-asm/85030
25329 * lra-constraints.c (process_alt_operands): Don't match BLKmode
25330 and non BLKmode operands.
25332 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25335 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
25336 Clean up attributes.
25338 2018-03-23 Richard Biener <rguenther@suse.de>
25341 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
25342 we are going to emit early debug for LTO.
25344 2018-03-23 Jakub Jelinek <jakub@redhat.com>
25346 PR inline-asm/85034
25347 * function.c (match_asm_constraints_1): Don't optimize if input
25348 doesn't satisfy general_operand predicate for output's mode.
25350 PR inline-asm/85022
25351 * alias.c (write_dependence_p): Don't require for x_canonicalized
25352 non-VOIDmode if x has VOIDmode.
25355 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
25356 just don't try to optimize it rather than assert it never happens.
25358 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
25360 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
25361 macro expansions for definition of ST_INTERNAL_<mode> and
25362 LD_INTERNAL_<mode> builtins.
25363 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
25365 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
25367 (altivec_expand_st_builtin): Likewise.
25368 (altivec_expand_builtin): Remove calls to deleted functions.
25369 (rs6000_address_for_altivec): Delete this function.
25370 * config/rs6000/vector.md: Remove expands for
25371 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
25373 2018-03-22 Sudakshina Das <sudi.das@arm.com>
25376 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
25377 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
25378 re-computing once computed.
25379 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
25380 (arm_init_machine_status): Initialize
25381 machine->static_chain_stack_bytes.
25383 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
25386 * doc/extend.texi: Add four new prototypes for vec_ld.
25387 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
25388 definitions for more logical presentation.
25389 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
25390 entries for V1TI variants of __builtin_altivec_ld builtin.
25391 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
25392 handling of V1TI variant of LVX icode pattern.
25393 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
25394 (rs6000_gimple_fold_builtin): Likewise.
25395 (altivec_init_builtins): Add code to define
25396 __builtin_altivec_lvx_v1ti function.
25398 2018-03-22 Jakub Jelinek <jakub@redhat.com>
25400 PR inline-asm/84941
25401 * function.c (match_asm_constraints_1): Don't do the optimization
25402 if input isn't a REG, SUBREG, MEM or constant.
25404 2018-03-22 Tom de Vries <tom@codesourcery.com>
25406 PR tree-optimization/84956
25407 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
25408 bb_has_abnormal_pred.
25410 2018-03-22 Jakub Jelinek <jakub@redhat.com>
25413 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
25414 DECL_INITIAL (decl) to decl at the end.
25415 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
25416 adjust the comment.
25418 2018-03-21 Joseph Myers <joseph@codesourcery.com>
25420 * doc/extend.texi (__builtin_tgmath): Document when complex
25421 integer types are treated as _Complex _Float64.
25423 2018-03-21 Tom de Vries <tom@codesourcery.com>
25425 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
25427 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25429 PR tree-optimization/84960
25430 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
25431 if it is ENTRY block, move them into single succ of ENTRY in that case.
25433 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
25435 PR tree-optimization/84811
25436 * poly-int.h (poly_span_traits): Remove the T3 parameter and
25437 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
25438 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
25439 (known_subrange_p): Update accordingly. Cast each value involved
25440 in the size comparison, rather than casting the result of the
25443 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25445 PR tree-optimization/84982
25446 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
25447 by flipping the least significant bit rather than all bits from
25448 bitpos to bitpos + bitsize - 1.
25450 2018-03-21 Nathan Sidwell <nathan@acm.org>
25452 * doc/extend.texi (Deprecated Features): Remove mention of
25453 long-deleted deprecations.
25455 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25458 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
25459 * configure: Regenerate.
25461 2018-03-21 Tom de Vries <tom@codesourcery.com>
25463 PR tree-optimization/83126
25464 * tree-parloops.c (num_phis): New function.
25465 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
25467 2018-03-21 Nathan Sidwell <nathan@acm.org>
25469 * doc/extend.texi (Deprecated Features): Update deprecated flags,
25470 mention anon-struct/union members and trailing attributes.
25472 2018-03-21 Bin Cheng <bin.cheng@arm.com>
25474 PR tree-optimization/84969
25475 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
25476 builtin memset partitions if they set different rhs values.
25478 2018-03-21 Jakub Jelinek <jakub@redhat.com>
25480 PR rtl-optimization/84989
25481 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
25482 VEC_DUPLICATE with scalar result mode.
25484 2018-03-21 Martin Liska <mliska@suse.cz>
25487 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
25488 not intended return statement.
25490 2018-03-21 Martin Liska <mliska@suse.cz>
25493 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
25494 (chkp_find_bound_slots_1): Limit number of iterations.
25496 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
25499 * Minor grammar fixes for x86 options.
25501 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25504 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
25505 holding REG_CFA_RESTORE notes, instead turn them into a USE.
25507 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
25510 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
25511 (altivec_lvx_<mode>_1op): Likewise.
25512 (altivec_stvx_<mode>_2op): Likewise.
25513 (altivec_stvx_<mode>_1op): Likewise.
25514 (altivec_lvx_<VM2:mode>): New define_expand.
25515 (altivec_stvx_<VM2:mode>): Likewise.
25516 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
25517 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
25518 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
25519 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
25520 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
25521 (rs6000_gen_lvx): Likewise.
25522 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
25523 (altivec_expand_stv_builtin): Likewise.
25524 (altivec_expand_builtin): Likewise.
25525 * config/rs6000/vector.md: Likewise.
25527 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25530 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
25533 2018-03-20 Richard Biener <rguenther@suse.de>
25536 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
25537 sign-conversions as zero, fall back to standard scalar_stmt
25540 2018-03-20 Martin Liska <mliska@suse.cz>
25543 * predict.c (rebuild_frequencies): Handle case when we have
25544 PROFILE_ABSENT, but flag_guess_branch_prob is false.
25546 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25549 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
25550 flag_section_anchors.
25551 * varasm.c (use_blocks_for_decl_p): Remove hack for
25552 dw2_force_const_mem.
25555 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
25557 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
25558 be created, use lowpart_subreg of operands[0] rather than operands[0]
25560 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
25561 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
25562 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
25563 and n constraint instead of aarch64_shift_imm_di and Usd.
25564 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
25565 (*aarch64_<optab>_reg_minus<mode>3): ... this.
25567 2018-03-20 Sudakshina Das <sudi.das@arm.com>
25570 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
25571 to favor GPR over NEON registers.
25572 (<shift>di3_neon): Likewise.
25574 2018-03-20 Tom de Vries <tom@codesourcery.com>
25577 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
25578 (nvptx_process_pars): Emit bar.sync asap and alap.
25580 2018-03-20 Tom de Vries <tom@codesourcery.com>
25583 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
25584 seen_label if seen_label is already set.
25586 2018-03-20 Jakub Jelinek <jakub@redhat.com>
25589 * config/i386/i386.c (fold_builtin_cpu): For features above 31
25590 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
25591 Use 1U instead of 1. Formatting fixes.
25594 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
25595 instead of TREE_TYPE (s1) for the return value.
25597 2018-03-19 Jakub Jelinek <jakub@redhat.com>
25599 PR tree-optimization/84946
25600 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
25601 bitsize + bitsize in poly_uint64 rather than poly_int64.
25604 * dwarf2asm.c: Include fold-const.c.
25605 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
25606 of decl rather than decl itself.
25608 PR rtl-optimization/84643
25609 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
25611 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
25614 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
25615 calling assemble_variable.
25617 2018-03-19 Sudakshina Das <sudi.das@arm.com>
25620 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
25621 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
25623 2018-03-19 Jim Wilson <jimw@sifive.com>
25626 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
25627 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
25628 (riscv_first_stack_step): Likewise.
25629 (riscv_option_override): Use STACK_BOUNDARY instead of
25630 MIN_STACK_BOUNDARY.
25631 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
25632 MIN_STACK_BOUNDARY.
25633 (BIGGEST_ALIGNMENT): Set to 128.
25634 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
25635 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
25638 2018-03-19 Richard Biener <rguenther@suse.de>
25640 PR tree-optimization/84933
25641 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
25642 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
25644 2018-03-19 Richard Biener <rguenther@suse.de>
25646 PR tree-optimization/84859
25647 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
25648 (cond_if_else_store_replacement): Perform sinking operation on
25649 single-store BBs regardless of MAX_STORES_TO_SINK setting.
25650 Generalize what a BB with a single eligible store is.
25652 2018-03-19 Richard Biener <rguenther@suse.de>
25654 PR tree-optimization/84929
25655 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
25656 chrec_is_positive against non-chrec arg.
25658 2018-03-19 Tamar Christina <tamar.christina@arm.com>
25661 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
25663 2018-03-18 Martin Liska <mliska@suse.cz>
25665 PR rtl-optimization/84635
25666 * regrename.c (build_def_use): Use matches_mode only when
25669 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
25671 PR tree-optimization/84913
25672 * tree-vect-loop.c (vectorizable_reduction): Don't try to
25673 vectorize chains of COND_EXPRs.
25675 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25677 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
25679 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25681 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
25683 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
25685 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
25687 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25688 Kito Cheng <kito.cheng@gmail.com>
25690 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
25691 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
25692 (nds32_adjust_reg_alloc_order): New function.
25693 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
25695 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
25697 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
25698 nds32_print_operand, nds32_print_operand_address): Use
25699 HOST_WIDE_INT_PRINT_DEC instead.
25701 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
25703 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
25705 2018-03-17 Jakub Jelinek <jakub@redhat.com>
25708 * config/i386/i386.c (initial_ix86_tune_features,
25709 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
25710 unsigned long long.
25711 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
25712 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
25713 rather than 1u << ix86_tune. Formatting fix.
25714 (ix86_option_override_internal): Change ix86_arch_mask from
25715 unsigned int to unsigned HOST_WIDE_INT, initialize to
25716 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
25717 (ix86_function_specific_restore): Likewise.
25719 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25722 * postreload.c (reload_combine_recognize_pattern): Perform
25723 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
25724 truncate_int_for_mode the result for the destination's mode.
25727 * hsa-gen.c (mem_type_for_type): Fix comment typo.
25728 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
25730 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
25733 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
25736 * lra-assigns.c (lra_split_hard_reg_for): Don't use
25737 regno_allocno_class_array and sorted_pseudos.
25738 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
25739 insns where regno is used.
25741 2018-03-16 Martin Liska <mliska@suse.cz>
25744 * multiple_target.c (create_dispatcher_calls): Redirect
25745 reference in the symbol table.
25747 2018-03-16 Martin Liska <mliska@suse.cz>
25750 * multiple_target.c (create_dispatcher_calls): Redirect also
25753 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25757 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
25758 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
25759 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
25761 2018-03-16 Julia Koval <julia.koval@intel.com>
25763 * doc/invoke.texi (Skylake Server): Add CLWB.
25764 Cannonlake): Remove CLWB.
25766 2018-03-16 Jakub Jelinek <jakub@redhat.com>
25768 PR tree-optimization/84841
25769 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
25771 (FLOAT_ONE_CONST_TYPE): Define.
25772 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
25773 (sort_by_operand_rank): Put entries with higher constant_type last
25774 rather than first to match comments.
25776 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
25778 * config/nios2/nios2.md (movsi_internal): Fix thinko in
25781 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25784 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
25785 check and use address of target always.
25787 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
25790 * config/i386/i386.c (indirect_thunk_needed): Update comments.
25791 (indirect_thunk_bnd_needed): Likewise.
25792 (indirect_thunks_used): Likewise.
25793 (indirect_thunks_bnd_used): Likewise.
25794 (indirect_return_needed): New.
25795 (indirect_return_bnd_needed): Likewise.
25796 (output_indirect_thunk_function): Add a bool argument for
25798 (output_indirect_thunk_function): Don't generate alias for
25799 function return thunk.
25800 (ix86_code_end): Call output_indirect_thunk_function to generate
25801 function return thunks.
25802 (ix86_output_function_return): Set indirect_return_bnd_needed
25803 and indirect_return_needed instead of indirect_thunk_bnd_needed
25804 and indirect_thunk_needed.
25806 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
25808 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
25809 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
25810 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
25812 2018-03-15 David Malcolm <dmalcolm@redhat.com>
25813 Paul Hua <paul.hua.gm@gmail.com>
25816 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
25818 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
25820 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
25821 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
25822 resp. SFmode cases.
25824 2018-03-15 Tamar Christina <tamar.christina@arm.com>
25827 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
25828 instead of GET_MODE_SIZE when comparing Units.
25830 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
25833 * varasm.c (hash_section): Return an unchangeble hash value
25834 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
25835 Return !aarch64_can_use_per_function_literal_pools_p ().
25837 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25840 * optabs.c (emit_conditional_move): Pass address of cmode's copy
25841 rather than address of cmode as last argument to prepare_cmp_insn.
25843 2018-03-15 Julia Koval <julia.koval@intel.com>
25845 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
25846 F_AVX512VNNI, F_AVX512BITALG): New.
25848 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
25851 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
25852 insn for floating-point loads and stores.
25854 2018-03-14 Carl Love <cel@us.ibm.com>
25856 * config/rs6000/rs6000-c.c: Add macro definitions for
25857 ALTIVEC_BUILTIN_VEC_PERMXOR.
25858 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
25859 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
25860 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
25861 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
25863 * config/doc/extend.texi: Add prototypes for vec_permxor.
25865 2018-03-14 David Malcolm <dmalcolm@redhat.com>
25868 * diagnostic-show-locus.c (class layout_point): Convert m_line
25869 from int to linenum_type.
25870 (line_span::comparator): Use linenum "compare" function when
25871 comparing line numbers.
25872 (test_line_span): New function.
25873 (layout_range::contains_point): Convert param "row" from int to
25875 (layout_range::intersects_line_p): Likewise.
25876 (layout::will_show_line_p): Likewise.
25877 (layout::print_source_line): Likewise.
25878 (layout::should_print_annotation_line_p): Likewise.
25879 (layout::print_annotation_line): Likewise.
25880 (layout::print_leading_fixits): Likewise.
25881 (layout::annotation_line_showed_range_p): Likewise.
25882 (struct line_corrections): Likewise for field m_row.
25883 (line_corrections::line_corrections): Likewise for param "row".
25884 (layout::print_trailing_fixits): Likewise.
25885 (layout::get_state_at_point): Likewise.
25886 (layout::get_x_bound_for_row): Likewise.
25887 (layout::print_line): Likewise.
25888 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
25889 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
25890 * input.c (selftest::test_linenum_comparisons): New function.
25891 (selftest::input_c_tests): Call it.
25892 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
25893 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
25894 * selftest.h (ASSERT_GT): New macro.
25895 (ASSERT_GT_AT): New macro.
25896 (ASSERT_LT): New macro.
25897 (ASSERT_LT_AT): New macro.
25899 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
25901 PR rtl-optimization/84780
25902 * combine.c (distribute_links): Don't make a link based on pc_rtx.
25904 2018-03-14 Martin Liska <mliska@suse.cz>
25906 * tree.c (record_node_allocation_statistics): Use
25907 get_stats_node_kind.
25908 (get_stats_node_kind): New function extracted from
25909 record_node_allocation_statistics.
25910 (free_node): Use get_stats_node_kind.
25912 2018-03-14 Richard Biener <rguenther@suse.de>
25914 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
25915 that the value-set of ANTIC_IN doesn't grow.
25918 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
25920 (BB_VISITED_WITH_VISITED_SUCCS): New define.
25921 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
25923 2018-03-14 Julia Koval <julia.koval@intel.com>
25925 * config.gcc (icelake-client, icelake-server): New.
25927 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
25928 (initial_ix86_arch_features): Ditto.
25929 (PTA_SKYLAKE): Add SGX.
25930 (PTA_ICELAKE): Remove.
25931 (PTA_ICELAKE_CLIENT): New.
25932 (PTA_ICELAKE_SERVER): New.
25933 (ix86_option_override_internal): Split up icelake on icelake client and
25935 (get_builtin_code_for_version): Ditto.
25936 (fold_builtin_cpu): Ditto.
25937 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
25938 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
25939 * config/i386/i386.h (processor_type): Ditto.
25940 * doc/invoke.texi: Ditto.
25942 2018-03-14 Jakub Jelinek <jakub@redhat.com>
25945 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
25946 INTEGER_CST offset, add it together with bitpos / 8 and
25947 sign extend based on POINTER_SIZE.
25951 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25954 * config/i386/constraints.md (Yc): New register constraint.
25955 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25956 Use Yc constraint for alternative 2 of operand 0. Remove
25957 preferred_for_speed attribute.
25959 2018-03-14 Richard Biener <rguenther@suse.de>
25961 PR tree-optimization/84830
25962 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
25963 with the old one to avoid oscillations.
25965 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
25968 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
25970 (assign_by_spills): Return a flag of reload assignment failure.
25971 Do not process the reload assignment failures. Do not spill other
25972 reload pseudos if they has the same reg class. Update n if
25974 (lra_assign): Add a return arg. Set up from the result of
25975 assign_by_spills call.
25976 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25977 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25978 usage_insns if it is not NULL.
25979 (spill_hard_reg_in_range): New function.
25980 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25981 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25982 function prototypes.
25983 (lra_assign): Change prototype.
25984 * lra.c (lra): Add code to deal with fails by splitting hard reg
25987 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
25989 * config/riscv/riscv.opt (mrelax): New option.
25990 * config/riscv/riscv.c (riscv_file_start): Emit ".option
25991 "norelax" when riscv_mrelax is disabled.
25992 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
25994 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25997 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
25998 reassociation for int modes.
26000 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26002 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
26003 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
26005 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
26006 * config/aarch64/aarch64-sve.md
26007 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
26008 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
26009 (*extend<mode><Vwide>2): Rename to...
26010 (aarch64_sve_extend<mode><Vwide>2): ...this.
26011 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
26012 renaming the old pattern to...
26013 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
26015 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
26016 define_expand, renaming the old pattern to...
26017 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
26018 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
26019 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
26020 account when deciding which SVE instruction the optab should use.
26021 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
26023 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26025 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
26026 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
26027 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
26028 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
26029 (tlsdesc_small_<mode>): Turn a define_expand and use
26030 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
26031 (tlsdesc_small_advsimd_<mode>): ...this.
26032 (tlsdesc_small_sve_<mode>): New pattern.
26034 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
26036 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
26037 (UNSPEC_UMUL_HIGHPART): New constants.
26038 (MUL_HIGHPART): New int iteraor.
26039 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
26040 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
26042 (*<su>mul<mode>3_highpart): New define_insn.
26044 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
26047 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
26050 2018-03-13 Martin Liska <mliska@suse.cz>
26053 * (sem_item_optimizer::sem_item_optimizer): Initialize new
26055 (sem_item_optimizer::~sem_item_optimizer): Release it.
26056 (sem_item_optimizer::merge_classes): Register variable aliases.
26057 (sem_item_optimizer::fixup_pt_set): New function.
26058 (sem_item_optimizer::fixup_points_to_sets): Likewise.
26059 * ipa-icf.h: Declare new variables and functions.
26061 2018-03-13 Jakub Jelinek <jakub@redhat.com>
26063 PR middle-end/84834
26064 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
26065 integer_pow2p@2 and test integer_pow2p in condition.
26066 (A < 0 ? C : 0): Similarly for @1.
26068 PR middle-end/84831
26069 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
26070 characters starting at p contain '\0' character, don't look beyond
26074 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
26075 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
26078 * reg-stack.c (change_stack): Change update_end var from int to
26079 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
26080 also call set_block_for_insn on the newly added insns and rescan.
26083 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
26084 on the last operand.
26087 * tree.c (stabilize_reference_1): Return save_expr (e) for
26088 STATEMENT_LIST even if it doesn't have side-effects.
26090 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
26092 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
26094 2018-03-12 Renlin Li <renlin.li@arm.com>
26096 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
26097 aarch64_output_scalar_simd_mov_immediate.
26099 2018-03-12 Martin Sebor <msebor@redhat.com>
26101 PR tree-optimization/83456
26102 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
26103 for perfectly overlapping calls to memcpy.
26104 (gimple_fold_builtin_memory_chk): Same.
26105 (gimple_fold_builtin_strcpy): Handle no-warning.
26106 (gimple_fold_builtin_stxcpy_chk): Same.
26107 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
26109 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
26111 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
26112 parameter. Use it for SFmode.
26113 (rs6000_function_arg_advance_1): Adjust.
26114 (rs6000_function_arg): Adjust.
26115 (rs6000_gimplify_va_arg): Pass false for that new parameter.
26117 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
26119 PR rtl-optimization/84169
26120 PR rtl-optimization/84780
26121 * combine.c (can_combine_p): Check for a 2-insn combination whether
26122 the destination register is used between the two insns, too.
26124 2018-03-12 Richard Biener <rguenther@suse.de>
26126 PR tree-optimization/84803
26127 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
26128 for refs DR analysis didn't process.
26130 2018-03-12 Richard Biener <rguenther@suse.de>
26132 PR tree-optimization/84777
26133 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
26134 force-vectorize loops ignore whether we are optimizing for size.
26136 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
26138 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
26139 (TARGET_MD_ASM_ADJUST): Define.
26141 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
26142 Kito Cheng <kito.cheng@gmail.com>
26143 Chung-Ju Wu <jasonwucj@gmail.com>
26145 * config/nds32/nds32.c (nds32_compute_stack_frame,
26146 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
26147 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
26148 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
26149 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
26150 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
26151 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
26152 * config/nds32/nds32.md (prologue, epilogue): Use macro
26153 NDS32_V3PUSH_AVAILABLE_P to do checking.
26155 2018-03-11 Jakub Jelinek <jakub@redhat.com>
26158 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
26159 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
26160 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
26161 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
26162 addition of most attributes on !orig_type_die or the attribute not
26163 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
26165 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26166 Chung-Ju Wu <jasonwucj@gmail.com>
26168 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
26169 __NDS32_VH__ macro.
26170 * config/nds32/nds32.opt (mvh): New option.
26172 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26173 Chung-Ju Wu <jasonwucj@gmail.com>
26175 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
26177 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
26178 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
26181 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
26182 Chung-Ju Wu <jasonwucj@gmail.com>
26184 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
26186 * config/nds32/nds32-multiple.md (strlensi): New pattern.
26187 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
26189 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
26190 Kito Cheng <kito.cheng@gmail.com>
26191 Chung-Ju Wu <jasonwucj@gmail.com>
26193 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
26194 UNSPEC_FFMISM and UNSPEC_FLMISM.
26195 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
26196 for ffb, ffmism and flmism.
26197 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
26198 (unspec_ffmism): Ditto.
26199 (unspec_flmism): Ditto.
26200 (nds32_expand_builtin_impl): Check if string extension is available.
26201 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
26202 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
26204 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
26207 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
26210 * lra-assigns.c (assign_by_spills): Return a flag of reload
26211 assignment failure. Do not process the reload assignment
26212 failures. Do not spill other reload pseudos if they has the same
26214 (lra_assign): Add a return arg. Set up from the result of
26215 assign_by_spills call.
26216 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26217 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26218 usage_insns if it is not NULL.
26219 (spill_hard_reg_in_range): New function.
26220 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26221 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26222 function prototypes.
26223 (lra_assign): Change prototype.
26224 * lra.c (lra): Add code to deal with fails by splitting hard reg
26227 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
26230 * config/i386/i386.opt: Replace Enforcment with Enforcement.
26232 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
26235 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
26236 (dw_val_node): Add val_symbolic_view.
26237 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
26238 (symview_upper_bound): New.
26239 (new_line_info_table): Initialize symviews_since_reset.
26240 (dwarf2out_source_line): Count symviews_since_reset and set
26241 symview_upper_bound.
26242 (dw_val_equal_p): Handle symview.
26243 (add_AT_symview): New.
26244 (print_dw_val): Handle symview.
26245 (attr_checksum, attr_checksum_ordered): Likewise.
26246 (same_dw_val_p, size_of_die): Likewise.
26247 (value_format, output_die): Likewise.
26248 (add_high_low_attributes): Use add_AT_symview for entry_view.
26249 (dwarf2out_finish): Reset symview_upper_bound, clear
26252 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
26255 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
26256 Add strict argument and use it.
26257 (rs6000_split_multireg_move): Update for new strict argument.
26258 (mem_operand_gpr): Disallow all non-offsettable addresses.
26259 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
26261 2018-03-09 Jakub Jelinek <jakub@redhat.com>
26264 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
26265 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
26266 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
26269 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
26270 decl, use remap_type if we want to use the type.
26272 2018-03-09 Martin Sebor <msebor@redhat.com>
26274 PR tree-optimization/84526
26275 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
26277 (builtin_access::generic_overlap): Be prepared to handle non-array
26280 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
26282 PR rtl-optimization/84682
26283 * lra-constraints.c (process_address_1): Check is_address flag
26284 for address constraints.
26285 (process_alt_operands): Likewise.
26286 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
26287 preprocess_constraints.
26288 * recog.h (preprocess_constraints): Add oploc parameter.
26290 * recog.c (preprocess_constraints): Test address_operand for
26291 CT_ADDRESS constraints.
26293 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
26296 * lra-assigns.c (assign_by_spills): Return a flag of reload
26297 assignment failure. Do not process the reload assignment
26298 failures. Do not spill other reload pseudos if they has the same
26300 (lra_assign): Add a return arg. Set up from the result of
26301 assign_by_spills call.
26302 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
26303 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
26304 usage_insns if it is not NULL.
26305 (spill_hard_reg_in_range): New function.
26306 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
26307 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
26308 function prototypes.
26309 (lra_assign): Change prototype.
26310 * lra.c (lra): Add code to deal with fails by splitting hard reg
26313 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26316 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
26317 Accept complain bool parameter. Only emit errors if it is true.
26318 (arm_parse_cpu_option_name): Likewise.
26319 (arm_target_thumb_only): Adjust callers of the above.
26320 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
26321 prototype to take a default true bool parameter.
26322 (arm_parse_arch_option_name): Likewise.
26324 2018-03-09 David Malcolm <dmalcolm@redhat.com>
26325 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
26329 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
26330 * configure: Regenerate.
26331 * configure.ac ("linker --version-script option"): New.
26332 ("linker soname option"): New.
26334 2018-03-09 Richard Biener <rguenther@suse.de>
26336 PR tree-optimization/84775
26337 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
26338 immediate uses of predicate stmts and mark them modified.
26341 PR tree-optimization/84178
26342 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
26344 (version_loop_for_if_conversion): Delay update_ssa call.
26345 (tree_if_conversion): Delay update_ssa until after predicate
26348 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
26351 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
26352 when the function accesses prior frames.
26354 2018-03-08 Jakub Jelinek <jakub@redhat.com>
26357 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
26358 gen_llsym, otherwise call maybe_gen_llsym.
26360 PR inline-asm/84742
26361 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
26362 has ',' character inside of it.
26364 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26367 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
26368 as clobbering CC_REGNUM.
26370 2018-03-08 Richard Biener <rguenther@suse.de>
26372 PR middle-end/84552
26373 * tree-scalar-evolution.c: Include tree-into-ssa.h.
26374 (follow_copies_to_constant): Do not follow SSA names registered
26377 2018-03-08 Richard Biener <rguenther@suse.de>
26379 PR tree-optimization/84178
26380 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
26382 (version_loop_for_if_conversion): Delay update_ssa call.
26383 (tree_if_conversion): Delay update_ssa until after predicate
26386 2018-03-08 David Malcolm <dmalcolm@redhat.com>
26388 PR tree-optimization/84178
26389 * tree-if-conv.c (release_bb_predicate): Remove the
26390 the assertion that the stmts have NULL use_ops.
26391 Discard the statements, asserting that they haven't
26392 yet been added to a BB.
26394 2018-03-08 Richard Biener <rguenther@suse.de>
26396 PR tree-optimization/84746
26397 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
26398 (phi_translate): Pass in destination ANTIC_OUT set.
26399 (phi_translate_1): Likewise. For a simplified result lookup
26400 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
26401 (phi_translate_set): Adjust.
26402 (do_pre_regular_insertion): Likewise.
26403 (do_pre_partial_partial_insertion): Likewise.
26405 2018-03-08 Martin Liska <mliska@suse.cz>
26407 PR gcov-profile/84735
26408 * doc/gcov.texi: Document usage of profile files.
26409 * gcov-io.h: Document changes in the format.
26411 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
26415 * dwarf2out.c (struct dw_line_info_table): Update comments for
26417 (FORCE_RESET_NEXT_VIEW): New.
26418 (FORCE_RESETTING_VIEW_P): New.
26419 (RESETTING_VIEW_P): Check for -1 too.
26420 (ZERO_VIEW_P): Likewise.
26421 (new_line_info_table): Force-reset next view.
26422 (dwarf2out_begin_function): Likewise.
26423 (dwarf2out_source_line): Simplify zero_view_p initialization.
26424 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
26425 view directly. Omit view when omitting .loc at line 0.
26427 2018-03-08 Jakub Jelinek <jakub@redhat.com>
26429 PR tree-optimization/84740
26430 * tree-switch-conversion.c (process_switch): Call build_constructors
26431 only if info.phi_count is non-zero.
26433 PR tree-optimization/84739
26434 * tree-tailcall.c (find_tail_calls): Check call arguments against
26435 DECL_ARGUMENTS (current_function_decl) rather than
26436 DECL_ARGUMENTS (func) when checking for tail recursion.
26438 2018-03-07 Jakub Jelinek <jakub@redhat.com>
26440 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
26441 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
26442 Volker Reichelt's entry and add entries for people that perform
26443 GCC fuzzy testing and report numerous bugs.
26445 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
26448 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
26449 readonly data in sdata, if that is disabled.
26450 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
26451 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
26452 -mreadonly-in-sdata option.
26454 2018-03-07 Martin Sebor <msebor@redhat.com>
26456 PR tree-optimization/84468
26457 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
26458 basic block when looking for nul assignment.
26460 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
26463 * except.h (output_function_exception_table): Adjust prototype.
26464 * except.c (output_function_exception_table): Remove FNNAME parameter
26465 and add SECTION parameter. Ouput one part of the table at a time.
26466 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
26467 the first part of the exception table and emit unwind directives.
26468 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
26469 (i386_pe_seh_cold_init): Likewise.
26470 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
26471 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
26472 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
26473 (ix86_output_call_insn): Emit a nop in one more case for SEH.
26474 * config/i386/winnt.c: Include except.h.
26475 (struct seh_frame_state): Add reg_offset, after_prologue and
26476 in_cold_section fields.
26477 (i386_pe_seh_end_prologue): Set seh->after_prologue.
26478 (i386_pe_seh_cold_init): New function.
26479 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
26480 to seh->in_cold_section.
26481 (seh_emit_push): Record the offset of the push.
26482 (seh_emit_save): Record the offet of the save.
26483 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
26484 Test seh->after_prologue to disregard the epilogue.
26485 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
26486 (i386_pe_end_cold_function): New function.
26488 2018-03-07 Jakub Jelinek <jakub@redhat.com>
26491 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
26492 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
26495 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
26497 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
26498 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
26500 PR middle-end/84723
26501 * multiple_target.c: Include tree-inline.h and intl.h.
26502 (expand_target_clones): Diagnose and fail if node->definition and
26503 !tree_versionable_function_p (node->decl).
26505 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
26507 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
26509 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
26510 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
26511 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
26513 2018-03-06 Jakub Jelinek <jakub@redhat.com>
26516 * combine.c (try_combine): Use reg_or_subregno instead of handling
26517 just paradoxical SUBREGs and REGs.
26519 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
26521 * config/arc/arc.c (arc_finalize_pic): Remove function.
26522 (arc_must_save_register): We use single base PIC register, remove
26523 checks to save/restore the PIC register.
26524 (arc_expand_prologue): Likewise.
26525 * config/arc/arc-protos.h (arc_set_default_type_attributes):
26527 (arc_verify_short): Likewise.
26528 (arc_attr_type): Likewise.
26529 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
26530 (walk_stores): Likewise.
26531 (arc_address_cost): Make it static.
26532 (arc_verify_short): Likewise.
26533 (branch_dest): Likewise.
26534 (arc_attr_type): Likewise.
26535 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
26536 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
26537 (arc_final_prescan_insn): Remove inserting the nops due to
26538 hardware hazards. It is done in reorg step.
26539 (insn_length_variant_t): Remove.
26540 (insn_length_parameters_t): Likewise.
26541 (arc_insn_length_parameters): Likewise.
26542 (arc_get_insn_variants): Likewise.
26543 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
26545 2018-03-06 Jakub Jelinek <jakub@redhat.com>
26547 PR inline-asm/84683
26548 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
26551 PR tree-optimization/84687
26552 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
26554 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
26556 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26558 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
26559 Rename to ppc_speculation_barrier.
26560 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
26561 __builtin_ppc_speculation_barrier.
26563 2018-03-05 Jakub Jelinek <jakub@redhat.com>
26566 * combine.c (combine_simplify_rtx): Don't try to simplify if
26567 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
26570 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
26572 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
26573 to 32 bytes when compiling for POWER9.
26575 2018-03-05 Jakub Jelinek <jakub@redhat.com>
26578 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
26579 regparm >= 3 with no arg reg available also for calls with
26580 flag_force_indirect_call. Pass decl to ix86_function_regparm.
26583 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
26585 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
26587 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
26590 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
26592 2018-03-05 Richard Biener <rguenther@suse.de>
26594 PR tree-optimization/84486
26595 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
26596 When inserting a __builtin_assume_aligned call set the LHS
26597 SSA name alignment info accordingly.
26599 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
26601 PR tree-optimization/84114
26602 * config/aarch64/aarch64.c (aarch64_reassociation_width)
26603 Avoid reassociation of FLOAT_MODE addition.
26605 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
26607 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
26608 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
26609 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
26610 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
26611 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
26612 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
26613 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
26615 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
26616 __builtin_ia32_wbinvd): New builtins.
26617 (SPECIAL_ARGS2): New.
26618 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
26619 (SPECIAL_ARGS2): New.
26620 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
26621 (ix86_valid_target_attribute_inner_p): Ditto.
26622 (ix86_init_mmx_sse_builtins): Add special_args2.
26623 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
26624 TARGET_WBNOINVD_P): New.
26625 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
26626 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
26627 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
26628 * config/i386/immintrin.h (_wbinvd): New intrinsic.
26629 * config/i386/pconfigintrin.h: New file.
26630 * config/i386/wbnoinvdintrin.h: Ditto.
26631 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
26633 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
26635 2018-03-05 Richard Biener <rguenther@suse.de>
26637 PR tree-optimization/84670
26638 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
26640 (BB_VISITED_WITH_VISITED_SUCCS): New define.
26641 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
26642 (compute_antic_aux): Only assert the number of values in ANTIC_IN
26643 doesn't grow if all successors (recursively) were visited at least
26646 2018-03-05 Richard Biener <rguenther@suse.de>
26648 PR tree-optimization/84650
26649 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
26650 if executed in the loop pipeline.
26652 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
26654 * doc/configfiles.texi (Configuration Files): Move info about
26655 conditionalizing $target-protos.h to...
26656 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
26657 differs from $target-protos.h.
26659 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
26660 Chung-Ju Wu <jasonwucj@gmail.com>
26662 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
26663 * config/nds32/nds32-multiple.md (setmemsi): Define.
26664 * config/nds32/nds32-memory-manipulation.c
26665 (nds32_gen_dup_4_byte_to_word_value): New.
26666 (emit_setmem_word_loop): New.
26667 (emit_setmem_byte_loop): New.
26668 (nds32_expand_setmem_loop): New.
26669 (nds32_expand_setmem_loop_v3m): New.
26670 (nds32_expand_setmem_unroll): New.
26671 (nds32_expand_setmem): New.
26673 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26674 Chung-Ju Wu <jasonwucj@gmail.com>
26676 * config/nds32/nds32-memory-manipulation.c
26677 (nds32_emit_load_store): New.
26678 (nds32_emit_post_inc_load_store): New.
26679 (nds32_emit_mem_move): New.
26680 (nds32_emit_mem_move_block): New.
26681 (nds32_expand_movmemsi_loop_unknown_size): New.
26682 (nds32_expand_movmemsi_loop_known_size): New.
26683 (nds32_expand_movmemsi_loop): New.
26684 (nds32_expand_movmemsi_unroll): New.
26685 (nds32_expand_movmemqi): Rename ...
26686 (nds32_expand_movmemsi): ... to this.
26687 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
26688 (movmemsi): ... to this.
26689 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
26690 (nds32_expand_movmemsi): ... to this.
26692 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26693 Monk Chiang <sh.chiang04@gmail.com>
26694 Chung-Ju Wu <jasonwucj@gmail.com>
26696 * config/nds32/nds32-protos.h
26697 (nds32_expand_load_multiple): New arguments.
26698 (nds32_expand_store_multiple): Ditto.
26699 (nds32_valid_multiple_load_store): Rename ...
26700 (nds32_valid_multiple_load_store_p): ... to this.
26701 * config/nds32/nds32-memory-manipulation.c
26702 (nds32_expand_load_multiple): Refine implementation.
26703 (nds32_expand_store_multiple): Ditto.
26704 * config/nds32/nds32-multiple.md
26705 (load_multiple): Update nds32_expand_load_multiple interface.
26706 (store_multiple): Update nds32_expand_store_multiple interface.
26707 * config/nds32/nds32-predicates.c
26708 (nds32_valid_multiple_load_store): Rename ...
26709 (nds32_valid_multiple_load_store_p): ... to this and refine
26711 * config/nds32/predicates.md
26712 (nds32_load_multiple_and_update_address_operation): New predicate.
26713 (nds32_store_multiple_and_update_address_operation): New predicate.
26715 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
26716 Chung-Ju Wu <jasonwucj@gmail.com>
26718 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
26719 (combo): New attribute.
26720 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
26722 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
26724 * config/nds32/nds32.opt: Change -mcmodel= default value.
26726 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
26727 Monk Chiang <sh.chiang04@gmail.com>
26728 Chung-Ju Wu <jasonwucj@gmail.com>
26730 * config/nds32/constants.md (unspec_element): New enum.
26731 * config/nds32/constraints.md (Umw): New constraint.
26732 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
26733 * config/nds32/nds32-intrinsic.md: Likewise.
26734 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
26735 (nds32_valid_smw_lwm_base_p): New.
26736 (nds32_output_smw_single_word): New.
26737 (nds32_output_lmw_single_word): New.
26738 (nds32_expand_unaligned_load): New.
26739 (nds32_expand_unaligned_store): New.
26740 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
26741 (nds32_output_smw_single_word): Declare.
26742 (nds32_output_lmw_single_word): Declare.
26743 (nds32_expand_unaligned_load): Declare.
26744 (nds32_expand_unaligned_store): Declare.
26745 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
26746 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
26747 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
26748 NDS32_BUILTIN_UASTORE_DW.
26749 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
26752 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
26753 Kito Cheng <kito.cheng@gmail.com>
26754 Chung-Ju Wu <jasonwucj@gmail.com>
26756 * config/nds32/nds32-intrinsic.c
26757 (nds32_expand_builtin_null_ftype_reg): Delete.
26758 (nds32_expand_builtin_reg_ftype_imm): Ditto.
26759 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
26760 (nds32_read_argument): New.
26761 (nds32_legitimize_target): Ditto.
26762 (nds32_legitimize_argument): Ditto.
26763 (nds32_check_constant_argument): Ditto.
26764 (nds32_expand_unop_builtin): Ditto.
26765 (nds32_expand_unopimm_builtin): Ditto.
26766 (nds32_expand_binop_builtin): Ditto.
26767 (nds32_builtin_decl_impl): Ditto.
26768 (builtin_description): Ditto.
26769 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
26770 (nds32_init_builtins_impl): Ditto.
26771 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
26772 (nds32_builtin_decl): New.
26773 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
26774 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
26776 2018-03-02 Jeff Law <law@redhat.com>
26778 * reorg.c (stop_search_p): Handle DEBUG_INSN.
26779 (redundant_insn, fill_simple_delay_slots): Likewise.
26780 (fill_slots_from_thread): Likewise.
26781 * resource.c (mark_referenced_resources): Likewise.
26782 (mark_set_resources, find_dead_or_set_registers): Likewise.
26784 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26786 * substring-locations.h (format_warning_va): Formatting fix for
26787 ATTRIBUTE_GCC_DIAG.
26788 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
26790 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
26791 * substring-locations.c: Include intl.h.
26792 (format_warning_va): Turned into small wrapper around
26793 format_warning_n_va, renamed to ...
26794 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
26795 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
26797 (format_warning_at_substring_n): New function.
26798 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
26799 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
26800 format_warning_at_substring with just a shorter name instead of
26801 const function pointer.
26802 (fmtwarn_n): New function.
26803 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
26804 appropriate, get rid of all the fmtstr temporaries, move conditionals
26805 with G_() wrapped string literals directly into fmtwarn arguments,
26806 cast dir.len to (int), formatting fixes.
26808 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
26810 * doc/invoke.texi: Remove "Cilk Plus" references.
26812 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26813 Richard Biener <rguenther@suse.de>
26816 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
26817 for error or warning attributes if CALL_FROM_THUNK_P is set.
26820 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26823 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
26824 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
26827 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
26828 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
26830 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
26831 instead of -1U in last predictors element's probability member.
26833 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
26836 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
26837 arguments if they are comparable.
26839 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26841 PR tree-optimization/84634
26842 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
26843 masks and masked_loop_p with a single loop_masks, making sure it's
26844 null for bb vectorization.
26846 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26848 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
26849 (vect_analyze_data_ref_access): Use loop->safe_len rather than
26850 loop->force_vectorize to check whether there is no alias.
26852 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26855 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
26857 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
26859 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
26860 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
26861 instead of a loop around prev_real_insn.
26862 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
26865 PR inline-asm/84625
26866 * config/i386/i386.c (ix86_print_operand): Use conditional
26867 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
26870 2018-03-02 Richard Biener <rguenther@suse.de>
26872 PR tree-optimization/84427
26873 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
26874 (bitmap_set_subtract_values): Rewrite to handle multiple
26877 (prune_clobbered_mems): Likewise.
26878 (phi_translate): Take edge instead of pred/phiblock.
26879 (phi_translate_1): Likewise.
26880 (phi_translate_set): Likewise. Insert all translated
26881 exprs for a value into the set, keeping possibly multiple
26882 expressions per value.
26883 (compute_antic_aux): Adjust for phi_translate changes.
26884 When intersecting union the expressions and prune those
26885 not in the final value set, keeping possibly multiple
26886 expressions per value. Do not use value-insertion
26887 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
26888 all expressions. Add verification that the value-sets
26889 only shrink during iteration.
26890 (compute_partial_antic_aux): Adjust for the phi_translate changes.
26891 (do_pre_regular_insertion): Likewise.
26892 (do_pre_partial_partial_insertion): Likewise.
26894 2018-03-02 Richard Biener <rguenther@suse.de>
26897 * config/darwin.c (saved_debug_info_level): New static global.
26898 (darwin_asm_lto_start): Disable debug info generation for LTO out.
26899 (darwin_asm_lto_end): Restore debug info generation settings.
26901 2018-03-01 Martin Liska <mliska@suse.cz>
26904 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
26905 volatile arguments.
26907 2018-03-01 Richard Biener <rguenther@suse.de>
26910 * dwarf2out.c (gen_variable_die): Properly handle late VLA
26911 type annotation with LTO when debug was disabled at compile-time.
26913 2018-03-01 Matthew Fortune <mfortune@gmail.com>
26915 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
26917 (mips_final_postscan_insn): Likewise.
26919 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
26921 PR rtl-optimization/84528
26922 * alias.c (init_alias_target): Add commentary.
26923 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
26924 a unique base value if the frame pointer is not eliminated
26925 to the stack pointer.
26927 2018-03-01 Tom de Vries <tom@codesourcery.com>
26929 PR rtl-optimization/83327
26930 * lra-int.h (hard_regs_spilled_into): Declare.
26931 * lra.c (hard_regs_spilled_into): Define.
26932 (init_reg_info): Init hard_regs_spilled_into.
26933 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
26934 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
26935 (process_bb_lives): Handle hard_regs_spilled_into.
26936 (lra_create_live_ranges_1): Before doing liveness propagation, clear
26937 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
26939 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
26941 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
26942 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
26943 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
26944 * config/rs6000/aix72.h: New file.
26946 2018-02-28 Jakub Jelinek <jakub@redhat.com>
26948 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
26949 instead of warning_at with conditional singular and plural messages
26953 * stor-layout.c (update_alignment_for_field): For
26954 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
26955 && !DECL_PACKED (field), do the alignment update, just use
26956 only desired_align instead of MAX (type_align, desired_align)
26958 (place_field): Don't do known_align < desired_align handling
26959 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
26960 is non-NULL, instead do it after rli->prev_field handling and
26961 only if not within a bitfield word. For DECL_PACKED (field)
26962 use type_align of BITS_PER_UNIT.
26964 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
26966 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
26967 superfluous parentheses and trailing spaces.
26969 2018-02-28 Richard Biener <rguenther@suse.de>
26971 PR tree-optimization/84584
26972 * graphite-scop-detection.c (scop_detection::add_scop): Discard
26973 SCoPs with fake exit edge.
26975 2018-02-28 Martin Liska <mliska@suse.cz>
26978 * timevar.c (timer::print): Fix format to properly print 100%
26981 2018-02-28 Richard Biener <rguenther@suse.de>
26983 PR middle-end/84607
26984 * genmatch.c (capture_info::walk_match): Do not mark
26985 captured expressions without operands as expr_p given
26986 they act more like predicates and should be subject to
26987 "lost tail" side-effect preserving.
26989 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
26991 PR rtl-optimization/81611
26992 * auto-inc-dec.c (attempt_change): Move dead note from
26993 mem_insn if it's the next use of regno
26994 (find_address): Take address use of reg holding
26995 non-incremented value. Add parm to limit search to the named
26997 (merge_in_block): Attempt to use a mem insn that is the next
26998 use of the original regno.
27000 2018-02-27 Martin Sebor <msebor@redhat.com>
27003 * doc/invoke.texi (-Wmissing-attributes): New option.
27004 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
27006 2018-02-27 Martin Sebor <msebor@redhat.com>
27008 PR translation/84207
27009 * diagnostic-core.h (warning_n, error_n, inform_n): Change
27010 n argument to unsigned HOST_WIDE_INT.
27011 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
27012 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
27013 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
27014 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
27016 2018-02-27 Richard Biener <rguenther@suse.de>
27018 PR tree-optimization/84512
27019 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
27020 Do not use the estimate returned from record_stmt_cost for
27021 the scalar iteration cost but sum properly using add_stmt_cost.
27023 2018-02-27 Richard Biener <rguenther@suse.de>
27025 PR tree-optimization/84466
27026 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27027 Adjust last change to less strictly validate use operands.
27029 2018-02-27 Martin Liska <mliska@suse.cz>
27031 PR gcov-profile/84548
27032 * gcov.c (process_file): Allow partial overlap and consider it
27033 also as group functions.
27034 (output_lines): Properly calculate range of lines for a group.
27036 2018-02-27 Martin Liska <mliska@suse.cz>
27038 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
27039 'ggc' suffixes. Change first column width.
27040 (timer::print): Fix formatting of the column.
27042 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
27044 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
27045 preserve inline entry blocks for the sake of debug inline
27046 entry point markers alone.
27047 (remove_unused_locals): Suggest in comments a better place to
27048 force the preservation of inline entry blocks that are
27049 otherwise unused, but do not preserve them.
27051 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27053 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
27055 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27058 * config/i386/constraints.md (Bs): Replace
27059 ix86_indirect_branch_register with
27060 TARGET_INDIRECT_BRANCH_REGISTER.
27062 * config/i386/i386.md (indirect_jump): Likewise.
27063 (tablejump): Likewise.
27064 (*sibcall_memory): Likewise.
27065 (*sibcall_value_memory): Likewise.
27066 Peepholes of indirect call and jump via memory: Likewise.
27067 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
27068 (*sibcall_value_GOT_32): Likewise.
27069 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27070 (GOT_memory_operand): Likewise.
27071 (call_insn_operand): Likewise.
27072 (sibcall_insn_operand): Likewise.
27073 (GOT32_symbol_operand): Likewise.
27074 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
27076 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
27078 PR rtl-optimization/83496
27079 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
27080 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
27081 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
27082 redundant insn, if any.
27083 (relax_delay_slots): Likewise.
27084 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
27086 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
27088 PR tree-optimization/83965
27089 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
27090 that grouped statements are part of a reduction chain. Return
27091 true if the statement is not marked as a reduction itself but
27092 is part of a group.
27093 (vect_recog_dot_prod_pattern): Don't check whether the statement
27094 is part of a group here.
27095 (vect_recog_sad_pattern): Likewise.
27096 (vect_recog_widen_sum_pattern): Likewise.
27098 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
27101 * final.c (rest_of_clean_state): Also look for calls inside sequences.
27103 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
27106 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
27108 (ix86_output_indirect_function_return): New prototype.
27109 (ix86_split_simple_return_pop_internal): Likewise.
27110 * config/i386/i386.c (indirect_return_via_cx): New.
27111 (indirect_return_via_cx_bnd): Likewise.
27112 (indirect_thunk_name): Handle return va CX_REG.
27113 (output_indirect_thunk_function): Create alias for
27114 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
27115 (ix86_output_indirect_jmp): Remove the bool argument.
27116 (ix86_output_indirect_function_return): New function.
27117 (ix86_split_simple_return_pop_internal): Likewise.
27118 * config/i386/i386.md (*indirect_jump): Don't pass false
27119 to ix86_output_indirect_jmp.
27120 (*tablejump_1): Likewise.
27121 (simple_return_pop_internal): Change it to define_insn_and_split.
27122 Call ix86_split_simple_return_pop_internal to split it for
27123 -mfunction-return=.
27124 (simple_return_indirect_internal): Call
27125 ix86_output_indirect_function_return instead of
27126 ix86_output_indirect_jmp.
27128 2018-02-26 Jakub Jelinek <jakub@redhat.com>
27131 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
27132 memset and value initialization afterwards.
27134 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
27136 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
27138 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
27141 * common/config/aarch64/aarch64-common.c
27142 (aarch_option_optimization_table[]): Switch
27143 off fomit-frame-pointer
27145 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
27146 Chung-Ju Wu <jasonwucj@gmail.com>
27148 * config/nds32/nds32-multiple.md (load_multiple): Disallow
27150 (store_multiple): Ditto.
27152 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
27154 * config.gcc: Add --with-cpu support for nds32 target.
27155 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
27156 * config/nds32/nds32.opt: Add -mcpu= option.
27158 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
27160 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
27161 isel=yes): Warn for these deprecated options.
27163 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
27165 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
27166 ISA_2_5_MASKS_EMBEDDED.
27168 2018-02-23 Jakub Jelinek <jakub@redhat.com>
27170 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
27171 p->max as pointers rather than using iterative_hash_expr.
27173 2018-02-23 Carl Love <cel@us.ibm.com>
27175 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
27176 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
27178 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
27179 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
27180 P8V_BUILTIN_VEC_VUNSIGNED2.
27182 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
27185 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
27186 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
27188 * lra-constraints.c (curr_insn_transform): Set up
27189 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
27191 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
27192 * lra-eliminations.c (spill_pseudos): Ditto.
27193 (process_insn_for_elimination): Ditto.
27194 * lra-lives.c (reg_early_clobber_p): Use the new macros.
27195 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
27196 LRA_NON_CLOBBERED_ALT.
27198 2018-02-22 Martin Sebor <msebor@redhat.com>
27200 PR tree-optimization/84480
27201 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
27202 to maybe_diag_stxncpy_trunc. Call it.
27203 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
27204 from gimple_fold_builtin_strcpy. Print inlining stack.
27205 (handle_builtin_stxncpy): Print inlining stack.
27206 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
27208 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
27211 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
27212 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
27213 and -fcheck-pointer-bounds are used together.
27214 (indirect_thunk_prefix): New enum.
27215 (indirect_thunk_need_prefix): New function.
27216 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
27217 "_nt" instead of "_bnd" for NOTRACK prefix.
27218 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
27219 (output_indirect_thunk_function): Likewise.
27221 (ix86_code_end): Update output_indirect_thunk_function calls.
27222 (ix86_output_indirect_branch_via_reg): Replace
27223 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
27224 (ix86_output_indirect_branch_via_push): Likewise.
27225 (ix86_output_function_return): Likewise.
27226 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
27227 incompatible with -fcf-protection=branch and
27228 -fcheck-pointer-bounds.
27230 2018-02-22 Steve Ellcey <sellcey@cavium.com>
27233 * config/aarch64/aarch64.c (aarch64_print_address_internal):
27234 Change gcc_assert call to output_operand_lossage.
27236 2018-02-22 Steve Ellcey <sellcey@cavium.com>
27238 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
27240 2018-02-22 DJ Delorie <dj@redhat.com>
27241 Sebastian Perta <sebastian.perta@renesas.com>
27242 Oleg Endo <olegendo@gcc.gnu.org>
27244 * config/rx/rx.c (rx_rtx_costs): New function.
27245 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
27247 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27249 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
27251 2018-02-22 Martin Liska <mliska@suse.cz>
27254 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
27255 Add "native" as a possible value.
27257 2018-02-22 Martin Liska <mliska@suse.cz>
27260 * config/i386/i386.c (ix86_option_override_internal):
27261 Add "native" as a possible value for -march and -mtune.
27263 2018-02-22 Jakub Jelinek <jakub@redhat.com>
27266 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
27267 to all type variants.
27269 PR tree-optimization/84503
27270 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
27271 width as info->bitpos + info->bitsize - start.
27272 (merged_store_group::merge_overlapping): Simplify width computation.
27273 (check_no_overlap): New function.
27274 (imm_store_chain_info::try_coalesce_bswap): Compute expected
27275 start + width and last_order of the group, fail if check_no_overlap
27277 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
27278 to group if check_no_overlap fails.
27280 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
27282 * config/rs6000/altivec.md: Delete contraint arguments to
27283 define_expand, define_split, and define_peephole2, and in
27284 define_insn_and_split if always unused.
27285 * config/rs6000/darwin.md: Ditto.
27286 * config/rs6000/dfp.md: Ditto.
27287 * config/rs6000/rs6000.md: Ditto.
27288 * config/rs6000/sync.md: Ditto.
27289 * config/rs6000/vector.md: Ditto.
27290 * config/rs6000/vsx.md: Ditto.
27292 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
27294 * config/rs6000/altivec.md: Write output control strings as braced
27295 blocks instead of double-quoted strings.
27296 * config/rs6000/darwin.md: Ditto.
27297 * config/rs6000/rs6000.md: Ditto.
27298 * config/rs6000/vector.md: Ditto.
27299 * config/rs6000/vsx.md: Ditto.
27301 2018-02-21 Jason Merrill <jason@redhat.com>
27303 PR c++/84314 - ICE with templates and fastcall attribute.
27304 * attribs.c (build_type_attribute_qual_variant): Remove assert.
27306 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
27308 * ipa-cp.c (determine_versionability): Fix comment typos.
27310 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
27313 * ipa-cp.c (determine_versionability): Do not version functions caling
27316 2018-02-21 Martin Liska <mliska@suse.cz>
27319 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
27320 Add "native" as a possible value.
27321 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
27322 the macro when native cpu detection is available.
27324 2018-02-21 Martin Liska <mliska@suse.cz>
27327 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
27328 Add "native" as a possible value.
27329 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
27330 when native cpu detection is available.
27332 2018-02-21 Jakub Jelinek <jakub@redhat.com>
27333 Martin Sebor <msebor@redhat.com>
27335 PR tree-optimization/84478
27336 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
27338 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
27339 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
27340 support which is conservatively correct, for 2 only stay conservative
27341 for maxlen. Formatting and comment capitalization fixes. Add STRICT
27342 argument to the 2 argument get_range_strlen, adjust 6 arg
27343 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
27345 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
27346 (gimple_fold_builtin_strlen): Pass true as last argument to
27349 2018-02-20 Martin Sebor <msebor@redhat.com>
27351 PR middle-end/84095
27352 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
27353 (builtin_memref::set_base_and_offset): Same. Handle inner references.
27354 (builtin_memref::builtin_memref): Factor out parts into
27355 set_base_and_offset and call it.
27357 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
27359 PR middle-end/84406
27360 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
27361 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
27362 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
27363 search at the associated MODE_INT.
27365 2018-02-20 Jeff Law <law@redhat.com>
27367 PR middle-end/82123
27368 PR tree-optimization/81592
27369 PR middle-end/79257
27370 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
27371 for range data rather than using global data.
27372 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
27373 range data rather than using global data.
27374 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
27375 pass it to children as needed.
27376 (struct directive::fmtresult): Similarly.
27377 (struct directive::set_width): Similarly.
27378 (struct directive::set_precision): Similarly.
27379 (format_integer, format_directive, parse_directive): Similarly.
27380 (format_none): Accept unnamed vr_values parameter.
27381 (format_percent, format_floating, format_character): Similarly.
27382 (format_string, format_plain): Similarly.
27383 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
27384 the EVRP range analyzer for range data rather than using global data.
27385 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
27386 gimple-ssa-evrp-analyze.h
27387 (class sprintf_dom_walker): Add after_dom_children member function.
27388 Add evrp_range_analyzer member.
27389 (sprintf_dom_walker::before_dom_children): Call into the EVRP
27390 range analyzer as needed.
27391 (sprintf_dom_walker::after_dom_children): New member function.
27392 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
27394 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
27395 (evrp_range_analyzer::pop_to_marker): Likewise.
27397 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
27399 PR tree-optimization/84419
27400 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
27401 with the required type if its current type is compatible but
27404 2018-02-20 Jakub Jelinek <jakub@redhat.com>
27406 PR middle-end/82004
27407 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
27408 after vectorization.
27410 2018-02-20 Martin Liska <mliska@suse.cz>
27413 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
27414 possible values if we don't have a hint.
27416 2018-02-20 Martin Liska <mliska@suse.cz>
27420 * final.c (shorten_branches): Build align_tab array with one
27422 * opts.c (finish_options): Add alignment option limit check.
27423 (MAX_CODE_ALIGN): Likewise.
27424 (MAX_CODE_ALIGN_VALUE): Likewise.
27425 * doc/invoke.texi: Document maximum allowed option value for
27426 all -falign-* options.
27428 2018-02-19 Jakub Jelinek <jakub@redhat.com>
27431 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
27432 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
27433 * var-tracking.c (emit_note_insn_var_location): Remove all references
27434 to NOTE_INSN_CALL_ARG_LOCATION.
27435 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
27436 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
27437 Use copy_rtx_if_shared.
27438 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
27439 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
27440 (dwarf2out_var_location): Remove handling of
27441 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
27443 * final.c (final_scan_insn): Remove all references to
27444 NOTE_INSN_CALL_ARG_LOCATION.
27445 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
27446 before dumping final insns.
27447 * except.c (emit_note_eh_region_end): Remove all references to
27448 NOTE_INSN_CALL_ARG_LOCATION.
27449 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
27450 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
27451 * config/arc/arc.c (hwloop_optimize): Likewise.
27452 * config/arm/arm.c (create_fix_barrier): Likewise.
27453 * config/s390/s390.c (s390_chunkify_start): Likewise.
27454 * config/sh/sh.c (find_barrier): Likewise.
27455 * config/i386/i386.c (rest_of_insert_endbranch,
27456 ix86_seh_fixup_eh_fallthru): Likewise.
27457 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
27458 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
27459 * config/frv/frv.c (frv_function_prologue): Likewise.
27460 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
27462 (note_outside_basic_block_p): Remove all references to
27463 NOTE_INSN_CALL_ARG_LOCATION.
27464 * gengtype.c (adjust_field_rtx_def): Likewise.
27465 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
27467 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
27468 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
27471 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
27474 PR tree-optimization/84452
27475 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
27476 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
27479 2018-02-19 Martin Liska <mliska@suse.cz>
27482 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
27484 2018-02-19 Martin Liska <mliska@suse.cz>
27485 Richard Sandiford <richard.sandiford@linaro.org>
27487 PR tree-optimization/82491
27488 * gimple-fold.c (get_base_constructor): Make earlier bail out
27491 2018-02-19 Carl Love <cel@us.ibm.com>
27493 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
27494 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
27496 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
27497 P8V_BUILTIN_VEC_NEG.
27499 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
27501 * config/rl78/rl78.md (movdf): New define expand.
27503 2018-02-19 Martin Liska <mliska@suse.cz>
27506 * doc/invoke.texi: Fix typo.
27507 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
27509 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
27511 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
27512 handle rs6000_single_float and rs6000_double_float specially for
27515 2018-02-16 Jeff Law <law@redhat.com>
27517 * config/rx/rx.c (add_pop_cfi_notes): New function.;
27518 (pop_regs): Use it.
27520 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27523 * ipa-inline.c (inline_small_functions): Fix a typo.
27525 2018-02-16 Nathan Sidwell <nathan@acm.org>
27527 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
27529 2018-02-16 Carl Love <cel@us.ibm.com>
27531 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
27532 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
27533 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
27534 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
27535 expansion to P8V_BUILTIN_VEC_FLOAT2.
27537 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
27539 PR rtl-optimization/70023
27540 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
27541 src_regno into account.
27543 2018-02-16 Carl Love <cel@us.ibm.com>
27545 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
27546 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
27547 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
27548 * config/rs6000/rs6000.c: Remove case statements for
27549 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
27550 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
27551 and P9V_BUILTIN_VEC_VINSERT4B.
27552 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
27553 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
27554 * config/rs6000/vsx.md:
27555 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
27558 2018-02-16 Carl Love <cel@us.ibm.com>
27560 * config/rs6000/altivec.h: Add builtin names vec_extract4b
27562 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
27564 * config/rs6000/rs6000-c.c: Add the definitions for
27565 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
27566 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
27567 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
27568 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
27569 definition for insert4b and define insn *insert3b_internal.
27570 * doc/extend.texi: Add documentation for vec_extract4b.
27572 2018-02-16 Nathan Sidwell <nathan@acm.org>
27574 * doc/extend.texi (Backwards Compatibility): Mention friend
27575 injection. Note for-scope is deprecated.
27576 * doc/invoke.texi (-ffriend-injection): Deprecate.
27578 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
27580 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
27581 that moved to I2, also allow destinations that are a paradoxical
27582 subreg (instead of a normal reg).
27584 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
27587 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
27590 2018-02-16 Richard Biener <rguenther@suse.de>
27592 PR tree-optimization/84037
27593 PR tree-optimization/84016
27595 * config/i386/i386.c (ix86_builtin_vectorization_cost):
27596 Adjust vec_construct for the fact we need additional higher latency
27597 128bit inserts for AVX256 and AVX512 vector builds.
27598 (ix86_add_stmt_cost): Scale vector construction cost for
27601 2018-02-16 Richard Biener <rguenther@suse.de>
27603 PR tree-optimization/84417
27604 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
27605 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
27606 (non_rewritable_lvalue_p): Likewise, use poly-ints.
27608 2018-02-16 Martin Liska <mliska@suse.cz>
27611 * internal-fn.def (ASAN_CHECK): Set proper flags.
27612 (ASAN_MARK): Likewise.
27614 2018-02-16 Julia Koval <julia.koval@intel.com>
27616 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
27617 from PTA_CANNONLAKE.
27619 2018-02-16 Jakub Jelinek <jakub@redhat.com>
27622 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
27623 Use ++iter rather than iter++ for std::list iterators.
27624 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
27625 defer deleting them until all nodes in the forest are processed. Do
27626 free even leaf nodes. Change to_process into auto_vec.
27629 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
27630 * vec.h (vec_default_construct): Use memset instead of placement new
27631 if BROKEN_VALUE_INITIALIZATION is defined.
27632 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
27633 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
27636 PR rtl-optimization/83723
27637 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
27638 * lra.c (lra_substitute_pseudo): Likewise. If true, use
27639 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
27641 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
27643 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
27645 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
27647 PR rtl-optimization/81443
27648 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
27649 from inner REGs to paradoxical SUBREGs.
27651 2018-02-16 Richard Biener <rguenther@suse.de>
27653 PR tree-optimization/84399
27654 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
27655 For operands we can analyze at their definition make sure we can
27656 analyze them at each use as well.
27658 2018-02-16 Richard Biener <rguenther@suse.de>
27660 PR tree-optimization/84190
27661 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
27662 volatile accesses if the decl isn't volatile.
27664 2018-02-15 Jason Merrill <jason@redhat.com>
27666 PR c++/84314 - ICE with templates and fastcall attribute.
27667 * attribs.c (build_type_attribute_qual_variant): Don't clobber
27668 TYPE_CANONICAL on an existing type.
27670 2018-02-15 Jakub Jelinek <jakub@redhat.com>
27672 PR tree-optimization/84383
27673 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
27674 dstoff nor call operand_equal_p if dstbase is NULL.
27676 PR tree-optimization/84334
27677 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
27678 also a CONSTANT_CLASS_P, punt.
27680 2018-02-14 Jim Wilson <jimw@sifive.com>
27682 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
27683 first SMALL_OPERAND check. New local min_second_step. Move assert
27684 to where locals are set. Add TARGET_RVC support.
27685 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
27687 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
27689 * doc/invoke.texi: Correct -Wformat-overflow code sample.
27691 2018-02-14 Martin Sebor <msebor@redhat.com>
27693 PR tree-optimization/83698
27694 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
27695 arrays constrain the offset range to their bounds.
27696 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
27697 (builtin_access::overlap): Avoid setting the size of overlap if it's
27699 (maybe_diag_overlap): Also consider arrays when deciding what values
27700 of offsets to include in diagnostics.
27702 2018-02-14 Martin Sebor <msebor@redhat.com>
27705 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
27706 that correspond to the kind of a declaration.
27708 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
27711 * config/pa/pa.md: Load address of PIC label using the linkage table
27712 if the label is nonlocal.
27714 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
27716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
27717 warning message if user requests -maltivec=be.
27718 * doc/invoke.texi: Document deprecation of -maltivec=be.
27720 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
27723 * config/rs6000/rs6000-c.c: Update definitions for
27724 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
27725 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
27727 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27730 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
27731 add _get_ssp intrinsics. Remove argument from
27732 __builtin_ia32_rdssp[d|q].
27733 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
27734 * config/i386/i386-builtin.def: Remove argument from
27735 __builtin_ia32_rdssp[d|q].
27736 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
27737 ix86_expand_special_args_builtin for _rdssp[d|q].
27738 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
27739 Clear register before usage.
27740 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
27741 Add documentation for new _get_ssp and _inc_ssp intrinsics.
27743 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
27745 PR tree-optimization/84357
27746 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
27747 operand 1 of an ARRAY_REF too.
27749 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
27752 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
27753 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27755 (set_of_reg): New struct.
27756 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
27757 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
27758 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
27760 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
27761 Split into bitclr, bitset, bitinvert patterns if appropriate.
27762 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
27763 use rx_fuse_in_memory_bitop.
27764 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
27765 to named insn, correct maximum insn length.
27767 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
27770 * machmode.def: Define a complex mode for PARTIAL_INT.
27771 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
27773 * doc/rtl.texi: Document CSPImode.
27774 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
27776 (msp430_hard_regno_nregs_with_padding): Likewise.
27778 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
27781 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
27783 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
27785 PR rtl-optimization/84169
27786 * combine.c (try_combine): New variable split_i2i3. Set it to true if
27787 we generated a parallel as new i3 and we split that to new i2 and i3
27788 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
27789 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
27790 those to i2, not i1. Partially rewrite this scan code.
27792 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27795 * stor-layout.c (place_field): For variable length fields, adjust
27796 offset_align afterwards not just based on the field's alignment,
27797 but also on the size.
27799 PR middle-end/84309
27800 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
27801 of exps and logs in the use_exp2 case.
27803 2018-02-13 Jeff Law <law@redhat.com>
27805 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
27806 entry for "vector".
27808 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
27811 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
27815 * common.opt (gas-loc-support, gas-locview-support): New.
27816 (ginline-points, ginternal-reset-location-views): New.
27817 * doc/invoke.texi: Document them. Use @itemx where intended.
27818 (gvariable-location-views): Adjust.
27819 * target.def (reset_location_view): New.
27820 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
27821 (TARGET_RESET_LOCATION_VIEW): New.
27822 * doc/tm.texi: Rebuilt.
27823 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
27824 (dwarf2out_default_as_locview_support): New.
27825 (output_asm_line_debug_info): Use option variables.
27826 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
27827 (output_loc_list): Likewise.
27828 (add_high_low_attributes): Check option variables.
27829 Don't output entry view attribute in strict mode.
27830 (gen_inlined_subroutine_die): Check option variables.
27831 (dwarf2out_inline_entry): Likewise.
27832 (init_sections_and_labels): Likewise.
27833 (dwarf2out_early_finish): Likewise.
27834 (maybe_reset_location_view): New, from...
27835 (dwarf2out_var_location): ... here. Call it.
27836 * debug.h (dwarf2out_default_as_loc_support): Declare.
27837 (dwarf2out_default_as_locview_support): Declare.
27838 * hooks.c (hook_int_rtx_insn_0): New.
27839 * hooks.h (hook_int_rtx_insn_0): Declare.
27840 * toplev.c (process_options): Take -gas-loc-support and
27841 -gas-locview-support from dwarf2out. Enable
27842 -gvariable-location-views by default only with locview
27843 assembler support. Enable -ginternal-reset-location-views by
27844 default only if the target defines the corresponding hook.
27845 Enable -ginline-points by default if location views are
27846 enabled; force it disabled if statement frontiers are
27848 * tree-inline.c (expand_call_inline): Check option variables.
27849 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
27851 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
27853 PR tree-optimization/84321
27854 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
27855 handling. Also check whether the anti-range contains any values
27856 that satisfy the mask; switch to a VR_RANGE if not.
27858 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
27861 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
27863 2018-02-13 Martin Jambor <mjambor@suse.cz>
27866 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
27867 of call statements, also set location of a load to a temporary.
27869 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
27871 * config/rl78/rl78.c (add_vector_labels): New function.
27872 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
27873 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
27874 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
27875 which checks that no arguments are passed.
27876 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
27877 * doc/extend.texi: Documentation for the new attribute.
27879 2018-02-13 Andreas Schwab <schwab@suse.de>
27881 * config/riscv/linux.h (CPP_SPEC): Define.
27883 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27886 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
27887 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
27888 OPTION_MASK_ISA_AES as first argument to def_builtin_const
27889 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
27890 instead of OPTION_MASK_ISA_PCLMUL as first argument to
27891 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
27892 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
27893 temporarily for AES and PCLMUL builtins.
27895 PR tree-optimization/84339
27896 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
27897 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
27900 PR middle-end/84309
27901 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
27902 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
27903 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
27905 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
27907 * omp-simd-clone.h: New file.
27908 * omp-simd-clone.c: Include omp-simd-clone.h.
27909 (expand_simd_clones): No longer static.
27910 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
27911 cgraph.h and omp-simd-clone.h.
27912 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
27913 (vect_recog_widen_shift_pattern): Formatting fix.
27914 (vect_pattern_recog_1): Don't check optab for calls.
27917 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
27918 operands[2] into a REG before using gen_lowpart on it.
27920 2018-02-12 Jeff Law <law@redhat.com>
27923 * config/sh/sh.c (find_barrier): Consider a sibling call
27926 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
27927 successfully back substituting a reg.
27929 2018-02-12 Richard Biener <rguenther@suse.de>
27931 PR tree-optimization/84037
27932 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
27933 parameter, move visited init to caller.
27934 (vect_slp_analyze_operations): Separate cost from validity
27935 check, initialize visited once for all instances.
27936 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
27938 * tree-vect-stmts.c (vect_model_simple_cost): Make early
27940 (vect_model_promotion_demotion_cost): Likewise.
27941 (vectorizable_bswap): Guard cost modeling with !slp_node
27942 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
27944 (vectorizable_call): Likewise.
27945 (vectorizable_conversion): Likewise.
27946 (vectorizable_assignment): Likewise.
27947 (vectorizable_shift): Likewise.
27948 (vectorizable_operation): Likewise.
27949 (vectorizable_store): Likewise.
27950 (vectorizable_load): Likewise.
27951 (vectorizable_condition): Likewise.
27952 (vectorizable_comparison): Likewise.
27954 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
27957 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
27958 (ASAN_MARK): Fix fnspec to account for return value, change pointer
27959 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
27961 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
27963 PR middle-end/83665
27964 * params.def (inline-min-speedup): Increase from 8 to 15.
27965 (max-inline-insns-auto): Decrease from 40 to 30.
27966 * ipa-split.c (consider_split): Add some buffer for function to
27967 be considered inlining candidate.
27968 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
27971 2018-02-12 Richard Biener <rguenther@suse.de>
27973 PR tree-optimization/84037
27974 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
27975 matched stmts if we cannot swap the non-matched ones.
27977 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
27979 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
27980 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
27981 _mm_maskz_scalef_round_ss): New intrinsics.
27982 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
27983 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
27984 __builtin_ia32_scalefss_round): Remove.
27985 (__builtin_ia32_scalefsd_mask_round,
27986 __builtin_ia32_scalefss_mask_round): New intrinsics.
27987 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
27988 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
27989 ((match_operand:VF_128 2 "<round_nimm_predicate>"
27990 "<round_constraint>")): Changed to ...
27991 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
27992 "<round_scalar_constraint>")): ... this.
27993 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
27994 %0, %1, %2<round_op3>}"): Changed to ...
27995 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
27996 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
27997 %2<round_scalar_mask_op3>}"): ... this.
27998 * config/i386/subst.md (round_scalar_nimm_predicate): New.
28000 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
28002 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
28003 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
28004 (_mm_maskz_sqrt_round_ss): New intrinsics.
28005 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
28006 (__builtin_ia32_sqrtsd_mask_round)
28007 (__builtin_ia32_sqrtss_mask_round): New builtins.
28008 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
28009 (__builtin_ia32_sqrtss_round): Remove.
28010 (__builtin_ia32_sqrtsd_mask_round)
28011 (__builtin_ia32_sqrtss_mask_round): New builtins.
28012 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
28013 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
28014 ((match_operand:VF_128 1 "vector_operand"
28015 "xBm,<round_constraint>")): Changed to ...
28016 ((match_operand:VF_128 1 "vector_operand"
28017 "xBm,<round_scalar_constraint>")): ... this.
28018 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
28019 %0, %2, %<iptr>1<round_op3>}): Changed to ...
28020 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
28021 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
28022 %<iptr>1<round_scalar_mask_op3>}): ... this.
28023 ((set_attr "prefix" "<round_prefix>")): Changed to ...
28024 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
28026 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
28029 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
28030 Cast vec_cmpeq result to correct type.
28031 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
28032 Cast vec_cmpgt result to correct type.
28034 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
28036 * final.c (final_scan_insn_1): Renamed from...
28037 (final_scan_insn): ... this. New wrapper, to recover
28038 seen from the outermost call in recursive ones.
28039 * config/sparc/sparc.c (output_return): Drop seen from call.
28040 (output_sibcall): Likewise.
28041 * config/visium/visium.c (output_branch): Likewise.
28043 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
28045 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
28048 2018-02-10 Alan Modra <amodra@gmail.com>
28051 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
28052 Specify LR as an input.
28054 2018-02-10 Jakub Jelinek <jakub@redhat.com>
28057 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
28058 remove_member_access_dummy_vars): New functions.
28059 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
28060 lower_omp_1, execute_lower_omp): Use them.
28062 PR rtl-optimization/84308
28063 * shrink-wrap.c (spread_components): Release todo vector.
28065 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
28067 PR rtl-optimization/57193
28068 * ira-color.c (struct allocno_color_data): Add member
28069 conflict_allocno_hard_prefs.
28070 (update_conflict_allocno_hard_prefs): New.
28071 (bucket_allocno_compare_func): Add a preference based on
28072 conflict_allocno_hard_prefs.
28073 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
28074 (color_allocnos): Remove a dead code. Initiate
28075 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
28077 2018-02-09 Jakub Jelinek <jakub@redhat.com>
28080 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
28081 constraint from =wa to wa. Avoid a subreg on the output operand,
28082 instead use a pseudo and subreg it in a move.
28083 (p9_xxbrd_<mode>): Changed to ...
28084 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
28085 (p9_xxbrd_v2df): New expander.
28086 (p9_xxbrw_<mode>): Changed to ...
28087 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
28088 (p9_xxbrw_v4sf): New expander.
28090 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
28092 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
28094 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
28097 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
28098 multiply in 32-bit mode.
28099 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
28100 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
28103 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
28105 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
28106 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
28107 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
28108 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
28110 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
28113 * dwarf2out.c (is_trivial_indirect_ref): New function.
28114 (dwarf2out_late_global_decl): Do not generate a location
28115 attribute for variables that have a non-trivial DECL_VALUE_EXPR
28116 and that are not defined in the current unit.
28118 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
28120 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
28121 instead of a libcall for UNORDERED.
28123 2018-02-09 Tamar Christina <tamar.christina@arm.com>
28126 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
28127 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
28129 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28132 * config/s390/s390.c (s390_set_current_function): Invoke
28133 s390_indirect_branch_settings also if fndecl didn't change.
28135 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
28137 * config/rs6000/rs6000.md (blockage): Set length to zero.
28139 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
28141 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
28143 2018-02-09 Jakub Jelinek <jakub@redhat.com>
28146 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
28147 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
28151 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
28152 PARALLEL incoming that failed vt_get_decl_and_offset check.
28154 PR middle-end/84237
28155 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
28156 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
28158 (get_variable_section): For decls in named .bss* sections pass true as
28159 second argument to bss_initializer_p.
28161 2018-02-09 Marek Polacek <polacek@redhat.com>
28162 Jakub Jelinek <jakub@redhat.com>
28165 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
28166 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
28167 Sync some changes from cxx_fold_indirect_ref.
28169 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
28171 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
28173 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
28174 (BLOCK_INLINE_ENTRY_LABEL): New.
28175 (dwarf2out_var_location): Disregard inline entry markers.
28176 (inline_entry_data): New struct.
28177 (inline_entry_data_hasher): New hashtable type.
28178 (inline_entry_data_hasher::hash): New.
28179 (inline_entry_data_hasher::equal): New.
28180 (inline_entry_data_table): New variable.
28181 (add_high_low_attributes): Add DW_AT_entry_pc and
28182 DW_AT_GNU_entry_view attributes if a pending entry is found
28183 in inline_entry_data_table. Add old entry_pc attribute only
28184 if debug nonbinding markers are disabled.
28185 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
28186 markers are enabled.
28187 (block_within_block_p, dwarf2out_inline_entry): New.
28188 (dwarf2out_finish): Check that no entries remained in
28189 inline_entry_data_table.
28190 * final.c (reemit_insn_block_notes): Handle inline entry notes.
28191 (final_scan_insn, notice_source_line): Likewise.
28192 (rest_of_clean_state): Skip inline entry markers.
28193 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
28195 * gimple.c (gimple_build_debug_inline_entry): New.
28196 * gimple.h (enum gimple_debug_subcode): Add
28197 GIMPLE_DEBUG_INLINE_ENTRY.
28198 (gimple_build_debug_inline_entry): Declare.
28199 (gimple_debug_inline_entry_p): New.
28200 (gimple_debug_nonbind_marker_p): Adjust.
28201 * insn-notes.def (INLINE_ENTRY): New.
28202 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
28203 inline entry marker notes.
28204 (print_insn): Likewise.
28205 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
28206 (INSN_DEBUG_MARKER_KIND): Likewise.
28207 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
28208 * tree-inline.c (expand_call_inline): Build and insert
28209 debug_inline_entry stmt.
28210 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
28211 inline entry blocks early, if nonbind markers are enabled.
28212 (dump_scope_block): Dump fragment info.
28213 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
28214 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
28215 (gimple_build_debug_inline_entry): New.
28216 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
28217 Enable/disable inline entry points too.
28218 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
28219 (DEBUG_INSN): Describe inline entry markers.
28221 * common.opt (gvariable-location-views): New.
28222 (gvariable-location-views=incompat5): New.
28223 * config.in: Rebuilt.
28224 * configure: Rebuilt.
28225 * configure.ac: Test assembler for view support.
28226 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
28227 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
28228 * dwarf2out.c (var_loc_view): New typedef.
28229 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
28230 (dwarf2out_locviews_in_attribute): New.
28231 (dwarf2out_locviews_in_loclist): New.
28232 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
28233 (enum dw_line_info_opcode): Add LI_adv_address.
28234 (struct dw_line_info_table): Add view.
28235 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
28236 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
28237 (zero_view_p): New variable.
28238 (ZERO_VIEW_P): New macro.
28239 (output_asm_line_debug_info): New.
28240 (struct var_loc_node): Add view.
28241 (add_AT_view_list, AT_loc_list): New.
28242 (add_var_loc_to_decl): Add view param. Test it against last.
28243 (new_loc_list): Add view params. Record them.
28244 (AT_loc_list_ptr): Handle loc and view lists.
28245 (view_list_to_loc_list_val_node): New.
28246 (print_dw_val): Handle dw_val_class_view_list.
28247 (size_of_die): Likewise.
28248 (value_format): Likewise.
28249 (loc_list_has_views): New.
28250 (gen_llsym): Set vl_symbol too.
28251 (maybe_gen_llsym, skip_loc_list_entry): New.
28252 (dwarf2out_maybe_output_loclist_view_pair): New.
28253 (output_loc_list): Output view list or entries too.
28254 (output_view_list_offset): New.
28255 (output_die): Handle dw_val_class_view_list.
28256 (output_dwarf_version): New.
28257 (output_compilation_unit_header): Use it.
28258 (output_skeleton_debug_sections): Likewise.
28259 (output_rnglists, output_line_info): Likewise.
28260 (output_pubnames, output_aranges): Update version comments.
28261 (output_one_line_info_table): Output view numbers in asm comments.
28262 (dw_loc_list): Determine current endview, pass it to new_loc_list.
28263 Call maybe_gen_llsym.
28264 (loc_list_from_tree_1): Adjust.
28265 (add_AT_location_description): Create view list attribute if
28266 needed, check it's absent otherwise.
28267 (convert_cfa_to_fb_loc_list): Adjust.
28268 (maybe_emit_file): Call output_asm_line_debug_info for test.
28269 (dwarf2out_var_location): Reset views as needed. Precompute
28270 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
28271 attribute. Set view.
28272 (new_line_info_table): Reset next view.
28273 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
28274 (dwarf2out_source_line): Likewise. Output view resets and labels to
28275 the assembler, or select appropriate line info opcodes.
28276 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
28277 (optimize_string_length): Catch it. Adjust.
28278 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
28279 dw_val_class_view_list, and remove it if no longer needed.
28280 (hash_loc_list): Hash view numbers.
28281 (loc_list_hasher::equal): Compare them.
28282 (optimize_location_lists): Check whether a view list symbol is
28283 needed, and whether the locview attribute is present, and
28284 whether they match. Remove the locview attribute if no longer
28286 (index_location_lists): Call skip_loc_list_entry for test.
28287 (dwarf2out_finish): Call output_asm_line_debug_info for test.
28288 Use output_dwarf_version.
28289 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
28290 (struct dw_val_node): Add val_view_list.
28291 * final.c (SEEN_NEXT_VIEW): New.
28292 (set_next_view_needed): New.
28293 (clear_next_view_needed): New.
28294 (maybe_output_next_view): New.
28295 (final_start_function): Rename to...
28296 (final_start_function_1): ... this. Take pointer to FIRST,
28297 add SEEN parameter. Emit param bindings in the initial view.
28298 (final_start_function): Reintroduce SEEN-less interface.
28299 (final): Rename to...
28300 (final_1): ... this. Take SEEN parameter. Output final pending
28301 next view at the end.
28302 (final): Reintroduce seen-less interface.
28303 (final_scan_insn): Output pending next view before switching
28304 sections or ending a block. Mark the next view as needed when
28305 outputting variable locations. Notify debug backend of section
28306 changes, and of location view changes.
28307 (rest_of_handle_final): Adjust.
28308 * toplev.c (process_options): Autodetect value for debug variable
28309 location views option. Warn on incompat5 without -gdwarf-5.
28310 * doc/invoke.texi (gvariable-location-views): New.
28311 (gvariable-location-views=incompat5): New.
28312 (gno-variable-location-views): New.
28314 2018-02-08 David Malcolm <dmalcolm@redhat.com>
28316 PR tree-optimization/84136
28317 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
28318 that the result of find_edge is non-NULL.
28320 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
28323 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
28324 storing integer register in SImode. Fix cost of 256 and 512
28325 byte aligned SSE register store.
28327 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
28329 * config/i386/i386.c (ix86_multiplication_cost): Fix
28330 multiplication cost for TARGET_AVX512DQ.
28332 2018-02-08 Marek Polacek <polacek@redhat.com>
28334 PR tree-optimization/84238
28335 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
28338 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28340 PR tree-optimization/84265
28341 * tree-vect-stmts.c (vectorizable_store): Don't treat
28342 VMAT_CONTIGUOUS accesses as grouped.
28343 (vectorizable_load): Likewise.
28345 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28347 PR tree-optimization/81635
28348 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
28349 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
28350 (test_round_for_mask): New functions.
28351 (wide_int_cc_tests): Call test_round_for_mask.
28352 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
28353 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
28354 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
28355 range returned by get_range_info.
28357 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
28360 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
28361 * symtab.c: Include builtins.h
28362 (symtab_node::output_to_lto_symbol_table_p): Move here
28363 from lto-streamer-out.c:output_symbol_p.
28364 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
28365 (output_symbol_p): Move all logic to symtab.c
28366 (produce_symtab): Update.
28368 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28370 * config/s390/s390-opts.h (enum indirect_branch): Define.
28371 * config/s390/s390-protos.h (s390_return_addr_from_memory)
28372 (s390_indirect_branch_via_thunk)
28373 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
28374 (enum s390_indirect_branch_type): Define.
28375 * config/s390/s390.c (struct s390_frame_layout, struct
28376 machine_function): Remove.
28377 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
28378 (indirect_branch_table_label_no, indirect_branch_table_name):
28380 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
28381 (enum s390_indirect_branch_option): Define.
28382 (s390_return_addr_from_memory): New function.
28383 (s390_handle_string_attribute): New function.
28384 (s390_attribute_table): Add new attribute handler.
28385 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
28386 (s390_indirect_branch_via_thunk): New function.
28387 (s390_indirect_branch_via_inline_thunk): New function.
28388 (s390_function_ok_for_sibcall): When jumping via thunk disallow
28389 sibling call optimization for non z10 compiles.
28390 (s390_emit_call): Force indirect branch target to be a single
28391 register. Add r1 clobber for non-z10 compiles.
28392 (s390_emit_epilogue): Emit return jump via return_use expander.
28393 (s390_reorg): Handle JUMP_INSNs as execute targets.
28394 (s390_option_override_internal): Perform validity checks for the
28395 new command line options.
28396 (s390_indirect_branch_attrvalue): New function.
28397 (s390_indirect_branch_settings): New function.
28398 (s390_set_current_function): Invoke s390_indirect_branch_settings.
28399 (s390_output_indirect_thunk_function): New function.
28400 (s390_code_end): Implement target hook.
28401 (s390_case_values_threshold): Implement target hook.
28402 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
28404 * config/s390/s390.h (struct s390_frame_layout)
28405 (struct machine_function): Move here from s390.c.
28406 (TARGET_INDIRECT_BRANCH_NOBP_RET)
28407 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
28408 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
28409 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
28410 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
28411 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
28412 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
28413 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
28414 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
28415 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
28416 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
28417 (mnemonic attribute): Add values which aren't recognized
28419 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
28420 pattern for branch conversion. Fix mnemonic attribute.
28421 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
28422 indirect branch via thunk if requested.
28423 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
28424 ("*indirect_jump"): Disable for branch conversion using out of
28426 ("indirect_jump_via_thunk<mode>_z10")
28427 ("indirect_jump_via_thunk<mode>")
28428 ("indirect_jump_via_inlinethunk<mode>_z10")
28429 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
28430 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
28431 ("casesi_jump_via_inlinethunk<mode>_z10")
28432 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
28433 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
28434 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
28435 ("*indirect2_jump"): Disable for branch conversion.
28436 ("casesi_jump"): Turn into expander and expand patterns for branch
28438 ("return_use"): New expander.
28439 ("*return"): Emit return via thunk and rename it to ...
28440 ("*return<mode>"): ... this one.
28441 * config/s390/s390.opt: Add new options and and enum for the
28444 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
28446 * lra-constraints.c (match_reload): Unconditionally use
28447 gen_lowpart_SUBREG, rather than selecting between that
28448 and equivalent gen_rtx_SUBREG code.
28450 2018-02-08 Richard Biener <rguenther@suse.de>
28452 PR tree-optimization/84233
28453 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
28454 changed flag instead of boguously re-using phi_inserted.
28456 2018-02-08 Martin Jambor <mjambor@suse.cz>
28458 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
28459 static local variables.
28461 2018-02-08 Richard Biener <rguenther@suse.de>
28463 PR tree-optimization/84278
28464 * tree-vect-stmts.c (vectorizable_store): When looking for
28465 smaller vector types to perform grouped strided loads/stores
28466 make sure the mode is supported by the target.
28467 (vectorizable_load): Likewise.
28469 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
28471 * config/aarch64/aarch64.c (aarch64_components_for_bb):
28472 Increase LDP/STP opportunities by adding adjacent callee-saves.
28474 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
28476 PR rtl-optimization/84068
28477 PR rtl-optimization/83459
28478 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
28480 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
28482 PR tree-optimization/84224
28483 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
28484 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
28485 non-zero arguments.
28487 2018-02-07 Iain Sandoe <iain@codesourcery.com>
28490 * config/rs6000/altivec.md (*restore_world): Remove LR use.
28491 * config/rs6000/predicates.md (restore_world_operation): Adjust op
28492 count, remove one USE.
28494 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
28496 * doc/install.texi (Configuration): Document the
28497 --with-long-double-format={ibm,ieee} PowerPC configuration
28501 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
28502 Convert from define_expand to be define_insn_and_split. Rework
28503 float/double/_Float128 conversions to QI/HI/SImode to work with
28504 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
28505 conversions to QI/HImode types did a store and then a load to
28506 truncate the value. For conversions to VSX registers, don't split
28507 the insn, instead emit the code directly. Use the code iterator
28508 any_fix to combine signed and unsigned conversions.
28509 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
28510 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
28511 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
28512 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
28513 (fix_<mode>di2_hw): Likewise.
28514 (fixuns_<mode>di2_hw): Likewise.
28515 (fix_<mode>si2_hw): Likewise.
28516 (fixuns_<mode>si2_hw): Likewise.
28517 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
28518 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
28519 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
28520 fix<uns>_trunc<SFDF:mode>si2_p8.
28521 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
28522 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
28523 (fix<uns>_<mode>_mem): Likewise.
28524 (fctiw<u>z_<mode>_mem): Likewise.
28525 (fix<uns>_<mode>_mem): Likewise.
28526 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
28527 the register allocator from doing a direct move to the GPRs to do
28528 a store, and instead use the ISA 3.0 store byte/half-word from
28529 vector register instruction. For IEEE 128-bit floating point,
28530 also optimize stores of 32-bit ints.
28531 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
28533 2018-02-07 Alan Hayward <alan.hayward@arm.com>
28535 * genextract.c (push_pathstr_operand): New function to support
28537 (walk_rtx): Call push_pathstr_operand.
28538 (print_path): Support [a-zA-Z].
28540 2018-02-07 Richard Biener <rguenther@suse.de>
28542 PR tree-optimization/84037
28543 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
28544 (cse_and_gimplify_to_preheader): Declare.
28545 (vect_get_place_in_interleaving_chain): Likewise.
28546 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
28548 (_loop_vec_info::~_loop_vec_info): Delete it.
28549 (cse_and_gimplify_to_preheader): New function.
28550 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
28551 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
28552 (vectorizable_load): Likewise. For grouped stores always base
28553 the IV on the first element.
28554 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
28555 condition before gimplifying.
28557 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28559 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
28560 *DIV_EXPR and *MOD_EXPR.
28562 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
28565 * config/i386/i386.c (ix86_option_override_internal): Mask out
28566 the CF_SET bit when checking -fcf-protection.
28568 2018-02-07 Tom de Vries <tom@codesourcery.com>
28571 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
28574 2018-02-07 Richard Biener <rguenther@suse.de>
28576 PR tree-optimization/84204
28577 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
28580 PR tree-optimization/84205
28581 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
28582 special-case isl_ast_op_zdiv_r.
28584 PR tree-optimization/84223
28585 * graphite-scop-detection.c (gather_bbs::before_dom_children):
28586 Only add conditions from within the region.
28587 (gather_bbs::after_dom_children): Adjust.
28589 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
28592 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
28593 * config/avr/avr.md: Only post-reload split REG-REG moves if
28594 either register is GENERAL_REG_P.
28596 2018-02-07 Jakub Jelinek <jakub@redhat.com>
28598 PR tree-optimization/84235
28599 * tree-ssa-scopedtables.c
28600 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
28601 if the subtraction is performed in floating point type where NaNs are
28602 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
28603 build 1. Formatting fix.
28605 2018-02-06 Jakub Jelinek <jakub@redhat.com>
28608 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
28609 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
28610 and skip it regardless of bb boundaries. Use CALL_P macro,
28611 don't test INSN_P (insn) together with CALL_P or JUMP_P check
28612 unnecessarily, formatting fix.
28614 2018-02-06 Michael Collison <michael.collison@arm.com>
28616 * config/arm/thumb2.md:
28617 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
28618 (*thumb_mov_notscc): Ditto.
28620 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
28623 * config/rs6000/rs6000.md (su code attribute): Use "u" for
28624 unsigned_fix, not "s".
28626 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28628 * configure.ac (gcc_fn_eh_frame_ro): New function.
28629 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
28630 correct .eh_frame permissions.
28631 * configure: Regenerate.
28633 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
28635 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
28636 irrelevant options.
28638 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28640 * config/rs6000/rs6000.c (rs6000_option_override_internal):
28641 Display warning message for -mno-speculate-indirect-jumps.
28643 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
28645 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
28647 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
28649 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
28651 PR tree-optimization/84225
28652 * tree-eh.c (find_trapping_overflow): Only call
28653 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
28655 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28658 * config/i386/i386.c: Reimplement the check of possible options
28659 -mibt/-mshstk conbination. Change error messages.
28660 * doc/invoke.texi: Fix a typo: remove extra '='.
28662 2018-02-06 Marek Polacek <polacek@redhat.com>
28664 PR tree-optimization/84228
28665 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
28667 2018-02-06 Tamar Christina <tamar.christina@arm.com>
28670 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
28671 emitted arch directives.
28672 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
28673 __ARM_FEATURE_COPROC before changing architectures.
28675 2018-02-06 Richard Biener <rguenther@suse.de>
28677 * config/i386/i386.c (print_reg): Fix typo.
28678 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
28680 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
28682 * configure: Regenerate.
28684 2018-02-05 Martin Sebor <msebor@redhat.com>
28686 PR tree-optimization/83369
28687 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
28690 2018-02-05 Martin Liska <mliska@suse.cz>
28692 * doc/invoke.texi: Cherry-pick upstream r323995.
28694 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
28696 * ira.c (ira_init_register_move_cost): Adjust comment.
28698 2018-02-05 Martin Liska <mliska@suse.cz>
28700 PR gcov-profile/84137
28701 * doc/gcov.texi: Fix typo in documentation.
28703 2018-02-05 Martin Liska <mliska@suse.cz>
28705 PR gcov-profile/83879
28706 * doc/gcov.texi: Document necessity of --dynamic-list-data when
28707 using dlopen functionality.
28709 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
28711 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
28712 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
28713 _mm_maskz_range_ss, _mm_mask_range_round_ss,
28714 _mm_maskz_range_round_ss): New intrinsics.
28715 (__builtin_ia32_rangesd128_round)
28716 (__builtin_ia32_rangess128_round): Remove.
28717 (__builtin_ia32_rangesd128_mask_round,
28718 __builtin_ia32_rangess128_mask_round): New builtins.
28719 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
28720 __builtin_ia32_rangess128_round): Remove.
28721 (__builtin_ia32_rangesd128_mask_round,
28722 __builtin_ia32_rangess128_mask_round): New builtins.
28723 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
28724 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
28725 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
28726 "<round_saeonly_constraint>")): Changed to ...
28727 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
28728 "<round_saeonly_scalar_constraint>")): ... this.
28729 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
28730 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
28731 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
28732 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
28733 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
28735 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
28737 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
28739 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
28740 Remove all values except native, 8540 and 8548.
28742 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
28744 * config/i386/i386.c (ix86_output_function_return): Pass
28745 INVALID_REGNUM, instead of -1, as invalid register number to
28746 indirect_thunk_name and output_indirect_thunk.
28748 2018-02-02 Julia Koval <julia.koval@intel.com>
28750 * config.gcc: Add -march=icelake.
28751 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
28752 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
28753 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
28754 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
28755 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
28756 (processor_target_table): Add icelake.
28757 (ix86_option_override_internal): Handle new PTAs.
28758 (get_builtin_code_for_version): Handle icelake.
28759 (M_INTEL_COREI7_ICELAKE): New.
28760 (fold_builtin_cpu): Handle icelake.
28761 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
28762 * doc/invoke.texi: Add -march=icelake.
28764 2018-02-02 Julia Koval <julia.koval@intel.com>
28766 * config/i386/i386.c (ix86_option_override_internal): Change flags type
28767 to wide_int_bitmask.
28768 * wide-int-bitmask.h: New.
28770 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
28773 * config/i386/i386.md: Replace Pmode with word_mode in
28774 builtin_setjmp_setup and builtin_longjmp to support x32.
28776 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
28780 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
28782 (rs6000_supported_cpu_names): New static variable.
28783 (linux_cpu_translation_table): Likewise.
28784 (elf_platform) <cpu>: Define new static variable and use it.
28785 Translate kernel AT_PLATFORM name to canonical name if needed.
28786 Error if platform name is unknown.
28788 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
28791 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
28793 2018-02-01 Jeff Law <law@redhat.com>
28796 * config/i386/i386.c (release_scratch_register_on_entry): Add new
28797 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
28798 the scratch if RELEASE_VIA_POP is false.
28799 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
28800 If we have to save a temporary register, decrement SIZE appropriately.
28801 Pass new arguments to release_scratch_register_on_entry.
28802 (ix86_adjust_stack_and_probe): Likewise.
28803 (ix86_emit_probe_stack_range): Pass new arguments to
28804 release_scratch_register_on_entry.
28806 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
28808 PR rtl-optimization/84157
28809 * combine.c (change_zero_ext): Use REG_P predicate in
28810 front of HARD_REGISTER_P predicate.
28812 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
28814 * config/avr/avr.c (avr_option_override): Move disabling of
28815 -fdelete-null-pointer-checks to...
28816 * common/config/avr/avr-common.c (avr_option_optimization_table):
28819 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28821 PR tree-optimization/81635
28822 * tree-data-ref.c (split_constant_offset_1): For types that
28823 wrap on overflow, try to use range info to prove that wrapping
28826 2018-02-01 Renlin Li <renlin.li@arm.com>
28829 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
28830 TAILCALL_ADDR_REGS.
28831 (aarch64_register_move_cost): Likewise.
28832 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
28833 TAILCALL_ADDR_REGS.
28834 (REG_CLASS_NAMES): Likewise.
28835 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
28836 TAILCALL_ADDR_REGS. Remove IP registers.
28837 * config/aarch64/aarch64.md (Ucs): Update register constraint.
28839 2018-02-01 Richard Biener <rguenther@suse.de>
28841 * domwalk.h (dom_walker::dom_walker): Add additional constructor
28842 for specifying RPO order and allow NULL for that.
28843 * domwalk.c (dom_walker::dom_walker): Likewise.
28844 (dom_walker::walk): Handle NULL RPO order.
28845 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
28847 (rewrite_update_dom_walker): Likewise.
28848 (mark_def_dom_walker): Likewise.
28850 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28852 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
28853 (aarch64_maybe_expand_sve_subreg_move): Declare.
28854 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
28855 * config/aarch64/predicates.md (aarch64_any_register_operand): New
28857 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
28858 that are semantically a reverse operation.
28859 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
28860 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
28861 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
28863 (aarch64_can_change_mode_class): For big-endian, forbid changes
28864 between two SVE modes if they have different element sizes.
28866 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28868 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
28869 the TImode handling for big-endian targets.
28871 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28873 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
28874 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
28876 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
28877 Remove BSWAP handing for big-endian targets and use the form of
28878 LD1RQ appropariate for the mode.
28880 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28882 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
28883 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
28884 duplicated element.
28886 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28889 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
28890 check for operands that need to go through aarch64_sve_reload_be.
28892 2018-02-01 Jakub Jelinek <jakub@redhat.com>
28894 PR tree-optimization/81661
28895 PR tree-optimization/84117
28896 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
28897 * tree-eh.c: Include gimplify.h.
28898 (find_trapping_overflow, replace_trapping_overflow,
28899 rewrite_to_non_trapping_overflow): New functions.
28900 * tree-vect-loop.c: Include tree-eh.h.
28901 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
28902 * tree-data-ref.c: Include tree-eh.h.
28903 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
28905 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
28907 PR rtl-optimization/84123
28908 * combine.c (change_zero_ext): Check if hard register satisfies
28909 can_change_dest_mode before calling gen_lowpart_SUBREG.
28911 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
28914 * ira.c (ira_init_register_move_cost): Remove assert.
28916 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28918 PR rtl-optimization/84071
28919 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
28920 * doc/tm.texi: Regenerate.
28922 2018-01-31 Richard Biener <rguenther@suse.de>
28924 PR tree-optimization/84132
28925 * tree-data-ref.c (analyze_miv_subscript): Properly
28926 check whether evolution_function_is_affine_multivariate_p
28927 before calling gcd_of_steps_may_divide_p.
28929 2018-01-31 Julia Koval <julia.koval@intel.com>
28932 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28933 * config/i386/i386.md (rdpid_rex64) New.
28934 (rdpid): Make 32bit only.
28936 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
28939 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
28940 an IDENTIFIER_NODE for FUNCTION_TYPE's.
28942 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28945 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
28947 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
28949 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28951 PR rtl-optimization/84071
28952 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
28953 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
28955 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
28957 * config/arc/arc.c (arc_handle_aux_attribute): New function.
28958 (arc_attribute_table): Add 'aux' attribute.
28959 (arc_in_small_data_p): Consider aux like variables.
28960 (arc_is_aux_reg_p): New function.
28961 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
28962 (arc_get_aux_arg): New function.
28963 (prepare_move_operands): Handle aux-register access.
28964 (arc_handle_aux_attribute): New function.
28965 * doc/extend.texi (ARC Variable attributes): Add subsection.
28967 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
28969 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
28970 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
28971 (arc_attribute_table): Add 'uncached' attribute.
28972 (arc_print_operand): Print '.di' flag for uncached memory
28974 (arc_in_small_data_p): Do not consider for small data the uncached
28976 (arc_is_uncached_mem_p): New function.
28977 * config/arc/predicates.md (compact_store_memory_operand): Check
28978 for uncached memory accesses.
28979 (nonvol_nonimm_operand): Likewise.
28980 * doc/extend.texi (ARC Type Attribute): New subsection.
28982 2018-01-31 Jakub Jelinek <jakub@redhat.com>
28985 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
28986 falign-loops=): Add Optimization flag.
28988 2018-01-30 Jeff Law <law@redhat.com>
28991 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
28992 INT_REGISTERS_SAVED. Check it prior to calling
28993 get_scratch_register_on_entry.
28994 (ix86_adjust_stack_and_probe): Similarly.
28995 (ix86_emit_probe_stack_range): Similarly.
28996 (ix86_expand_prologue): Corresponding changes.
28998 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29001 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
29002 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
29004 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
29007 * lra-constraints.c (curr_insn_transform): Process AND in the
29010 2018-01-30 Jakub Jelinek <jakub@redhat.com>
29012 PR rtl-optimization/83986
29013 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
29014 dependence against last_pending_memory_flush in addition to
29015 pending_jump_insns.
29017 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
29019 PR tree-optimization/81611
29020 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
29023 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29026 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
29029 2018-01-30 Richard Biener <rguenther@suse.de>
29030 Jakub Jelinek <jakub@redhat.com>
29032 PR tree-optimization/84111
29033 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
29034 inner loops added during recursion, as they don't have up-to-date
29037 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
29040 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
29041 (can_inline_edge_by_limits_p): ... here.
29042 (can_early_inline_edge_p, check_callers,
29043 update_caller_keys, update_callee_keys, recursive_inlining,
29044 add_new_edges_to_heap, speculation_useful_p,
29045 inline_small_functions,
29046 inline_small_functions, flatten_function,
29047 inline_to_all_callers_1): Update.
29049 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
29051 * profile-count.c (profile_count::combine_with_ipa_count): Handle
29054 2018-01-30 Richard Biener <rguenther@suse.de>
29056 PR tree-optimization/83008
29057 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
29058 invariant and constant vector uses in stmts when they need
29059 more than one stmt.
29061 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29064 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
29065 * configure: Regenerate.
29067 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29069 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
29071 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
29072 Use gen_rtx_REG rather than gen_lowpart.
29074 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29076 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
29077 rather than 0 when creating partial subregs.
29079 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
29081 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
29084 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
29087 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
29088 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
29089 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
29090 flags. This restores the settings used before the 2017-07-24.
29091 Turning off pre increment/decrement/modify allows IVOPTS to
29092 optimize DF/SF loops where the index is an int.
29094 2018-01-29 Richard Biener <rguenther@suse.de>
29095 Kelvin Nilsen <kelvin@gcc.gnu.org>
29098 * tree-vect-stmts.c (vectorizable_call): Don't call
29099 targetm.vectorize_builtin_md_vectorized_function if callee is
29102 2018-01-22 Carl Love <cel@us.ibm.com>
29104 * doc/extend.tex: Fix typo in second arg in
29105 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
29107 2018-01-29 Richard Biener <rguenther@suse.de>
29109 PR tree-optimization/84086
29110 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
29111 (flush_ssaname_freelist): When SSA names were released reset
29112 the SCEV hash table.
29114 2018-01-29 Richard Biener <rguenther@suse.de>
29116 PR tree-optimization/84057
29117 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
29118 removed paths when removing edges.
29120 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
29122 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
29123 -mfunction-return=@var{choice}.
29125 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
29127 PR diagnostic/84034
29128 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
29129 Handle CR like TAB.
29130 (layout::print_source_line): Likewise.
29131 (test_get_line_width_without_trailing_whitespace): Add test cases.
29133 2018-01-27 Jakub Jelinek <jakub@redhat.com>
29135 PR middle-end/84040
29136 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
29139 2018-01-26 Jim Wilson <jimw@sifive.com>
29141 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
29143 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
29146 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29148 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
29149 and CMP + SUB-immediate -> SUBS.
29151 2018-01-26 Martin Sebor <msebor@redhat.com>
29153 PR tree-optimization/83896
29154 * tree-ssa-strlen.c (get_string_len): Rename...
29155 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
29156 Avoid assuming length is constant.
29157 (handle_char_store): Use HOST_WIDE_INT for string length.
29159 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
29162 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
29163 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
29165 2018-01-26 Richard Biener <rguenther@suse.de>
29167 PR rtl-optimization/84003
29168 * dse.c (record_store): Only record redundant stores when
29169 the earlier store aliases at least all accesses the later one does.
29171 2018-01-26 Jakub Jelinek <jakub@redhat.com>
29173 PR rtl-optimization/83985
29174 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
29175 REG_CFA_RESTORE insns.
29176 (delete_unmarked_insns): Don't ignore separate shrink wrapping
29177 REG_CFA_RESTORE insns here.
29180 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
29181 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
29183 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29185 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
29186 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
29187 (arc_init): Likewise.
29188 (arc_override_options): Likewise.
29189 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
29191 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
29193 * config/arc/arc.h (TARGET_DBNZ): Define.
29194 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
29195 properly set the tune attribute.
29196 (dbnz): Use TARGET_DBNZ guard.
29197 * config/arc/arc.opt (mtune): Add core3 option.
29199 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29201 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
29202 recognize new pic like addresses.
29203 (arc_delegitimize_address): Clean up.
29205 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29207 * config/arc/arc-arches.def: Option mrf16 valid for all
29209 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
29210 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
29211 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
29212 * config/arc/arc-tables.opt: Regenerate.
29213 * config/arc/arc.c (arc_conditional_register_usage): Handle
29214 reduced register file case.
29215 (arc_file_start): Set must have build attributes.
29216 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
29217 mrf16 option value.
29218 * config/arc/arc.opt (mrf16): Add new option.
29219 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
29220 * config/arc/genmultilib.awk: Handle new mrf16 option.
29221 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
29222 * config/arc/t-multilib: Regenerate.
29223 * doc/invoke.texi (ARC Options): Document mrf16 option.
29225 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29227 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
29228 * config/arc/arc.c (arc_handle_secure_attribute): New function.
29229 (arc_attribute_table): Add 'secure_call' attribute.
29230 (arc_print_operand): Print secure call operand.
29231 (arc_function_ok_for_sibcall): Don't optimize tail calls when
29233 (arc_is_secure_call_p): New function. * config/arc/arc.md
29234 (call_i): Add support for sjli instruction.
29235 (call_value_i): Likewise.
29236 * config/arc/constraints.md (Csc): New constraint.
29238 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
29239 John Eric Martin <John.Martin@emmicro-us.com>
29241 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
29242 * config/arc/arc.c (_arc_jli_section): New struct.
29243 (arc_jli_section): New type.
29244 (rc_jli_sections): New static variable.
29245 (arc_handle_jli_attribute): New function.
29246 (arc_attribute_table): Add jli_always and jli_fixed attribute.
29247 (arc_file_end): New function.
29248 (TARGET_ASM_FILE_END): Define.
29249 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
29250 (arc_add_jli_section): New function.
29251 (jli_call_scan): Likewise.
29252 (arc_reorg): Call jli_call_scan.
29253 (arc_output_addsi): Remove 'S' from printing asm operand.
29254 (arc_is_jli_call_p): New function.
29255 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
29257 (movhi_insn): Likewise.
29258 (movsi_insn): Likewise.
29259 (movsi_set_cc_insn): Likewise.
29260 (loadqi_update): Likewise.
29261 (load_zeroextendqisi_update): Likewise.
29262 (load_signextendqisi_update): Likewise.
29263 (loadhi_update): Likewise.
29264 (load_zeroextendhisi_update): Likewise.
29265 (load_signextendhisi_update): Likewise.
29266 (loadsi_update): Likewise.
29267 (loadsf_update): Likewise.
29268 (movsicc_insn): Likewise.
29269 (bset_insn): Likewise.
29270 (bxor_insn): Likewise.
29271 (bclr_insn): Likewise.
29272 (bmsk_insn): Likewise.
29273 (bicsi3_insn): Likewise.
29274 (cmpsi_cc_c_insn): Likewise.
29275 (movsi_ne): Likewise.
29276 (movsi_cond_exec): Likewise.
29277 (clrsbsi2): Likewise.
29278 (norm_f): Likewise.
29288 (call_i): Remove 'S' asm letter, add jli instruction.
29289 (call_value_i): Likewise.
29290 * config/arc/arc.op (mjli-always): New option.
29291 * config/arc/constraints.md (Cji): New constraint.
29292 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
29294 (subsf3_fpx): Likewise.
29295 (mulsf3_fpx): Likewise.
29296 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
29298 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
29299 function attrbutes.
29300 * doc/invoke.texi (ARC): Document mjli-always option.
29302 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
29304 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
29305 avoid addition with 0 and use incw and decw where possible.
29307 2018-01-26 Richard Biener <rguenther@suse.de>
29309 PR tree-optimization/81082
29310 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
29311 association if it requires casting to unsigned.
29312 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
29313 from fold_plusminus_mult_expr to catch important cases late when
29314 range info is available.
29316 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29318 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
29319 * configure.ac (hidden_linkonce): New test.
29320 * configure: Regenerate.
29321 * config.in: Regenerate.
29323 2018-01-26 Julia Koval <julia.koval@intel.com>
29325 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
29326 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
29327 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
29328 _mm_mask_bitshuffle_epi64_mask): Fix type.
29329 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
29330 USI_FTYPE_V4DI_V4DI_USI): Remove.
29331 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
29332 __builtin_ia32_vpshufbitqmb256_mask,
29333 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
29334 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
29335 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
29337 2018-01-26 Alan Modra <amodra@gmail.com>
29340 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
29341 UNSPEC_VBPERMQ. Sort other unspecs.
29343 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
29345 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
29347 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
29349 PR middle-end/83055
29350 * predict.c (drop_profile): Do not push/pop cfun; update also
29352 (handle_missing_profiles): Fix logic looking for zero profiles.
29354 2018-01-25 Jakub Jelinek <jakub@redhat.com>
29356 PR middle-end/83977
29357 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
29358 on functions with #pragma omp declare simd or functions with simd
29360 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
29361 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
29362 Remove trailing \n from warning_at calls.
29364 2018-01-25 Tom de Vries <tom@codesourcery.com>
29367 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
29368 for neutered workers.
29370 2018-01-24 Joseph Myers <joseph@codesourcery.com>
29373 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
29374 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
29376 2018-01-24 Jeff Law <law@redhat.com>
29379 * i386.c (get_probe_interval): Move to earlier point.
29380 (ix86_compute_frame_layout): If -fstack-clash-protection and
29381 the frame is larger than the probe interval, then use pushes
29382 to save registers rather than reg->mem moves.
29383 (ix86_expand_prologue): Remove conditional for int_registers_saved
29386 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
29389 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
29390 min/max for never referenced object.
29392 2018-01-24 Jakub Jelinek <jakub@redhat.com>
29394 PR middle-end/83977
29395 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
29397 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
29398 attributes from DECL_ATTRIBUTES (decl) without affecting
29399 DECL_ATTRIBUTES (current_function_decl).
29400 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
29401 functions with non-NULL DECL_ABSTRACT_ORIGIN.
29403 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
29405 PR tree-optimization/83979
29406 * fold-const.c (fold_comparison): Use constant_boolean_node
29407 instead of boolean_{true,false}_node.
29409 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
29411 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
29414 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29416 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29417 Simplify the clause that sets the length attribute.
29418 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29419 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
29420 clause that sets the length attribute.
29421 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29423 2018-01-24 Tom de Vries <tom@codesourcery.com>
29426 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
29427 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
29428 Add strict parameter.
29429 (prevent_branch_around_nothing): Insert dummy insn between branch to
29430 label and label with no ptx insn inbetween.
29431 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
29433 2018-01-24 Tom de Vries <tom@codesourcery.com>
29436 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
29437 for neutered threads in warp.
29438 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
29440 2018-01-24 Richard Biener <rguenther@suse.de>
29442 PR tree-optimization/83176
29443 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
29446 2018-01-24 Richard Biener <rguenther@suse.de>
29448 PR tree-optimization/82819
29449 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
29450 code generating pluses that are no-ops in the target precision.
29452 2018-01-24 Richard Biener <rguenther@suse.de>
29454 PR middle-end/84000
29455 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
29457 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29459 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
29460 to merge probabilities.
29461 * predict.c (probably_never_executed): Also mark as cold functions
29462 with global 0 profile and guessed local profile.
29463 * profile-count.c (profile_probability::combine_with_count): New
29465 * profile-count.h (profile_probability::operator*,
29466 profile_probability::operator*=, profile_probability::operator/,
29467 profile_probability::operator/=): Reduce precision to adjusted
29468 and set value to guessed on contradictory divisions.
29469 (profile_probability::combine_with_freq): Remove.
29470 (profile_probability::combine_wiht_count): Declare.
29471 (profile_count::force_nonzero):: Set to adjusted.
29472 (profile_count::probability_in):: Set quality to adjusted.
29473 * tree-ssa-tail-merge.c (replace_block_by): Use
29474 combine_with_count.
29476 2018-01-23 Andrew Waterman <andrew@sifive.com>
29477 Jim Wilson <jimw@sifive.com>
29479 * config/riscv/riscv.c (riscv_stack_boundary): New.
29480 (riscv_option_override): Set riscv_stack_boundary. Handle
29481 riscv_preferred_stack_boundary_arg.
29482 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
29483 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
29484 (STACK_BOUNDARY): Set to riscv_stack_boundary.
29485 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
29486 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
29487 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
29489 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
29492 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
29493 of struct ix86_frame.
29494 (ix86_expand_epilogue): Likewise. Add a local variable for
29495 the reg_save_offset field in struct ix86_frame.
29497 2018-01-23 Bin Cheng <bin.cheng@arm.com>
29499 PR tree-optimization/82604
29500 * tree-loop-distribution.c (enum partition_kind): New enum item
29501 PKIND_PARTIAL_MEMSET.
29502 (partition_builtin_p): Support above new enum item.
29503 (generate_code_for_partition): Ditto.
29504 (compute_access_range): Differentiate cases that equality can be
29505 proven at all loops, the innermost loops or no loops.
29506 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
29507 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
29508 (finalize_partitions, distribute_loop): Don't fuse partition of
29509 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
29510 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
29511 parloop is enabled.
29513 2018-01-23 Martin Liska <mliska@suse.cz>
29515 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
29516 order to ignore the predictor.
29517 (PRED_POLYMORPHIC_CALL): Likewise.
29518 (PRED_RECURSIVE_CALL): Likewise.
29520 2018-01-23 Martin Liska <mliska@suse.cz>
29522 * tree-profile.c (tree_profiling): Print function header to
29523 aware reader which function we are working on.
29524 * value-prof.c (gimple_find_values_to_profile): Do not print
29525 not interesting value histograms.
29527 2018-01-23 Martin Liska <mliska@suse.cz>
29529 * profile-count.h (enum profile_quality): Add
29530 profile_uninitialized as the first value. Do not number values
29531 as they are zero based.
29532 (profile_count::verify): Update sanity check.
29533 (profile_probability::verify): Likewise.
29535 2018-01-23 Nathan Sidwell <nathan@acm.org>
29537 * doc/invoke.texi (ffor-scope): Deprecate.
29539 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29541 PR tree-optimization/83510
29542 * domwalk.c (set_all_edges_as_executable): New function.
29543 (dom_walker::dom_walker): Convert bool param
29544 "skip_unreachable_blocks" to enum reachability. Move setup of
29545 edge flags to set_all_edges_as_executable and only do it when
29546 reachability is REACHABLE_BLOCKS.
29547 * domwalk.h (enum dom_walker::reachability): New enum.
29548 (dom_walker::dom_walker): Convert bool param
29549 "skip_unreachable_blocks" to enum reachability.
29550 (set_all_edges_as_executable): New decl.
29551 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
29552 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
29554 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
29555 but converting true to REACHABLE_BLOCKS.
29556 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
29558 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
29559 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
29560 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
29562 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
29563 if check_all_array_refs will be called.
29565 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29567 * tree.c (selftest::test_location_wrappers): Add more test
29570 2018-01-23 David Malcolm <dmalcolm@redhat.com>
29572 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
29573 (selftest::test_bit_in_range): Likewise.
29575 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
29578 * doc/sourcebuild.texi (vect_float): Say that the selector
29579 only describes the situation when -funsafe-math-optimizations is on.
29580 (vect_float_strict): Document.
29582 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
29584 PR tree-optimization/83965
29585 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
29586 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
29587 instead of checking only for a reduction.
29588 (vect_recog_widen_sum_pattern): Likewise.
29590 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29592 * predict.c (probably_never_executed): Only use precise profile info.
29593 (compute_function_frequency): Skip after inlining hack since we now
29594 have quality checking.
29596 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
29598 * profile-count.h (profile_probability::very_unlikely,
29599 profile_probability::unlikely, profile_probability::even): Set
29600 precision to guessed.
29602 2018-01-23 Richard Biener <rguenther@suse.de>
29604 PR tree-optimization/83963
29605 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29606 Properly terminate dominator walk when crossing the exit edge not
29607 when visiting its source block.
29609 2018-01-23 Jakub Jelinek <jakub@redhat.com>
29612 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
29613 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
29615 2018-01-22 Jakub Jelinek <jakub@redhat.com>
29617 PR tree-optimization/83957
29618 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
29619 semicolon after for body surrounded by braces.
29621 PR tree-optimization/83081
29622 * profile-count.h (profile_probability::split): New method.
29623 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
29624 Use profile_probability::split.
29625 (do_compare_rtx_and_jump): Fix adjustment of probabilities
29626 when splitting a single conditional jump into 2.
29628 2018-01-22 David Malcolm <dmalcolm@redhat.com>
29630 PR tree-optimization/69452
29631 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
29634 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29636 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
29637 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
29638 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
29640 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29642 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
29644 * config/rl78/rl78.md (movdi): New define_expand.
29645 * config/rl78/rl78.c (rl78_split_movdi): New function.
29647 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
29650 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
29652 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
29653 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
29654 128-bit to produce an UNSPEC move to get the double word with the
29655 signbit and then a shift directly to do signbit.
29656 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
29657 implementation with a new version that just does either a direct
29658 move or a regular move. Move memory interface to separate insns.
29659 Move insns so they are next to the expander.
29660 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
29661 with signbit move. Split big and little endian case.
29662 (signbit<mode>2_dm_mem_le): Likewise.
29663 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
29664 (signbit<mode>2_dm2): Likewise.
29666 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29668 * config/rl78/rl78.md (anddi3): New define_expand.
29670 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29672 * config/rl78/rl78.md (umindi3): New define_expand.
29674 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29676 * config/rl78/rl78.md (smindi3): New define_expand.
29678 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29680 * config/rl78/rl78.md (smaxdi3): New define_expand.
29682 2018-01-22 Carl Love <cel@us.ibm.com>
29684 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
29685 LVX_V1TI): Add macro expansion.
29686 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
29687 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
29688 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
29689 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
29690 Change check to determine if the instruction is a byte reversing
29691 entry. Fix typo in comment.
29692 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
29693 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
29694 Add def_builtin calls for new builtins.
29695 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
29696 Add define_insn expansion.
29698 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29700 * config/rl78/rl78.md (umaxdi3): New define_expand.
29702 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
29704 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
29705 for non-QImode registers.
29707 2018-01-22 Richard Biener <rguenther@suse.de>
29709 PR tree-optimization/83963
29710 * graphite-scop-detection.c (scop_detection::get_sese): Delay
29711 including the loop exit block.
29712 (scop_detection::merge_sese): Likewise.
29713 (scop_detection::add_scop): Do it here instead.
29715 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29717 * doc/sourcebuild.texi (arm_softfloat): Document.
29719 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
29722 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
29723 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
29724 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
29726 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29727 David Edelsohn <dje.gcc@gmail.com>
29730 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29731 Change "crset eq" to "crset 2".
29732 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29733 (*call_indirect_aix<mode>_nospec): Likewise.
29734 (*call_value_indirect_aix<mode>_nospec): Likewise.
29735 (*call_indirect_elfv2<mode>_nospec): Likewise.
29736 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
29737 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
29738 change assembly output from . to $.
29739 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29740 (indirect_jump<mode>_nospec): Change assembly output from . to $.
29741 (*tablejump<mode>_internal1_nospec): Likewise.
29743 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
29746 * config/sh/sh_optimize_sett_clrt.cc:
29747 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
29749 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
29751 PR tree-optimization/83940
29752 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
29753 offset_dt to vect_constant_def rather than vect_unknown_def_type.
29754 (vect_check_load_store_mask): Add a mask_dt_out parameter and
29755 use it to pass back the definition type.
29756 (vect_check_store_rhs): Likewise rhs_dt_out.
29757 (vect_build_gather_load_calls): Add a mask_dt argument and use
29758 it instead of a call to vect_is_simple_use.
29759 (vectorizable_store): Update calls to vect_check_load_store_mask
29760 and vect_check_store_rhs. Use the dt returned by the latter instead
29761 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
29762 instead of calls to vect_is_simple_use. Pass the scalar rather
29763 than the vector operand to vect_is_simple_use when handling
29764 second and subsequent copies of an rhs value.
29765 (vectorizable_load): Update calls to vect_check_load_store_mask
29766 and vect_build_gather_load_calls. Use the cached mask_dt and
29767 gs_info.offset_dt instead of calls to vect_is_simple_use.
29769 2018-01-20 Jakub Jelinek <jakub@redhat.com>
29771 PR middle-end/83945
29772 * tree-emutls.c: Include gimplify.h.
29773 (lower_emutls_2): New function.
29774 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
29775 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
29776 it before further processing.
29779 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
29780 UINTVAL (trueop1) instead of INTVAL (op1).
29782 2018-01-19 Jakub Jelinek <jakub@redhat.com>
29786 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
29787 INCOMING_FRAME_SP_OFFSET if not defined.
29788 (scan_trace): Add ENTRY argument. If true and
29789 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
29790 emit a note to adjust the CFA offset.
29791 (create_cfi_notes): Adjust scan_trace callers.
29792 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
29793 INCOMING_FRAME_SP_OFFSET in the CIE.
29794 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
29795 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
29797 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
29798 * doc/tm.texi: Regenerated.
29800 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29802 PR rtl-optimization/83147
29803 * lra-constraints.c (remove_inheritance_pseudos): Use
29804 lra_substitute_pseudo_within_insn.
29806 2018-01-19 Tom de Vries <tom@codesourcery.com>
29807 Cesar Philippidis <cesar@codesourcery.com>
29810 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
29812 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
29815 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
29816 spaces for function labels.
29818 2018-01-19 Martin Liska <mliska@suse.cz>
29820 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
29821 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
29822 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
29823 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
29824 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
29825 (PRED_CONST_RETURN): Change from 69 to 65.
29826 (PRED_NULL_RETURN): Change from 91 to 71.
29827 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
29828 (PRED_LOOP_GUARD): Change from 66 to 73.
29830 2018-01-19 Martin Liska <mliska@suse.cz>
29832 * predict.c (predict_insn_def): Add new assert.
29833 (struct branch_predictor): Change type to signed integer.
29834 (test_prediction_value_range): Amend test to cover
29835 PROB_UNINITIALIZED.
29836 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
29837 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
29838 (PRED_LOOP_ITERATIONS_MAX): Likewise.
29839 (PRED_LOOP_IV_COMPARE): Likewise.
29840 * predict.h (PROB_UNINITIALIZED): Define new constant.
29842 2018-01-19 Martin Liska <mliska@suse.cz>
29844 * predict.c (dump_prediction): Add new format for
29845 analyze_brprob.py script which is enabled with -details
29847 * profile-count.h (precise_p): New function.
29849 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29851 PR tree-optimization/83922
29852 * tree-vect-loop.c (vect_verify_full_masking): Return false if
29853 there are no statements that need masking.
29854 (vect_active_double_reduction_p): New function.
29855 (vect_analyze_loop_operations): Use it when handling phis that
29856 are not in the loop header.
29858 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29860 PR tree-optimization/83914
29861 * tree-vect-loop.c (vectorizable_induction): Don't convert
29862 init_expr or apply the peeling adjustment for inductions
29863 that are nested within the vectorized loop.
29865 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29867 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
29870 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29874 * function.h (gimplify_parameters): Add gimple_seq * argument.
29875 * function.c: Include gimple.h and options.h.
29876 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
29877 for the added local temporaries if needed.
29878 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
29879 if there are any parameter cleanups, wrap whole body into a
29880 try/finally with the cleanups.
29882 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29885 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
29886 Use GET_MODE_CLASS for scalar floating point.
29888 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29892 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
29893 Fix call of call_cgraph_insertion_hooks.
29895 2018-01-18 Martin Sebor <msebor@redhat.com>
29897 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
29899 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29902 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
29905 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
29908 * common.opt: (-ffile-prefix-map): New option.
29909 * opts.c (common_handle_option): Defer it.
29910 * opts-global.c (handle_common_deferred_options): Handle it.
29911 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
29912 * file-prefix-map.h: New file.
29913 (remap_debug_filename, add_debug_prefix_map): ...here.
29914 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
29915 * final.c (debug_prefix_map, add_debug_prefix_map
29916 remap_debug_filename): Move to...
29917 * file-prefix-map.c: New file.
29918 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
29919 generalize, get rid of alloca(), use strrchr() instead of strchr().
29920 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
29921 Implement in terms of add_prefix_map().
29922 (remap_macro_filename, remap_debug_filename): Implement in term of
29924 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
29925 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
29926 * dbxout.c: Include file-prefix-map.h.
29927 * varasm.c: Likewise.
29928 * vmsdbgout.c: Likewise.
29929 * xcoffout.c: Likewise.
29930 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
29931 * doc/cppopts.texi (-fmacro-prefix-map): Document.
29932 * doc/invoke.texi (-ffile-prefix-map): Document.
29933 (-fdebug-prefix-map): Update description.
29935 2018-01-18 Martin Liska <mliska@suse.cz>
29937 * config/i386/i386.c (indirect_thunk_name): Document that also
29939 (output_indirect_thunk): Document why both instructions
29940 (pause and lfence) are generated.
29942 2018-01-18 Richard Biener <rguenther@suse.de>
29944 PR tree-optimization/83887
29945 * graphite-scop-detection.c
29946 (scop_detection::get_nearest_dom_with_single_entry): Remove.
29947 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
29948 (scop_detection::merge_sese): Re-implement with a flood-fill
29949 algorithm that properly finds a SESE region if it exists.
29951 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29954 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
29955 pointer_diff optimizations use view_convert instead of convert.
29957 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29959 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29960 Generate different code for -mno-speculate-indirect-jumps.
29961 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29962 (*call_indirect_aix<mode>): Disable for
29963 -mno-speculate-indirect-jumps.
29964 (*call_indirect_aix<mode>_nospec): New define_insn.
29965 (*call_value_indirect_aix<mode>): Disable for
29966 -mno-speculate-indirect-jumps.
29967 (*call_value_indirect_aix<mode>_nospec): New define_insn.
29968 (*sibcall_nonlocal_sysv<mode>): Generate different code for
29969 -mno-speculate-indirect-jumps.
29970 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29972 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
29974 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
29975 long double type, set the flags for noting the default long double
29976 type, even if we don't pass or return a long double type.
29978 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
29981 * ipa-inline.c (flatten_function): Do not overwrite final inlining
29984 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
29986 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
29987 support for merge[hl].
29988 (fold_mergehl_helper): New helper function.
29989 (tree-vector-builder.h): New #include for tree_vector_builder usage.
29990 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
29991 (altivec_vmrglw_direct): Add xxmrglw insn.
29993 2018-01-17 Andrew Waterman <andrew@sifive.com>
29995 * config/riscv/riscv.c (riscv_conditional_register_usage): If
29996 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
29998 2018-01-17 David Malcolm <dmalcolm@redhat.com>
30001 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
30002 call the lto_location_cache before reading the
30003 DECL_SOURCE_LOCATION of the types.
30005 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
30006 Richard Sandiford <richard.sandiford@linaro.org>
30008 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
30009 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
30010 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
30011 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
30012 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
30014 * config/aarch64/constraints.md (aarch64_movti_operand):
30016 * config/aarch64/predicates.md (Uti): Add new constraint.
30018 2018-01-17 Carl Love <cel@us.ibm.com>
30020 * config/rs6000/vsx.md (define_expand xl_len_r,
30021 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
30022 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
30024 (define_expand, define_insn): Move the shift left from the
30025 define_insn to the define_expand for lxvl and stxvl instructions.
30026 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
30027 and XL_LEN_R definitions to PURE.
30029 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
30031 * config/i386/i386.c (indirect_thunk_name): Declare regno
30032 as unsigned int. Compare regno with INVALID_REGNUM.
30033 (output_indirect_thunk): Ditto.
30034 (output_indirect_thunk_function): Ditto.
30035 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
30036 in the call to output_indirect_thunk_function.
30038 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
30040 PR middle-end/83884
30041 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
30042 rather than the size of inner_type to determine the stack slot size
30043 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
30045 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
30048 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
30051 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
30053 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
30054 endian Linux systems to optionally enable multilibs for selecting
30055 the long double type if the user configured an explicit type.
30056 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
30057 have no long double multilibs if not defined.
30058 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
30059 warn if the user used -mabi={ieee,ibm}longdouble and we built
30060 multilibs for long double.
30061 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
30062 appropriate multilib option.
30063 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
30065 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
30066 for building long double multilibs.
30067 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
30069 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
30071 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
30074 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
30076 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
30079 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
30082 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
30085 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
30087 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
30088 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
30090 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
30092 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
30093 different rtl trees depending on TARGET_64BIT.
30094 (rs6000_gen_lvx): Likewise.
30096 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
30098 * config/visium/visium.md (nop): Tweak comment.
30099 (hazard_nop): Likewise.
30101 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30103 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
30104 -mspeculate-indirect-jumps.
30105 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
30106 for -mno-speculate-indirect-jumps.
30107 (*call_indirect_elfv2<mode>_nospec): New define_insn.
30108 (*call_value_indirect_elfv2<mode>): Disable for
30109 -mno-speculate-indirect-jumps.
30110 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
30111 (indirect_jump): Emit different RTL for
30112 -mno-speculate-indirect-jumps.
30113 (*indirect_jump<mode>): Disable for
30114 -mno-speculate-indirect-jumps.
30115 (*indirect_jump<mode>_nospec): New define_insn.
30116 (tablejump): Emit different RTL for
30117 -mno-speculate-indirect-jumps.
30118 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
30119 (tablejumpsi_nospec): New define_expand.
30120 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
30121 (tablejumpdi_nospec): New define_expand.
30122 (*tablejump<mode>_internal1): Disable for
30123 -mno-speculate-indirect-jumps.
30124 (*tablejump<mode>_internal1_nospec): New define_insn.
30125 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
30128 2018-01-16 Artyom Skrobov tyomitch@gmail.com
30130 * caller-save.c (insert_save): Drop unnecessary parameter. All
30133 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30134 Richard Biener <rguenth@suse.de>
30137 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
30138 return early, inline manually is_gimple_sizepos. Make sure if we
30139 call gimplify_expr we don't end up with a gimple constant.
30140 * tree.c (variably_modified_type_p): Don't return true for
30141 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
30142 * gimplify.h (is_gimple_sizepos): Remove.
30144 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30146 PR tree-optimization/83857
30147 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
30148 vectorizable_live_operation for pure SLP statements.
30149 (vectorizable_live_operation): Handle PHIs.
30151 2018-01-16 Richard Biener <rguenther@suse.de>
30153 PR tree-optimization/83867
30154 * tree-vect-stmts.c (vect_transform_stmt): Precompute
30155 nested_in_vect_loop_p since the scalar stmt may get invalidated.
30157 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30160 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
30161 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
30162 If off is not INTEGER_CST, issue a may not be aligned warning
30163 rather than isn't aligned. Use isn%'t rather than isn't.
30164 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
30166 <case MULT_EXPR>: Improve the case when bottom and one of the
30167 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
30168 operand, in that case check if the other operand is multiple of
30169 bottom divided by the INTEGER_CST operand.
30171 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30174 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
30175 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
30176 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
30177 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
30178 * config/pa/pa.c (pa_function_arg_advance): Likewise.
30179 (pa_function_arg, pa_arg_partial_bytes): Likewise.
30180 (pa_function_arg_size): New function.
30182 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30184 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
30185 in a separate statement.
30187 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
30189 PR tree-optimization/83847
30190 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
30191 group gathers and scatters.
30193 2018-01-16 Jakub Jelinek <jakub@redhat.com>
30195 PR rtl-optimization/86620
30196 * params.def (max-sched-ready-insns): Bump minimum value to 1.
30198 PR rtl-optimization/83213
30199 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
30200 to last if both are JUMP_INSNs.
30202 PR tree-optimization/83843
30203 * gimple-ssa-store-merging.c
30204 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
30205 store_immediate_info for bswap/nop orig_stores.
30207 2018-01-15 Andrew Waterman <andrew@sifive.com>
30209 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
30211 <UDIV>: Increase cost if !TARGET_DIV.
30213 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
30215 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
30216 (define_attr "cr_logical_3op"): New.
30217 (cceq_ior_compare): Adjust.
30218 (cceq_ior_compare_complement): Adjust.
30219 (*cceq_rev_compare): Adjust.
30220 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
30221 (is_cracked_insn): Adjust.
30222 (insn_must_be_first_in_group): Adjust.
30223 * config/rs6000/40x.md: Adjust.
30224 * config/rs6000/440.md: Adjust.
30225 * config/rs6000/476.md: Adjust.
30226 * config/rs6000/601.md: Adjust.
30227 * config/rs6000/603.md: Adjust.
30228 * config/rs6000/6xx.md: Adjust.
30229 * config/rs6000/7450.md: Adjust.
30230 * config/rs6000/7xx.md: Adjust.
30231 * config/rs6000/8540.md: Adjust.
30232 * config/rs6000/cell.md: Adjust.
30233 * config/rs6000/e300c2c3.md: Adjust.
30234 * config/rs6000/e500mc.md: Adjust.
30235 * config/rs6000/e500mc64.md: Adjust.
30236 * config/rs6000/e5500.md: Adjust.
30237 * config/rs6000/e6500.md: Adjust.
30238 * config/rs6000/mpc.md: Adjust.
30239 * config/rs6000/power4.md: Adjust.
30240 * config/rs6000/power5.md: Adjust.
30241 * config/rs6000/power6.md: Adjust.
30242 * config/rs6000/power7.md: Adjust.
30243 * config/rs6000/power8.md: Adjust.
30244 * config/rs6000/power9.md: Adjust.
30245 * config/rs6000/rs64.md: Adjust.
30246 * config/rs6000/titan.md: Adjust.
30248 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30250 * config/i386/predicates.md (indirect_branch_operand): Rewrite
30251 ix86_indirect_branch_register logic.
30253 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30255 * config/i386/constraints.md (Bs): Update
30256 ix86_indirect_branch_register check. Don't check
30257 ix86_indirect_branch_register with GOT_memory_operand.
30259 * config/i386/predicates.md (GOT_memory_operand): Don't check
30260 ix86_indirect_branch_register here.
30261 (GOT32_symbol_operand): Likewise.
30263 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30265 * config/i386/predicates.md (constant_call_address_operand):
30266 Rewrite ix86_indirect_branch_register logic.
30267 (sibcall_insn_operand): Likewise.
30269 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30271 * config/i386/constraints.md (Bs): Replace
30272 ix86_indirect_branch_thunk_register with
30273 ix86_indirect_branch_register.
30275 * config/i386/i386.md (indirect_jump): Likewise.
30276 (tablejump): Likewise.
30277 (*sibcall_memory): Likewise.
30278 (*sibcall_value_memory): Likewise.
30279 Peepholes of indirect call and jump via memory: Likewise.
30280 * config/i386/i386.opt: Likewise.
30281 * config/i386/predicates.md (indirect_branch_operand): Likewise.
30282 (GOT_memory_operand): Likewise.
30283 (call_insn_operand): Likewise.
30284 (sibcall_insn_operand): Likewise.
30285 (GOT32_symbol_operand): Likewise.
30287 2018-01-15 Jakub Jelinek <jakub@redhat.com>
30289 PR middle-end/83837
30290 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
30291 type rather than type addr's type points to.
30292 (expand_omp_atomic_mutex): Likewise.
30293 (expand_omp_atomic): Likewise.
30295 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
30298 * config/i386/i386.c (output_indirect_thunk_function): Use
30299 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
30300 for __x86_return_thunk.
30302 2018-01-15 Richard Biener <rguenther@suse.de>
30304 PR middle-end/83850
30305 * expmed.c (extract_bit_field_1): Fix typo.
30307 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30310 * config/arm/iterators.md (VF): New mode iterator.
30311 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
30312 Remove integer-related logic from pattern.
30313 (neon_vabd<mode>_3): Likewise.
30315 2018-01-15 Jakub Jelinek <jakub@redhat.com>
30317 PR middle-end/82694
30318 * common.opt (fstrict-overflow): No longer an alias.
30319 (fwrapv-pointer): New option.
30320 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
30321 also for pointer types based on flag_wrapv_pointer.
30322 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
30323 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
30324 opts->x_flag_wrapv got set.
30325 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
30326 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
30327 POINTER_TYPE_OVERFLOW_UNDEFINED.
30328 * match.pd: Likewise in address comparison pattern.
30329 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
30331 2018-01-15 Richard Biener <rguenther@suse.de>
30334 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
30335 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
30336 Reset type names to their identifier if their TYPE_DECL doesn't
30337 have linkage (and thus is used for ODR and devirt).
30338 (save_debug_info_for_decl): Remove.
30339 (save_debug_info_for_type): Likewise.
30340 (add_tree_to_fld_list): Adjust.
30341 * tree-pretty-print.c (dump_generic_node): Make dumping of
30342 type names more robust.
30344 2018-01-15 Richard Biener <rguenther@suse.de>
30346 * BASE-VER: Bump to 8.0.1.
30348 2018-01-14 Martin Sebor <msebor@redhat.com>
30351 * builtins.c (check_access): Avoid warning when the no-warning bit
30354 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
30356 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
30357 * ira-color (allocno_hard_regs_compare): Likewise.
30359 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
30362 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
30363 Use .pushsection/.popsection.
30365 2018-01-14 Martin Sebor <msebor@redhat.com>
30368 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
30370 2018-01-14 Jakub Jelinek <jakub@redhat.com>
30372 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
30373 entry from extra_headers.
30374 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
30375 extra_headers, make the list bitwise identical to the i?86-*-* one.
30377 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30379 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
30380 -mcmodel=large with -mindirect-branch=thunk,
30381 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
30382 -mfunction-return=thunk-extern.
30383 * doc/invoke.texi: Document -mcmodel=large is incompatible with
30384 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
30385 -mfunction-return=thunk and -mfunction-return=thunk-extern.
30387 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30389 * config/i386/i386.c (print_reg): Print the name of the full
30390 integer register without '%'.
30391 (ix86_print_operand): Handle 'V'.
30392 * doc/extend.texi: Document 'V' modifier.
30394 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30396 * config/i386/constraints.md (Bs): Disallow memory operand for
30397 -mindirect-branch-register.
30399 * config/i386/predicates.md (indirect_branch_operand): Likewise.
30400 (GOT_memory_operand): Likewise.
30401 (call_insn_operand): Likewise.
30402 (sibcall_insn_operand): Likewise.
30403 (GOT32_symbol_operand): Likewise.
30404 * config/i386/i386.md (indirect_jump): Call convert_memory_address
30405 for -mindirect-branch-register.
30406 (tablejump): Likewise.
30407 (*sibcall_memory): Likewise.
30408 (*sibcall_value_memory): Likewise.
30409 Disallow peepholes of indirect call and jump via memory for
30410 -mindirect-branch-register.
30411 (*call_pop): Replace m with Bw.
30412 (*call_value_pop): Likewise.
30413 (*sibcall_pop_memory): Replace m with Bs.
30414 * config/i386/i386.opt (mindirect-branch-register): New option.
30415 * doc/invoke.texi: Document -mindirect-branch-register option.
30417 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30419 * config/i386/i386-protos.h (ix86_output_function_return): New.
30420 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
30421 set function_return_type.
30422 (indirect_thunk_name): Add ret_p to indicate thunk for function
30424 (output_indirect_thunk_function): Pass false to
30425 indirect_thunk_name.
30426 (ix86_output_indirect_branch_via_reg): Likewise.
30427 (ix86_output_indirect_branch_via_push): Likewise.
30428 (output_indirect_thunk_function): Create alias for function
30429 return thunk if regno < 0.
30430 (ix86_output_function_return): New function.
30431 (ix86_handle_fndecl_attribute): Handle function_return.
30432 (ix86_attribute_table): Add function_return.
30433 * config/i386/i386.h (machine_function): Add
30434 function_return_type.
30435 * config/i386/i386.md (simple_return_internal): Use
30436 ix86_output_function_return.
30437 (simple_return_internal_long): Likewise.
30438 * config/i386/i386.opt (mfunction-return=): New option.
30439 (indirect_branch): Mention -mfunction-return=.
30440 * doc/extend.texi: Document function_return function attribute.
30441 * doc/invoke.texi: Document -mfunction-return= option.
30443 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
30445 * config/i386/i386-opts.h (indirect_branch): New.
30446 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
30447 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
30448 with local indirect jump when converting indirect call and jump.
30449 (ix86_set_indirect_branch_type): New.
30450 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
30451 (indirectlabelno): New.
30452 (indirect_thunk_needed): Likewise.
30453 (indirect_thunk_bnd_needed): Likewise.
30454 (indirect_thunks_used): Likewise.
30455 (indirect_thunks_bnd_used): Likewise.
30456 (INDIRECT_LABEL): Likewise.
30457 (indirect_thunk_name): Likewise.
30458 (output_indirect_thunk): Likewise.
30459 (output_indirect_thunk_function): Likewise.
30460 (ix86_output_indirect_branch_via_reg): Likewise.
30461 (ix86_output_indirect_branch_via_push): Likewise.
30462 (ix86_output_indirect_branch): Likewise.
30463 (ix86_output_indirect_jmp): Likewise.
30464 (ix86_code_end): Call output_indirect_thunk_function if needed.
30465 (ix86_output_call_insn): Call ix86_output_indirect_branch if
30467 (ix86_handle_fndecl_attribute): Handle indirect_branch.
30468 (ix86_attribute_table): Add indirect_branch.
30469 * config/i386/i386.h (machine_function): Add indirect_branch_type
30470 and has_local_indirect_jump.
30471 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
30473 (tablejump): Likewise.
30474 (*indirect_jump): Use ix86_output_indirect_jmp.
30475 (*tablejump_1): Likewise.
30476 (simple_return_indirect_internal): Likewise.
30477 * config/i386/i386.opt (mindirect-branch=): New option.
30478 (indirect_branch): New.
30481 (thunk-inline): Likewise.
30482 (thunk-extern): Likewise.
30483 * doc/extend.texi: Document indirect_branch function attribute.
30484 * doc/invoke.texi: Document -mindirect-branch= option.
30486 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
30489 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
30491 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
30493 * ipa-inline.c (want_inline_small_function_p): Return false if
30494 inlining has already failed with CIF_FINAL_ERROR.
30495 (update_caller_keys): Call want_inline_small_function_p before
30497 (update_callee_keys): Likewise.
30499 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
30501 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
30503 (rs6000_quadword_masked_address_p): Likewise.
30504 (quad_aligned_load_p): Likewise.
30505 (quad_aligned_store_p): Likewise.
30506 (const_load_sequence_p): Add comment to describe the outer-most loop.
30507 (mimic_memory_attributes_and_flags): New function.
30508 (rs6000_gen_stvx): Likewise.
30509 (replace_swapped_aligned_store): Likewise.
30510 (rs6000_gen_lvx): Likewise.
30511 (replace_swapped_aligned_load): Likewise.
30512 (replace_swapped_load_constant): Capitalize argument name in
30513 comment describing this function.
30514 (rs6000_analyze_swaps): Add a third pass to search for vector loads
30515 and stores that access quad-word aligned addresses and replace
30516 with stvx or lvx instructions when appropriate.
30517 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
30518 New function prototype.
30519 (rs6000_quadword_masked_address_p): Likewise.
30520 (rs6000_gen_lvx): Likewise.
30521 (rs6000_gen_stvx): Likewise.
30522 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
30523 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
30524 when memory address is aligned.
30525 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
30526 this split to select lvx instruction when memory address is aligned.
30527 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
30528 instruction when memory address is aligned.
30529 (*vsx_le_perm_load_v16qi): Likewise.
30530 (four unnamed splitters): Modify to select the stvx instruction
30531 when memory is aligned.
30533 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
30535 * predict.c (determine_unlikely_bbs): Handle correctly BBs
30536 which appears in the queue multiple times.
30538 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30539 Alan Hayward <alan.hayward@arm.com>
30540 David Sherwood <david.sherwood@arm.com>
30542 * tree-vectorizer.h (vec_lower_bound): New structure.
30543 (_loop_vec_info): Add check_nonzero and lower_bounds.
30544 (LOOP_VINFO_CHECK_NONZERO): New macro.
30545 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
30546 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
30547 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
30548 fields. Make seg_len the distance travelled, not including the
30550 (dr_direction_indicator): Declare.
30551 (dr_zero_step_indicator): Likewise.
30552 (dr_known_forward_stride_p): Likewise.
30553 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
30555 (runtime_alias_check_p): Allow runtime alias checks with
30557 (operator ==): Compare access_size and align.
30558 (prune_runtime_alias_test_list): Rework for new distinction between
30559 the access_size and seg_len.
30560 (create_intersect_range_checks_index): Likewise. Cope with polynomial
30562 (get_segment_min_max): New function.
30563 (create_intersect_range_checks): Use it.
30564 (dr_step_indicator): New function.
30565 (dr_direction_indicator): Likewise.
30566 (dr_zero_step_indicator): Likewise.
30567 (dr_known_forward_stride_p): Likewise.
30568 * tree-loop-distribution.c (data_ref_segment_size): Return
30569 DR_STEP * (niters - 1).
30570 (compute_alias_check_pairs): Update call to the dr_with_seg_len
30572 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
30573 (vect_preserves_scalar_order_p): New function, split out from...
30574 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
30575 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
30576 (vect_vfa_access_size): New function.
30577 (vect_vfa_align): Likewise.
30578 (vect_compile_time_alias): Take access_size_a and access_b arguments.
30579 (dump_lower_bound): New function.
30580 (vect_check_lower_bound): Likewise.
30581 (vect_small_gap_p): Likewise.
30582 (vectorizable_with_step_bound_p): Likewise.
30583 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
30584 depencies if the vectorization factor is 1. Convert the checks
30585 for nonzero steps into checks on the bounds of DR_STEP. Try using
30586 a bunds check for variable steps if the minimum required step is
30587 relatively small. Update calls to the dr_with_seg_len
30588 constructor and to vect_compile_time_alias.
30589 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
30591 (vect_loop_versioning): Call it.
30592 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
30594 (vect_estimate_min_profitable_iters): Account for any bounds checks.
30596 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30597 Alan Hayward <alan.hayward@arm.com>
30598 David Sherwood <david.sherwood@arm.com>
30600 * doc/sourcebuild.texi (vect_scatter_store): Document.
30601 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
30603 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
30605 * genopinit.c (main): Add supports_vec_scatter_store and
30606 supports_vec_scatter_store_cached to target_optabs.
30607 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
30608 IFN_MASK_SCATTER_STORE.
30609 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
30611 * internal-fn.h (internal_store_fn_p): Declare.
30612 (internal_fn_stored_value_index): Likewise.
30613 * internal-fn.c (scatter_store_direct): New macro.
30614 (expand_scatter_store_optab_fn): New function.
30615 (direct_scatter_store_optab_supported_p): New macro.
30616 (internal_store_fn_p): New function.
30617 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
30618 IFN_MASK_SCATTER_STORE.
30619 (internal_fn_mask_index): Likewise.
30620 (internal_fn_stored_value_index): New function.
30621 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
30622 for scatter stores.
30623 * optabs-query.h (supports_vec_scatter_store_p): Declare.
30624 * optabs-query.c (supports_vec_scatter_store_p): New function.
30625 * tree-vectorizer.h (vect_get_store_rhs): Declare.
30626 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
30627 true for scatter stores.
30628 (vect_gather_scatter_fn_p): Handle scatter stores too.
30629 (vect_check_gather_scatter): Consider using scatter stores if
30630 supports_vec_scatter_store_p.
30631 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
30632 scatter stores too.
30633 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30634 internal_fn_stored_value_index.
30635 (check_load_store_masking): Handle scatter stores too.
30636 (vect_get_store_rhs): Make public.
30637 (vectorizable_call): Use internal_store_fn_p.
30638 (vectorizable_store): Handle scatter store internal functions.
30639 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
30640 when deciding whether the end of the group has been reached.
30641 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
30642 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
30643 (mask_scatter_store<mode>): New insns.
30645 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30646 Alan Hayward <alan.hayward@arm.com>
30647 David Sherwood <david.sherwood@arm.com>
30649 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
30650 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
30651 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
30653 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
30654 Use vect_truncate_gather_scatter_offset if we can't treat the
30655 operation as a normal gather load or scatter store.
30656 (get_group_load_store_type): Take the gather_scatter_info
30657 as argument. Try using a gather load or scatter store for
30658 single-element groups.
30659 (get_load_store_type): Update calls to get_group_load_store_type
30660 and vect_use_strided_gather_scatters_p.
30662 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30663 Alan Hayward <alan.hayward@arm.com>
30664 David Sherwood <david.sherwood@arm.com>
30666 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
30667 optional tree argument.
30668 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
30670 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
30671 but continue to use the current value as a fallback.
30672 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
30673 to compare the updates.
30674 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
30675 (get_load_store_type): Use it when handling a strided access.
30676 (vect_get_strided_load_store_ops): New function.
30677 (vect_get_data_ptr_increment): Likewise.
30678 (vectorizable_load): Handle strided gather loads. Always pass
30679 a step to vect_create_data_ref_ptr and bump_vector_ptr.
30681 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30682 Alan Hayward <alan.hayward@arm.com>
30683 David Sherwood <david.sherwood@arm.com>
30685 * doc/md.texi (gather_load@var{m}): Document.
30686 (mask_gather_load@var{m}): Likewise.
30687 * genopinit.c (main): Add supports_vec_gather_load and
30688 supports_vec_gather_load_cached to target_optabs.
30689 * optabs-tree.c (init_tree_optimization_optabs): Use
30690 ggc_cleared_alloc to allocate target_optabs.
30691 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
30692 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
30694 * internal-fn.h (internal_load_fn_p): Declare.
30695 (internal_gather_scatter_fn_p): Likewise.
30696 (internal_fn_mask_index): Likewise.
30697 (internal_gather_scatter_fn_supported_p): Likewise.
30698 * internal-fn.c (gather_load_direct): New macro.
30699 (expand_gather_load_optab_fn): New function.
30700 (direct_gather_load_optab_supported_p): New macro.
30701 (direct_internal_fn_optab): New function.
30702 (internal_load_fn_p): Likewise.
30703 (internal_gather_scatter_fn_p): Likewise.
30704 (internal_fn_mask_index): Likewise.
30705 (internal_gather_scatter_fn_supported_p): Likewise.
30706 * optabs-query.c (supports_at_least_one_mode_p): New function.
30707 (supports_vec_gather_load_p): Likewise.
30708 * optabs-query.h (supports_vec_gather_load_p): Declare.
30709 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
30710 and memory_type field.
30711 (NUM_PATTERNS): Bump to 15.
30712 * tree-vect-data-refs.c: Include internal-fn.h.
30713 (vect_gather_scatter_fn_p): New function.
30714 (vect_describe_gather_scatter_call): Likewise.
30715 (vect_check_gather_scatter): Try using internal functions for
30716 gather loads. Recognize existing calls to a gather load function.
30717 (vect_analyze_data_refs): Consider using gather loads if
30718 supports_vec_gather_load_p.
30719 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
30720 (vect_get_gather_scatter_offset_type): Likewise.
30721 (vect_convert_mask_for_vectype): Likewise.
30722 (vect_add_conversion_to_patterm): Likewise.
30723 (vect_try_gather_scatter_pattern): Likewise.
30724 (vect_recog_gather_scatter_pattern): New pattern recognizer.
30725 (vect_vect_recog_func_ptrs): Add it.
30726 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
30727 internal_fn_mask_index and internal_gather_scatter_fn_p.
30728 (check_load_store_masking): Take the gather_scatter_info as an
30729 argument and handle gather loads.
30730 (vect_get_gather_scatter_ops): New function.
30731 (vectorizable_call): Check internal_load_fn_p.
30732 (vectorizable_load): Likewise. Handle gather load internal
30734 (vectorizable_store): Update call to check_load_store_masking.
30735 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
30736 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
30737 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
30738 (aarch64_gather_scale_operand_d): New predicates.
30739 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
30740 (mask_gather_load<mode>): New insns.
30742 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30743 Alan Hayward <alan.hayward@arm.com>
30744 David Sherwood <david.sherwood@arm.com>
30746 * optabs.def (fold_left_plus_optab): New optab.
30747 * doc/md.texi (fold_left_plus_@var{m}): Document.
30748 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
30749 * internal-fn.c (fold_left_direct): Define.
30750 (expand_fold_left_optab_fn): Likewise.
30751 (direct_fold_left_optab_supported_p): Likewise.
30752 * fold-const-call.c (fold_const_fold_left): New function.
30753 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
30754 * tree-parloops.c (valid_reduction_p): New function.
30755 (gather_scalar_reductions): Use it.
30756 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
30757 (vect_finish_replace_stmt): Declare.
30758 * tree-vect-loop.c (fold_left_reduction_fn): New function.
30759 (needs_fold_left_reduction_p): New function, split out from...
30760 (vect_is_simple_reduction): ...here. Accept reductions that
30761 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
30762 (vect_force_simple_reduction): Also store the reduction type in
30763 the assignment's STMT_VINFO_REDUC_TYPE.
30764 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
30765 (merge_with_identity): New function.
30766 (vect_expand_fold_left): Likewise.
30767 (vectorize_fold_left_reduction): Likewise.
30768 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
30769 scalar phi in place for it. Check for target support and reject
30770 cases that would reassociate the operation. Defer the transform
30771 phase to vectorize_fold_left_reduction.
30772 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
30773 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
30774 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
30776 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30778 * tree-if-conv.c (predicate_mem_writes): Remove redundant
30779 call to ifc_temp_var.
30781 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30782 Alan Hayward <alan.hayward@arm.com>
30783 David Sherwood <david.sherwood@arm.com>
30785 * target.def (legitimize_address_displacement): Take the original
30786 offset as a poly_int.
30787 * targhooks.h (default_legitimize_address_displacement): Update
30789 * targhooks.c (default_legitimize_address_displacement): Likewise.
30790 * doc/tm.texi: Regenerate.
30791 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
30792 as an argument, moving assert of ad->disp == ad->disp_term to...
30793 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
30794 Try calling targetm.legitimize_address_displacement before expanding
30795 the address rather than afterwards, and adjust for the new interface.
30796 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
30797 Match the new hook interface. Handle SVE addresses.
30798 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
30799 new hook interface.
30801 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30803 * Makefile.in (OBJS): Add early-remat.o.
30804 * target.def (select_early_remat_modes): New hook.
30805 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
30806 * doc/tm.texi: Regenerate.
30807 * targhooks.h (default_select_early_remat_modes): Declare.
30808 * targhooks.c (default_select_early_remat_modes): New function.
30809 * timevar.def (TV_EARLY_REMAT): New timevar.
30810 * passes.def (pass_early_remat): New pass.
30811 * tree-pass.h (make_pass_early_remat): Declare.
30812 * early-remat.c: New file.
30813 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
30815 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
30817 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30818 Alan Hayward <alan.hayward@arm.com>
30819 David Sherwood <david.sherwood@arm.com>
30821 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
30822 vfm1 with a bound_epilog parameter.
30823 (vect_do_peeling): Update calls accordingly, and move the prologue
30824 call earlier in the function. Treat the base bound_epilog as 0 for
30825 fully-masked loops and retain vf - 1 for other loops. Add 1 to
30826 this base when peeling for gaps.
30827 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
30828 with fully-masked loops.
30829 (vect_estimate_min_profitable_iters): Handle the single peeled
30830 iteration in that case.
30832 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30833 Alan Hayward <alan.hayward@arm.com>
30834 David Sherwood <david.sherwood@arm.com>
30836 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
30837 single-element interleaving even if the size is not a power of 2.
30838 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
30839 accesses for single-element interleaving if the group size is
30842 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30843 Alan Hayward <alan.hayward@arm.com>
30844 David Sherwood <david.sherwood@arm.com>
30846 * doc/md.texi (fold_extract_last_@var{m}): Document.
30847 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
30848 * optabs.def (fold_extract_last_optab): New optab.
30849 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
30850 * internal-fn.c (fold_extract_direct): New macro.
30851 (expand_fold_extract_optab_fn): Likewise.
30852 (direct_fold_extract_optab_supported_p): Likewise.
30853 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
30854 * tree-vect-loop.c (vect_model_reduction_cost): Handle
30855 EXTRACT_LAST_REDUCTION.
30856 (get_initial_def_for_reduction): Do not create an initial vector
30857 for EXTRACT_LAST_REDUCTION reductions.
30858 (vectorizable_reduction): Leave the scalar phi in place for
30859 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
30860 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
30861 epilogue code for EXTRACT_LAST_REDUCTION and defer the
30862 transform phase to vectorizable_condition.
30863 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
30865 (vect_finish_stmt_generation): ...here.
30866 (vect_finish_replace_stmt): New function.
30867 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
30868 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
30870 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
30872 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30873 Alan Hayward <alan.hayward@arm.com>
30874 David Sherwood <david.sherwood@arm.com>
30876 * doc/md.texi (extract_last_@var{m}): Document.
30877 * optabs.def (extract_last_optab): New optab.
30878 * internal-fn.def (EXTRACT_LAST): New internal function.
30879 * internal-fn.c (cond_unary_direct): New macro.
30880 (expand_cond_unary_optab_fn): Likewise.
30881 (direct_cond_unary_optab_supported_p): Likewise.
30882 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
30883 loops using EXTRACT_LAST.
30884 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
30885 (extract_last_<mode>): ...this optab.
30886 (vec_extract<mode><Vel>): Update accordingly.
30888 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30889 Alan Hayward <alan.hayward@arm.com>
30890 David Sherwood <david.sherwood@arm.com>
30892 * target.def (empty_mask_is_expensive): New hook.
30893 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
30894 * doc/tm.texi: Regenerate.
30895 * targhooks.h (default_empty_mask_is_expensive): Declare.
30896 * targhooks.c (default_empty_mask_is_expensive): New function.
30897 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
30898 if the target says that empty masks are expensive.
30899 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
30901 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
30903 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30904 Alan Hayward <alan.hayward@arm.com>
30905 David Sherwood <david.sherwood@arm.com>
30907 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
30908 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
30909 (vect_use_loop_mask_for_alignment_p): New function.
30910 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
30911 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
30912 niters_skip argument. Make sure that the first niters_skip elements
30913 of the first iteration are inactive.
30914 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
30915 Update call to vect_set_loop_masks_directly.
30916 (get_misalign_in_elems): New function, split out from...
30917 (vect_gen_prolog_loop_niters): ...here.
30918 (vect_update_init_of_dr): Take a code argument that specifies whether
30919 the adjustment should be added or subtracted.
30920 (vect_update_init_of_drs): Likewise.
30921 (vect_prepare_for_masked_peels): New function.
30922 (vect_do_peeling): Skip prologue peeling if we're using a mask
30923 instead. Update call to vect_update_inits_of_drs.
30924 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
30926 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
30927 alignment. Do not include the number of peeled iterations in
30928 the minimum threshold in that case.
30929 (vectorizable_induction): Adjust the start value down by
30930 LOOP_VINFO_MASK_SKIP_NITERS iterations.
30931 (vect_transform_loop): Call vect_prepare_for_masked_peels.
30932 Take the number of skipped iterations into account when calculating
30934 * tree-vect-stmts.c (vect_gen_while_not): New function.
30936 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30937 Alan Hayward <alan.hayward@arm.com>
30938 David Sherwood <david.sherwood@arm.com>
30940 * doc/sourcebuild.texi (vect_fully_masked): Document.
30941 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
30942 default value to 0.
30943 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
30945 (vect_analyze_loop_2): ...here. Don't check the vectorization
30946 factor against the number of loop iterations if the loop is
30949 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30950 Alan Hayward <alan.hayward@arm.com>
30951 David Sherwood <david.sherwood@arm.com>
30953 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
30954 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
30955 (dump_groups): Update accordingly.
30956 (iv_use::mem_type): New member variable.
30957 (address_p): New function.
30958 (record_use): Add a mem_type argument and initialize the new
30960 (record_group_use): Add a mem_type argument. Use address_p.
30961 Remove obsolete null checks of base_object. Update call to record_use.
30962 (find_interesting_uses_op): Update call to record_group_use.
30963 (find_interesting_uses_cond): Likewise.
30964 (find_interesting_uses_address): Likewise.
30965 (get_mem_type_for_internal_fn): New function.
30966 (find_address_like_use): Likewise.
30967 (find_interesting_uses_stmt): Try find_address_like_use before
30968 calling find_interesting_uses_op.
30969 (addr_offset_valid_p): Use the iv mem_type field as the type
30970 of the addressed memory.
30971 (add_autoinc_candidates): Likewise.
30972 (get_address_cost): Likewise.
30973 (split_small_address_groups_p): Use address_p.
30974 (split_address_groups): Likewise.
30975 (add_iv_candidate_for_use): Likewise.
30976 (autoinc_possible_for_pair): Likewise.
30977 (rewrite_groups): Likewise.
30978 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
30979 (determine_group_iv_cost): Update after split of USE_ADDRESS.
30980 (get_alias_ptr_type_for_ptr_address): New function.
30981 (rewrite_use_address): Rewrite address uses in calls that were
30982 identified by find_address_like_use.
30984 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30985 Alan Hayward <alan.hayward@arm.com>
30986 David Sherwood <david.sherwood@arm.com>
30988 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
30990 * gimple-expr.h (is_gimple_addressable: Likewise.
30991 * gimple-expr.c (is_gimple_address): Likewise.
30992 * internal-fn.c (expand_call_mem_ref): New function.
30993 (expand_mask_load_optab_fn): Use it.
30994 (expand_mask_store_optab_fn): Likewise.
30996 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30997 Alan Hayward <alan.hayward@arm.com>
30998 David Sherwood <david.sherwood@arm.com>
31000 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
31001 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
31002 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
31003 (cond_umax@var{mode}): Document.
31004 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
31005 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
31006 (cond_umin_optab, cond_umax_optab): New optabs.
31007 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
31008 (COND_IOR, COND_XOR): New internal functions.
31009 * internal-fn.h (get_conditional_internal_fn): Declare.
31010 * internal-fn.c (cond_binary_direct): New macro.
31011 (expand_cond_binary_optab_fn): Likewise.
31012 (direct_cond_binary_optab_supported_p): Likewise.
31013 (get_conditional_internal_fn): New function.
31014 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
31015 Cope with reduction statements that are vectorized as calls rather
31017 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
31018 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
31019 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
31020 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
31021 (UNSPEC_COND_EOR): New unspecs.
31022 (optab): Add mappings for them.
31023 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
31024 (sve_int_op, sve_fp_op): New int attributes.
31026 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31027 Alan Hayward <alan.hayward@arm.com>
31028 David Sherwood <david.sherwood@arm.com>
31030 * optabs.def (while_ult_optab): New optab.
31031 * doc/md.texi (while_ult@var{m}@var{n}): Document.
31032 * internal-fn.def (WHILE_ULT): New internal function.
31033 * internal-fn.h (direct_internal_fn_supported_p): New override
31034 that takes two types as argument.
31035 * internal-fn.c (while_direct): New macro.
31036 (expand_while_optab_fn): New function.
31037 (convert_optab_supported_p): Likewise.
31038 (direct_while_optab_supported_p): New macro.
31039 * wide-int.h (wi::udiv_ceil): New function.
31040 * tree-vectorizer.h (rgroup_masks): New structure.
31041 (vec_loop_masks): New typedef.
31042 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
31043 and fully_masked_p.
31044 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
31045 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
31046 (vect_max_vf): New function.
31047 (slpeel_make_loop_iterate_ntimes): Delete.
31048 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
31049 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
31050 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
31051 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
31052 internal-fn.h, stor-layout.h and optabs-query.h.
31053 (vect_set_loop_mask): New function.
31054 (add_preheader_seq): Likewise.
31055 (add_header_seq): Likewise.
31056 (interleave_supported_p): Likewise.
31057 (vect_maybe_permute_loop_masks): Likewise.
31058 (vect_set_loop_masks_directly): Likewise.
31059 (vect_set_loop_condition_masked): Likewise.
31060 (vect_set_loop_condition_unmasked): New function, split out from
31061 slpeel_make_loop_iterate_ntimes.
31062 (slpeel_make_loop_iterate_ntimes): Rename to..
31063 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
31064 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
31065 (vect_do_peeling): Update call accordingly.
31066 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
31068 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
31069 mask_compare_type, can_fully_mask_p and fully_masked_p.
31070 (release_vec_loop_masks): New function.
31071 (_loop_vec_info): Use it to free the loop masks.
31072 (can_produce_all_loop_masks_p): New function.
31073 (vect_get_max_nscalars_per_iter): Likewise.
31074 (vect_verify_full_masking): Likewise.
31075 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
31076 retries, and free the mask rgroups before retrying. Check loop-wide
31077 reasons for disallowing fully-masked loops. Make the final decision
31078 about whether use a fully-masked loop or not.
31079 (vect_estimate_min_profitable_iters): Do not assume that peeling
31080 for the number of iterations will be needed for fully-masked loops.
31081 (vectorizable_reduction): Disable fully-masked loops.
31082 (vectorizable_live_operation): Likewise.
31083 (vect_halve_mask_nunits): New function.
31084 (vect_double_mask_nunits): Likewise.
31085 (vect_record_loop_mask): Likewise.
31086 (vect_get_loop_mask): Likewise.
31087 (vect_transform_loop): Handle the case in which the final loop
31088 iteration might handle a partial vector. Call vect_set_loop_condition
31089 instead of slpeel_make_loop_iterate_ntimes.
31090 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
31091 (check_load_store_masking): New function.
31092 (prepare_load_store_mask): Likewise.
31093 (vectorizable_store): Handle fully-masked loops.
31094 (vectorizable_load): Likewise.
31095 (supportable_widening_operation): Use vect_halve_mask_nunits for
31097 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
31098 (vect_gen_while): New function.
31099 * config/aarch64/aarch64.md (umax<mode>3): New expander.
31100 (aarch64_uqdec<mode>): New insn.
31102 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31103 Alan Hayward <alan.hayward@arm.com>
31104 David Sherwood <david.sherwood@arm.com>
31106 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
31107 (reduc_xor_scal_optab): New optabs.
31108 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
31109 (reduc_xor_scal_@var{m}): Document.
31110 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
31111 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
31112 internal functions.
31113 * fold-const-call.c (fold_const_call): Handle them.
31114 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
31115 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
31116 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
31117 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
31118 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
31119 (UNSPEC_XORV): New unspecs.
31120 (optab): Add entries for them.
31121 (BITWISEV): New int iterator.
31122 (bit_reduc_op): New int attributes.
31124 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31125 Alan Hayward <alan.hayward@arm.com>
31126 David Sherwood <david.sherwood@arm.com>
31128 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
31129 * internal-fn.def (VEC_SHL_INSERT): New internal function.
31130 * optabs.def (vec_shl_insert_optab): New optab.
31131 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
31132 (duplicate_and_interleave): Likewise.
31133 * tree-vect-loop.c: Include internal-fn.h.
31134 (neutral_op_for_slp_reduction): New function, split out from
31135 get_initial_defs_for_reduction.
31136 (get_initial_def_for_reduction): Handle option 2 for variable-length
31137 vectors by loading the neutral value into a vector and then shifting
31138 the initial value into element 0.
31139 (get_initial_defs_for_reduction): Replace the code argument with
31140 the neutral value calculated by neutral_op_for_slp_reduction.
31141 Use gimple_build_vector for constant-length vectors.
31142 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
31143 but the first group_size elements have a neutral value.
31144 Use duplicate_and_interleave otherwise.
31145 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
31146 Update call to get_initial_defs_for_reduction. Handle SLP
31147 reductions for variable-length vectors by creating one vector
31148 result for each scalar result, with the elements associated
31149 with other scalar results stubbed out with the neutral value.
31150 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
31151 Require IFN_VEC_SHL_INSERT for double reductions on
31152 variable-length vectors, or SLP reductions that have
31153 a neutral value. Require can_duplicate_and_interleave_p
31154 support for variable-length unchained SLP reductions if there
31155 is no neutral value, such as for MIN/MAX reductions. Also require
31156 the number of vector elements to be a multiple of the number of
31157 SLP statements when doing variable-length unchained SLP reductions.
31158 Update call to vect_create_epilog_for_reduction.
31159 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
31160 and remove initial values.
31161 (duplicate_and_interleave): Make public.
31162 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
31163 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
31165 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31166 Alan Hayward <alan.hayward@arm.com>
31167 David Sherwood <david.sherwood@arm.com>
31169 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
31170 (can_duplicate_and_interleave_p): New function.
31171 (vect_get_and_check_slp_defs): Take the vector of statements
31172 rather than just the current one. Remove excess parentheses.
31173 Restriction rejectinon of vect_constant_def and vect_external_def
31174 for variable-length vectors to boolean types, or types for which
31175 can_duplicate_and_interleave_p is false.
31176 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
31177 (duplicate_and_interleave): New function.
31178 (vect_get_constant_vectors): Use gimple_build_vector for
31179 constant-length vectors and suitable variable-length constant
31180 vectors. Use duplicate_and_interleave for other variable-length
31181 vectors. Don't defer the update when inserting new statements.
31183 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31184 Alan Hayward <alan.hayward@arm.com>
31185 David Sherwood <david.sherwood@arm.com>
31187 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
31188 min_profitable_iters doesn't go negative.
31190 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31191 Alan Hayward <alan.hayward@arm.com>
31192 David Sherwood <david.sherwood@arm.com>
31194 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
31195 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
31196 * optabs.def (vec_mask_load_lanes_optab): New optab.
31197 (vec_mask_store_lanes_optab): Likewise.
31198 * internal-fn.def (MASK_LOAD_LANES): New internal function.
31199 (MASK_STORE_LANES): Likewise.
31200 * internal-fn.c (mask_load_lanes_direct): New macro.
31201 (mask_store_lanes_direct): Likewise.
31202 (expand_mask_load_optab_fn): Handle masked operations.
31203 (expand_mask_load_lanes_optab_fn): New macro.
31204 (expand_mask_store_optab_fn): Handle masked operations.
31205 (expand_mask_store_lanes_optab_fn): New macro.
31206 (direct_mask_load_lanes_optab_supported_p): Likewise.
31207 (direct_mask_store_lanes_optab_supported_p): Likewise.
31208 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
31210 (vect_load_lanes_supported): Likewise.
31211 * tree-vect-data-refs.c (strip_conversion): New function.
31212 (can_group_stmts_p): Likewise.
31213 (vect_analyze_data_ref_accesses): Use it instead of checking
31214 for a pair of assignments.
31215 (vect_store_lanes_supported): Take a masked_p parameter.
31216 (vect_load_lanes_supported): Likewise.
31217 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
31218 vect_store_lanes_supported and vect_load_lanes_supported.
31219 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
31220 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
31221 parameter. Don't allow gaps for masked accesses.
31222 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
31223 and vect_load_lanes_supported.
31224 (get_load_store_type): Take a masked_p parameter and update
31225 call to get_group_load_store_type.
31226 (vectorizable_store): Update call to get_load_store_type.
31227 Handle IFN_MASK_STORE_LANES.
31228 (vectorizable_load): Update call to get_load_store_type.
31229 Handle IFN_MASK_LOAD_LANES.
31231 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31232 Alan Hayward <alan.hayward@arm.com>
31233 David Sherwood <david.sherwood@arm.com>
31235 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
31237 * config/aarch64/aarch64-protos.h
31238 (aarch64_sve_struct_memory_operand_p): Declare.
31239 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
31240 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
31241 (VPRED, vpred): Handle SVE structure modes.
31242 * config/aarch64/constraints.md (Utx): New constraint.
31243 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
31244 (aarch64_sve_struct_nonimmediate_operand): New predicates.
31245 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
31246 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
31247 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
31248 structure modes. Split into pieces after RA.
31249 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
31250 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
31252 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
31253 SVE structure modes.
31254 (aarch64_classify_address): Likewise.
31255 (sizetochar): Move earlier in file.
31256 (aarch64_print_operand): Handle SVE register lists.
31257 (aarch64_array_mode): New function.
31258 (aarch64_sve_struct_memory_operand_p): Likewise.
31259 (TARGET_ARRAY_MODE): Redefine.
31261 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31262 Alan Hayward <alan.hayward@arm.com>
31263 David Sherwood <david.sherwood@arm.com>
31265 * target.def (array_mode): New target hook.
31266 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
31267 * doc/tm.texi: Regenerate.
31268 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
31269 * hooks.c (hook_optmode_mode_uhwi_none): New function.
31270 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
31271 targetm.array_mode.
31272 * stor-layout.c (mode_for_array): Likewise. Support polynomial
31275 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31276 Alan Hayward <alan.hayward@arm.com>
31277 David Sherwood <david.sherwood@arm.com>
31279 * fold-const.c (fold_binary_loc): Check the argument types
31280 rather than the result type when testing for a vector operation.
31282 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31284 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
31285 * doc/tm.texi: Regenerate.
31287 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31288 Alan Hayward <alan.hayward@arm.com>
31289 David Sherwood <david.sherwood@arm.com>
31291 * doc/invoke.texi (-msve-vector-bits=): Document new option.
31292 (sve): Document new AArch64 extension.
31293 * doc/md.texi (w): Extend the description of the AArch64
31294 constraint to include SVE vectors.
31295 (Upl, Upa): Document new AArch64 predicate constraints.
31296 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
31298 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
31299 (msve-vector-bits=): New option.
31300 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
31301 SVE when these are disabled.
31302 (sve): New extension.
31303 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
31304 modes. Adjust their number of units based on aarch64_sve_vg.
31305 (MAX_BITSIZE_MODE_ANY_MODE): Define.
31306 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
31307 aarch64_addr_query_type.
31308 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
31309 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
31310 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
31311 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
31312 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
31313 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
31314 (aarch64_simd_imm_zero_p): Delete.
31315 (aarch64_check_zero_based_sve_index_immediate): Declare.
31316 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
31317 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
31318 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
31319 (aarch64_sve_float_mul_immediate_p): Likewise.
31320 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
31321 rather than an rtx.
31322 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
31323 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
31324 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
31325 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
31326 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
31327 (aarch64_regmode_natural_size): Likewise.
31328 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
31329 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
31331 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
31332 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
31333 for VG and the SVE predicate registers.
31334 (V_ALIASES): Add a "z"-prefixed alias.
31335 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
31336 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
31337 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
31338 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
31339 (REG_CLASS_NAMES): Add entries for them.
31340 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
31341 and the predicate registers.
31342 (aarch64_sve_vg): Declare.
31343 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
31344 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
31345 (REGMODE_NATURAL_SIZE): Define.
31346 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
31348 * config/aarch64/aarch64.c: Include cfgrtl.h.
31349 (simd_immediate_info): Add a constructor for series vectors,
31350 and an associated step field.
31351 (aarch64_sve_vg): New variable.
31352 (aarch64_dbx_register_number): Handle VG and the predicate registers.
31353 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
31354 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
31355 (VEC_ANY_DATA, VEC_STRUCT): New constants.
31356 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
31357 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
31358 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
31359 (aarch64_get_mask_mode): New functions.
31360 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
31361 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
31362 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
31363 predicate modes and predicate registers. Explicitly restrict
31364 GPRs to modes of 16 bytes or smaller. Only allow FP registers
31365 to store a vector mode if it is recognized by
31366 aarch64_classify_vector_mode.
31367 (aarch64_regmode_natural_size): New function.
31368 (aarch64_hard_regno_caller_save_mode): Return the original mode
31370 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
31371 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
31372 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
31373 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
31375 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
31376 does not overlap dest if the function is frame-related. Handle
31378 (aarch64_split_add_offset): New function.
31379 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
31380 them aarch64_add_offset.
31381 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
31382 and update call to aarch64_sub_sp.
31383 (aarch64_add_cfa_expression): New function.
31384 (aarch64_expand_prologue): Pass extra temporary registers to the
31385 functions above. Handle the case in which we need to emit new
31386 DW_CFA_expressions for registers that were originally saved
31387 relative to the stack pointer, but now have to be expressed
31388 relative to the frame pointer.
31389 (aarch64_output_mi_thunk): Pass extra temporary registers to the
31391 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
31392 IP0 and IP1 values for SVE frames.
31393 (aarch64_expand_vec_series): New function.
31394 (aarch64_expand_sve_widened_duplicate): Likewise.
31395 (aarch64_expand_sve_const_vector): Likewise.
31396 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
31397 Handle SVE constants. Use emit_move_insn to move a force_const_mem
31398 into the register, rather than emitting a SET directly.
31399 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
31400 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
31401 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
31402 (offset_9bit_signed_scaled_p): New functions.
31403 (aarch64_replicate_bitmask_imm): New function.
31404 (aarch64_bitmask_imm): Use it.
31405 (aarch64_cannot_force_const_mem): Reject expressions involving
31406 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
31407 (aarch64_classify_index): Handle SVE indices, by requiring
31408 a plain register index with a scale that matches the element size.
31409 (aarch64_classify_address): Handle SVE addresses. Assert that
31410 the mode of the address is VOIDmode or an integer mode.
31411 Update call to aarch64_classify_symbol.
31412 (aarch64_classify_symbolic_expression): Update call to
31413 aarch64_classify_symbol.
31414 (aarch64_const_vec_all_in_range_p): New function.
31415 (aarch64_print_vector_float_operand): Likewise.
31416 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
31417 "vN" for FP registers with SVE modes. Handle (const ...) vectors
31418 and the FP immediates 1.0 and 0.5.
31419 (aarch64_print_address_internal): Handle SVE addresses.
31420 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
31421 (aarch64_regno_regclass): Handle predicate registers.
31422 (aarch64_secondary_reload): Handle big-endian reloads of SVE
31424 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
31425 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
31426 (aarch64_convert_sve_vector_bits): New function.
31427 (aarch64_override_options): Use it to handle -msve-vector-bits=.
31428 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
31429 rather than an rtx.
31430 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
31431 Handle SVE vector and predicate modes. Accept VL-based constants
31432 that need only one temporary register, and VL offsets that require
31433 no temporary registers.
31434 (aarch64_conditional_register_usage): Mark the predicate registers
31435 as fixed if SVE isn't available.
31436 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
31437 Return true for SVE vector and predicate modes.
31438 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
31439 rather than an unsigned int. Handle SVE modes.
31440 (aarch64_preferred_simd_mode): Update call accordingly. Handle
31442 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
31444 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
31445 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
31446 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
31447 (aarch64_sve_float_mul_immediate_p): New functions.
31448 (aarch64_sve_valid_immediate): New function.
31449 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
31450 Explicitly reject structure modes. Check for INDEX constants.
31451 Handle PTRUE and PFALSE constants.
31452 (aarch64_check_zero_based_sve_index_immediate): New function.
31453 (aarch64_simd_imm_zero_p): Delete.
31454 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
31455 vector modes. Accept constants in the range of CNT[BHWD].
31456 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
31457 ask for an Advanced SIMD mode.
31458 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
31459 (aarch64_simd_vector_alignment): Handle SVE predicates.
31460 (aarch64_vectorize_preferred_vector_alignment): New function.
31461 (aarch64_simd_vector_alignment_reachable): Use it instead of
31463 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
31464 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
31466 (MAX_VECT_LEN): Delete.
31467 (expand_vec_perm_d): Add a vec_flags field.
31468 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
31469 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
31470 (aarch64_evpc_ext): Don't apply a big-endian lane correction
31472 (aarch64_evpc_rev): Rename to...
31473 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
31474 (aarch64_evpc_rev_global): New function.
31475 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
31476 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
31478 (aarch64_evpc_sve_tbl): New function.
31479 (aarch64_expand_vec_perm_const_1): Update after rename of
31480 aarch64_evpc_rev. Handle SVE permutes too, trying
31481 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
31482 than aarch64_evpc_tbl.
31483 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
31484 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
31485 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
31486 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
31487 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
31488 (aarch64_expand_sve_vcond): New functions.
31489 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
31490 of aarch64_vector_mode_p.
31491 (aarch64_dwarf_poly_indeterminate_value): New function.
31492 (aarch64_compute_pressure_classes): Likewise.
31493 (aarch64_can_change_mode_class): Likewise.
31494 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
31495 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
31496 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
31497 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
31498 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
31499 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
31500 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
31501 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
31503 (Dn, Dl, Dr): Accept const as well as const_vector.
31504 (Dz): Likewise. Compare against CONST0_RTX.
31505 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
31506 of "vector" where appropriate.
31507 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
31508 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
31509 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
31510 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
31511 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
31512 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
31513 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
31514 (v_int_equiv): Extend to SVE modes.
31515 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
31517 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
31518 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
31519 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
31520 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
31521 (SVE_COND_FP_CMP): New int iterators.
31522 (perm_hilo): Handle the new unpack unspecs.
31523 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
31525 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
31526 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
31527 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
31528 (aarch64_equality_operator, aarch64_constant_vector_operand)
31529 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
31530 (aarch64_sve_nonimmediate_operand): Likewise.
31531 (aarch64_sve_general_operand): Likewise.
31532 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
31533 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
31534 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
31535 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
31536 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
31537 (aarch64_sve_float_arith_immediate): Likewise.
31538 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
31539 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
31540 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
31541 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
31542 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
31543 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
31544 (aarch64_sve_float_arith_operand): Likewise.
31545 (aarch64_sve_float_arith_with_sub_operand): Likewise.
31546 (aarch64_sve_float_mul_operand): Likewise.
31547 (aarch64_sve_vec_perm_operand): Likewise.
31548 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
31549 (aarch64_mov_operand): Accept const_poly_int and const_vector.
31550 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
31551 as well as const_vector.
31552 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
31553 in file. Use CONST0_RTX and CONSTM1_RTX.
31554 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
31555 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
31556 Use aarch64_simd_imm_zero.
31557 * config/aarch64/aarch64-sve.md: New file.
31558 * config/aarch64/aarch64.md: Include it.
31559 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
31560 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
31561 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
31562 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
31563 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
31564 (sve): New attribute.
31565 (enabled): Disable instructions with the sve attribute unless
31567 (movqi, movhi): Pass CONST_POLY_INT operaneds through
31568 aarch64_expand_mov_immediate.
31569 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
31570 CNT[BHSD] immediates.
31571 (movti): Split CONST_POLY_INT moves into two halves.
31572 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
31573 Split additions that need a temporary here if the destination
31574 is the stack pointer.
31575 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
31576 (*add<mode>3_poly_1): New instruction.
31577 (set_clobber_cc): New expander.
31579 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31581 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
31582 parameter and use it instead of GET_MODE_SIZE (innermode). Use
31583 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
31584 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
31585 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
31586 Change innermode from fixed_mode_size to machine_mode.
31587 (simplify_subreg): Update call accordingly. Handle a constant-sized
31588 subreg of a variable-length CONST_VECTOR.
31590 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
31591 Alan Hayward <alan.hayward@arm.com>
31592 David Sherwood <david.sherwood@arm.com>
31594 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
31595 (add_offset_to_base): New function, split out from...
31596 (create_mem_ref): ...here. When handling a scale other than 1,
31597 check first whether the address is valid without the offset.
31598 Add it into the base if so, leaving the index and scale as-is.
31600 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31603 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
31604 fold_for_warn before checking if arg2 is INTEGER_CST.
31606 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
31608 * config/rs6000/predicates.md (load_multiple_operation): Delete.
31609 (store_multiple_operation): Delete.
31610 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
31611 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
31612 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
31613 guarded by TARGET_STRING.
31614 (rs6000_output_load_multiple): Delete.
31615 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
31616 OPTION_MASK_STRING / TARGET_STRING handling.
31617 (print_operand) <'N', 'O'>: Add comment that these are unused now.
31618 (const rs6000_opt_masks) <"string">: Change mask to 0.
31619 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
31620 (MASK_STRING): Delete.
31621 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
31623 (load_multiple): Delete.
31630 (store_multiple): Delete.
31637 (movmemsi_8reg): Delete.
31638 (corresponding unnamed define_insn): Delete.
31639 (movmemsi_6reg): Delete.
31640 (corresponding unnamed define_insn): Delete.
31641 (movmemsi_4reg): Delete.
31642 (corresponding unnamed define_insn): Delete.
31643 (movmemsi_2reg): Delete.
31644 (corresponding unnamed define_insn): Delete.
31645 (movmemsi_1reg): Delete.
31646 (corresponding unnamed define_insn): Delete.
31647 * config/rs6000/rs6000.opt (mno-string): New.
31648 (mstring): Replace by deprecation warning stub.
31649 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
31651 2018-01-12 Jakub Jelinek <jakub@redhat.com>
31653 * regrename.c (regrename_do_replace): If replacing the same
31654 reg multiple times, try to reuse last created gen_raw_REG.
31657 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
31658 main to workaround a bug in GDB.
31660 2018-01-12 Tom de Vries <tom@codesourcery.com>
31663 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
31665 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
31667 PR rtl-optimization/80481
31668 * ira-color.c (get_cap_member): New function.
31669 (allocnos_conflict_by_live_ranges_p): Use it.
31670 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
31671 (setup_slot_coalesced_allocno_live_ranges): Ditto.
31673 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
31676 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
31677 (*saddl_se_1): Ditto.
31678 (*ssubsi_1): Ditto.
31679 (*ssubl_se_1): Ditto.
31681 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31683 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
31684 rather than wi::to_widest for DR_INITs.
31685 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
31686 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
31687 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
31689 (vect_analyze_group_access_1): Note that here.
31691 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31693 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
31694 polynomial type sizes.
31696 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
31698 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
31699 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
31700 (gimple_add_tmp_var): Likewise.
31702 2018-01-12 Martin Liska <mliska@suse.cz>
31704 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
31705 (gimple_alloc_sizes): Likewise.
31706 (dump_gimple_statistics): Use PRIu64 in printf format.
31707 * gimple.h: Change uint64_t to int.
31709 2018-01-12 Martin Liska <mliska@suse.cz>
31711 * tree-core.h: Use uint64_t instead of int.
31712 * tree.c (tree_node_counts): Likewise.
31713 (tree_node_sizes): Likewise.
31714 (dump_tree_statistics): Use PRIu64 in printf format.
31716 2018-01-12 Martin Liska <mliska@suse.cz>
31718 * Makefile.in: As qsort_chk is implemented in vec.c, add
31719 vec.o to linkage of gencfn-macros.
31720 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
31721 passing the info to record_node_allocation_statistics.
31722 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
31724 * ggc-common.c (struct ggc_usage): Add operator== and use
31725 it in operator< and compare function.
31726 * mem-stats.h (struct mem_usage): Likewise.
31727 * vec.c (struct vec_usage): Remove operator< and compare
31728 function. Can be simply inherited.
31730 2018-01-12 Martin Jambor <mjambor@suse.cz>
31733 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
31734 * tree-ssa-math-opts.c: Include domwalk.h.
31735 (convert_mult_to_fma_1): New function.
31736 (fma_transformation_info): New type.
31737 (fma_deferring_state): Likewise.
31738 (cancel_fma_deferring): New function.
31739 (result_of_phi): Likewise.
31740 (last_fma_candidate_feeds_initial_phi): Likewise.
31741 (convert_mult_to_fma): Added deferring logic, split actual
31742 transformation to convert_mult_to_fma_1.
31743 (math_opts_dom_walker): New type.
31744 (math_opts_dom_walker::after_dom_children): New method, body moved
31745 here from pass_optimize_widening_mul::execute, added deferring logic
31747 (pass_optimize_widening_mul::execute): Moved most of code to
31748 math_opts_dom_walker::after_dom_children.
31749 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
31750 * config/i386/i386.c (ix86_option_override_internal): Added
31751 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
31753 2018-01-12 Richard Biener <rguenther@suse.de>
31756 * dwarf2out.c (gen_variable_die): Do not reset old_die for
31757 inline instance vars.
31759 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
31762 * config/rx/rx.c (rx_is_restricted_memory_address):
31763 Handle SUBREG case.
31765 2018-01-12 Richard Biener <rguenther@suse.de>
31767 PR tree-optimization/80846
31768 * target.def (split_reduction): New target hook.
31769 * targhooks.c (default_split_reduction): New function.
31770 * targhooks.h (default_split_reduction): Declare.
31771 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
31772 target requests first reduce vectors by combining low and high
31774 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
31775 (get_vectype_for_scalar_type_and_size): Export.
31776 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
31777 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
31778 * doc/tm.texi: Regenerate.
31779 * config/i386/i386.c (ix86_split_reduction): Implement
31780 TARGET_VECTORIZE_SPLIT_REDUCTION.
31782 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31785 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
31786 in PIC mode except for TARGET_VXWORKS_RTP.
31787 * config/sparc/sparc.c: Include cfgrtl.h.
31788 (TARGET_INIT_PIC_REG): Define.
31789 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
31790 (sparc_pic_register_p): New predicate.
31791 (sparc_legitimate_address_p): Use it.
31792 (sparc_legitimize_pic_address): Likewise.
31793 (sparc_delegitimize_address): Likewise.
31794 (sparc_mode_dependent_address_p): Likewise.
31795 (gen_load_pcrel_sym): Remove 4th parameter.
31796 (load_got_register): Adjust call to above. Remove obsolete stuff.
31797 (sparc_expand_prologue): Do not call load_got_register here.
31798 (sparc_flat_expand_prologue): Likewise.
31799 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
31800 (sparc_use_pseudo_pic_reg): New function.
31801 (sparc_init_pic_reg): Likewise.
31802 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
31803 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
31805 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
31807 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
31808 Add item for branch_cost.
31810 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
31812 PR rtl-optimization/83565
31813 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
31814 not extend the result to a larger mode for rotate operations.
31815 (num_sign_bit_copies1): Likewise.
31817 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31820 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
31822 Use values-Xc.o for -pedantic.
31823 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
31825 2018-01-12 Martin Liska <mliska@suse.cz>
31828 * ipa-devirt.c (final_warning_record::grow_type_warnings):
31830 (possible_polymorphic_call_targets): Use it.
31831 (ipa_devirt): Likewise.
31833 2018-01-12 Martin Liska <mliska@suse.cz>
31835 * profile-count.h (enum profile_quality): Use 0 as invalid
31836 enum value of profile_quality.
31838 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
31840 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
31841 -mext-string options.
31843 2018-01-12 Richard Biener <rguenther@suse.de>
31845 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
31846 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
31847 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
31849 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
31851 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31853 * configure.ac (--with-long-double-format): Add support for the
31854 configuration option to change the default long double format on
31856 * config.gcc (powerpc*-linux*-*): Likewise.
31857 * configure: Regenerate.
31858 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
31859 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
31860 used without modification.
31862 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31864 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
31865 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
31866 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
31867 MISC_BUILTIN_SPEC_BARRIER.
31868 (rs6000_init_builtins): Likewise.
31869 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
31871 (speculation_barrier): New define_insn.
31872 * doc/extend.texi: Document __builtin_speculation_barrier.
31874 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31877 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
31878 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
31879 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
31881 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
31882 integral modes instead of "ss" and "sd".
31883 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
31884 vectors with 32-bit and 64-bit elements.
31885 (vecdupssescalarmodesuffix): New mode attribute.
31886 (vec_dup<mode>): Use it.
31888 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
31891 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
31892 frame if argument is passed on stack.
31894 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31897 * ree.c (combine_reaching_defs): Optimize also
31898 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
31899 reg2=any_extend(exp); reg1=reg2;, formatting fix.
31901 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31903 PR middle-end/83189
31904 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
31906 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31908 PR middle-end/83718
31909 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
31910 after they are computed.
31912 2018-01-11 Bin Cheng <bin.cheng@arm.com>
31914 PR tree-optimization/83695
31915 * gimple-loop-linterchange.cc
31916 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
31917 reset cached scev information after interchange.
31918 (pass_linterchange::execute): Remove call to scev_reset_htab.
31920 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31922 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
31923 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
31924 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
31925 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
31926 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
31927 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
31928 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
31929 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
31930 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
31931 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
31932 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
31933 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
31934 (V_lane_reg): Likewise.
31935 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
31937 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
31938 (vfmal_lane_low<mode>_intrinsic,
31939 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
31940 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
31941 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
31942 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
31943 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
31944 vfmsl_lane_high<mode>_intrinsic): New define_insns.
31946 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31948 * config/arm/arm-cpus.in (fp16fml): New feature.
31949 (ALL_SIMD): Add fp16fml.
31950 (armv8.2-a): Add fp16fml as an option.
31951 (armv8.3-a): Likewise.
31952 (armv8.4-a): Add fp16fml as part of fp16.
31953 * config/arm/arm.h (TARGET_FP16FML): Define.
31954 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
31956 * config/arm/arm-modes.def (V2HF): Define.
31957 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
31958 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
31959 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
31960 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
31961 vfmsl_low, vfmsl_high): New set of builtins.
31962 * config/arm/iterators.md (PLUSMINUS): New code iterator.
31963 (vfml_op): New code attribute.
31964 (VFMLHALVES): New int iterator.
31965 (VFML, VFMLSEL): New mode attributes.
31966 (V_reg): Define mapping for V2HF.
31967 (V_hi, V_lo): New mode attributes.
31968 (VF_constraint): Likewise.
31969 (vfml_half, vfml_half_selector): New int attributes.
31970 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
31972 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
31973 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
31975 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
31976 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
31977 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
31978 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
31980 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
31981 Document new effective target and option set.
31983 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31985 * config/arm/arm-cpus.in (armv8_4): New feature.
31986 (ARMv8_4a): New fgroup.
31987 (armv8.4-a): New arch.
31988 * config/arm/arm-tables.opt: Regenerate.
31989 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
31990 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
31991 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
31992 Add matching rules for -march=armv8.4-a and extensions.
31993 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
31995 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
31998 * config/rx/rx.md (BW): New mode attribute.
31999 (sync_lock_test_and_setsi): Add mode suffix to insn output.
32001 2018-01-11 Richard Biener <rguenther@suse.de>
32003 PR tree-optimization/83435
32004 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
32005 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
32006 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
32008 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32009 Alan Hayward <alan.hayward@arm.com>
32010 David Sherwood <david.sherwood@arm.com>
32012 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
32014 (aarch64_classify_address): Initialize it. Track polynomial offsets.
32015 (aarch64_print_address_internal): Use it to check for a zero offset.
32017 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32018 Alan Hayward <alan.hayward@arm.com>
32019 David Sherwood <david.sherwood@arm.com>
32021 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
32022 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
32023 Return a poly_int64 rather than a HOST_WIDE_INT.
32024 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
32025 rather than a HOST_WIDE_INT.
32026 * config/aarch64/aarch64.h (aarch64_frame): Protect with
32027 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
32028 hard_fp_offset, frame_size, initial_adjust, callee_offset and
32029 final_offset from HOST_WIDE_INT to poly_int64.
32030 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
32031 to_constant when getting the number of units in an Advanced SIMD
32033 (aarch64_builtin_vectorized_function): Check for a constant number
32035 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
32037 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
32038 attribute instead of GET_MODE_NUNITS.
32039 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
32040 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
32041 GET_MODE_SIZE for fixed-size registers.
32042 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
32043 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
32044 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
32045 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
32046 (aarch64_print_operand, aarch64_print_address_internal)
32047 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
32048 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
32049 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
32050 Handle polynomial GET_MODE_SIZE.
32051 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
32052 wider than SImode without modification.
32053 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
32054 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
32055 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
32056 passing and returning SVE modes.
32057 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
32058 rather than GEN_INT.
32059 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
32060 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
32061 (aarch64_allocate_and_probe_stack_space): Likewise.
32062 (aarch64_layout_frame): Cope with polynomial offsets.
32063 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
32064 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
32065 polynomial offsets.
32066 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
32067 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
32068 poly_int64 rather than a HOST_WIDE_INT.
32069 (aarch64_get_separate_components, aarch64_process_components)
32070 (aarch64_expand_prologue, aarch64_expand_epilogue)
32071 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
32072 (aarch64_anchor_offset): New function, split out from...
32073 (aarch64_legitimize_address): ...here.
32074 (aarch64_builtin_vectorization_cost): Handle polynomial
32075 TYPE_VECTOR_SUBPARTS.
32076 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
32078 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
32079 number of elements from the PARALLEL rather than the mode.
32080 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
32081 rather than GET_MODE_BITSIZE.
32082 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
32083 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
32084 (aarch64_expand_vec_perm_const_1): Handle polynomial
32085 d->perm.length () and d->perm elements.
32086 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
32087 Apply to_constant to d->perm elements.
32088 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
32089 polynomial CONST_VECTOR_NUNITS.
32090 (aarch64_move_pointer): Take amount as a poly_int64 rather
32092 (aarch64_progress_pointer): Avoid temporary variable.
32093 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
32094 the mode attribute instead of GET_MODE.
32096 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32097 Alan Hayward <alan.hayward@arm.com>
32098 David Sherwood <david.sherwood@arm.com>
32100 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
32101 x exists before using it.
32102 (aarch64_add_constant_internal): Rename to...
32103 (aarch64_add_offset_1): ...this. Replace regnum with separate
32104 src and dest rtxes. Handle the case in which they're different,
32105 including when the offset is zero. Replace scratchreg with an rtx.
32106 Use 2 additions if there is no spare register into which we can
32107 move a 16-bit constant.
32108 (aarch64_add_constant): Delete.
32109 (aarch64_add_offset): Replace reg with separate src and dest
32110 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
32111 Use aarch64_add_offset_1.
32112 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
32113 an rtx rather than an int. Take the delta as a poly_int64
32114 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
32115 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
32116 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
32117 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
32118 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
32119 and aarch64_add_sp.
32120 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
32121 aarch64_add_constant.
32123 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32125 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
32126 Use scalar_float_mode.
32128 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32130 * config/aarch64/aarch64-simd.md
32131 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
32132 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
32133 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
32134 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
32135 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
32136 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
32137 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
32138 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
32139 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
32140 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
32142 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32145 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
32146 targ_options->x_arm_arch_string is non NULL.
32148 2018-01-11 Tamar Christina <tamar.christina@arm.com>
32150 * config/aarch64/aarch64.h
32151 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
32153 2018-01-11 Sudakshina Das <sudi.das@arm.com>
32156 * expmed.c (emit_store_flag_force): Swap if const op0
32157 and change VOIDmode to mode of op0.
32159 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
32161 PR rtl-optimization/83761
32162 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
32163 than bytes to mode_for_size.
32165 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32167 PR middle-end/83189
32168 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
32169 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
32172 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32174 PR middle-end/83575
32175 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
32176 when in layout mode.
32177 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
32178 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
32181 2018-01-10 Michael Collison <michael.collison@arm.com>
32183 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
32184 * config/aarch64/aarch64-option-extension.def: Add
32185 AARCH64_OPT_EXTENSION of 'fp16fml'.
32186 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32187 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
32188 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
32189 * config/aarch64/constraints.md (Ui7): New constraint.
32190 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
32191 (VFMLA_SEL_W): Ditto.
32194 (VFMLA16_LOW): New int iterator.
32195 (VFMLA16_HIGH): Ditto.
32196 (UNSPEC_FMLAL): New unspec.
32197 (UNSPEC_FMLSL): Ditto.
32198 (UNSPEC_FMLAL2): Ditto.
32199 (UNSPEC_FMLSL2): Ditto.
32200 (f16mac): New code attribute.
32201 * config/aarch64/aarch64-simd-builtins.def
32202 (aarch64_fmlal_lowv2sf): Ditto.
32203 (aarch64_fmlsl_lowv2sf): Ditto.
32204 (aarch64_fmlalq_lowv4sf): Ditto.
32205 (aarch64_fmlslq_lowv4sf): Ditto.
32206 (aarch64_fmlal_highv2sf): Ditto.
32207 (aarch64_fmlsl_highv2sf): Ditto.
32208 (aarch64_fmlalq_highv4sf): Ditto.
32209 (aarch64_fmlslq_highv4sf): Ditto.
32210 (aarch64_fmlal_lane_lowv2sf): Ditto.
32211 (aarch64_fmlsl_lane_lowv2sf): Ditto.
32212 (aarch64_fmlal_laneq_lowv2sf): Ditto.
32213 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
32214 (aarch64_fmlalq_lane_lowv4sf): Ditto.
32215 (aarch64_fmlsl_lane_lowv4sf): Ditto.
32216 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
32217 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
32218 (aarch64_fmlal_lane_highv2sf): Ditto.
32219 (aarch64_fmlsl_lane_highv2sf): Ditto.
32220 (aarch64_fmlal_laneq_highv2sf): Ditto.
32221 (aarch64_fmlsl_laneq_highv2sf): Ditto.
32222 (aarch64_fmlalq_lane_highv4sf): Ditto.
32223 (aarch64_fmlsl_lane_highv4sf): Ditto.
32224 (aarch64_fmlalq_laneq_highv4sf): Ditto.
32225 (aarch64_fmlsl_laneq_highv4sf): Ditto.
32226 * config/aarch64/aarch64-simd.md:
32227 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
32228 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
32229 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
32230 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
32231 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
32232 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
32233 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
32234 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
32235 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
32236 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
32237 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
32238 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
32239 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
32240 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
32241 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
32242 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
32243 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
32244 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
32245 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
32246 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
32247 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
32248 (vfmlsl_low_u32): Ditto.
32249 (vfmlalq_low_u32): Ditto.
32250 (vfmlslq_low_u32): Ditto.
32251 (vfmlal_high_u32): Ditto.
32252 (vfmlsl_high_u32): Ditto.
32253 (vfmlalq_high_u32): Ditto.
32254 (vfmlslq_high_u32): Ditto.
32255 (vfmlal_lane_low_u32): Ditto.
32256 (vfmlsl_lane_low_u32): Ditto.
32257 (vfmlal_laneq_low_u32): Ditto.
32258 (vfmlsl_laneq_low_u32): Ditto.
32259 (vfmlalq_lane_low_u32): Ditto.
32260 (vfmlslq_lane_low_u32): Ditto.
32261 (vfmlalq_laneq_low_u32): Ditto.
32262 (vfmlslq_laneq_low_u32): Ditto.
32263 (vfmlal_lane_high_u32): Ditto.
32264 (vfmlsl_lane_high_u32): Ditto.
32265 (vfmlal_laneq_high_u32): Ditto.
32266 (vfmlsl_laneq_high_u32): Ditto.
32267 (vfmlalq_lane_high_u32): Ditto.
32268 (vfmlslq_lane_high_u32): Ditto.
32269 (vfmlalq_laneq_high_u32): Ditto.
32270 (vfmlslq_laneq_high_u32): Ditto.
32271 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
32272 (AARCH64_FL_FOR_ARCH8_4): New.
32273 (AARCH64_ISA_F16FML): New ISA flag.
32274 (TARGET_F16FML): New feature flag for fp16fml.
32275 (doc/invoke.texi): Document new fp16fml option.
32277 2018-01-10 Michael Collison <michael.collison@arm.com>
32279 * config/aarch64/aarch64-builtins.c:
32280 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
32281 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32282 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
32283 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
32284 (AARCH64_ISA_SHA3): New ISA flag.
32285 (TARGET_SHA3): New feature flag for sha3.
32286 * config/aarch64/iterators.md (sha512_op): New int attribute.
32287 (CRYPTO_SHA512): New int iterator.
32288 (UNSPEC_SHA512H): New unspec.
32289 (UNSPEC_SHA512H2): Ditto.
32290 (UNSPEC_SHA512SU0): Ditto.
32291 (UNSPEC_SHA512SU1): Ditto.
32292 * config/aarch64/aarch64-simd-builtins.def
32293 (aarch64_crypto_sha512hqv2di): New builtin.
32294 (aarch64_crypto_sha512h2qv2di): Ditto.
32295 (aarch64_crypto_sha512su0qv2di): Ditto.
32296 (aarch64_crypto_sha512su1qv2di): Ditto.
32297 (aarch64_eor3qv8hi): Ditto.
32298 (aarch64_rax1qv2di): Ditto.
32299 (aarch64_xarqv2di): Ditto.
32300 (aarch64_bcaxqv8hi): Ditto.
32301 * config/aarch64/aarch64-simd.md:
32302 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
32303 (aarch64_crypto_sha512su0qv2di): Ditto.
32304 (aarch64_crypto_sha512su1qv2di): Ditto.
32305 (aarch64_eor3qv8hi): Ditto.
32306 (aarch64_rax1qv2di): Ditto.
32307 (aarch64_xarqv2di): Ditto.
32308 (aarch64_bcaxqv8hi): Ditto.
32309 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
32310 (vsha512h2q_u64): Ditto.
32311 (vsha512su0q_u64): Ditto.
32312 (vsha512su1q_u64): Ditto.
32313 (veor3q_u16): Ditto.
32314 (vrax1q_u64): Ditto.
32315 (vxarq_u64): Ditto.
32316 (vbcaxq_u16): Ditto.
32317 * config/arm/types.md (crypto_sha512): New type attribute.
32318 (crypto_sha3): Ditto.
32319 (doc/invoke.texi): Document new sha3 option.
32321 2018-01-10 Michael Collison <michael.collison@arm.com>
32323 * config/aarch64/aarch64-builtins.c:
32324 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
32325 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32326 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
32327 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
32328 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
32329 (AARCH64_ISA_SM4): New ISA flag.
32330 (TARGET_SM4): New feature flag for sm4.
32331 * config/aarch64/aarch64-simd-builtins.def
32332 (aarch64_sm3ss1qv4si): Ditto.
32333 (aarch64_sm3tt1aq4si): Ditto.
32334 (aarch64_sm3tt1bq4si): Ditto.
32335 (aarch64_sm3tt2aq4si): Ditto.
32336 (aarch64_sm3tt2bq4si): Ditto.
32337 (aarch64_sm3partw1qv4si): Ditto.
32338 (aarch64_sm3partw2qv4si): Ditto.
32339 (aarch64_sm4eqv4si): Ditto.
32340 (aarch64_sm4ekeyqv4si): Ditto.
32341 * config/aarch64/aarch64-simd.md:
32342 (aarch64_sm3ss1qv4si): Ditto.
32343 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
32344 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
32345 (aarch64_sm4eqv4si): Ditto.
32346 (aarch64_sm4ekeyqv4si): Ditto.
32347 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
32348 (sm3part_op): Ditto.
32349 (CRYPTO_SM3TT): Ditto.
32350 (CRYPTO_SM3PART): Ditto.
32351 (UNSPEC_SM3SS1): New unspec.
32352 (UNSPEC_SM3TT1A): Ditto.
32353 (UNSPEC_SM3TT1B): Ditto.
32354 (UNSPEC_SM3TT2A): Ditto.
32355 (UNSPEC_SM3TT2B): Ditto.
32356 (UNSPEC_SM3PARTW1): Ditto.
32357 (UNSPEC_SM3PARTW2): Ditto.
32358 (UNSPEC_SM4E): Ditto.
32359 (UNSPEC_SM4EKEY): Ditto.
32360 * config/aarch64/constraints.md (Ui2): New constraint.
32361 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
32362 * config/arm/types.md (crypto_sm3): New type attribute.
32363 (crypto_sm4): Ditto.
32364 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
32365 (vsm3tt1aq_u32): Ditto.
32366 (vsm3tt1bq_u32): Ditto.
32367 (vsm3tt2aq_u32): Ditto.
32368 (vsm3tt2bq_u32): Ditto.
32369 (vsm3partw1q_u32): Ditto.
32370 (vsm3partw2q_u32): Ditto.
32371 (vsm4eq_u32): Ditto.
32372 (vsm4ekeyq_u32): Ditto.
32373 (doc/invoke.texi): Document new sm4 option.
32375 2018-01-10 Michael Collison <michael.collison@arm.com>
32377 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
32378 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
32379 (AARCH64_FL_FOR_ARCH8_4): New.
32380 (AARCH64_FL_V8_4): New flag.
32381 (doc/invoke.texi): Document new armv8.4-a option.
32383 2018-01-10 Michael Collison <michael.collison@arm.com>
32385 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
32386 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
32387 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
32388 * config/aarch64/aarch64-option-extension.def: Add
32389 AARCH64_OPT_EXTENSION of 'sha2'.
32390 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
32391 (crypto): Disable sha2 and aes if crypto disabled.
32392 (crypto): Enable aes and sha2 if enabled.
32393 (simd): Disable sha2 and aes if simd disabled.
32394 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
32396 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
32397 (TARGET_SHA2): New feature flag for sha2.
32398 (TARGET_AES): New feature flag for aes.
32399 * config/aarch64/aarch64-simd.md:
32400 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
32401 conditional on TARGET_AES.
32402 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
32403 (aarch64_crypto_sha1hsi): Make pattern conditional
32405 (aarch64_crypto_sha1hv4si): Ditto.
32406 (aarch64_be_crypto_sha1hv4si): Ditto.
32407 (aarch64_crypto_sha1su1v4si): Ditto.
32408 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
32409 (aarch64_crypto_sha1su0v4si): Ditto.
32410 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
32411 (aarch64_crypto_sha256su0v4si): Ditto.
32412 (aarch64_crypto_sha256su1v4si): Ditto.
32413 (doc/invoke.texi): Document new aes and sha2 options.
32415 2018-01-10 Martin Sebor <msebor@redhat.com>
32417 PR tree-optimization/83781
32418 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
32421 2018-01-11 Martin Sebor <msebor@gmail.com>
32422 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32424 PR tree-optimization/83501
32425 PR tree-optimization/81703
32427 * tree-ssa-strlen.c (get_string_cst): Rename...
32428 (get_string_len): ...to this. Handle global constants.
32429 (handle_char_store): Adjust.
32431 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
32432 Jim Wilson <jimw@sifive.com>
32434 * config/riscv/riscv-protos.h (riscv_output_return): New.
32435 * config/riscv/riscv.c (struct machine_function): New naked_p field.
32436 (riscv_attribute_table, riscv_output_return),
32437 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
32438 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
32439 (riscv_compute_frame_info): Only compute frame->mask if not a naked
32441 (riscv_expand_prologue): Add early return for naked function.
32442 (riscv_expand_epilogue): Likewise.
32443 (riscv_function_ok_for_sibcall): Return false for naked function.
32444 (riscv_set_current_function): New.
32445 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
32446 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
32447 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
32448 * doc/extend.texi (RISC-V Function Attributes): New.
32450 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
32452 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
32453 check for 128-bit long double before checking TCmode.
32454 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
32455 128-bit long doubles before checking TFmode or TCmode.
32456 (FLOAT128_IBM_P): Likewise.
32458 2018-01-10 Martin Sebor <msebor@redhat.com>
32460 PR tree-optimization/83671
32461 * builtins.c (c_strlen): Unconditionally return zero for the empty
32463 Use -Warray-bounds for warnings.
32464 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
32465 for non-constant array indices with COMPONENT_REF, arrays of
32466 arrays, and pointers to arrays.
32467 (gimple_fold_builtin_strlen): Determine and set length range for
32468 non-constant character arrays.
32470 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
32472 PR middle-end/81897
32473 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
32476 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
32478 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
32480 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
32483 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
32484 VECTOR_MEM_ALTIVEC_OR_VSX_P.
32485 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
32486 indexed_or_indirect_operand predicate.
32487 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
32488 (*vsx_le_perm_load_v8hi): Likewise.
32489 (*vsx_le_perm_load_v16qi): Likewise.
32490 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
32491 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
32492 (*vsx_le_perm_store_v8hi): Likewise.
32493 (*vsx_le_perm_store_v16qi): Likewise.
32494 (eight unnamed splitters): Likewise.
32496 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
32498 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
32499 * config/rs6000/emmintrin.h: Likewise.
32500 * config/rs6000/mmintrin.h: Likewise.
32501 * config/rs6000/xmmintrin.h: Likewise.
32503 2018-01-10 David Malcolm <dmalcolm@redhat.com>
32506 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
32508 * tree.c (tree_nop_conversion): Return true for location wrapper
32510 (maybe_wrap_with_location): New function.
32511 (selftest::check_strip_nops): New function.
32512 (selftest::test_location_wrappers): New function.
32513 (selftest::tree_c_tests): Call it.
32514 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
32515 (maybe_wrap_with_location): New decl.
32516 (EXPR_LOCATION_WRAPPER_P): New macro.
32517 (location_wrapper_p): New inline function.
32518 (tree_strip_any_location_wrapper): New inline function.
32520 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
32523 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
32524 stack_realign_offset for the largest alignment of stack slot
32526 (ix86_find_max_used_stack_alignment): New function.
32527 (ix86_finalize_stack_frame_flags): Use it. Set
32528 max_used_stack_alignment if we don't realign stack.
32529 * config/i386/i386.h (machine_function): Add
32530 max_used_stack_alignment.
32532 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
32534 * config/arm/arm.opt (-mbranch-cost): New option.
32535 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
32538 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
32541 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
32542 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
32544 2018-01-10 Richard Biener <rguenther@suse.de>
32547 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
32548 early out so it also covers the case where we have a non-NULL
32551 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
32553 PR tree-optimization/83753
32554 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
32555 for non-strided grouped accesses if the number of elements is 1.
32557 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
32560 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
32561 * i386.h (TARGET_USE_GATHER): Define.
32562 * x86-tune.def (X86_TUNE_USE_GATHER): New.
32564 2018-01-10 Martin Liska <mliska@suse.cz>
32567 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
32568 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
32570 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
32571 CLEANUP_NO_PARTITIONING is not set.
32573 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
32575 * doc/rtl.texi: Remove documentation of (const ...) wrappers
32576 for vectors, as a partial revert of r254296.
32577 * rtl.h (const_vec_p): Delete.
32578 (const_vec_duplicate_p): Don't test for vector CONSTs.
32579 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
32580 * expmed.c (make_tree): Likewise.
32583 * common.md (E, F): Use CONSTANT_P instead of checking for
32585 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
32586 checking for CONST_VECTOR.
32588 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32590 PR middle-end/83575
32591 * predict.c (force_edge_cold): Handle in more sane way edges
32592 with no prediction.
32594 2018-01-09 Carl Love <cel@us.ibm.com>
32596 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
32598 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
32599 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
32600 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
32601 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
32602 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
32603 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
32604 * config/rs6000/rs6000-protos.h: Add extern defition for
32605 rs6000_generate_float2_double_code.
32606 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
32608 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
32609 (float2_v2df): Add define_expand.
32611 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
32614 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
32615 op_mode in the force_to_mode call.
32617 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32619 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
32620 instead of checking each element individually.
32621 (aarch64_evpc_uzp): Likewise.
32622 (aarch64_evpc_zip): Likewise.
32623 (aarch64_evpc_ext): Likewise.
32624 (aarch64_evpc_rev): Likewise.
32625 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
32626 instead of checking each element individually. Return true without
32628 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
32629 whether all selected elements come from the same input, instead of
32630 checking each element individually. Remove calls to gen_rtx_REG,
32631 start_sequence and end_sequence and instead assert that no rtl is
32634 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32636 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
32637 order of HIGH and CONST checks.
32639 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
32641 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
32642 if the destination isn't an SSA_NAME.
32644 2018-01-09 Richard Biener <rguenther@suse.de>
32646 PR tree-optimization/83668
32647 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
32649 (canonicalize_loop_form): ... here, renamed from ...
32650 (canonicalize_loop_closed_ssa_form): ... this and amended to
32651 swap successor edges for loop exit blocks to make us use
32652 the RPO order we need for initial schedule generation.
32654 2018-01-09 Joseph Myers <joseph@codesourcery.com>
32656 PR tree-optimization/64811
32657 * match.pd: When optimizing comparisons with Inf, avoid
32658 introducing or losing exceptions from comparisons with NaN.
32660 2018-01-09 Martin Liska <mliska@suse.cz>
32663 * asan.c (shadow_mem_size): Add gcc_assert.
32665 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
32667 Don't save registers in main().
32670 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
32671 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
32672 * config/avr/avr.c (avr_set_current_function): Don't error if
32673 naked, OS_task or OS_main are specified at the same time.
32674 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
32676 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
32678 * common/config/avr/avr-common.c (avr_option_optimization_table):
32679 Switch on -mmain-is-OS_task for optimizing compilations.
32681 2018-01-09 Richard Biener <rguenther@suse.de>
32683 PR tree-optimization/83572
32684 * graphite.c: Include cfganal.h.
32685 (graphite_transform_loops): Connect infinite loops to exit
32686 and remove fake edges at the end.
32688 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32690 * ipa-inline.c (edge_badness): Revert accidental checkin.
32692 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
32695 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
32696 symbols; not inline clones.
32698 2018-01-09 Jakub Jelinek <jakub@redhat.com>
32701 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
32702 hard registers. Formatting fixes.
32704 PR preprocessor/83722
32705 * gcc.c (try_generate_repro): Pass
32706 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
32707 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
32710 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
32711 Kito Cheng <kito.cheng@gmail.com>
32713 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
32714 (riscv_leaf_function_p): Delete.
32715 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
32717 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32719 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
32721 (do_ifelse): New function.
32722 (do_isel): New function.
32723 (do_sub3): New function.
32724 (do_add3): New function.
32725 (do_load_mask_compare): New function.
32726 (do_overlap_load_compare): New function.
32727 (expand_compare_loop): New function.
32728 (expand_block_compare): Call expand_compare_loop() when appropriate.
32729 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
32730 option description.
32731 (-mblock-compare-inline-loop-limit): New option.
32733 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32736 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
32737 Reverse order of second and third operands in first alternative.
32738 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
32739 of first and second elements in UNSPEC_VPERMR vector.
32740 (altivec_expand_vec_perm_le): Likewise.
32742 2018-01-08 Jeff Law <law@redhat.com>
32744 PR rtl-optimizatin/81308
32745 * tree-switch-conversion.c (cfg_altered): New file scoped static.
32746 (process_switch): If group_case_labels makes a change, then set
32748 (pass_convert_switch::execute): If a switch is converted, then
32749 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
32751 PR rtl-optimization/81308
32752 * recog.c (split_all_insns): Conditionally cleanup the CFG after
32755 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
32757 PR target/83663 - Revert r255946
32758 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
32759 generation for cases where splatting a value is not useful.
32760 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
32761 across a vec_duplicate and a paradoxical subreg forming a vector
32762 mode to a vec_concat.
32764 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32766 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
32767 -march=armv8.3-a variants.
32768 * config/arm/t-multilib: Likewise.
32769 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
32771 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32773 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
32775 (cceq_ior_compare_complement): Give it a name so I can use it, and
32776 change boolean_or_operator predicate to boolean_operator so it can
32777 be used to generate a crand.
32778 (eqne): New code iterator.
32779 (bd/bd_neg): New code_attrs.
32780 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
32781 a single define_insn.
32782 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
32783 decrement (bdnzt/bdnzf/bdzt/bdzf).
32784 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
32785 with the new names of the branch decrement patterns, and added the
32786 names of the branch decrement conditional patterns.
32788 2018-01-08 Richard Biener <rguenther@suse.de>
32790 PR tree-optimization/83563
32791 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
32794 2018-01-08 Richard Biener <rguenther@suse.de>
32796 PR middle-end/83713
32797 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
32799 2018-01-08 Richard Biener <rguenther@suse.de>
32801 PR tree-optimization/83685
32802 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
32803 references to abnormals.
32805 2018-01-08 Richard Biener <rguenther@suse.de>
32808 * dwarf2out.c (output_indirect_strings): Handle empty
32809 skeleton_debug_str_hash.
32810 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
32812 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32814 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
32815 (emit_store_direct): Likewise.
32816 (arc_trampoline_adjust_address): Likewise.
32817 (arc_asm_trampoline_template): New function.
32818 (arc_initialize_trampoline): Use asm_trampoline_template.
32819 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
32820 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
32821 * config/arc/arc.md (flush_icache): Delete pattern.
32823 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32825 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
32826 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
32829 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32832 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
32833 by not USED_FOR_TARGET.
32834 (make_pass_resolve_sw_modes): Likewise.
32836 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32838 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
32841 2018-01-08 Richard Biener <rguenther@suse.de>
32843 PR middle-end/83580
32844 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
32846 2018-01-08 Richard Biener <rguenther@suse.de>
32848 PR middle-end/83517
32849 * match.pd ((t * 2) / 2) -> t): Add missing :c.
32851 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
32853 PR middle-end/81897
32854 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
32855 basic blocks with a small number of successors.
32856 (convert_control_dep_chain_into_preds): Improve handling of
32858 (dump_predicates): Split apart into...
32859 (dump_pred_chain): ...here...
32860 (dump_pred_info): ...and here.
32861 (can_one_predicate_be_invalidated_p): Add debugging printfs.
32862 (can_chain_union_be_invalidated_p): Improve check for invalidation
32864 (uninit_uses_cannot_happen): Avoid unnecessary if
32865 convert_control_dep_chain_into_preds yielded nothing.
32867 2018-01-06 Martin Sebor <msebor@redhat.com>
32869 PR tree-optimization/83640
32870 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
32871 subtracting negative offset from size.
32872 (builtin_access::overlap): Adjust offset bounds of the access to fall
32873 within the size of the object if possible.
32875 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
32877 PR rtl-optimization/83699
32878 * expmed.c (extract_bit_field_1): Restrict the vector usage of
32879 extract_bit_field_as_subreg to cases in which the extracted
32880 value is also a vector.
32882 * lra-constraints.c (process_alt_operands): Test for the equivalence
32883 substitutions when detecting a possible reload cycle.
32885 2018-01-06 Jakub Jelinek <jakub@redhat.com>
32888 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
32889 by default if flag_selective_schedling{,2}. Formatting fixes.
32891 PR rtl-optimization/83682
32892 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
32893 if it has non-VECTOR_MODE element mode.
32894 (vec_duplicate_p): Likewise.
32896 PR middle-end/83694
32897 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
32898 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
32900 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32903 * config/i386/i386-builtin.def
32904 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
32905 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
32906 Require also OPTION_MASK_ISA_AVX512F in addition to
32907 OPTION_MASK_ISA_GFNI.
32908 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
32909 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
32910 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
32911 to OPTION_MASK_ISA_GFNI.
32912 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
32913 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
32914 OPTION_MASK_ISA_AVX512BW.
32915 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
32916 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
32917 addition to OPTION_MASK_ISA_GFNI.
32918 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
32919 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
32920 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
32921 to OPTION_MASK_ISA_GFNI.
32922 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
32923 a requirement for all ISAs rather than any of them with a few
32925 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
32927 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
32928 bitmasks to be enabled with 3 exceptions, instead of requiring any
32929 enabled ISA with lots of exceptions.
32930 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
32931 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
32932 Change avx512bw in isa attribute to avx512f.
32933 * config/i386/sgxintrin.h: Add license boilerplate.
32934 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
32935 to __AVX512F__ and __AVX512VL to __AVX512VL__.
32936 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
32937 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
32939 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
32940 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
32941 temporarily sse2 rather than sse if not enabled already.
32944 * config/i386/sse.md (VI248_VLBW): Rename to ...
32945 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
32946 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
32947 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
32948 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
32949 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
32950 mode iterator instead of VI248_VLBW.
32952 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
32954 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
32955 (record_modified): Skip clobbers; add debug output.
32956 (param_change_prob): Use sreal frequencies.
32958 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
32960 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
32961 punt for user-aligned variables.
32963 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
32965 * tree-chrec.c (chrec_contains_symbols): Return true for
32968 2018-01-05 Sudakshina Das <sudi.das@arm.com>
32971 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
32972 of (x|y) == x for BICS pattern.
32974 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32976 PR tree-optimization/83605
32977 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
32978 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
32981 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
32983 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
32984 * config/epiphany/rtems.h: New file.
32986 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32987 Uros Bizjak <ubizjak@gmail.com>
32990 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
32991 QIreg_operand instead of register_operand predicate.
32992 * config/i386/i386.c (ix86_rop_should_change_byte_p,
32993 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
32994 comments instead of -fmitigate[-_]rop.
32996 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32999 * cgraphunit.c (symbol_table::compile): Switch to text_section
33000 before calling assembly_start debug hook.
33001 * run-rtl-passes.c (run_rtl_passes): Likewise.
33004 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33006 * tree-vrp.c (extract_range_from_binary_expr_1): Check
33007 range_int_cst_p rather than !symbolic_range_p before calling
33008 extract_range_from_multiplicative_op_1.
33010 2018-01-04 Jeff Law <law@redhat.com>
33012 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
33013 redundant test in assertion.
33015 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33017 * doc/rtl.texi: Document machine_mode wrapper classes.
33019 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33021 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
33022 using tree_to_uhwi.
33024 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33026 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
33027 the VEC_PERM_EXPR fold to fail.
33029 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33032 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
33033 to switched_sections.
33035 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33038 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
33039 test for d.testing.
33041 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
33044 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
33045 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
33047 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33050 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
33051 is BLKmode and bitpos not zero or mode change is needed.
33053 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
33056 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
33059 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
33062 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
33063 instead of MULT rtx. Update all corresponding splitters.
33064 (*saddl_se): Ditto.
33065 (*ssub<modesuffix>): Ditto.
33066 (*ssubl_se): Ditto.
33067 (*cmp_sadd_di): Update split patterns.
33068 (*cmp_sadd_si): Ditto.
33069 (*cmp_sadd_sidi): Ditto.
33070 (*cmp_ssub_di): Ditto.
33071 (*cmp_ssub_si): Ditto.
33072 (*cmp_ssub_sidi): Ditto.
33073 * config/alpha/predicates.md (const23_operand): New predicate.
33074 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
33075 Look for ASHIFT, not MULT inner operand.
33076 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
33078 2018-01-04 Martin Liska <mliska@suse.cz>
33080 PR gcov-profile/83669
33081 * gcov.c (output_intermediate_file): Add version to intermediate
33083 * doc/gcov.texi: Document new field 'version' in intermediate
33084 file format. Fix location of '-k' option of gcov command.
33086 2018-01-04 Martin Liska <mliska@suse.cz>
33089 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
33091 2018-01-04 Jakub Jelinek <jakub@redhat.com>
33093 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
33095 2018-01-03 Martin Sebor <msebor@redhat.com>
33097 PR tree-optimization/83655
33098 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
33099 checking calls with invalid arguments.
33101 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33103 * tree-vect-stmts.c (vect_get_store_rhs): New function.
33104 (vectorizable_mask_load_store): Delete.
33105 (vectorizable_call): Return false for masked loads and stores.
33106 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
33107 instead of gimple_assign_rhs1.
33108 (vectorizable_load): Handle IFN_MASK_LOAD.
33109 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
33111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33113 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
33115 (vectorizable_mask_load_store): ...here.
33116 (vectorizable_load): ...and here.
33118 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33120 * tree-vect-stmts.c (vect_build_all_ones_mask)
33121 (vect_build_zero_merge_argument): New functions, split out from...
33122 (vectorizable_load): ...here.
33124 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33126 * tree-vect-stmts.c (vect_check_store_rhs): New function,
33128 (vectorizable_mask_load_store): ...here.
33129 (vectorizable_store): ...and here.
33131 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33133 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
33135 (vectorizable_mask_load_store): ...here.
33137 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33139 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
33140 (vect_model_store_cost): Take a vec_load_store_type instead of a
33142 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
33143 (vect_model_store_cost): Take a vec_load_store_type instead of a
33145 (vectorizable_mask_load_store): Update accordingly.
33146 (vectorizable_store): Likewise.
33147 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
33149 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33151 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
33152 IFN_MASK_LOAD calls here rather than...
33153 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
33155 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33156 Alan Hayward <alan.hayward@arm.com>
33157 David Sherwood <david.sherwood@arm.com>
33159 * expmed.c (extract_bit_field_1): For vector extracts,
33160 fall back to extract_bit_field_as_subreg if vec_extract
33163 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33164 Alan Hayward <alan.hayward@arm.com>
33165 David Sherwood <david.sherwood@arm.com>
33167 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
33168 they are variable or constant sized.
33169 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
33170 slots for constant-sized data.
33172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33173 Alan Hayward <alan.hayward@arm.com>
33174 David Sherwood <david.sherwood@arm.com>
33176 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
33177 handling COND_EXPRs with boolean comparisons, try to find a better
33178 basis for the mask type than the boolean itself.
33180 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33182 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
33183 is calculated and how it can be overridden.
33184 * genmodes.c (max_bitsize_mode_any_mode): New variable.
33185 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
33187 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
33190 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33191 Alan Hayward <alan.hayward@arm.com>
33192 David Sherwood <david.sherwood@arm.com>
33194 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
33195 Remove the mode argument.
33196 (aarch64_simd_valid_immediate): Remove the mode and inverse
33198 * config/aarch64/iterators.md (bitsize): New iterator.
33199 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
33200 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
33201 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
33202 aarch64_simd_valid_immediate.
33203 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
33204 (aarch64_reg_or_bic_imm): Likewise.
33205 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
33206 with an insn_type enum and msl with a modifier_type enum.
33207 Replace element_width with a scalar_mode. Change the shift
33208 to unsigned int. Add constructors for scalar_float_mode and
33209 scalar_int_mode elements.
33210 (aarch64_vect_float_const_representable_p): Delete.
33211 (aarch64_can_const_movi_rtx_p)
33212 (aarch64_simd_scalar_immediate_valid_for_move)
33213 (aarch64_simd_make_constant): Update call to
33214 aarch64_simd_valid_immediate.
33215 (aarch64_advsimd_valid_immediate_hs): New function.
33216 (aarch64_advsimd_valid_immediate): Likewise.
33217 (aarch64_simd_valid_immediate): Remove mode and inverse
33218 arguments. Rewrite to use the above. Use const_vec_duplicate_p
33219 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
33220 and aarch64_float_const_representable_p on the result.
33221 (aarch64_output_simd_mov_immediate): Remove mode argument.
33222 Update call to aarch64_simd_valid_immediate and use of
33223 simd_immediate_info.
33224 (aarch64_output_scalar_simd_mov_immediate): Update call
33227 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33228 Alan Hayward <alan.hayward@arm.com>
33229 David Sherwood <david.sherwood@arm.com>
33231 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
33232 (mode_nunits): Likewise CONST_MODE_NUNITS.
33233 * machmode.def (ADJUST_NUNITS): Document.
33234 * genmodes.c (mode_data::need_nunits_adj): New field.
33235 (blank_mode): Update accordingly.
33236 (adj_nunits): New variable.
33237 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
33239 (emit_mode_size_inline): Set need_bytesize_adj for all modes
33240 listed in adj_nunits.
33241 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
33242 listed in adj_nunits. Don't emit case statements for such modes.
33243 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
33244 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
33245 nothing if adj_nunits is nonnull.
33246 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
33247 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
33248 (emit_mode_fbit): Update use of print_maybe_const_decl.
33249 (emit_move_size): Likewise. Treat the array as non-const
33251 (emit_mode_adjustments): Handle adj_nunits.
33253 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33255 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
33256 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
33257 (VECTOR_MODES): Use it.
33258 (make_vector_modes): Take the prefix as an argument.
33260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33261 Alan Hayward <alan.hayward@arm.com>
33262 David Sherwood <david.sherwood@arm.com>
33264 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
33265 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
33266 for MODE_VECTOR_BOOL.
33267 * machmode.def (VECTOR_BOOL_MODE): Document.
33268 * genmodes.c (VECTOR_BOOL_MODE): New macro.
33269 (make_vector_bool_mode): New function.
33270 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
33272 * lto-streamer-in.c (lto_input_mode_table): Likewise.
33273 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
33275 * stor-layout.c (int_mode_for_mode): Likewise.
33276 * tree.c (build_vector_type_for_mode): Likewise.
33277 * varasm.c (output_constant_pool_2): Likewise.
33278 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
33279 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
33280 for MODE_VECTOR_BOOL.
33281 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
33282 of mode class checks.
33283 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
33284 instead of a list of mode class checks.
33285 (expand_vector_scalar_condition): Likewise.
33286 (type_for_widest_vector_mode): Handle BImode as an inner mode.
33288 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33289 Alan Hayward <alan.hayward@arm.com>
33290 David Sherwood <david.sherwood@arm.com>
33292 * machmode.h (mode_size): Change from unsigned short to
33294 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
33295 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
33296 or if measurement_type is not polynomial.
33297 (fixed_size_mode::includes_p): Check for constant-sized modes.
33298 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
33299 return a poly_uint16 rather than an unsigned short.
33300 (emit_mode_size): Change the type of mode_size from unsigned short
33301 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
33302 (emit_mode_adjustments): Cope with polynomial vector sizes.
33303 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33305 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33307 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
33308 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
33309 * caller-save.c (setup_save_areas): Likewise.
33310 (replace_reg_with_saved_mem): Likewise.
33311 * calls.c (emit_library_call_value_1): Likewise.
33312 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
33313 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
33314 (gen_lowpart_for_combine): Likewise.
33315 * convert.c (convert_to_integer_1): Likewise.
33316 * cse.c (equiv_constant, cse_insn): Likewise.
33317 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
33318 (cselib_subst_to_values): Likewise.
33319 * dce.c (word_dce_process_block): Likewise.
33320 * df-problems.c (df_word_lr_mark_ref): Likewise.
33321 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
33322 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
33323 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
33324 (rtl_for_decl_location): Likewise.
33325 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
33326 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
33327 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
33328 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
33329 (expand_expr_real_1): Likewise.
33330 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
33331 (pad_below): Likewise.
33332 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
33333 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
33334 * ira.c (get_subreg_tracking_sizes): Likewise.
33335 * ira-build.c (ira_create_allocno_objects): Likewise.
33336 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
33337 (ira_sort_regnos_for_alter_reg): Likewise.
33338 * ira-costs.c (record_operand_costs): Likewise.
33339 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
33340 (resolve_simple_move): Likewise.
33341 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
33342 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
33343 (lra_constraints): Likewise.
33344 (CONST_POOL_OK_P): Reject variable-sized modes.
33345 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
33346 (add_pseudo_to_slot, lra_spill): Likewise.
33347 * omp-low.c (omp_clause_aligned_alignment): Likewise.
33348 * optabs-query.c (get_best_extraction_insn): Likewise.
33349 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
33350 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
33351 (expand_mult_highpart, valid_multiword_target_p): Likewise.
33352 * recog.c (offsettable_address_addr_space_p): Likewise.
33353 * regcprop.c (maybe_mode_change): Likewise.
33354 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
33355 * regrename.c (build_def_use): Likewise.
33356 * regstat.c (dump_reg_info): Likewise.
33357 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
33358 (find_reloads, find_reloads_subreg_address): Likewise.
33359 * reload1.c (eliminate_regs_1): Likewise.
33360 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
33361 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
33362 (simplify_binary_operation_1, simplify_subreg): Likewise.
33363 * targhooks.c (default_function_arg_padding): Likewise.
33364 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
33365 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
33366 (verify_gimple_assign_ternary): Likewise.
33367 * tree-inline.c (estimate_move_cost): Likewise.
33368 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33369 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
33370 (get_address_cost_ainc): Likewise.
33371 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33372 (vect_supportable_dr_alignment): Likewise.
33373 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33374 (vectorizable_reduction): Likewise.
33375 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
33376 (vectorizable_operation, vectorizable_load): Likewise.
33377 * tree.c (build_same_sized_truth_vector_type): Likewise.
33378 * valtrack.c (cleanup_auto_inc_dec): Likewise.
33379 * var-tracking.c (emit_note_insn_var_location): Likewise.
33380 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
33381 (ADDR_VEC_ALIGN): Likewise.
33383 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33384 Alan Hayward <alan.hayward@arm.com>
33385 David Sherwood <david.sherwood@arm.com>
33387 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
33389 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
33390 or if measurement_type is polynomial.
33391 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
33392 * combine.c (make_extraction): Likewise.
33393 * dse.c (find_shift_sequence): Likewise.
33394 * dwarf2out.c (mem_loc_descriptor): Likewise.
33395 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
33396 (extract_bit_field, extract_low_bits): Likewise.
33397 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
33398 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
33399 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
33400 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
33401 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
33402 * reload.c (find_reloads): Likewise.
33403 * reload1.c (alter_reg): Likewise.
33404 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
33405 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
33406 * tree-if-conv.c (predicate_mem_writes): Likewise.
33407 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
33408 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
33409 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
33410 * valtrack.c (dead_debug_insert_temp): Likewise.
33411 * varasm.c (mergeable_constant_section): Likewise.
33412 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
33414 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33415 Alan Hayward <alan.hayward@arm.com>
33416 David Sherwood <david.sherwood@arm.com>
33418 * expr.c (expand_assignment): Cope with polynomial mode sizes
33419 when assigning to a CONCAT.
33421 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33422 Alan Hayward <alan.hayward@arm.com>
33423 David Sherwood <david.sherwood@arm.com>
33425 * machmode.h (mode_precision): Change from unsigned short to
33427 (mode_to_precision): Return a poly_uint16 rather than an unsigned
33429 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
33430 or if measurement_type is not polynomial.
33431 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
33432 in which the mode is already known to be a scalar_int_mode.
33433 * genmodes.c (emit_mode_precision): Change the type of mode_precision
33434 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
33436 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33437 for GET_MODE_PRECISION.
33438 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33439 for GET_MODE_PRECISION.
33440 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
33442 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
33443 (expand_field_assignment, make_extraction): Likewise.
33444 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
33445 (get_last_value): Likewise.
33446 * convert.c (convert_to_integer_1): Likewise.
33447 * cse.c (cse_insn): Likewise.
33448 * expr.c (expand_expr_real_1): Likewise.
33449 * lra-constraints.c (simplify_operand_subreg): Likewise.
33450 * optabs-query.c (can_atomic_load_p): Likewise.
33451 * optabs.c (expand_atomic_load): Likewise.
33452 (expand_atomic_store): Likewise.
33453 * ree.c (combine_reaching_defs): Likewise.
33454 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
33455 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
33456 * tree.h (type_has_mode_precision_p): Likewise.
33457 * ubsan.c (instrument_si_overflow): Likewise.
33459 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33460 Alan Hayward <alan.hayward@arm.com>
33461 David Sherwood <david.sherwood@arm.com>
33463 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
33464 polynomial numbers of units.
33465 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
33466 (valid_vector_subparts_p): New function.
33467 (build_vector_type): Remove temporary shim and take the number
33468 of units as a poly_uint64 rather than an int.
33469 (build_opaque_vector_type): Take the number of units as a
33470 poly_uint64 rather than an int.
33471 * tree.c (build_vector_from_ctor): Handle polynomial
33472 TYPE_VECTOR_SUBPARTS.
33473 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
33474 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
33475 (build_vector_from_val): If the number of units is variable,
33476 use build_vec_duplicate_cst for constant operands and
33477 VEC_DUPLICATE_EXPR otherwise.
33478 (make_vector_type): Remove temporary is_constant ().
33479 (build_vector_type, build_opaque_vector_type): Take the number of
33480 units as a poly_uint64 rather than an int.
33481 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
33483 * cfgexpand.c (expand_debug_expr): Likewise.
33484 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
33485 (store_constructor, expand_expr_real_1): Likewise.
33486 (const_scalar_mask_from_tree): Likewise.
33487 * fold-const-call.c (fold_const_reduction): Likewise.
33488 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
33489 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
33490 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
33491 (fold_relational_const): Likewise.
33492 (native_interpret_vector): Likewise. Change the size from an
33493 int to an unsigned int.
33494 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
33495 TYPE_VECTOR_SUBPARTS.
33496 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
33497 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
33498 duplicating a non-constant operand into a variable-length vector.
33499 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
33500 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
33501 * ipa-icf.c (sem_variable::equals): Likewise.
33502 * match.pd: Likewise.
33503 * omp-simd-clone.c (simd_clone_subparts): Likewise.
33504 * print-tree.c (print_node): Likewise.
33505 * stor-layout.c (layout_type): Likewise.
33506 * targhooks.c (default_builtin_vectorization_cost): Likewise.
33507 * tree-cfg.c (verify_gimple_comparison): Likewise.
33508 (verify_gimple_assign_binary): Likewise.
33509 (verify_gimple_assign_ternary): Likewise.
33510 (verify_gimple_assign_single): Likewise.
33511 * tree-pretty-print.c (dump_generic_node): Likewise.
33512 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33513 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
33514 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
33515 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
33516 (vect_shift_permute_load_chain): Likewise.
33517 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
33518 (expand_vector_condition, optimize_vector_constructor): Likewise.
33519 (lower_vec_perm, get_compute_type): Likewise.
33520 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
33521 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
33522 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
33523 (vect_recog_mask_conversion_pattern): Likewise.
33524 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
33525 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
33526 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33527 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
33528 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
33529 (vectorizable_shift, vectorizable_operation, vectorizable_store)
33530 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
33531 (supportable_widening_operation): Likewise.
33532 (supportable_narrowing_operation): Likewise.
33533 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
33535 * varasm.c (output_constant): Likewise.
33537 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33538 Alan Hayward <alan.hayward@arm.com>
33539 David Sherwood <david.sherwood@arm.com>
33541 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
33542 so that both the length == 3 and length != 3 cases set up their
33543 own permute vectors. Add comments explaining why we know the
33544 number of elements is constant.
33545 (vect_permute_load_chain): Likewise.
33547 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33548 Alan Hayward <alan.hayward@arm.com>
33549 David Sherwood <david.sherwood@arm.com>
33551 * machmode.h (mode_nunits): Change from unsigned char to
33553 (ONLY_FIXED_SIZE_MODES): New macro.
33554 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
33555 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
33556 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
33558 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
33559 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
33560 or if measurement_type is not polynomial.
33561 * genmodes.c (ZERO_COEFFS): New macro.
33562 (emit_mode_nunits_inline): Make mode_nunits_inline return a
33564 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
33565 Use ZERO_COEFFS when emitting initializers.
33566 * data-streamer.h (bp_pack_poly_value): New function.
33567 (bp_unpack_poly_value): Likewise.
33568 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
33569 for GET_MODE_NUNITS.
33570 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
33571 for GET_MODE_NUNITS.
33572 * tree.c (make_vector_type): Remove temporary shim and make
33573 the real function take the number of units as a poly_uint64
33574 rather than an int.
33575 (build_vector_type_for_mode): Handle polynomial nunits.
33576 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
33577 * emit-rtl.c (const_vec_series_p_1): Likewise.
33578 (gen_rtx_CONST_VECTOR): Likewise.
33579 * fold-const.c (test_vec_duplicate_folding): Likewise.
33580 * genrecog.c (validate_pattern): Likewise.
33581 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
33582 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
33583 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
33584 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
33585 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
33586 * rtlanal.c (subreg_get_info): Likewise.
33587 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33588 (vect_grouped_load_supported): Likewise.
33589 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
33590 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
33591 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
33592 (simplify_const_unary_operation, simplify_binary_operation_1)
33593 (simplify_const_binary_operation, simplify_ternary_operation)
33594 (test_vector_ops_duplicate, test_vector_ops): Likewise.
33595 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
33596 instead of CONST_VECTOR_NUNITS.
33597 * varasm.c (output_constant_pool_2): Likewise.
33598 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
33599 explicit-encoded elements in the XVEC for variable-length vectors.
33601 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33603 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
33605 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33606 Alan Hayward <alan.hayward@arm.com>
33607 David Sherwood <david.sherwood@arm.com>
33609 * coretypes.h (fixed_size_mode): Declare.
33610 (fixed_size_mode_pod): New typedef.
33611 * builtins.h (target_builtins::x_apply_args_mode)
33612 (target_builtins::x_apply_result_mode): Change type to
33613 fixed_size_mode_pod.
33614 * builtins.c (apply_args_size, apply_result_size, result_vector)
33615 (expand_builtin_apply_args_1, expand_builtin_apply)
33616 (expand_builtin_return): Update accordingly.
33618 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33620 * cse.c (hash_rtx_cb): Hash only the encoded elements.
33621 * cselib.c (cselib_hash_rtx): Likewise.
33622 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
33623 CONST_VECTOR encoding.
33625 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33626 Jeff Law <law@redhat.com>
33629 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
33630 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
33631 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
33632 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
33635 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
33636 explicitly probe *sp in a noreturn function if there were any callee
33637 register saves or frame pointer is needed.
33639 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33642 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
33643 BLKmode for ternary, binary or unary expressions.
33646 * var-tracking.c (delete_vta_debug_insn): New inline function.
33647 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
33648 insns from get_insns () to NULL instead of each bb separately.
33649 Use delete_vta_debug_insn. No longer static.
33650 (vt_debug_insns_local, variable_tracking_main_1): Adjust
33651 delete_vta_debug_insns callers.
33652 * rtl.h (delete_vta_debug_insns): Declare.
33653 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
33654 instead of variable_tracking_main.
33656 2018-01-03 Martin Sebor <msebor@redhat.com>
33658 PR tree-optimization/83603
33659 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
33660 arguments past the endof the argument list in functions declared
33661 without a prototype.
33662 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
33663 Avoid checking when arguments are null.
33665 2018-01-03 Martin Sebor <msebor@redhat.com>
33668 * doc/extend.texi (attribute const): Fix a typo.
33669 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
33670 issuing -Wsuggest-attribute for void functions.
33672 2018-01-03 Martin Sebor <msebor@redhat.com>
33674 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
33675 offset_int::from instead of wide_int::to_shwi.
33676 (maybe_diag_overlap): Remove assertion.
33677 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
33678 * gimple-ssa-sprintf.c (format_directive): Same.
33679 (parse_directive): Same.
33680 (sprintf_dom_walker::compute_format_length): Same.
33681 (try_substitute_return_value): Same.
33683 2018-01-03 Jeff Law <law@redhat.com>
33685 PR middle-end/83654
33686 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
33687 non-constant residual for zero at runtime and avoid probing in
33688 that case. Reorganize code for trailing problem to mirror handling
33691 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33693 PR tree-optimization/83501
33694 * tree-ssa-strlen.c (get_string_cst): New.
33695 (handle_char_store): Call get_string_cst.
33697 2018-01-03 Martin Liska <mliska@suse.cz>
33699 PR tree-optimization/83593
33700 * tree-ssa-strlen.c: Include tree-cfg.h.
33701 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
33702 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
33703 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
33705 (strlen_dom_walker::before_dom_children): Call
33706 gimple_purge_dead_eh_edges. Dump tranformation with details
33708 (strlen_dom_walker::before_dom_children): Update call by adding
33709 new argument cleanup_eh.
33710 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
33712 2018-01-03 Martin Liska <mliska@suse.cz>
33715 * cif-code.def (VARIADIC_THUNK): New enum value.
33716 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
33719 2018-01-03 Jan Beulich <jbeulich@suse.com>
33721 * sse.md (mov<mode>_internal): Tighten condition for when to use
33722 vmovdqu<ssescalarsize> for TI and OI modes.
33724 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33726 Update copyright years.
33728 2018-01-03 Martin Liska <mliska@suse.cz>
33731 * ipa-visibility.c (function_and_variable_visibility): Skip
33732 functions with noipa attribure.
33734 2018-01-03 Jakub Jelinek <jakub@redhat.com>
33736 * gcc.c (process_command): Update copyright notice dates.
33737 * gcov-dump.c (print_version): Ditto.
33738 * gcov.c (print_version): Ditto.
33739 * gcov-tool.c (print_version): Ditto.
33740 * gengtype.c (create_file): Ditto.
33741 * doc/cpp.texi: Bump @copying's copyright year.
33742 * doc/cppinternals.texi: Ditto.
33743 * doc/gcc.texi: Ditto.
33744 * doc/gccint.texi: Ditto.
33745 * doc/gcov.texi: Ditto.
33746 * doc/install.texi: Ditto.
33747 * doc/invoke.texi: Ditto.
33749 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33751 * vector-builder.h (vector_builder::m_full_nelts): Change from
33752 unsigned int to poly_uint64.
33753 (vector_builder::full_nelts): Update prototype accordingly.
33754 (vector_builder::new_vector): Likewise.
33755 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
33756 (vector_builder::operator ==): Likewise.
33757 (vector_builder::finalize): Likewise.
33758 * int-vector-builder.h (int_vector_builder::int_vector_builder):
33759 Take the number of elements as a poly_uint64 rather than an
33761 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
33762 from unsigned int to poly_uint64.
33763 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
33764 (vec_perm_indices::new_vector): Likewise.
33765 (vec_perm_indices::length): Likewise.
33766 (vec_perm_indices::nelts_per_input): Likewise.
33767 (vec_perm_indices::input_nelts): Likewise.
33768 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
33769 number of elements per input as a poly_uint64 rather than an
33770 unsigned int. Use the original encoding for variable-length
33771 vectors, rather than clamping each individual element.
33772 For the second and subsequent elements in each pattern,
33773 clamp the step and base before clamping their sum.
33774 (vec_perm_indices::series_p): Handle polynomial element counts.
33775 (vec_perm_indices::all_in_range_p): Likewise.
33776 (vec_perm_indices_to_tree): Likewise.
33777 (vec_perm_indices_to_rtx): Likewise.
33778 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
33779 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
33780 (tree_vector_builder::new_binary_operation): Handle polynomial
33781 element counts. Return false if we need to know the number
33782 of elements at compile time.
33783 * fold-const.c (fold_vec_perm): Punt if the number of elements
33784 isn't known at compile time.
33786 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33788 * vec-perm-indices.h (vec_perm_builder): Change element type
33789 from HOST_WIDE_INT to poly_int64.
33790 (vec_perm_indices::element_type): Update accordingly.
33791 (vec_perm_indices::clamp): Handle polynomial element_types.
33792 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33793 (vec_perm_indices::all_in_range_p): Likewise.
33794 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
33796 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
33797 polynomial vec_perm_indices element types.
33798 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
33799 * fold-const.c (fold_vec_perm): Likewise.
33800 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
33801 * tree-vect-generic.c (lower_vec_perm): Likewise.
33802 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33803 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
33804 element type to HOST_WIDE_INT.
33806 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33807 Alan Hayward <alan.hayward@arm.com>
33808 David Sherwood <david.sherwood@arm.com>
33810 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
33811 rather than an int. Use plus_constant.
33812 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
33813 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
33815 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33816 Alan Hayward <alan.hayward@arm.com>
33817 David Sherwood <david.sherwood@arm.com>
33819 * calls.c (emit_call_1, expand_call): Change struct_value_size from
33820 a HOST_WIDE_INT to a poly_int64.
33822 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33823 Alan Hayward <alan.hayward@arm.com>
33824 David Sherwood <david.sherwood@arm.com>
33826 * calls.c (load_register_parameters): Cope with polynomial
33827 mode sizes. Require a constant size for BLKmode parameters
33828 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
33829 forces a parameter to be padded at the lsb end in order to
33830 fill a complete number of words, require the parameter size
33831 to be ordered wrt UNITS_PER_WORD.
33833 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33834 Alan Hayward <alan.hayward@arm.com>
33835 David Sherwood <david.sherwood@arm.com>
33837 * reload1.c (spill_stack_slot_width): Change element type
33838 from unsigned int to poly_uint64_pod.
33839 (alter_reg): Treat mode sizes as polynomial.
33841 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33842 Alan Hayward <alan.hayward@arm.com>
33843 David Sherwood <david.sherwood@arm.com>
33845 * reload.c (complex_word_subreg_p): New function.
33846 (reload_inner_reg_of_subreg, push_reload): Use it.
33848 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33849 Alan Hayward <alan.hayward@arm.com>
33850 David Sherwood <david.sherwood@arm.com>
33852 * lra-constraints.c (process_alt_operands): Reject matched
33853 operands whose sizes aren't ordered.
33854 (match_reload): Refer to this check here.
33856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33857 Alan Hayward <alan.hayward@arm.com>
33858 David Sherwood <david.sherwood@arm.com>
33860 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
33861 that the mode size is in the set {1, 2, 4, 8, 16}.
33863 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33864 Alan Hayward <alan.hayward@arm.com>
33865 David Sherwood <david.sherwood@arm.com>
33867 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
33868 Use plus_constant instead of gen_rtx_PLUS.
33870 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33871 Alan Hayward <alan.hayward@arm.com>
33872 David Sherwood <david.sherwood@arm.com>
33874 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
33875 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
33876 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
33877 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
33878 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
33879 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
33880 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
33881 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
33882 * config/i386/i386.c (ix86_push_rounding): ...this new function.
33883 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
33885 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
33886 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
33887 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
33888 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
33889 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
33890 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
33891 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
33892 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
33893 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
33894 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
33896 * expr.c (emit_move_resolve_push): Treat the input and result
33897 of PUSH_ROUNDING as a poly_int64.
33898 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
33899 (emit_push_insn): Likewise.
33900 * lra-eliminations.c (mark_not_eliminable): Likewise.
33901 * recog.c (push_operand): Likewise.
33902 * reload1.c (elimination_effects): Likewise.
33903 * rtlanal.c (nonzero_bits1): Likewise.
33904 * calls.c (store_one_arg): Likewise. Require the padding to be
33905 known at compile time.
33907 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33908 Alan Hayward <alan.hayward@arm.com>
33909 David Sherwood <david.sherwood@arm.com>
33911 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
33912 Use plus_constant instead of gen_rtx_PLUS.
33914 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33915 Alan Hayward <alan.hayward@arm.com>
33916 David Sherwood <david.sherwood@arm.com>
33918 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
33919 rather than an int.
33921 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33922 Alan Hayward <alan.hayward@arm.com>
33923 David Sherwood <david.sherwood@arm.com>
33925 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
33926 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
33927 via stack temporaries. Treat the mode size as polynomial too.
33929 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33930 Alan Hayward <alan.hayward@arm.com>
33931 David Sherwood <david.sherwood@arm.com>
33933 * expr.c (expand_expr_real_2): When handling conversions involving
33934 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
33935 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
33936 as a poly_uint64 too.
33938 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33939 Alan Hayward <alan.hayward@arm.com>
33940 David Sherwood <david.sherwood@arm.com>
33942 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
33944 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33945 Alan Hayward <alan.hayward@arm.com>
33946 David Sherwood <david.sherwood@arm.com>
33948 * combine.c (can_change_dest_mode): Handle polynomial
33949 REGMODE_NATURAL_SIZE.
33950 * expmed.c (store_bit_field_1): Likewise.
33951 * expr.c (store_constructor): Likewise.
33952 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
33953 and polynomial REGMODE_NATURAL_SIZE.
33954 (gen_lowpart_common): Likewise.
33955 * reginfo.c (record_subregs_of_mode): Likewise.
33956 * rtlanal.c (read_modify_subreg_p): Likewise.
33958 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33959 Alan Hayward <alan.hayward@arm.com>
33960 David Sherwood <david.sherwood@arm.com>
33962 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
33963 numbers of elements.
33965 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33966 Alan Hayward <alan.hayward@arm.com>
33967 David Sherwood <david.sherwood@arm.com>
33969 * match.pd: Cope with polynomial numbers of vector elements.
33971 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33972 Alan Hayward <alan.hayward@arm.com>
33973 David Sherwood <david.sherwood@arm.com>
33975 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
33976 in a POINTER_PLUS_EXPR.
33978 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33979 Alan Hayward <alan.hayward@arm.com>
33980 David Sherwood <david.sherwood@arm.com>
33982 * omp-simd-clone.c (simd_clone_subparts): New function.
33983 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
33984 (ipa_simd_modify_function_body): Likewise.
33986 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33987 Alan Hayward <alan.hayward@arm.com>
33988 David Sherwood <david.sherwood@arm.com>
33990 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
33991 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
33992 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
33993 (expand_vector_condition, vector_element): Likewise.
33994 (subparts_gt): New function.
33995 (get_compute_type): Use subparts_gt.
33996 (count_type_subparts): Delete.
33997 (expand_vector_operations_1): Use subparts_gt instead of
33998 count_type_subparts.
34000 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34001 Alan Hayward <alan.hayward@arm.com>
34002 David Sherwood <david.sherwood@arm.com>
34004 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
34005 (vect_compile_time_alias): ...this new function. Do the calculation
34006 on poly_ints rather than trees.
34007 (vect_prune_runtime_alias_test_list): Update call accordingly.
34009 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34010 Alan Hayward <alan.hayward@arm.com>
34011 David Sherwood <david.sherwood@arm.com>
34013 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
34015 (vect_schedule_slp_instance): Likewise.
34017 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34018 Alan Hayward <alan.hayward@arm.com>
34019 David Sherwood <david.sherwood@arm.com>
34021 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
34022 constant and extern definitions for variable-length vectors.
34023 (vect_get_constant_vectors): Note that the number of units
34024 is known to be constant.
34026 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34027 Alan Hayward <alan.hayward@arm.com>
34028 David Sherwood <david.sherwood@arm.com>
34030 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
34031 of units as polynomial. Choose between WIDE and NARROW based
34034 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34035 Alan Hayward <alan.hayward@arm.com>
34036 David Sherwood <david.sherwood@arm.com>
34038 * tree-vect-stmts.c (simd_clone_subparts): New function.
34039 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
34041 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34042 Alan Hayward <alan.hayward@arm.com>
34043 David Sherwood <david.sherwood@arm.com>
34045 * tree-vect-stmts.c (vectorizable_call): Treat the number of
34046 vectors as polynomial. Use build_index_vector for
34047 IFN_GOMP_SIMD_LANE.
34049 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34050 Alan Hayward <alan.hayward@arm.com>
34051 David Sherwood <david.sherwood@arm.com>
34053 * tree-vect-stmts.c (get_load_store_type): Treat the number of
34054 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
34055 for variable-length vectors.
34056 (vectorizable_mask_load_store): Treat the number of units as
34057 polynomial, asserting that it is constant if the condition has
34058 already been enforced.
34059 (vectorizable_store, vectorizable_load): Likewise.
34061 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34062 Alan Hayward <alan.hayward@arm.com>
34063 David Sherwood <david.sherwood@arm.com>
34065 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
34066 of units as polynomial. Punt if we can't tell at compile time
34067 which vector contains the final result.
34069 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34070 Alan Hayward <alan.hayward@arm.com>
34071 David Sherwood <david.sherwood@arm.com>
34073 * tree-vect-loop.c (vectorizable_induction): Treat the number
34074 of units as polynomial. Punt on SLP inductions. Use an integer
34075 VEC_SERIES_EXPR for variable-length integer reductions. Use a
34076 cast of such a series for variable-length floating-point
34079 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34080 Alan Hayward <alan.hayward@arm.com>
34081 David Sherwood <david.sherwood@arm.com>
34083 * tree.h (build_index_vector): Declare.
34084 * tree.c (build_index_vector): New function.
34085 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
34086 of units as polynomial, forcibly converting it to a constant if
34087 vectorizable_reduction has already enforced the condition.
34088 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
34089 to create a {1,2,3,...} vector.
34090 (vectorizable_reduction): Treat the number of units as polynomial.
34091 Choose vectype_in based on the largest scalar element size rather
34092 than the smallest number of units. Enforce the restrictions
34095 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34096 Alan Hayward <alan.hayward@arm.com>
34097 David Sherwood <david.sherwood@arm.com>
34099 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
34100 number of units as polynomial.
34102 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34103 Alan Hayward <alan.hayward@arm.com>
34104 David Sherwood <david.sherwood@arm.com>
34106 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
34107 * target.def (autovectorize_vector_sizes): Return the vector sizes
34108 by pointer, using vector_sizes rather than a bitmask.
34109 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
34110 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
34111 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
34113 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
34114 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
34115 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
34116 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
34117 * omp-general.c (omp_max_vf): Likewise.
34118 * omp-low.c (omp_clause_aligned_alignment): Likewise.
34119 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
34120 * tree-vect-loop.c (vect_analyze_loop): Likewise.
34121 * tree-vect-slp.c (vect_slp_bb): Likewise.
34122 * doc/tm.texi: Regenerate.
34123 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
34125 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
34126 the vector size as a poly_uint64 rather than an unsigned int.
34127 (current_vector_size): Change from an unsigned int to a poly_uint64.
34128 (get_vectype_for_scalar_type): Update accordingly.
34129 * tree.h (build_truth_vector_type): Take the size and number of
34130 units as a poly_uint64 rather than an unsigned int.
34131 (build_vector_type): Add a temporary overload that takes
34132 the number of units as a poly_uint64 rather than an unsigned int.
34133 * tree.c (make_vector_type): Likewise.
34134 (build_truth_vector_type): Take the number of units as a poly_uint64
34135 rather than an unsigned int.
34137 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34138 Alan Hayward <alan.hayward@arm.com>
34139 David Sherwood <david.sherwood@arm.com>
34141 * target.def (get_mask_mode): Take the number of units and length
34142 as poly_uint64s rather than unsigned ints.
34143 * targhooks.h (default_get_mask_mode): Update accordingly.
34144 * targhooks.c (default_get_mask_mode): Likewise.
34145 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
34146 * doc/tm.texi: Regenerate.
34148 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34149 Alan Hayward <alan.hayward@arm.com>
34150 David Sherwood <david.sherwood@arm.com>
34152 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
34153 * omp-general.c (omp_max_vf): Likewise.
34154 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
34155 (expand_omp_simd): Handle polynomial safelen.
34156 * omp-low.c (omplow_simd_context): Add a default constructor.
34157 (omplow_simd_context::max_vf): Change from int to poly_uint64.
34158 (lower_rec_simd_input_clauses): Update accordingly.
34159 (lower_rec_input_clauses): Likewise.
34161 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34162 Alan Hayward <alan.hayward@arm.com>
34163 David Sherwood <david.sherwood@arm.com>
34165 * tree-vectorizer.h (vect_nunits_for_cost): New function.
34166 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
34167 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
34168 (vect_analyze_slp_cost): Likewise.
34169 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
34170 (vect_model_load_cost): Likewise.
34172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34173 Alan Hayward <alan.hayward@arm.com>
34174 David Sherwood <david.sherwood@arm.com>
34176 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
34177 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
34178 from an unsigned int * to a poly_uint64_pod *.
34179 (calculate_unrolling_factor): New function.
34180 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
34182 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34183 Alan Hayward <alan.hayward@arm.com>
34184 David Sherwood <david.sherwood@arm.com>
34186 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
34187 from an unsigned int to a poly_uint64.
34188 (_loop_vec_info::slp_unrolling_factor): Likewise.
34189 (_loop_vec_info::vectorization_factor): Change from an int
34191 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
34192 (vect_get_num_vectors): New function.
34193 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
34194 (vect_get_num_copies): Use vect_get_num_vectors.
34195 (vect_analyze_data_ref_dependences): Change max_vf from an int *
34196 to an unsigned int *.
34197 (vect_analyze_data_refs): Change min_vf from an int * to a
34199 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
34200 than an unsigned HOST_WIDE_INT.
34201 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
34202 (vect_analyze_data_ref_dependence): Change max_vf from an int *
34203 to an unsigned int *.
34204 (vect_analyze_data_ref_dependences): Likewise.
34205 (vect_compute_data_ref_alignment): Handle polynomial vf.
34206 (vect_enhance_data_refs_alignment): Likewise.
34207 (vect_prune_runtime_alias_test_list): Likewise.
34208 (vect_shift_permute_load_chain): Likewise.
34209 (vect_supportable_dr_alignment): Likewise.
34210 (dependence_distance_ge_vf): Take the vectorization factor as a
34211 poly_uint64 rather than an unsigned HOST_WIDE_INT.
34212 (vect_analyze_data_refs): Change min_vf from an int * to a
34214 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
34215 vfm1 as a poly_uint64 rather than an int. Make the same change
34216 for the returned bound_scalar.
34217 (vect_gen_vector_loop_niters): Handle polynomial vf.
34218 (vect_do_peeling): Likewise. Update call to
34219 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
34220 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
34222 * tree-vect-loop.c (vect_determine_vectorization_factor)
34223 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
34224 (vect_get_known_peeling_cost): Likewise.
34225 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
34226 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
34227 (vect_transform_loop): Likewise. Use the lowest possible VF when
34228 updating the upper bounds of the loop.
34229 (vect_min_worthwhile_factor): Make static. Return an unsigned int
34230 rather than an int.
34231 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
34232 polynomial unroll factors.
34233 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
34234 (vect_make_slp_decision): Likewise.
34235 (vect_supported_load_permutation_p): Likewise, and polynomial
34237 (vect_analyze_slp_cost): Handle polynomial vf.
34238 (vect_slp_analyze_node_operations): Likewise.
34239 (vect_slp_analyze_bb_1): Likewise.
34240 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
34241 than an unsigned HOST_WIDE_INT.
34242 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
34243 (vectorizable_load): Handle polynomial vf.
34244 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
34246 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
34248 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34249 Alan Hayward <alan.hayward@arm.com>
34250 David Sherwood <david.sherwood@arm.com>
34252 * match.pd: Handle bit operations involving three constants
34253 and try to fold one pair.
34255 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34257 * tree-vect-loop-manip.c: Include gimple-fold.h.
34258 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
34259 niters_maybe_zero parameters. Handle other cases besides a step of 1.
34260 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
34261 Add a path that uses a step of VF instead of 1, but disable it
34263 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
34264 and niters_no_overflow parameters. Update calls to
34265 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
34266 Create a new SSA name if the latter choses to use a ste other
34267 than zero, and return it via niters_vector_mult_vf_var.
34268 * tree-vect-loop.c (vect_transform_loop): Update calls to
34269 vect_do_peeling, vect_gen_vector_loop_niters and
34270 slpeel_make_loop_iterate_ntimes.
34271 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
34272 (vect_gen_vector_loop_niters): Update declarations after above changes.
34274 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
34276 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
34277 128-bit round to integer instructions.
34278 (ceil<mode>2): Likewise.
34279 (btrunc<mode>2): Likewise.
34280 (round<mode>2): Likewise.
34282 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34284 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
34285 unaligned VSX load/store on P8/P9.
34286 (expand_block_clear): Allow the use of unaligned VSX
34287 load/store on P8/P9.
34289 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34291 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
34293 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
34294 swap associated with both a load and a store.
34296 2018-01-02 Andrew Waterman <andrew@sifive.com>
34298 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
34299 * config/riscv/riscv.md (clear_cache): Use it.
34301 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
34303 * web.c: Remove out-of-date comment.
34305 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34307 * expr.c (fixup_args_size_notes): Check that any existing
34308 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
34309 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
34310 (emit_single_push_insn): ...here.
34312 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34314 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
34315 (const_vector_encoded_nelts): New function.
34316 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
34317 (const_vector_int_elt, const_vector_elt): Declare.
34318 * emit-rtl.c (const_vector_int_elt_1): New function.
34319 (const_vector_elt): Likewise.
34320 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
34321 of CONST_VECTOR_ELT.
34323 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34325 * expr.c: Include rtx-vector-builder.h.
34326 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
34327 directly on the tree encoding.
34328 (const_vector_from_tree): Likewise.
34329 * optabs.c: Include rtx-vector-builder.h.
34330 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
34331 sequence of "u" values.
34332 * vec-perm-indices.c: Include rtx-vector-builder.h.
34333 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
34334 directly on the vec_perm_indices encoding.
34336 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34338 * doc/rtl.texi (const_vector): Describe new encoding scheme.
34339 * Makefile.in (OBJS): Add rtx-vector-builder.o.
34340 * rtx-vector-builder.h: New file.
34341 * rtx-vector-builder.c: Likewise.
34342 * rtl.h (rtx_def::u2): Add a const_vector field.
34343 (CONST_VECTOR_NPATTERNS): New macro.
34344 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
34345 (CONST_VECTOR_DUPLICATE_P): Likewise.
34346 (CONST_VECTOR_STEPPED_P): Likewise.
34347 (CONST_VECTOR_ENCODED_ELT): Likewise.
34348 (const_vec_duplicate_p): Check for a duplicated vector encoding.
34349 (unwrap_const_vec_duplicate): Likewise.
34350 (const_vec_series_p): Check for a non-duplicated vector encoding.
34351 Say that the function only returns true for integer vectors.
34352 * emit-rtl.c: Include rtx-vector-builder.h.
34353 (gen_const_vec_duplicate_1): Delete.
34354 (gen_const_vector): Call gen_const_vec_duplicate instead of
34355 gen_const_vec_duplicate_1.
34356 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
34357 (gen_const_vec_duplicate): Use rtx_vector_builder.
34358 (gen_const_vec_series): Likewise.
34359 (gen_rtx_CONST_VECTOR): Likewise.
34360 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
34361 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
34362 Build a new vector rather than modifying a CONST_VECTOR in-place.
34363 (handle_special_swappables): Update call accordingly.
34364 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
34365 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
34366 Build a new vector rather than modifying a CONST_VECTOR in-place.
34367 (handle_special_swappables): Update call accordingly.
34369 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34371 * simplify-rtx.c (simplify_const_binary_operation): Use
34372 CONST_VECTOR_ELT instead of XVECEXP.
34374 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34376 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
34377 the selector elements to be different from the data elements
34378 if the selector is a VECTOR_CST.
34379 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
34380 ssizetype for the selector.
34382 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34384 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
34385 before testing each element individually.
34386 * tree-vect-generic.c (lower_vec_perm): Likewise.
34388 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34390 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
34391 * selftest-run-tests.c (selftest::run_tests): Call it.
34392 * vector-builder.h (vector_builder::operator ==): New function.
34393 (vector_builder::operator !=): Likewise.
34394 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
34395 (vec_perm_indices::all_from_input_p): New function.
34396 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
34397 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
34398 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
34399 instead of reading the VECTOR_CST directly. Detect whether both
34400 vector inputs are the same before constructing the vec_perm_indices,
34401 and update the number of inputs argument accordingly. Use the
34402 utility functions added above. Only construct sel2 if we need to.
34404 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34406 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
34407 the broadcast of the low byte.
34408 (expand_mult_highpart): Use an explicit encoding for the permutes.
34409 * optabs-query.c (can_mult_highpart_p): Likewise.
34410 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
34411 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34412 (vectorizable_bswap): Likewise.
34413 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
34414 explicit encoding for the power-of-2 permutes.
34415 (vect_permute_store_chain): Likewise.
34416 (vect_grouped_load_supported): Likewise.
34417 (vect_permute_load_chain): Likewise.
34419 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34421 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
34422 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
34423 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
34424 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
34425 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
34426 (vect_gen_perm_mask_any): Likewise.
34428 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34430 * int-vector-builder.h: New file.
34431 * vec-perm-indices.h: Include int-vector-builder.h.
34432 (vec_perm_indices): Redefine as an int_vector_builder.
34433 (auto_vec_perm_indices): Delete.
34434 (vec_perm_builder): Redefine as a stand-alone class.
34435 (vec_perm_indices::vec_perm_indices): New function.
34436 (vec_perm_indices::clamp): Likewise.
34437 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
34438 (vec_perm_indices::new_vector): New function.
34439 (vec_perm_indices::new_expanded_vector): Update for new
34440 vec_perm_indices class.
34441 (vec_perm_indices::rotate_inputs): New function.
34442 (vec_perm_indices::all_in_range_p): Operate directly on the
34443 encoded form, without computing elided elements.
34444 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
34445 encoding. Update for new vec_perm_indices class.
34446 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
34447 the given vec_perm_builder.
34448 (expand_vec_perm_var): Update vec_perm_builder constructor.
34449 (expand_mult_highpart): Use vec_perm_builder instead of
34450 auto_vec_perm_indices.
34451 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
34452 vec_perm_indices instead of auto_vec_perm_indices. Use a single
34453 or double series encoding as appropriate.
34454 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
34455 vec_perm_indices instead of auto_vec_perm_indices.
34456 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34457 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34458 (vect_permute_store_chain): Likewise.
34459 (vect_grouped_load_supported): Likewise.
34460 (vect_permute_load_chain): Likewise.
34461 (vect_shift_permute_load_chain): Likewise.
34462 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34463 (vect_transform_slp_perm_load): Likewise.
34464 (vect_schedule_slp_instance): Likewise.
34465 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34466 (vectorizable_mask_load_store): Likewise.
34467 (vectorizable_bswap): Likewise.
34468 (vectorizable_store): Likewise.
34469 (vectorizable_load): Likewise.
34470 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
34471 vec_perm_indices instead of auto_vec_perm_indices. Use
34472 tree_to_vec_perm_builder to read the vector from a tree.
34473 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
34474 vec_perm_builder instead of a vec_perm_indices.
34475 (have_whole_vector_shift): Use vec_perm_builder and
34476 vec_perm_indices instead of auto_vec_perm_indices. Leave the
34477 truncation to calc_vec_perm_mask_for_shift.
34478 (vect_create_epilog_for_reduction): Likewise.
34479 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
34480 from auto_vec_perm_indices to vec_perm_indices.
34481 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
34482 instead of changing individual elements.
34483 (aarch64_vectorize_vec_perm_const): Use new_vector to install
34484 the vector in d.perm.
34485 * config/arm/arm.c (expand_vec_perm_d::perm): Change
34486 from auto_vec_perm_indices to vec_perm_indices.
34487 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
34488 instead of changing individual elements.
34489 (arm_vectorize_vec_perm_const): Use new_vector to install
34490 the vector in d.perm.
34491 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
34492 Update vec_perm_builder constructor.
34493 (rs6000_expand_interleave): Likewise.
34494 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
34495 (rs6000_expand_interleave): Likewise.
34497 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34499 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
34500 to qimode could truncate the indices.
34501 * optabs.c (expand_vec_perm_var): Likewise.
34503 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34505 * Makefile.in (OBJS): Add vec-perm-indices.o.
34506 * vec-perm-indices.h: New file.
34507 * vec-perm-indices.c: Likewise.
34508 * target.h (vec_perm_indices): Replace with a forward class
34510 (auto_vec_perm_indices): Move to vec-perm-indices.h.
34511 * optabs.h: Include vec-perm-indices.h.
34512 (expand_vec_perm): Delete.
34513 (selector_fits_mode_p, expand_vec_perm_var): Declare.
34514 (expand_vec_perm_const): Declare.
34515 * target.def (vec_perm_const_ok): Replace with...
34516 (vec_perm_const): ...this new hook.
34517 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
34518 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
34519 * doc/tm.texi: Regenerate.
34520 * optabs.def (vec_perm_const): Delete.
34521 * doc/md.texi (vec_perm_const): Likewise.
34522 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
34523 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
34524 expand_vec_perm for constant permutation vectors. Assert that
34525 the mode of variable permutation vectors is the integer equivalent
34526 of the mode that is being permuted.
34527 * optabs-query.h (selector_fits_mode_p): Declare.
34528 * optabs-query.c: Include vec-perm-indices.h.
34529 (selector_fits_mode_p): New function.
34530 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
34531 is defined, instead of checking whether the vec_perm_const_optab
34532 exists. Use targetm.vectorize.vec_perm_const instead of
34533 targetm.vectorize.vec_perm_const_ok. Check whether the indices
34534 fit in the vector mode before using a variable permute.
34535 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
34536 vec_perm_indices instead of an rtx.
34537 (expand_vec_perm): Replace with...
34538 (expand_vec_perm_const): ...this new function. Take the selector
34539 as a vec_perm_indices rather than an rtx. Also take the mode of
34540 the selector. Update call to shift_amt_for_vec_perm_mask.
34541 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
34542 Use vec_perm_indices::new_expanded_vector to expand the original
34543 selector into bytes. Check whether the indices fit in the vector
34544 mode before using a variable permute.
34545 (expand_vec_perm_var): Make global.
34546 (expand_mult_highpart): Use expand_vec_perm_const.
34547 * fold-const.c: Includes vec-perm-indices.h.
34548 * tree-ssa-forwprop.c: Likewise.
34549 * tree-vect-data-refs.c: Likewise.
34550 * tree-vect-generic.c: Likewise.
34551 * tree-vect-loop.c: Likewise.
34552 * tree-vect-slp.c: Likewise.
34553 * tree-vect-stmts.c: Likewise.
34554 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
34556 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
34557 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
34558 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
34559 (aarch64_vectorize_vec_perm_const): ...this new function.
34560 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34561 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34562 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
34563 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
34564 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34565 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34566 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
34568 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
34569 check for NEON modes.
34570 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
34571 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
34572 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
34573 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
34575 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
34576 the old VEC_PERM_CONST conditions.
34577 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
34578 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
34579 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
34580 (ia64_vectorize_vec_perm_const_ok): Merge into...
34581 (ia64_vectorize_vec_perm_const): ...this new function.
34582 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
34583 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
34584 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
34585 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
34586 * config/mips/mips.c (mips_expand_vec_perm_const)
34587 (mips_vectorize_vec_perm_const_ok): Merge into...
34588 (mips_vectorize_vec_perm_const): ...this new function.
34589 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
34590 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
34591 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
34592 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
34593 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
34594 (rs6000_expand_vec_perm_const): Delete.
34595 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
34597 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34598 (altivec_expand_vec_perm_const_le): Take each operand individually.
34599 Operate on constant selectors rather than rtxes.
34600 (altivec_expand_vec_perm_const): Likewise. Update call to
34601 altivec_expand_vec_perm_const_le.
34602 (rs6000_expand_vec_perm_const): Delete.
34603 (rs6000_vectorize_vec_perm_const_ok): Delete.
34604 (rs6000_vectorize_vec_perm_const): New function.
34605 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34606 an element count and rtx array.
34607 (rs6000_expand_extract_even): Update call accordingly.
34608 (rs6000_expand_interleave): Likewise.
34609 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
34610 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
34611 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
34612 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
34613 (rs6000_expand_vec_perm_const): Delete.
34614 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
34615 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34616 (altivec_expand_vec_perm_const_le): Take each operand individually.
34617 Operate on constant selectors rather than rtxes.
34618 (altivec_expand_vec_perm_const): Likewise. Update call to
34619 altivec_expand_vec_perm_const_le.
34620 (rs6000_expand_vec_perm_const): Delete.
34621 (rs6000_vectorize_vec_perm_const_ok): Delete.
34622 (rs6000_vectorize_vec_perm_const): New function. Remove stray
34623 reference to the SPE evmerge intructions.
34624 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
34625 an element count and rtx array.
34626 (rs6000_expand_extract_even): Update call accordingly.
34627 (rs6000_expand_interleave): Likewise.
34628 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
34629 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
34631 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
34633 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34635 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
34636 vector mode and that that mode matches the mode of the data
34638 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
34639 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
34640 directly using expand_vec_perm_1 when forcing selectors into
34642 (expand_vec_perm_var): New function, split out from expand_vec_perm.
34644 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34646 * optabs-query.h (can_vec_perm_p): Delete.
34647 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
34648 * optabs-query.c (can_vec_perm_p): Split into...
34649 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
34650 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
34651 particular selector is valid.
34652 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
34653 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
34654 (vect_grouped_load_supported): Likewise.
34655 (vect_shift_permute_load_chain): Likewise.
34656 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
34657 (vect_transform_slp_perm_load): Likewise.
34658 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
34659 (vectorizable_bswap): Likewise.
34660 (vect_gen_perm_mask_checked): Likewise.
34661 * fold-const.c (fold_ternary_loc): Likewise. Don't take
34662 implementations of variable permutation vectors into account
34663 when deciding which selector to use.
34664 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
34665 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
34666 with a false third argument.
34667 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
34668 to test whether the constant selector is valid and can_vec_perm_var_p
34669 to test whether a variable selector is valid.
34671 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34673 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
34674 * optabs-query.c (can_vec_perm_p): Likewise.
34675 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
34676 instead of vec_perm_indices.
34677 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
34678 (vect_gen_perm_mask_checked): Likewise,
34679 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
34680 (vect_gen_perm_mask_checked): Likewise,
34682 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
34684 * optabs-query.h (qimode_for_vec_perm): Declare.
34685 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
34686 (qimode_for_vec_perm): ...this new function.
34687 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
34689 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34691 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
34692 does not have a conditional at the top.
34694 2018-01-02 Richard Biener <rguenther@suse.de>
34696 * ipa-inline.c (big_speedup_p): Fix expression.
34698 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34701 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
34704 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
34708 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
34709 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
34710 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
34711 cond_taken_branch_cost 3->4.
34713 2018-01-01 Jakub Jelinek <jakub@redhat.com>
34715 PR tree-optimization/83581
34716 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
34717 TODO_cleanup_cfg if any changes have been made.
34719 PR middle-end/83608
34720 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
34721 convert_modes if target mode has the right side, but different mode
34724 PR middle-end/83609
34725 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
34726 last argument when extracting from CONCAT. If either from_real or
34727 from_imag is NULL, use expansion through memory. If result is not
34728 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
34729 the parts directly to inner mode, if even that fails, use expansion
34732 PR middle-end/83623
34733 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
34734 check for bswap in mode rather than HImode and use that in expand_unop
34737 Copyright (C) 2018 Free Software Foundation, Inc.
34739 Copying and distribution of this file, with or without modification,
34740 are permitted in any medium without royalty provided the copyright
34741 notice and this notice are preserved.